Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Battery-less detection and recording of tamper activity along with wireless interrogation
(USC Thesis Other)
Battery-less detection and recording of tamper activity along with wireless interrogation
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
BATTERY-LESS DETECTION AND RECORDING OF TAMPER ACTIVITY ALONG
WITH WIRELESS INTERROGATION
by
Matin Barekatain
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(ELECTRICAL ENGINEERING)
May 2024
Copyright 2024 Matin Barekatain
ii
Acknowledgments
The Microelectromechanical Systems (MEMS) group of the Ming Hsieh Department of Electrical
and Computer Engineering (ECE) at the University of Southern California (USC) has provided
me with a stimulating environment for investigating hardware solutions based on novel sensors
and microactuators. I want to express my deep gratitude to Professor Eun Sok Kim, who served as
my mentor and the principal investigator of our research team. Not only did he welcome me into
the group, but he also provided consistent and invaluable support throughout my Ph.D. program.
His guidance played a crucial role in shaping my research, involving insightful discussions on
scientific and engineering challenges, encouragement to explore new research paths, and thorough
reviews of my publications. Additionally, Professor Kim's generous assistance went beyond
academics, helping me navigate the challenges of settling in a new country as a Ph.D. student. His
dedication to research, meticulous attention to detail, and genuine kindness were constant sources
of inspiration, motivating me to strive for excellence and personal growth.
I want to thank Professor Wei Wu and Professor Qifa Zhou, who have been valuable members
of my dissertation committee and provided insightful discussions and advice toward improving
this thesis.
I am incredibly grateful to the former members of the USC MEMS group, including Dr. Anton
Shkel, Dr. Hai Liu, Dr. Jaehoon Lee, Prof. Joung-Hu Park, Dr. Yongkui Tang, Dr. Lurui Zhao, Dr.
Song Liu, and Dr. Yunqi Cao, who have provided the foundation for my research, lab training, and
mentorship. I would like to thank my current group members, Kianoush Sadeghian Esfahani,
Akash Roy, Anik Sengupta, Dr. Baptiste Neff, Junyi Wang, Hongxiang Gao, Dr. Kunfeng Wang,
iii
Aobo Zhang, and Diana Cantini for their valuable advice, discussion, and friendship. I am also
very thankful to all my friends at USC and elsewhere for their invaluable support.
My sincerest thanks go to Dr. Donghai Zhu, Alfonso Jimenez, Dr. Shiva Bhaskaran, and Joey
Vo for their continued support of the state-of-the-art clean-room facilities at USC. Their efforts
have been essential in keeping all the equipment up and running through the years.
I worked with Dr. Susan Schober as her TA for multiple years and learned from her insights
in teaching electronics fundamentals. I am so thankful for her.
Finally, I am eternally grateful to my family, whose advice, encouragement, and support in
providing me with the best opportunities made everything I have done possible. Their unwavering,
unconditional love, support, and encouragement have been a constant source of strength
throughout my life and this Ph.D. journey.
iv
Table of Contents
Acknowledgments........................................................................................................................................ii
List of Tables...............................................................................................................................................vi
List of Figures.............................................................................................................................................vii
Abstract.....................................................................................................................................................xvii
Chapter 1: Introduction.................................................................................................................................1
1.1. Motivation......................................................................................................................................1
1.2. Problem Statement.........................................................................................................................3
1.3. Scope of Work ...............................................................................................................................4
1.4. Overview of Chapters....................................................................................................................5
Chapter 2: Wireless and Battery-less Tamper Detector for Semiconductor Supply Chain..........................8
2.1. Background....................................................................................................................................8
2.2. Project Phases and Sensor Design ...............................................................................................15
2.3. Summary......................................................................................................................................19
Chapter 3: Pyroelectric Energy Harvesters (PEH) .....................................................................................20
3.1. Background..................................................................................................................................20
3.2. Design, Modeling, and Fabrication .............................................................................................24
3.3. Experimental Setups....................................................................................................................31
3.4. Results..........................................................................................................................................35
3.5. Summary......................................................................................................................................40
Chapter 4: Bulk Acoustic Wave (BAW) Resonators .................................................................................42
4.1. Background.......................................................................................................................................42
4.2. Film Bulk Acoustic Resonator (FBAR) ...........................................................................................44
4.2.1. Design and Modeling.................................................................................................................44
4.2.2. Fabrication .................................................................................................................................47
4.2.3. Results........................................................................................................................................48
4.3. High-Overtone Bulk Acoustic Resonator (HBAR)..........................................................................51
4.3.1. Design and Fabrication..............................................................................................................51
4.3.2. Results........................................................................................................................................53
4.4. Summary...........................................................................................................................................56
Chapter 5: Tamper Detection with PREAT Detector.................................................................................57
5.1. Overview of Tamper Detector System .............................................................................................57
5.2. HBAR’s Breakdown Characteristics................................................................................................61
5.2.1. Breakdown Characterization Setup ...........................................................................................61
5.2.2. Breakdown Characterization Results.........................................................................................62
5.2.3. MEMS Switch and Shock Tests................................................................................................67
5.3. RFID Antennas and Wireless Experiments......................................................................................70
5.3.1. Antenna Design and Modeling ..................................................................................................70
v
5.3.2. Antenna Characterization ..........................................................................................................72
5.3.3. PREAT Wireless Characterization ............................................................................................74
5.4. Summary...........................................................................................................................................78
Chapter 6: Piezoelectric MEMS Vibrational Energy Harvesting from Mechanical Shocks......................80
6.1. Background.......................................................................................................................................80
6.2. MEMS VEH with Sub-kHz Resonant Frequencies..........................................................................81
6.2.1. Design and Modeling.................................................................................................................81
6.2.2. Fabrication .................................................................................................................................85
6.3. Experimental Results........................................................................................................................87
6.4. Summary...........................................................................................................................................95
Chapter 7: Non-resonant Vibration Energy Harvesting from Human Walking Motion ............................96
7.1. Background.......................................................................................................................................96
7.2. Design and Fabrication.....................................................................................................................99
7.3. Experimental Setup and Results.....................................................................................................107
7.4. Summary.........................................................................................................................................110
Chapter 8: Conclusion and Future Direction ............................................................................................111
8.1. Tamper Detector Chip ....................................................................................................................111
8.2. Vibration Energy Harvesting..........................................................................................................113
Bibliography .............................................................................................................................................116
vi
List of Tables
Table 3.1: Comparison of properties of various pyroelectric materials [35]. ............................... 23
Table 3.2: Material properties of LiNbO3 [40], [41], [42]. .......................................................... 27
Table 3.3: Summary of the equivalent charges and voltages generated by PEH, derived from
the measured short-circuit current of the PEH.............................................................................. 39
Table 5.1: Measured electrical properties of HBAR at each breakdown phase. .......................... 65
Table 6.1: Piezoelectric materials properties [59]. ....................................................................... 85
Table 6.2: Dimension and measured properties of two fabricated VEHs..................................... 91
Table 7.1: Key parameters of the non-resonant VEH based on rectangular magnets (OMC
design)......................................................................................................................................... 106
Table 7.2: Key parameters of the non-resonant VEH based on square magnets (CMC design).106
vii
List of Figures
Figure 1.1: Illustration of an anti-tamper chip inside an IC package and being interrogated by
an RFID reader................................................................................................................................ 4
Figure 2.1: Vulnerabilities in the semiconductor supply chain, with the tamper detector’s
functionality potentially protecting against recycling [9]............................................................. 10
Figure 2.2: Hierarchical classification of the various counterfeit detection techniques [13]. ...... 12
Figure 2.3: Overview of ScatterVerif: verification of electronic boards using reflection
response of power distribution network [14]. ............................................................................... 12
Figure 2.4: Schematic diagram of the active tamper detection system [16]. ................................ 13
Figure 2.5: MEMS-assisted anti-tamper (MAAT) system with three primary components:
critical program information (CPI), the MEMS array, and the MEMS monitoring engine
(MME) [21]................................................................................................................................... 14
Figure 2.6: Results from Dynasolve 750 testing. Both samples were soaked for 45 minutes in
Dynasolve 750 at 105 ˚C [20]. As can be seen, scratch marks clearly appear on the tampered
IC after the test.............................................................................................................................. 14
Figure 2.7: Wireless detection system concept: (Top) before the HBAR RFID tag is broken
and (Bottom) after the HBAR RFID tag is electrically broken by tamper activity, which is
composed of two sequential events of de-soldering followed by mechanical banging................ 16
Figure 2.8: Tamper activity composed of two sequential events, one after the other, in that
order and both (not just one): (a) de-soldering, which requires about 300 °C temperature rise,
followed by (b) mechanical banging, which makes the MEMS acceleration switch to be
turned on to deliver pyroelectric voltage/charge from LiNbO3 to HBAR RFID tag. .................. 17
Figure 2.9: Prototype wireless tamper detection system with a wireless interrogator
transmitting a signal and measuring the backscattered signal from the tamper detector. The
viii
strength of the backscattered signal depends on the HBAR RFID tag that is connected (1)
directly to a microstrip patch antenna and (2) to a pyroelectric energy converter (PEC) via a
cantilever-based acceleration switch. The PEC generates a large voltage (and charge) in
response to temperature change due to a de-soldering process, while the cantilever switch is
turned on by mechanical banging (which follows the de-soldering) to let a large voltage and
charge to flow from the PEC to HBAR to break down the HBAR permanently and thus to
alter the characteristic of the backscattered signal........................................................................ 18
Figure 2.10: Schematic of the front and back sides of a PREAT chip along with a top view of
a high-overtone bulk acoustic resonator (HBAR) used in the PREAT. ....................................... 18
Figure 3.1: Published papers from 1999 to 2019 on pyroelectric materials vs. papers
investigating their energy harvesting applications (in red) [28]. .................................................. 22
Figure 3.2: Photo of a single 50 mm diameter × 1 mm thick LiTaO3 crystal generating
multi-kV from 75˚C temperature variations [33].......................................................................... 23
Figure 3.3: Illustration of pyroelectric mechanism for LiNbO3 crystal working as PEH. ........... 24
Figure 3.4: Equivalent electrical circuit of PEH before the switch is activated. .......................... 26
Figure 3.5: Theoretical PEH’s open-circuit voltage vs. frequency showing the effects of the
electrical and thermal time constants at the low and high frequencies, respectively.................... 29
Figure 3.6: Photo of 2510 mm3 crystal LiNbO3-based PEH with a copper-Kapton
cantilever switch. .......................................................................................................................... 30
Figure 3.7: Miniature PEH (left) Cross-sectional diagram of miniature PEH composed of
LiNbO3 substrate and aluminum layers with device areas from 4 to 16 mm2, (right)
fabricated devices on the substrate before dicing. ........................................................................ 30
Figure 3.8: Photos of a 550.15 mm3 LiNbO3-based PEC (integrated with a copper-Kapton
cantilever switch having a large air gap due to the warping of the cantilevers) placed on a
hotplate, as the temperature on the hotplate is raised from room temperature to 300 ˚,
showing the air gap (between the top and bottom beams of the switch) becoming less due to
the temperature-induced voltage on the PEC. At 300 ˚C, a narrow gap (~1 mm) remains
between
ix
the top and bottom beams of the cantilever switch that can be zero-ed by applying a
mechanical shock. ......................................................................................................................... 32
Figure 3.9: Schematic of the measurement set-up showing the LiNbO3-based PEH with
Kapton-Copper cantilever switch and two wires over a hot plate. ............................................... 32
Figure 3.10: Experimental setup for characterizing PEH’s ground truth response to
temperature variation with a 2-channel digital temperature measurement unit (HH506RA)
and a soldering hot gun at specific gaps for various max target temperatures under different
airflow levels................................................................................................................................. 33
Figure 3.11: Experimental setup for PEH’s short circuit current measurement with a Keithley
6485 pico-ammeter and a LabJack T7-Pro DAQ connected to a computer. The electrodes on
PEH are accessed by gold-plated tungsten needle probes inside a metal EMI shielding box.
The tip of a hot gun is placed inside the metal box. ..................................................................... 34
Figure 3.12: Experimental setup for characterizing the delivery of PEH voltage and charge to
a low impedance (10 MΩ) load when a mechanical switch connects the PEH to the load after
the temperature reaches the maximum by a hot gun placed 1 cm above the PEH inside a metal
box. An oscilloscope with a forced trigger captures the fast spike............................................... 35
Figure 3.13: (a) Hotplate temperature vs time, indicating the temperature ramp at 2 ˚C/sec. (b)
Measured voltage over the 8 MΩ resistor upon the switch connection after the voltage from
the PEH rises high enough due to temperature change followed by a mechanical shock. ........... 36
Figure 3.14: Measured temperatures (with two thermocouple probes placed at the top and
bottom of PEH) vs. time due to applied heat by a soldering hot gun that is turned on and off
with (1) three different hot air flow levels 4, 6, and 8, (2) the distance between the hot gun
and PEH at 1 and 2 cm, and (3) the max target temperature of 250 and 427 ˚C. The mid-point
temperature of the PEH estimated by taking the average value of the two probes is taken as
the ground truth temperature of the PEH. Experiments for 427 ˚C target temperature are
stopped sooner to prevent damaging the setup due to too high temperature. ............................... 37
Figure 3.15: (Top) The generated current and (Bottom) charges accumulated on an 8 mm2
PEH with respect to the PEH’s measured temperature at the PEH’s top and bottom surfaces
with a hot gun at a 1cm gap with a 250 ˚C maximum temperature. The equivalent open
circuit voltage by the accumulated charge peaks at ~1,100 volts based on the capacitance of
the PEH. ........................................................................................................................................ 38
x
Figure 3.16: Steady-state finite element analysis (FEA) simulations of (left) temperature and
(right) open-circuit voltage of a 4 mm2 LiNbO3 crystal (500 µm thick) under applied
temperature in the experimental setup with a hot gun at a 1 cm gap. The peak open-circuit
voltage surpasses 900 volts when the top surface temperature is 250 ˚C. .................................... 39
Figure 3.17: Measured voltage vs time by a PEH placed 1 cm under a hot gun with two
different target temperatures (top) and by two different PEHs with varying sizes under the
same target temperature (bottom) when a switch connects the PEH to a 10 MΩ load. The
maximum amplitude and total delivered charges are related to the PEH area and the hot gun's
temperature. When the voltage exceeds 3 volts, several nCs of charges are delivered to the
load, sufficient for HBAR’s dielectric breakdown. ...................................................................... 40
Figure 4.1: Schematic of longitudinal wave generation and propagation in an acoustic
resonator by an electric field in the thickness direction [46]. ....................................................... 44
Figure 4.2: Equivalent BVD model for FBAR. ............................................................................ 45
Figure 4.3: MATLAB simulation results showing S11 values of FBAR with 0.3 µm thick
ZnO film in log scale and Smith chart. ......................................................................................... 47
Figure 4.4: Cross-section diagram of the FBAR device, which is composed of silicon nitride
(SixNy), zinc oxide, and aluminum layers on a silicon wafer. ...................................................... 48
Figure 4.5: Photograph of the FBAR under the microscope. ....................................................... 48
Figure 4.6: S11 vs. frequency of fabricated FBAR at the fundamental resonance frequency. ...... 49
Figure 4.7: Phase vs. frequency of fabricated FBAR at the fundamental resonance frequency... 49
Figure 4.8: Reactance vs. frequency of fabricated FBAR at the fundamental resonance
frequency....................................................................................................................................... 50
Figure 4.9: Resistance vs. frequency of fabricated FBAR at the fundamental resonance
frequency....................................................................................................................................... 50
xi
Figure 4.10: Quality factor (Q) vs. frequency of fabricated FBAR at the fundamental
resonance frequency...................................................................................................................... 51
Figure 4.11: (a) Cross-sectional diagram of HBAR composed of the sapphire substrate, zinc
oxide film, and aluminum layers. (b) Photo of a fabricated HBAR on a sapphire substrate........ 52
Figure 4.12: Photo of the measurement setup for HBAR using a Cascade microprobe............... 53
Figure 4.13: Measured S11 parameter of HBAR over a wide 3 GHz range with 1.87 MHz
resolution....................................................................................................................................... 54
Figure 4.14: Measured S11 parameter of HBAR over a very narrow frequency range with
19 kHz resolution.......................................................................................................................... 54
Figure 4.15: Real part of the measured HBAR’s impedance........................................................ 55
Figure 4.16: Imaginary part of the measured HBAR’s impedance. ............................................. 55
Figure 4.17: Quality factor of HBAR at ~ 7.55 GHz.................................................................... 56
Figure 5.1: Proof-of-concept bulky wireless tamper detection system with a wireless
interrogator transmitting a signal and measuring the backscattered signal from the tamper
detector. The strength of the backscattered signal depends on the HBAR RFID tag that is
connected (1) directly to a microstrip patch antenna and (2) to a pyroelectric energy converter
(PEC) via a cantilever-based acceleration switch. The PEC generates a large voltage (and
charge) in response to temperature change due to a de-soldering process, while the cantilever
switch is turned on by mechanical banging (which follows the de-soldering) to let a large
voltage and charge flow from the PEC to HBAR to break down the HBAR permanently and
thus to alter the characteristic of the backscattered signal. ........................................................... 58
Figure 5.2: Schematic of the front and back sides of a PREAT chip along with a top view of a
high-overtone bulk acoustic resonator (HBAR) used in the PREAT. .......................................... 59
Figure 5.3: Schematic showing the components on the top and bottom sides of PREAT PCB
made of an adhesive-less polyimide-based substrate. The shock switch (HT-Micro AT-50-T)
and the antenna (Kyocera A1001312) share the same side of the PCB, while the HBAR and
the PEH are on opposite sides. The top and bottom pads are connected through vias. ................ 60
xii
Figure 5.4: Photo of a PREAT chip inside a laser-engraved IC package. (Middle) Illustration
of a PREAT chip inside an IC package and being interrogated by an RFID reader..................... 61
Figure 5.5: Experimental setup for characterizing HBAR’s electrical breakdown with a
programmable power supply; the voltage across the top and bottom electrodes of HBAR is
increased from 0 to 4.2 volts while the current through the HBAR is measured with a digital
multimeter. The power supply and multimeter are controlled via the PyVisa library.................. 62
Figure 5.6: Measured I-V characteristics for seven HBARs of various areas under DC voltage
applied between the top and bottom electrodes of HBAR. The measurements show no sign of
breakdown for HBARs up to 1 VDC............................................................................................ 63
Figure 5.7: Measured I-V characteristics for three HBARs with areas of (Top) 15,200 µm2,
(Middle) 8,550 µm2
, (Bottom) 3,800 µm2
, a DC voltage over 0 - 4.2 V is applied between the
top and bottom electrodes three times. During the first sweeps, the initial and second
permanent breakdowns occur around 2 – 3 V, making the HBAR’s DC resistance from
several MΩs down to about 1 kΩ. ................................................................................................ 64
Figure 5.8: A closer look at the first sweeps on the three HBARs shows that pre-breakdown
DC resistances depend on the HBAR’s top-view areas, while the post-breakdown resistances
are not related to the area, as the dielectric breakdown happens over weak spots that are
random. However, due to the breakdown, the quality factors (Qs) of the HBARs drop by
several factors. .............................................................................................................................. 65
Figure 5.9: Measured HBAR Q’s before and after the first breakdown phase over (a) a wide
2.5 GHz range with 1.9 MHz resolution and (b) a narrow 10 MHz range with 19 kHz
resolution. Note that the actual peak for Q before the breakdown is more than 2,000
(Fig. 10b), which is not shown in Fig. 10a over the wide range due to the network analyzer's
limited number of data acquisition points. The peak Q is less than 500 after the Phase I
breakdown..................................................................................................................................... 66
Figure 5.10: Illustration of a mechanical shock table with (1) an electrically controlled
magnetic gripper that grips a metallic plate holding a miniature shock switch (HT-Micro
AT-50) and a 3-axis high-g accelerometer (Analog Devices ADXL372) and (2) a
programmable linear actuator (Aerotech ACT115DL) which lifts up the plate. At an elevated
point, the gripper is turned off; the plate falls onto the table’s base, and a shock, which makes
the shock switch be on. An oscilloscope reads the signal passed through the switch, while the
shock acceleration is recorded through an Arduino Zero. ............................................................ 67
xiii
Figure 5.11: Measured Z-axis shock accelerations due to the impact of the metallic plate on
the base when the plate is released from various heights. Shocks beyond 40g are observed
from a free-fall distance greater than 6 mm.................................................................................. 68
Figure 5.12: Transient waveforms of the signal passing through the shock switch due to the
free falls producing more than 40g shock. Higher acceleration makes a more extended
transient switch on time. In all cases, the average on-time is beyond 5 ms, sufficiently long
for delivering all the generated charge from the PEH into the HBAR according to the
waveforms shown in Fig. 3.17...................................................................................................... 69
Figure 5.13: Measured 3-axis accelerations due to hand banging of the metallic plate onto a
hard object twenty different times, showing substantial accelerations directed towards the
other two axes (X-axis and Y-axis) when the banging is mainly toward the Z-axis. ................... 69
Figure 5.14: Perspective view of the micro-strip patch antenna, (b) top-view design of the
antenna with key dimensions. ....................................................................................................... 71
Figure 5.15: Simulated S11 parameter of the antenna showing the maximum absorption of
-30 dB at the target operating frequency and -10 dB bandwidth of 420 MHz. ............................ 72
Figure 5.16: Photo of E-shaped microstrip patch antenna on Cirlex® substrate next to a U.S.
quarter coin. .................................................................................................................................. 73
Figure 5.17: (Left) Measured S11 parameter of the antenna showing the maximum absorption
of -18 dB at the target operating frequency and 280 MHz -10 dB bandwidth. (Right)
Calculated S12 of the antenna into air, based on the measured S11. .............................................. 73
Figure 5.18: Measured S11’s (a) of the interrogator’s antenna (Kyocera 1005194 as a
transmitting antenna) and PREATS antenna (Kyocera A1001312 chip antenna as a receiving
antenna) mounted on the PCB. (b) of the two antennas on the Smith chart. ................................ 74
Figure 5.19: (a) Schematic of the measurement set-up showing two MSPAs facing each other
with the top MSPA connected to a Vector Network Analyzer and the bottom MSPA
connected to an HBAR RFID tag through a Cascade microprobe. (b) Actual photo of two
MSPAs separated by an air gap (and facing each other) during the wireless test. ....................... 75
xiv
Figure 5.20: (a) Measured S11 magnitude and (b) phase of HBAR obtained through the
wireless test over 1 MHz range with 14 kHz resolution. (c) and (d) Imaginary and real parts
of the wirelessly measured HBAR’s impedance, respectively. .................................................... 76
Figure 5.21: (a) Measured S11 magnitude and (b) phase of a broken HBAR obtained through
the wireless test over a 30 MHz with 19 kHz resolution. (c) Imaginary and (d) real parts of
the wirelessly measured impedance of the broken HBAR, respectively. ..................................... 77
Figure 5.22: Measured S11 parameter’s magnitude and phase of the HBAR across a zoomed
window of 1.6 MHz through wired and wireless setups before the breakdown while having
an fQ >1013 compared to the dead signal after HBAR’s breakdown............................................ 78
Figure 6.1: Illustration of the PZT bimorph structure, composed of two PZT layers covered
by metal electrodes, sandwiching a brass layer. ........................................................................... 82
Figure 6.2: Illustration of the ZnO-based unimorph cantilever structure. The ZnO thickness
is set at 0.3 µm while the cantilever is 1.5 µm thick..................................................................... 82
Figure 6.3: Illustration of the same PZT bimorph structure with an added proof mass at the
tip of the cantilever. ...................................................................................................................... 84
Figure 6.4: (a) Brief fabrication steps and (b) top view schematic of the vibration-energy
harvester (VEH). ........................................................................................................................... 86
Figure 6.5: Stress compensation through another deposition of low-stress silicon nitride after
micromachining a diaphragm [65]................................................................................................ 86
Figure 6.6: (a) Top view photo of the fabricated VEH and (b) photo of two diced VEHs with
a US quarter coin. ......................................................................................................................... 87
Figure 6.7: Photograph a bimorph cantilever, clamping system, and shaker. The cantilever
has a total dimension of 11 mm 1 mm 0.6 mm. ..................................................................... 88
Figure 6.8: Setup for measuring output voltage from VEHs with vibration applied by a
shaker [66]. ................................................................................................................................... 88
xv
Figure 6.9: Generated voltage from bimorph PZT cantilever while applying 2.6 g acceleration. 89
Figure 6.10: Photograph the bimorph cantilever with an added 200 mg proof mass at the tip. ... 90
Figure 6.11: Generated voltage from bimorph PZT cantilever (Left) without and (Right) with
a 200 mg proof mass while applying a 0.91 g acceleration. After adding the proof mass, the
generated voltage amplitude changed from 250 mv to 740 mv.................................................... 90
Figure 6.12: Output voltage of 2 mm 2 mm VEH vs. frequency as a function of applied
acceleration showing the softening of the spring with increasing vibrational amplitude............. 91
Figure 6.13: Output voltage vs time with 60 g shock applied periodically. ................................. 92
Figure 6.14: Output voltage vs. applied acceleration as a function of frequency......................... 93
Figure 6.15: Schematic of the circuit designed to build up charges on a capacitor, as the VEH
generates voltage and charge in response to applied acceleration. ............................................... 93
Figure 6.16: Measured voltage on the capacitor (b) as the VEH generates voltage and charge
due to 30-50g shocks applied to the VEH at 2 Hz (a). ................................................................. 94
Figure 7.1: A 3D-printed harvester based on a V-shaped spring [72]. ......................................... 97
Figure 7.2: Schematic and photo of a low resonant frequency VEH based on a ferrofluid
liquid spring with a cylindrical magnet array and a flexible coil. The volume and weight are
1.8 cc and 5 g, respectively [77]. .................................................................................................. 98
Figure 7.3: Microfabricated planar coils for electromagnetic energy harvesting [67]. ................ 99
Figure 7.4: Illustrations of the non-resonant vibrational energy harvester (VEH) composed
of five coils on the top and bottom sides of an acrylic chamber containing four rectangular
(left) and square (right) magnets................................................................................................. 100
Figure 7.5: Simulated magnetic field around the four magnets inside the acrylic chamber of
VEH. ........................................................................................................................................... 100
xvi
Figure 7.6: Photo of the VEH next to Samsung Galaxy Watch 4............................................... 101
Figure 7.7: (Top) Photo showing the hydrophobicity of a coil plate with a ferrofluid and
water droplet on the surface with contact angles of more than 90˚. ........................................... 102
Figure 7.8: Top-view photo of the VEH without the top plate................................................... 102
Figure 7.9: Cross-sectional-view photo of the VEH with five-coil arrays at the top and
bottom of the acrylic chamber, along with four rectangular magnets inside the chamber. ........ 103
Figure 7.10: Photos of (a) ACME AEX01 coil winding machine customized for
subminiature coils, (b) customized tooling to hold a cylindrical spool (1.8 mm in height and
1.3 mm in diameter), (c) coil winding over the spool, and fabricated (d) OMC and (e) CMC
along with the acrylic spools (used for the coil winding) and the magnets (to be used with
the fabricated coils)..................................................................................................................... 104
Figure 7.11: Calculated parametric and unit-less voltage (top) and power to a matched load
(bottom) by VEHs vs the number of turns in the coil; a larger number of turns means a
larger average distance between the coil (as the coil becomes thicker/taller with higher
turns) and the magnets. The CMC’s peak power is ~25% of the OMC peak power for an
applied acceleration of 1 g at 4 Hz, as the CMC and OMC have 200 and 300 turns for the
coils, respectively........................................................................................................................ 105
Figure 7.12: Experimental setup for VEH characterization........................................................ 107
Figure 7.13: Measured voltages from the bottom, top, and combined coil arrays when VEH
is driven with 2g acceleration at 2 Hz......................................................................................... 108
Figure 7.14: Measured power vs. acceleration as a function of frequency for the VEH with
a 250 Ω source resistance............................................................................................................ 109
Figure 7.15: Power vs. frequency for the two different designs (Fig. 7.4). ................................ 110
xvii
Abstract
The research elaborated in this thesis revolves around the design and application of multiple microelectromechanical systems (MEMS) that employ piezoelectric and electromagnetic sensors and
actuators, which are essential in developing highly efficient sensing systems for applications with
limited power resources.
The study features the design, simulation, and experimental analysis of a zero-power wireless
authentication system. This system utilizes a High-Overtone Bulk Acoustic Resonator (HBAR) as
an RFID tag for passive detection of target tampering activity, i.e., temperature elevation for desoldering followed by mechanical shocks for detaching integrated circuits (ICs) from printed
circuit boards (PCBs). The novel system operates at a frequency of 7.56 GHz with an fQ product
of more than 1013 and includes an energy harvester that generates a 6V pulse capable of
permanently changing the RFID tag's RF spectral properties.
Various energy harvesters have been developed using piezoelectric and pyroelectric properties
on multiple substrates, including bulk ceramics and bimorph structures, and through thin films.
These energy harvesters have been evaluated thoroughly to ascertain their effectiveness in
converting extreme thermal and mechanical excitations into electrical energy.
Further, the dissertation explores a compact wearable energy harvester that utilizes a nonresonant electromagnetic energy harvesting modality. This device, composed of wound microcoils and a magnet array suspended in ferrofluid within an acrylic chamber, has been fine-tuned to
xviii
generate power from low-frequency movements, such as human walking, despite its minimal form
factor.
In summary, this thesis presents a suite of innovative low-power solutions enabled by MEMS
resonators and piezoelectric thin films, suitable for various applications including but not limited
to secure wireless authentication of ICs and health monitoring using wearables.
1
Chapter 1: Introduction
The presented work demonstrates several MEMS devices based on piezoelectric, pyroelectric, and
electromagnetic properties, offering impressive performance for proposed ultra-low-power
sensing and actuating applications. This thesis outlines the development of multiple ultra-low
power devices in the realm of energy harvesters, MEMS resonators, and microactuators.
Specifically, it investigates a wireless and battery-less sensing system for securing the
semiconductor supply chain from counterfeit electronics through backscattering RFID
authentication with zero-power power constraints.
1.1. Motivation
Microsystems with power constraints require accurate design considerations to ensure they operate
effectively while managing minimal energy resources. These constraints are particularly relevant
for wireless or battery-less applications, where access to power is limited or wired power delivery
correlates with the system's ineffectiveness, and for devices that need to function with extremely
limited power resources or extended battery life duration. Wireless sensing and microactuation
involve using wireless technology to collect data from sensors or drive microactuators distributed
over a particular area or within specific objects without physical connections like wires or cables.
This technology has a wide range of applications, from supply chain security to healthcare,
industrial automation, and beyond. Here are some key aspects and applications of wireless sensing:
2
• Semiconductor supply chain security: Unfortunately, the ongoing chip shortage has
created a massive opportunity for counterfeit electronics to grow, posing a significant
threat across the semiconductor supply chain. Protecting ICs from tampering by
embedding wireless and battery-less backscatter tamper detectors into the IC packaging
could be a solution. An embedded sensor with dimension constraints set by various IC
packaging can keep records of tampering activities and reduce the profit margins of
counterfeiters by combatting illegal recycling, cloning, reverse engineering, and HW
trojans.
• Wearables batteries: The importance of energy harvesting in increasing the battery life
of wearable devices cannot be overstated, as it lies at the intersection of sustainability
and technical innovation. With their small size and ubiquitous nature, wearables
sometimes encounter considerable battery capacity limits, affecting operating
longevity and user experience. Wearable devices enhance convenience, connectivity,
productivity, and health monitoring for users' daily lives. Efforts to extend the battery
life of wearable devices while maintaining device functionality are a primary priority
in developing wearable technology.
The development of ultra-low power devices is a response to these challenges, where sensors
are designed to operate without an external power supply or a battery, instead relying on energy
harvesting methods or wireless power delivery. These sensors can derive power from external
wireless physical sources such as thermal gradients, RF signals, or mechanical vibrations, enabling
a much smaller form factor and extended operating life cycles while reducing their dependency on
bulky electrical power sources. Integrating such low-power sensors into microsystems allows for
passive data collection and continuous monitoring without needing battery replacements or wired
3
power sources, revolutionizing how we deploy sensors in inaccessible or harsh environments and
paving the way for a new generation of autonomous microsystems.
1.2. Problem Statement
This dissertation introduces a variety of innovative devices and techniques utilizing MEMS-based
sensors and microactuators to overcome the difficulties encountered in sensing applications for
power-constrained, remote, and wearable systems. Although each device and method serve a
distinct role, they all share a common principle: leveraging the inherent ability of MEMS devices
to generate or process signals passively and for integration into microfabricated systems in fields
traditionally reliant on actively powered devices.
Specifically, the research results for an application of a wireless and battery-less tamper
detection sensor, shown in Fig. 1.1, for enhancing the authenticity and security of the
semiconductor supply chain against counterfeit electronics, are explained in detail.
The IC tampering activity this project mainly targets is the cheap recycling of in-use or obsolete
electronic components from PCBs. The target tampering activity is performed by desoldering
ICs followed by mechanical banging of the PCB, resulting in scavenging numerous tampered ICs
in a short period, providing a considerable margin of profit for malicious counterfeiters. Through
an explanation of the details of the achievements in this work, explorations regarding various
energy harvesting techniques and materials required for this application, as well as other
microsystems and wearable devices in general, are expanded.
4
Figure 1.1: Illustration of an anti-tamper chip inside an IC package and being interrogated by an RFID
reader.
This thesis also proposes a complete design, device fabrication, and experimental evaluation
of energy harvesters utilizing pyroelectric, piezoelectric, and electromagnetic modalities with their
applications.
1.3. Scope of Work
A Passive RFID-Based Embedded Anti-Tamper (PREAT) chip for IC authentication in the
semiconductor supply chain has been developed, characterized, and tested in multiple phases.
Initially, the general concept of wireless authentication based on bulk acoustic resonators was
explored. A film bulk acoustic resonator (FBAR) operating at GHz is shown to drastically change
RF characteristics behavior before and after delamination caused by an unlimited amount of
supplied power delivered to it. Later, a new approach for detecting tampering activities (i.e., the
sequence of temperature elevation followed by mechanical banging) was proposed. The system
contained a bulky proof-of-concept pyroelectric energy harvester (PEH) that generates large
electrical voltage and charges upon facing a temperature gradient. The PEH delivers charge to a
high quality-factor (Q) high-overtone bulk acoustic resonance (HBAR), breaking it permanently
5
and making it non-interrogable through RF interrogation. The intact HBAR, operating at 7.56
GHz, was fabricated and tested through wired and wireless characterization. Automated dielectric
breakdown experiments were explored. The Last phase discusses the implementation of a
miniature version of the proof-of-concept sensor with dimensions that can fit into the packaging
of regular ICs. The bulky mechanical switch and antennas were also replaced by a miniature
version of such components, all placed on a flex polyimide PCB to form the miniature PREAT
chip.
Various pyroelectric, piezoelectric, and electromagnetic energy harvesting techniques were
experimentally investigated. All three modalities were designed, fabricated, and characterized.
Cantilever-based vibrational energy harvesters were developed using piezoelectric bulk ceramics
and thin films to generate charges in response to mechanical vibrations and shocks to charge a
capacitor gradually. A pyroelectric energy harvester was built to create the required charge and
voltage to break the HBAR tag when desoldering occurs at temperatures beyond 250 ˚C. A nonresonant electromagnetic vibrational energy harvester that produces decent power from human
walking motion under accelerations as low as 2g and sub 4 Hz bandwidth.
1.4. Overview of Chapters
In Chapter 1, an introduction, the motivation, and the scope of work for the accomplished research
are discussed at a high level, and the general description of tamper detection, energy harvesters,
and low-power auscultation applications is discussed.
6
Chapter 2 expands on the concept of tamper detection and its necessity in today’s IC supply
chain, the detailed background of existing approaches to combat counterfeit electronics, and the
PREAT microsystem's mechanism to record tampering activities.
Chapter 3 demonstrates the design, modeling, integration, and characterization of
pyroelectric energy harvesters based on Lithium Niobate (LiNbO3). A series of thermal
experiments have been exhibited in this chapter, showing the exceptional thermal performance of
LiNbO3 crystals and their capability to harvest large voltages excited by temperature gradients.
In Chapter 4, the design, fabrication, and high-frequency performance of FBAR and HBAR
are explored. The resonators’ RF performances are compared, and it is demonstrated that using
HBAR has an edge over FBAR due to its higher quality factor, specifically at higher frequencies.
Chapter 5 explores the breakdown characterization of the RFID tags and the wireless
performance variations of broken vs intact tags. The wireless experiments include interrogating
antennas connected to a vector network analyzer (VNA) communicating with PREAT antenna to
interrogate RF MEMS tag. The design, simulation, implementation, and characterization of
various antennas used in the tamper detector have been included. The electromechanical benchtop
characterization of a miniature MEMS switch is also discussed through a setup that features a
customized programmable shock table. The MEMS switch detects shocks beyond 40 g and
transfers generated charges by PEH to the HBAR in response to mechanical shocks. Finally, the
implementation of the miniature PREAT chip on a polyimide flex substrate is shown.
Chapter 6 is dedicated to exploring an alternative modality for energy harvesting through
piezoelectric energy harvesters using bimorph ceramic (PZT) and unimorph thin films (ZnO).
Piezoelectric cantilever-based vibrational energy harvesters are designed, fabricated, and
7
characterized to produce significant voltage spikes when facing mechanical shocks, which can
ultimately charge a capacitor.
In Chapter 7, a wearable non-resonant vibrational electromagnetic energy harvester is
designed, implemented, and characterized for power harvesting from human walking motion. The
demonstrated set of experiments and results for multiple designs, fabricated and miniaturized to
produce maximum power in the smallest feasible form factor, will offer an overview of
engineering trade-offs and suggest potential applications for such energy harvesters.
In Chapter 8, the conclusions of the presented research are discussed, and future directions
are explained.
8
Chapter 2: Wireless and Battery-less Tamper Detector for
Semiconductor Supply Chain
This chapter describes the critical need to enhance the security of the semiconductor supply chain
to safeguard against tampering. A comprehensive background of the research undertaken in this
field to address the counterfeiting issue in the semiconductor supply chain is first presented. Then
is shown a novel approach through wireless interrogation of ICs based on an embedded anti-tamper
chip which detects and records, without battery, cheap tampering activity that counterfeiters use
to detach a large number of chips from printed circuit board (PCB) by de-soldering followed by
mechanical banging of the PCB. The necessity and trade-offs of this approach are discussed at the
end of the chapter.
2.1. Background
In recent years, both the US and the global semiconductor industry have witnessed a troubling
increase in counterfeit electronic components. This rise is fueled by rising demand for electronic
devices, complex supply chains, and easier access to advanced manufacturing methods.
Counterfeit semiconductors, which mimic real ones but are often of inferior quality, bring
significant risks to mission-critical applications. As those lead to product failures, security
vulnerabilities, and safety problems [1], [2], the supply chain needs to be secure through
authentication technologies [2]. Hence, a hardware solution is required to protect the supply chain
9
of semiconductor components. The best way to deal with fake electronic components is to buy
items directly from the original manufacturer, affiliate, or post-retail provider approved by the
manufacturer. However, this is not always plausible or profitable in many electronic supply chains,
resulting in purchases through many independent retailers [3].
The Semiconductor Industry Associates (SIA) estimates that counterfeit semiconductor
components cost U.S. IC companies around $7.5 billion annually [4], [5]. The estimated global
counterfeiting of electronic components, including ICs, can cost the semiconductor industry $100
billion annually [6]. According to the European Union Intellectual Property Office (EUIPO),
counterfeiting of electronics, especially mobile phones and components, is on the rise, targeting
online markets and exploiting the chip shortage. Fake products infringe on IP, raise safety
concerns, and require urgent action [7].
Among the five most frequently counterfeited semiconductors, analog ICs topped the list,
comprising 25% of the reported cases. Microprocessor ICs and memory ICs followed closely, with
counterfeit rates of 13.4% and 13.1%, respectively. Programmable logic ICs comprised 8.3% of
the reports, while transistors were at the bottom with 7.6% [8].
In the semiconductor supply chain (Fig. 2.1), there are vulnerabilities of hardware trojan
and implants at various stages of the semiconductor supply chain [9]. In this manuscript, we deal
with the last stage of the supply chain and present a low-cost technique that will ensure the delivery
of authentic semiconductor chips to users by embedding into a semiconductor package a tamper
detector (1) capable of recording tamper activity (i.e., de-soldering followed by mechanical
banging) without battery and (2) wirelessly interrogatable of the recording.
10
Figure 2.1: Vulnerabilities in the semiconductor supply chain, with the tamper detector’s functionality
potentially protecting against recycling [9].
Counterfeit activities involve manipulative interventions at different manufacturing and
distribution sections, encompassing the introduction of hardware trojans and implants [10], the
misrepresentation of ICs during their transportation and distribution, as well as the illegal recycling
and re-utilizing of chips during or after their operational lifecycle [11]. A secure supply chain,
though, is challenging due to the intricate nature of the global supply chain market, where many
companies may buy ICs from third-party suppliers, especially when demand exceeds the available
stock in the primary suppliers. Though trustworthy third-party suppliers may reduce the risk of
counterfeit ICs, the semiconductor supply chain can be much more secure through a low-cost
tamper-detection technology to authenticate ICs.
At the late stages of the semiconductor supply chain, counterfeiters acquire tampered or
low-quality semiconductor components through various methods, including but not limited to [12]:
11
• Recycling: Recycling stands as the most common type of counterfeiting, involving the
resale of used ICs as new ones, purportedly by the original maker. This practice frequently
leads to diminished performance and a shorter operational life for the components,
attributed to possible harm incurred during recycling. This process includes the detachment
of components at high temperatures and the risk of introducing flaws while the ICs are
being cleaned and repackaged.
• Cloning: Illegal cloning refers to the unauthorized duplication and production of
semiconductor devices. This practice involves reproducing an original IC's design and
functionality without the permission of the copyright or patent holder. Cloning can range
from producing exact replicas to modifying certain aspects to avoid direct copyright
infringement while still capitalizing on the original design's performance characteristics.
• Re-marking: Re-marking involves altering the external markings on integrated circuits to
misrepresent their identity, performance specifications, or origin. This can include
changing the part number, manufacturer logo, or batch number to make a lower-grade or
counterfeit chip appear as a higher-grade, genuine component from a reputable
manufacturer. The remarking process often involves erasing the original markings through
methods like sanding or chemical etching, followed by the application of new, fraudulent
markings.
Various approaches have recently been introduced to combat tamper activities and trojan
implantation at different supply chain stages, as the anti-tamper approach depends on the
counterfeit strategy and the stage in which either IC or Printed Circuit Board (PCB) is under attack.
Fig. 2.2 depicts a hierarchical classification of these detection techniques [13].
12
Figure 2.2: Hierarchical classification of the various counterfeit detection techniques [13].
For instance, Mosavirik et al. employed an on-chip Field Programmable Gate Array
(FPGA)-based vector network analyzer (VNA) to characterize the impedance of the power
distribution network (PDN) within the system (Fig. 2.3). They also relied on utilizing machine
learning to detect traces of tampering by measuring scattering parameters of the entire PCB with
a single measurement and show that the counterfeiting activities affect the overall impedance of a
PCB differently in various frequency ranges [14], [15].
Figure 2.3: Overview of ScatterVerif: verification of electronic boards using reflection response of power
distribution network [14].
Vasile et al. [16] used a unique conductive mesh that detects intrusions and an active tamper
detection circuit that examines the mesh using signals that counterfeiters cannot replicate (Fig.
2.4). Other approaches include the usage of an integrated photodetector to detect crucial program
data loss during tamper activities [17], recording the history of IC usage and a built-in ring
13
oscillator frequency in non-volatile memory [18], and X-ray tomography to investigate traces of
tampering activities [19].
Figure 2.4: Schematic diagram of the active tamper detection system [16].
In case of relabeled IC counterfeits, the Dynasolve test [20] may work, as Dynasolve dissolves
cured urethanes, silicones, anhydride-cure epoxies, etc., and can remove the blacktopping used to
relabel ICs (Fig. 2.6). However, the Dynasolve test fails to detect counterfeit ICs that have not
been relabeled or have advanced blacktopping materials. Additionally, one can use abrasive acids
to etch the surface of the IC package to expose the IC die (within the package) for inspection via
microscopy. Or the IC die may be exposed by removing the lid of the IC package via mechanical
grinding. These tests, though, are not only destructive but also misses to catch many scavenged
ICs. Using cryptographic keys with random numbers, such as those based on tamper-sensitive
MEMS arrays [21] (Fig. 2.5), is an approach that is helpful in preventing reverse engineering but
not in detecting repackaging attempts. Electrical tests can catch some fake ICs but miss the
counterfeits that have the correct functionality (or the same IC chips) and are time-consuming and
14
destructive in most scenarios. These safeguards against counterfeiting are either ineffective in
detecting sophisticated counterfeits or too expensive/intrusive, and in general, they all need
electrical power to detect and record tampering.
Figure 2.5: MEMS-assisted anti-tamper (MAAT) system with three primary components: critical program
information (CPI), the MEMS array, and the MEMS monitoring engine (MME) [21].
So far, very few companies or labs have been working on advanced techniques to detect
IC-package tampering that does not require any electrical power or battery in order to deter IC
counterfeiting through scavenging of used IC chips from PCBs, though a dominant counterfeit
practice is to scavenge old/used semiconductor chips from PCBs and sell those as if they were
new.
Figure 2.6: Results from Dynasolve 750 testing. Both samples were soaked for 45 minutes in Dynasolve
750 at 105 ˚C [20]. As can be seen, scratch marks clearly appear on the tampered IC after the test.
15
To scavenge IC chips, PCBs are heated to 250 – 400 ˚C (to melt the solder) and then hit
against a hard object (to dismount the ICs that used to be soldered on the PCBs). The harvested
ICs are then relabeled, repackaged, or cosmetically cleaned before being sold as new ICs [22].
This practice leads to not only lost revenue for IC companies but also to increased failure risk in
operationally critical systems.
Since embedding a mm-sized battery-less tamper detector inside an IC package and
wirelessly interrogating the detector is a highly effective, as a low-cost anti-counterfeiting
approach, this thesis describes the proof-of-concept mm-sized prototypes and experimental results
that show the feasibility of benefiting from such tamper detectors. Specifically described is a
design based on four key components: (1) Pyroelectric Energy Harvesters (PEH) capable of
producing voltage and charge in response to de-soldering attempts, (2) a High-overtone Bulk
Acoustic Resonator (HBAR) working as a radio frequency identification (RFID) tag that can be
damaged by the voltage and charge produced by the pyroelectric energy harvester and that can
also be wirelessly interrogated, (3) an acceleration switch, and (4) multi-GHz antennas for wireless
interrogation. In this manuscript, the words pyroelectric energy harvester (PEH) and pyroelectric
energy converter (PEC) are used interchangeably.
2.2. Project Phases and Sensor Design
The proposed mechanism of our tamper detector is depicted in Fig. 2.7. Before targeted tampering
activity happens, the HBAR RFID tag is intact and can be wirelessly interrogated through the
RFID antenna using a wireless interrogator. After the tampering activity, the HBAR tag is broken,
and the tag is unidentifiable to the interrogator.
16
Figure 2.7: Wireless detection system concept: (Top) before the HBAR RFID tag is broken and (Bottom)
after the HBAR RFID tag is electrically broken by tamper activity, which is composed of two sequential
events of de-soldering followed by mechanical banging.
For tamper detection, banging with 40g shocks alone should not break the RFID tag since
semiconductor chips face shocks as large as thousands of g during sustainability tests. Nor
temperature rise alone should break the RFID tag, as the chips experience 250 - 300 ˚C while being
soldered on PCB. Thus, the tamper detector must detect the exact sequence of temperature rise
(due to de-soldering activity) followed by 40 g banging actions. The proposed system is based on
a PEH (that provides enough voltage and charge when there is a temperature rise of 250 - 300 ˚C)
and a MEMS acceleration switch (that makes a connection between the PEH and the RFID tag due
to 40 g shocks). The exact sequence of the two events and both of the two events happening in less
than tens of seconds are the necessary conditions for the RFID tag to be permanently broken, as
17
illustrated in Fig. 2.8. The change of the impedance and Q of HBAR due to the tampering activity
can be used to determine whether there has been a tamper activity.
Figure 2.8: Tamper activity composed of two sequential events, one after the other, in that order and both
(not just one): (a) de-soldering, which requires about 300 °C temperature rise, followed by (b) mechanical
banging, which makes the MEMS acceleration switch to be turned on to deliver pyroelectric
voltage/charge from LiNbO3 to HBAR RFID tag.
In the next chapter, a bulky proof-of-concept sensor with components as large as a cmscale is first described, followed by a miniature mm-scale design of the PREAT chip. The proofof-concept tamper detector is composed of a PEH, a mechanical switch, and a high-quality (Q)
HBAR coupled to a microstrip patch antenna; the four elements to build a prototype system shown
in Fig. 2.9. The HBAR on Sapphire is chosen because of the precision of the resonant frequencies
down to 3 – 4 significant digits due to its Q being greater than 2,000 at resonant frequencies in
GHz, in order to make it extremely difficult for counterfeiters to replicate HBARs and replace the
HBARs that have been damaged by the tamper activity. The proof-of-concept system includes a
LiNbO3 crystal PEH capable of generating large voltage and charge in response to temperature
change due to de-soldering activity, a cantilever-based switch that delivers the accumulated charge
and voltage (i.e., power) from PEH to HBAR-based RFID tag, and a microstrip patch antenna
coupled with the RFID tag.
18
Figure 2.9: Prototype wireless tamper detection system with a wireless interrogator transmitting a signal
and measuring the backscattered signal from the tamper detector. The strength of the backscattered signal
depends on the HBAR RFID tag that is connected (1) directly to a microstrip patch antenna and (2) to a
pyroelectric energy converter (PEC) via a cantilever-based acceleration switch. The PEC generates a large
voltage (and charge) in response to temperature change due to a de-soldering process, while the cantilever
switch is turned on by mechanical banging (which follows the de-soldering) to let a large voltage and
charge to flow from the PEC to HBAR to break down the HBAR permanently and thus to alter the
characteristic of the backscattered signal.
After the proof-of-concept tamper detector (which is not an integrated detector that could fit
into an IC package), the miniature PREAT detector (Fig. 2.10), an integrated device that can be fit
into an IC package, is presented along with the experimental characterization of the components
used in a PREAT detector.
Figure 2.10: Schematic of the front and back sides of a PREAT chip along with a top view of a highovertone bulk acoustic resonator (HBAR) used in the PREAT.
19
The experimental results presented in each of the following chapters indicate that PREAT
offers a reliable and low-cost authentication of semiconductor chips. In the following chapters, we
start with the design, implementation, and characterization of PEH and the RFID tags, followed
by their assembly with the switch and antenna to perform tamper detection.
2.3. Summary
This chapter describes the existing semiconductor tamper detection methods, including a system
impedance verification, electrical and Dynasolve testing, X-ray tomography, and MEMS-based
cryptographic keys. It highlights how these current techniques fail to detect tampering activities
effectively due to their low accuracy and time-consuming requirements.
In response to these inadequacies, the chapter introduces a novel approach that employs
battery-free wireless RFID sensing based on MEMS devices, mainly pyroelectric energy
harvesters and bulk acoustic resonators, integrated into a miniature embedded form factor. The
tamper detector is designed to detect a specific tamper activity that involves the detachment of ICs
from PCBs through de-soldering followed by mechanical banging. The chapter further outlines the
development stages of this detector, beginning with an initial bulky proof-of-concept system
followed by a miniaturization of the system to a millimeter scale so that it may be incorporated
within IC packaging.
20
Chapter 3: Pyroelectric Energy Harvesters (PEH)
This chapter describes the development of pyroelectric energy harvesters (PEHs) that produce the
required voltage and charges to break the dielectric film inside the MEMS RFID tag. Lithium
Niobate (LiNbO3) substrate is used for PEHs due to its exceptional pyroelectric, thermal, and
electrical properties, as it effectively produces electrical charge and voltage from a slowing rising
temperature. Beginning with an overview of the background of pyroelectric energy techniques and
the diverse materials currently employed, the chapter extends to the theoretical analysis, modeling,
and implementation of PEH that leads to the design of effective pyroelectric energy harvesters.
Finally, the experimental studies assessing the performance of the fabricated PEHs are presented.
3.1. Background
In order to accumulate enough charge and deliver a large voltage to break the RFID tag inside the
temper detector, we need to have a source of power generation from the heat applied to printed
circuit boards (PCBs) during the de-soldering of ICs on the PCBs. For that purpose, we rely on a
pyroelectric energy harvester that can endure high temperatures (as high as 500 ˚C) and produce
sufficient voltage and charges.
Pyroelectricity is a unique phenomenon found in certain crystalline materials that changes
their electric polarization in response to variations in temperature [23]. Pyroelectricity is different
21
from thermoelectricity (also known as the Peltier-Seebeck effect) [24], and the two concepts
should never be interchanged as each works based on different physical principles. In pyroelectric
materials, the temperature change alters the material's polarization, producing an electric field.
This effect is reversible, and the direction of the electric field changes with the sign of the
temperature change. On the other hand, thermoelectricity is based on the Peltier-Seebeck effect,
where a voltage is generated across two different conductive materials when there is a temperature
gradient between them. In other words, pyroelectricity typically relies on temporal temperature
changes, while thermoelectricity relies on spatial temperature variations. The pyroelectric
phenomenon is closely related to the piezoelectric effect, which produces electric charge or voltage
in response to a mechanical strain and/or stress [25].
The ability of pyroelectric materials to convert thermal energy into electrical energy allows
for creating systems that can harvest power from waste heat or ambient temperature gradient [26],
[27]. Although the energy harvesting concept based on pyroelectricity was first introduced in 1960,
in the past decades, only about 7% of published work (Fig. 3.1 [28]) on pyroelectric materials are
on their energy harvesting capabilities.
22
Figure 3.1: Published papers from 1999 to 2019 on pyroelectric materials vs. papers investigating their
energy harvesting applications (in red) [28].
Various techniques and materials have been investigated to create pyroelectric energy
harvesters [29, 30, 31, 32]. For instance, Antolak et al. studied a Lithium Tantalate-based
pyroelectric high-voltage pulser for producing pulse of up to 86 kV, where ~50 mJ energy was
extracted from the crystal over a temperature variation of 75 °C (Fig. 3.2) [33]. Patel et al. explored
high energy–density pyroelectric materials (called PNZST) and achieved a large energy density of
1.0 MJ/m3 per cycle with temperature cycles from 303 K to 403 K and with electric fields as large
as 9 kV/mm [34].
23
Figure 3.2: Photo of a single 50 mm diameter × 1 mm thick LiTaO3 crystal generating multi-kV from
75˚C temperature variations [33].
Table 3.1: Comparison of properties of various pyroelectric materials [35].
Table 3.1 [35] summarizes the pyroelectric and thermal properties of several pyroelectric
substrates and shows that Lithium Niobate and Lithium Tantalate can be used in PREAT chips due
to their high Curie temperature and pyroelectric coefficient. The Curie temperature is the specific
temperature at which pyroelectric and piezoelectric materials undergo a critical phase transition,
24
losing their spontaneous ferroelectricity and transitioning from a ferroelectric state to a paraelectric
state [36].
3.2. Design, Modeling, and Fabrication
The polarization in a pyroelectric material varies as temperature changes. When metal electrodes
are deposited across a LiNbO3 crystal, a PEH is formed for which the short-circuit pyroelectric
current flow due to temperature T change in time t is [37]:
=
(3.1)
where and A are pyroelectric coefficient and active area, respectively.
In our case, the temperature change is produced by pointing a soldering hot gun perpendicular to
the PEH substrate. The heat-up and cool-down processes induce accumulated charges (Fig. 3.3),
resulting in a large open circuit voltage across the PEH, which gradually discharges.
Figure 3.3: Illustration of pyroelectric mechanism for LiNbO3 crystal working as PEH.
25
Several parasitic and intrinsic electrical loads play a role in discharging accumulated
charge across an open loop circuit of PEH. Among them are the impedance of the PEH, the
impedance of the PREATS packaging, and the ambient parasitic load.
The equivalent open-circuit modeling of the PEH shown in Fig. 3.4 leads to an equation
for a low roll-off frequency, which is related to the inverse of the electrical time constant that is
impacted by the abovementioned loads. On the other hand, the conservation of energy rule for the
heat transfer indicates that the total amount of heat delivered to the sample should equal the sum
of the power the PEH can generate and the heat loss. This defines the concept of thermal time
constant, which is the thermal capacity of the PEH divided by its thermal conductivity. It is
experimentally observed that the electrical time constant is much larger than the thermal time
constant for LiNbO3, bringing the electrical roll-off frequency below the thermal roll-off
frequency. In the case of sinusoidally varying temperature with frequency , the voltage V induced
by the PEH is [37]:
() = 0
√(1 +
22)√(1 +
22)
(3.2)
where 0
is determined by the peak-to-peak temperature change, while and are the thermal
and electrical time constants of the PEH, respectively. The following two equations can estimate
these two parameters:
= (||||) ( + + ) (3.3)
26
=
(3.4)
where and are the equivalent capacitance resistance of the PREATS packaging, respectively,
while and are the equivalent capacitance and resistance of the ambient, respectively; and
and are the thermal capacity and thermal conductance of the PEH, respectively, and are affected
by the ambient and the packaging.
For the charge leakage during heating/cooling cycles to be negligible so that the roll-off
frequency is orders of magnitude lower than 1 Hz, and (which depend on packaging
conditions and environmental humidity) must be large, and will be indeed large in the PEH’s
intended application where the tamper detector is placed inside a hermetically-sealed IC package
which also provides the needed heat conductance to the PEH. Many IC packages are made with
materials of high electrical resistivity, which can potentially contain electrically insulating fillers
that offer high thermal conductivity (in the range of several −1
−1
) [38], [39], and thus, we
can ignore the potential impact of on increasing the roll-off frequency.
Figure 3.4: Equivalent electrical circuit of PEH before the switch is activated.
27
By neglecting ambient and packaging charge leakage, the charge-leak time constant (due
to resistance RPEH and capacitance CPEH) is calculated to be 47.75 minutes (i.e., the lower roll-off
frequency of 55 µHz) for LiNbO3, based on the resistivity and dielectric constant shown in
Table 3.2.
3 =
1
2
=
1
2
= 55 µ (3.5)
Table 3.2: Material properties of LiNbO3 [40], [41], [42].
Relative Dielectric
Constant (r) 85.2 Bulk Resistivity () 3.81014 Ω.cm
Z-cut
Wave Speed
7,316 m/sec Pyroelectric Constant (p’) -5 to -3.7 µC/m2
˚C
Curie Temperature 1,210 ˚C Melting Point 1,253 ˚C
With such a large RC time constant, there is minimal charge leakage between PEH’s top and
bottom electrodes over minutes. In the case of an open circuit, the electrical displacement D is
related to the electrical field E and temperature change T as follows:
Δ⃗ = 0Δ⃗ + Δ = 0 (3.6)
where and 0 are the pyroelectric coefficients and dielectric permittivity, respectively. And
the pyroelectric coefficient along the thickness direction (p3), we have
28
Δ3
Δ
= −
1
0
3
ϵr
(3.7)
Thus, the voltage changes due to the temperature change when the frequency of the temperature
change is much larger than 55 µHz is:
Δ ≈ ∫ −
1
0
3
ϵr
Δ =
0
3
ϵr
Δ
(3.8)
where d is the thickness of the pyroelectric crystal. The charge accumulated on the PEC is:
Δ = Δ =
0
0
3
ϵr
Δ = 3A ΔT
(3.9)
The expected frequency response Bode plot from Eq. 3.2 is shown in Fig. 3.5 with a
passband between the two roll-off frequencies (i.e., electrical and thermal roll-off) where the
voltage response peaks. By selecting LiNbO3 and ensuring good isolation by the packaging and
ambient, the energy harvesting design is optimized to fall in this range [40], [41], [42].
29
Figure 3.5: Theoretical PEH’s open-circuit voltage vs. frequency showing the effects of the electrical and
thermal time constants at the low and high frequencies, respectively.
For a preliminary bulky proof-of-concept design, a LiNbO3 crystal, not polished on any
sides, Z-cut, with 1025 mm3 dimension, as the PEH, is connected to a 51 cm2
cantilever
(emulating as a mechanical switch) made of a commercially available Kapton-Copper flexible
sheet (Fig. 3.6). Also, another PEH made of a Z-cut LiNbO3 substrate wafer (550.15 mm3
) is
connected to a 173 mm2
cantilever made of the same Kapton-Copper flexible sheet (DuPont
LF8510R composed of 25 µm thick Kapton, 25 µm thick adhesive, and 18 µm thick copper). The
electrodes of the cantilever are attached to the LiNbO3 using silver conductive epoxy, which is
cured at 65 ˚C in an oven for 15 minutes.
30
Figure 3.6: Photo of 2510 mm3
crystal LiNbO3-based PEH with a copper-Kapton cantilever switch.
The last design explored and used in the miniature PREAT chip ultimately is miniature
PEH with areas as small as 4 mm2
. The miniature PEHs are fabricated on a 500 µm thick X-cut
LiNbO3 substrate (3” wafer). Two hundred nm thick sputtered Aluminum electrodes are deposited
on the top and bottom surfaces of the substrate and are patterned (Fig. 3.7). A dicing saw machine
is used to dice the wafer into individual chips with areas ranging from 4–16 mm2
.
Figure 3.7: Miniature PEH (left) Cross-sectional diagram of miniature PEH composed of LiNbO3
substrate and aluminum layers with device areas from 4 to 16 mm2
, (right) fabricated devices on the
substrate before dicing.
The generated charge and voltage of the PEH due to the heat associated with de-soldering
is delivered to the RFID tag through a momentary (non-latching) mechanical switch that
31
temporarily establishes a connection between its two poles due to mechanical shock larger than
40-50g. We have integrated a micro shock switch (HT-Micro AT-50-T, HT-Micro Inc.) with
1.841.841.3 mm3 dimensions and a weight less than 25 mg into the PREAT chip for fast charge
delivery while providing extremely high insulation resistance when no acceleration is applied to
the PREAT chip. Note that the acceleration alone (without prior temperature elevation) cannot
break the HBAR, as sufficient charge should be built on the PEH before the acceleration is sensed.
3.3. Experimental Setups
To assess PEH’s feasibility of voltage and charge delivery to a multi-MΩ load in the proof-ofconcept bulky design, we start with bulky PEHs and eventually transition to experiments on the
miniature PEHs. Both the 2510 mm3
crystal LiNbO3 PEH and the 550.15 mm3 Z-cut substrate
PEH with a copper-Kapton cantilever switch are tested.
The PEHs are heated with a digital hot plate at a temperature ramp of about 2 ˚C/sec after
being taped to the hotplate’s surface with Kapton tape. As the PEH produces a voltage in
proportion to temperature, the air gap in the switch becomes narrower as the PEH temperature
increases (Fig. 3.8). When the temperature of the PEH is increased by 6˚ and 275 ˚C for the bulk
crystal and thin substrate design, respectively, the air gap becomes very narrow (< 1 mm), which
can be zero-ed by applying a mechanical shock.
32
Figure 3.8: Photos of a 550.15 mm3 LiNbO3-based PEC (integrated with a copper-Kapton cantilever
switch having a large air gap due to the warping of the cantilevers) placed on a hotplate, as the
temperature on the hotplate is raised from room temperature to 300 ˚, showing the air gap (between the
top and bottom beams of the switch) becoming less due to the temperature-induced voltage on the PEC.
At 300 ˚C, a narrow gap (~1 mm) remains between the top and bottom beams of the cantilever switch that
can be zero-ed by applying a mechanical shock.
To characterize the PEH’s charge and peak voltage as a function of temperature, we
connect an 8 MΩ resistor in series to the switch (similar to HBAR’s impedance as HBAR’s DC
resistance is about 5-10 MΩ) and measure the voltage drop across the resistor, as the temperature
is raised, in the setup illustrated in Fig. 3.9. A high-input-resistance (>1 GΩ) pre-amplifier is used
to minimize the loading effect of the oscilloscope (which has 1 MΩ input resistance).
Figure 3.9: Schematic of the measurement set-up showing the LiNbO3-based PEH with Kapton-Copper
cantilever switch and two wires over a hot plate.
33
After the feasibility assessment of the bulky PEHs, an in-depth evaluation of the miniature
PEHs is performed. This time, we use a soldering/de-soldering hot gun to apply heat to the
samples, which mimics similar tools that counterfeiters use in tampering. The current delivery
capability of PEHs is investigated in both short circuit (zero-load) and loaded configurations.
The temperature rise that PEH will experience during the de-soldering step is caused by a
hot gun placed at a distance from PEH, as shown in Fig. 3.10. The temperature at the front and
back surfaces of the PEH is measured by two thermocouple probes in contact with the surfaces.
The two probes are connected to a digital 2-channel thermocouple (HH506RA), and we take the
average of the two probes as the temperature at the PEH. With the low and high de-soldering
temperatures set at 250 and 427 ˚C, respectively, the target temperature at the PEH is varied by
varying the distance between the tip of the hot gun and the PEH (1 - 2 cm) and the airflow level of
the hot gun (flow levels 4, 6, and 8).
Figure 3.10: Experimental setup for characterizing PEH’s ground truth response to temperature variation
with a 2-channel digital temperature measurement unit (HH506RA) and a soldering hot gun at specific
gaps for various max target temperatures under different airflow levels.
34
The current generation capability of PEH is characterized by a pico-ammeter (Keithley
6485), as shown in Fig. 3.11, through two gold-plated tungsten needle probes contacting PEH’s
top and bottom electrodes. The PEH, hot gun tip, electrical probes, and lines are all placed in a
metal shielding box to minimize electromagnetic interference (EMI). A data acquisition (DAQ)
system (LabJack T7-Pro) is connected to the pico-ammeter to record the measured current values.
Figure 3.11: Experimental setup for PEH’s short circuit current measurement with a Keithley 6485 picoammeter and a LabJack T7-Pro DAQ connected to a computer. The electrodes on PEH are accessed by
gold-plated tungsten needle probes inside a metal EMI shielding box. The tip of a hot gun is placed inside
the metal box.
Finally, the charge delivery from PEH to HBAR is characterized by a switch connecting
the PEH (inside the metal box) to a 10 MΩ resistor (emulating HBAR), as shown in Fig. 3.12. The
switch is connected once the PEH reaches the maximum temperature change, and the fast voltage
spike is captured via the oscilloscope’s forced triggering option.
35
Figure 3.12: Experimental setup for characterizing the delivery of PEH voltage and charge to a low
impedance (10 MΩ) load when a mechanical switch connects the PEH to the load after the temperature
reaches the maximum by a hot gun placed 1 cm above the PEH inside a metal box. An oscilloscope with a
forced trigger captures the fast spike.
3.4. Results
This section starts with the experimental results obtained with the bulky proof-of-concept
PEHs and then describes the results obtained with the miniature PEHs. For the bulky designs, the
measured voltage across the 8 M resistor before and after the copper-Kapton switch is connected
(Fig. 3.13) shows a peak voltage of about 9.6 V, much higher than the 2.2 V needed to break down
the HBAR’s piezoelectric film (HBAR’s breakdown properties and phases are discussed in detail
in Chapter 5). From the current (obtained from the measured voltage) vs time, the total charge that
flows through the resistor from the PEH is about 1.36 nC, large enough to ensure thermal runaway
on the HBAR’s piezoelectric film.
36
The 550.15 mm3 PEH generates sufficient voltage and charge to break the HBAR when
the temperature is raised from room temperature to almost 300 ˚C (based on HBAR breakdown
characteristics in Chapter 5). The PEH will generate similar voltage and charge when the
temperature is lowered from 300 °C to room temperature (or room temperature to -250 °C) but
will not break the HBAR, similar to the case of the temperature rise unless sufficient mechanical
shocks are applied (to activate the acceleration switch), right when both the voltage and charge are
sufficiently high. Also, the largest crystal PEH showed similar results with only a 6 ˚C temperature
change due to its larger area and increased thickness.
Figure 3.13: (a) Hotplate temperature vs time, indicating the temperature ramp at 2 ˚C/sec. (b) Measured
voltage over the 8 MΩ resistor upon the switch connection after the voltage from the PEH rises high
enough due to temperature change followed by a mechanical shock.
For transitioning to a more controlled evaluation of miniature PEHs performances for the
PREAT chip, the initial experiment aims to characterize the ground truth temperature reference
for the applied heat and cooldown cycles sensed by the PEHs based on 0.5 mm-thick X-cut LiNbO3
substrates. As can be seen in Fig. 3.14, the steady-state temperature at PEH is in the range of 130
- 350 ˚C, with the gap between the hot gun and the PEH playing the pivotal role. Based on these,
the gap and the airflow are fixed at 1 cm and the maximum level of 8, respectively, for the
experiments targeting the temperature to be 200 - 350 ˚C.
37
Figure 3.14: Measured temperatures (with two thermocouple probes placed at the top and bottom of PEH)
vs. time due to applied heat by a soldering hot gun that is turned on and off with (1) three different hot air
flow levels 4, 6, and 8, (2) the distance between the hot gun and PEH at 1 and 2 cm, and (3) the max
target temperature of 250 and 427 ˚C. The mid-point temperature of the PEH estimated by taking the
average value of the two probes is taken as the ground truth temperature of the PEH. Experiments for 427
˚C target temperature are stopped sooner to prevent damaging the setup due to too high temperature.
The following experiment evaluated the short-circuit current generation of miniature
PEHs. The low-frequency current generation due to heat is obtained through digital signal
processing (DSP). A moving median window algorithm extracts the low-frequency signal with
higher accuracy than a regular low-pass filter, as shown in Fig. 3.15. The processed signal indicates
that a PEH with 8 mm2
in the lateral area produces a peak short-circuit current of ~0.45 nA during
thermal cycles targeting a peak temperature of 250 ˚C.
38
Figure 3.15: (Top) The generated current and (Bottom) charges accumulated on an 8 mm2 PEH with
respect to the PEH’s measured temperature at the PEH’s top and bottom surfaces with a hot gun at a 1cm
gap with a 250 ˚C maximum temperature. The equivalent open circuit voltage by the accumulated charge
peaks at ~1,100 volts based on the capacitance of the PEH.
As shown in Fig. 3.15, the equivalent charge that PEH can produce is obtained by
integrating the current over time. The open-circuit voltage at PEH is estimated by dividing the
charge by the PEH’s measured capacitance. Table 3.3 summarizes the charges and voltages
(obtained from the measures of short-circuit currents) of three different PEHs under various
thermal conditions. Also, the open circuit voltage is simulated through a finite element analysis
39
(shown in Fig. 3.16) by setting boundary condition temperatures to the measured values shown in
Fig. 3.14. The simulation results for a 4 mm2 PEH almost match the experimental data.
Table 3.3: Summary of the equivalent charges and voltages generated by PEH, derived from the measured
short-circuit current of the PEH.
Figure 3.16: Steady-state finite element analysis (FEA) simulations of (left) temperature and (right) opencircuit voltage of a 4 mm2 LiNbO3 crystal (500 µm thick) under applied temperature in the experimental
setup with a hot gun at a 1 cm gap. The peak open-circuit voltage surpasses 900 volts when the top
surface temperature is 250 ˚C.
Finally, the charge delivery of PEHs to a pseudo-HBAR load is investigated. As can be
seen in Fig. 3.17, the measured peak voltages are 6 and 14 V for 4 mm2 PEH and 8 mm2 PEH,
respectively. The charges delivered into the 10 MΩ resistor during the period when the voltage is
40
larger than 3 V are at least 2 nC, sufficient to break the HBAR according to previous feasibility
assessments through the bulky designs and HBAR’s breakdown characteristics, which will be
discussed in Chapter 5.
Figure 3.17: Measured voltage vs. time by a PEH placed 1 cm under a hot gun with two different target
temperatures (top) and by two different PEHs with varying sizes under the same target temperature
(bottom) when a switch connects the PEH to a 10 MΩ load. The maximum amplitude and total delivered
charges are related to the PEH area and the hot gun's temperature. When the voltage exceeds 3 volts,
several nCs of charges are delivered to the load, sufficient for HBAR’s dielectric breakdown.
3.5. Summary
This chapter offers an insight into the development of Pyroelectric Energy Harvesters (PEHs)
using lithium niobate (LiNbO₃), with a focus on their application to the tamper detector chips.
Theoretical basics and modeling of pyroelectricity, providing content for understanding how
41
pyroelectric effects can be utilized in energy harvesting, are described. Then implementations of
PEHs are presented, starting with the bulky PEHs followed by the miniature PEHs that are
particularly suited for tamper detector chips.
The presented experimental setups and results show the feasibility of implementing PEHs
for the PREAT chip as the PEHs generated voltage and charge in response to temperature change
due to de-soldering and the pyroelectric material (1) was able to endure de-soldering temperature
(about 300 ˚C), (2) had a high resistivity to prevent charge leakage during the de-soldering process
(taking many seconds), and (3) had a high pyroelectric coefficient.
After explaining BAW resonators in Chapter 4, the four main key components are
combined to form a PREAT chip, and the system’s performance is analyzed in Chapter 5.
42
Chapter 4: Bulk Acoustic Wave (BAW) Resonators
Bulk Acoustic Wave (BAW) Resonators have been used for filtering and oscillation for RF
circuits. By leveraging the electromechanical properties of piezoelectric thin films, film bulk
acoustic-wave resonators (FBAR) offer several advantages over traditional electronic
counterparts, such as low insertion loss due to high quality factor (Q). This chapter explores the
background, design, fabrication, and characterization of two standard BAW resonators, i.e., FBAR
and High-Overtone Bulk Acoustic Resonator (HBAR), as an HBAR is ideally suited for a passive
RFID tag in the GHz frequency range due to its extremely high Q at frequencies as high as 10
GHz. For tamper detection, HBAR is preferred over FBAR, as this chapter shows.
4.1. Background
Bulk Acoustic Wave (BAW) resonators are based on acoustic waves propagating through solids
and being reflected at two interfaces where the acoustic-impedance mismatch is large so that the
waves may resonate between the two interfaces [43]. Both FBAR and HBAR use a piezoelectric
thin film to convert electrical signals into acoustic waves and vice versa [44]. As BAW resonators
deal with acoustic waves within the bulk, they experience less wave attenuation compared to
Surface Acoustic Wave (SAW) resonators, resulting in a higher Q [45]. BAW resonators are used
to filter out unwanted frequencies in mobile phones, base stations, and other wireless
43
communication devices. Additionally, due to their high Q factor and stability over a broad range
of temperatures [46], [47], BAW resonators are also used in precision timing devices, such as
oscillators [48], in industrial and military applications.
In BAW resonators, stress and strain coefficients along each cartesian axis are coupled to
electrical variables. For ZnO (with a hexagonal symmetry), the mechanical and electrical
parameters are coupled through the following matrix equation [47]:
31
1
31 2
x
33 3
y
15 4
z
5
15
6
0 0
0 0
0 0
0 0
0 0
T
0 0 0
e T
e T
E
e T
E
e T
E
e T
=
(4.1)
where Ti, eij, and Ei are stress, piezoelectric coefficients, and electrical field, respectively. As can
be seen, longitudinal acoustic waves propagating toward the z-axis can be generated because of
the z-directed external electric field. For BAW resonators, we are specifically interested in T3 =
e33 Ez, i.e., longitudinal waves propagating into the bulk, as shown in Fig. 4.1.
44
Figure 4.1: Schematic of longitudinal wave generation and propagation in an acoustic resonator by an
electric field in the thickness direction [46].
4.2. Film Bulk Acoustic Resonator (FBAR)
4.2.1. Design and Modeling
As shown in Fig 4.1, the piezoelectric thickness must be around a few µm for GHz resonances in
an FBAR, first introduced in the ’80s and commercialized by Avago Inc. in the early 2000s [46].
With a few microns thick piezoelectric film such as ZnO and AlN, an FBAR with a high Q-factor
(> 1,000) can easily be made to resonate at GHz.
However, for breaking the dielectric film inside the resonator using the PEH that has limited
current generation capability, the thickness of ZnO needs to be as thin as possible. A 0.3 µm thick
ZnO film is known to have a good piezoelectric property for resonance at 9 GHz.
45
A Butterworth-Van Dyke (BVD) model can be used to analyze and optimize the design of
FBAR (Fig 4.2). The FBAR features series and parallel resonant frequencies determined by the
following equations:
Figure 4.2: Equivalent BVD model for FBAR.
=
1
2√
(4.2)
=
1
2√
1
+
1
0
(4.3)
where , , and are motional inductance, capacitance, and resistance, respectively, while
0 and Rs are the clamp capacitance and series resistance, respectively. The electromechanical
coupling coefficient of the resonator is determined by
2 =
2
( −
)
4
(4.4)
46
The equivalent BVD model parameters can then be extracted through the following equations
[49]:
=
1
42
2
(4.5)
=
16
2
2
0
(4.6)
=
(4.7)
0 =
(4.8)
=
2
2
(4.9)
where is the active area; is the thickness; is the dielectric constant of ZnO; is viscosity;
is the velocity of the acoustic wave;
is the quality factor at series resonance; and is the
mass density.
While in filter applications, the electromechanical coupling coefficient is one of the key
parameters for FBAR’s performance, in the tamper detection, the Q factor should be maximized.
In FBAR, the Q-factor is directly related to the acoustic quality of ZnO film. As we intend to keep
the thickness of ZnO to a minimum reliable value (i.e., 0.3 µm), we need to evaluate whether such
FBAR shows decent RF performance.
The Q factor for resonators can also be obtained from the gradient of the device impedance’s
phase (Φ) [50]:
47
=
2
dΦ
(4.10)
In order to have a good impedance matching with a clear S11 dip, the active area of FBAR
needs to be finetuned. The area of the electrode, as shown in the above equations, plays a primary
role in determining equivalent BVD model component values. The effect of this parameter was
simulated in MATLAB (Fig. 4.3). For such a thin thickness of ZnO, the resonance circles in the
Smith chart remain in the capacitive region, making the S11 dips shallower. Nevertheless, the
simulation shows that an active area of around 1,000 µm2 offers a better matching.
Figure 4.3: MATLAB simulation results showing S11 values of FBAR with 0.3 µm thick ZnO film in log
scale and Smith chart.
4.2.2. Fabrication
An FBAR with 0.3 µm thick ZnO film is fabricated on a silicon wafer coated with LPCVD Silicon
Nitride (SixNy) (Fig. 4.4). The silicon nitride serves as an etch mask for KOH wet etching on the
bottom of the wafer and as a supporting diaphragm for FBAR. After creating the diaphragm, a
bottom electrode is deposited with 0.2 µm thick evaporated aluminum and patterned. A
48
piezoelectric ZnO layer is then deposited with RF sputtering such that the thickness is half of the
wavelength of the target resonant frequency for longitudinal waves traveling in bulk ZnO. Finally,
an aluminum top electrode is deposited and is patterned along with ZnO. The active portion of the
resonator, which sits on the supporting diaphragm, has a pentagonal shape in the lateral plane to
minimize the lateral resonances and spurious modes. A top-view photograph of a completed FBAR
is shown in Figure 4.5.
Figure 4.4: Cross-section diagram of the FBAR device, which is composed of silicon nitride (SixNy), zinc
oxide, and aluminum layers on a silicon wafer.
Figure 4.5: Photograph of the FBAR under the microscope.
4.2.3. Results
The FBAR is characterized using a 37347 Anritsu vector network analyzer (VNA) with a Cascade
microprobe of Ground/Signal/Ground calibrated to cancel parasitic loads through the probe's tip.
The experimental results on FBARs with active area sizes ranging from 250 to 15,200 µm2
and
49
with 0.3 µm ZnO show poor performance at the fundamental resonant frequency (around 7 GHz)
with a peak Q less than 20. It is not clear why the fQ product for FBARs at frequencies larger than
5 GHz turns out to be poor. Figures 4.6 – 4.10 depict the sub-par performance of the fabricated
FBARs obtained through the S11 measurements using the VNA.
Figure 4.6: S11 vs. frequency of fabricated FBAR at the fundamental resonance frequency.
Figure 4.7: Phase vs. frequency of fabricated FBAR at the fundamental resonance frequency.
50
Figure 4.8: Reactance vs. frequency of fabricated FBAR at the fundamental resonance frequency.
Figure 4.9: Resistance vs. frequency of fabricated FBAR at the fundamental resonance frequency.
51
Figure 4.10: Quality factor (Q) vs. frequency of fabricated FBAR at the fundamental resonance
frequency.
4.3. High-Overtone Bulk Acoustic Resonator (HBAR)
4.3.1. Design and Fabrication
An HBAR consists of a piezoelectric thin film (with top and bottom electrodes) on a double-sidepolished substrate with low acoustic loss (such as sapphire, silicon, or fused silica) [51]. The
acoustic waves (generated by an AC voltage applied between the two electrodes sandwiching a
piezoelectric ZnO film) propagate into the substrate and resonate between the two air-solid
interfaces at harmonics of the fundamental thick-mode resonant frequency determined by the
substrate’s thickness, as illustrated in Fig. 4.11a. The substrate is hundreds of times thicker than
the piezoelectric film, so almost all the acoustic energy is stored there. Hence, the quality factor is
mainly determined by the substrate's acoustic quality.
52
Figure 4.11: (a) Cross-sectional diagram of HBAR composed of the sapphire substrate, zinc oxide film,
and aluminum layers. (b) Photo of a fabricated HBAR on a sapphire substrate.
The fundamental thickness-mode resonance of HBAR occurs when the wavelength of the
standing wave is half of the substrate’s thickness, and the Nth harmonic frequency (fN) is at an Nfold of the fundament resonant frequency. With negligible thicknesses of piezoelectric and
electrode layers compared to the substrate’s thickness, the acoustic velocity vs in the substrate is:
= 2
(+1 − ) (4.11)
where and +1 are the substrate thickness and (N+1)th harmonic resonant frequency,
respectively [51].
Dielectric breakdown of a thin film in HBAR is typically followed by thermal runaway,
provided that sufficient current or charge is flowing, resulting in an identifiable change in HBAR’s
spectral shape. In order to make the breakdown feasible, the thickness of HBAR’s ZnO film needs
to be as low as possible, but decreasing the thickness increases the risk of pinholes in the film.
Moreover, reducing ZnO thickness to less than 300 nm may degrade the piezoelectric properties.
Note that the relatively low resistance of HBAR compared to PEH will not cause charge leakage
during the charge accumulation process since the acceleration switch is open, and the resistance
connected to the PEH is almost infinite.
53
HBARs are fabricated on a 330 µm thick sapphire wafer by depositing and patterning a 150
nm thick sputtered Al film as the bottom electrode [52], followed by deposition of 350 nm thick
ZnO film and then deposition and patterning of a top Al electrode with the same thickness as the
bottom one. The size of the active area (Fig. 4.11b) varied from 3,800 to 15,200 µm2
. Finally, the
access to the bottom Al electrodes is opened by patterning the ZnO film.
4.3.2. Results
Using the 37347a Anritsu network analyzer, we characterize HBAR with a Cascade
microprobe of Ground/Signal/Ground (Fig. 4.12). The measured S11 shows the strongest
resonances at around 7.5 GHz (Fig. 4.14) because the piezoelectric film (350 nm thick) is most
effective in generating acoustic waves at 7.5 GHz (corresponding to the acoustic wavelength of
2350 nm). The measured S11 offers HBAR’s impedance (Z) vs. frequency (f), and from the
gradient of the impedance’s phase (Φ), we obtain quality factor (Q) through Eq. 4.10.
Figure 4.12: Photo of the measurement setup for HBAR using a Cascade microprobe.
54
Figure 4.13: Measured S11 parameter of HBAR over a wide 3 GHz range with 1.87 MHz resolution.
The highest Q of the HBAR shows more than 2,500 at its resonant frequencies (Fig. 4.17) with
product being 1.8751013 Hz. The resistance and reactance parts of the HBAR’s impedance
(ZHBAR) vary as a function of frequency, as shown in Fig. 4.15 and 4.16. If the HBAR is not broken,
the backscattered signal pattern (from the HBAR and antenna) contains a sharp drop at the resonant
frequency of 7.549 GHz (being exact to the third sub-decimal point due to the high Q).
Figure 4.14: Measured S11 parameter of HBAR over a very narrow frequency range with 19 kHz
resolution.
55
Figure 4.15: Real part of the measured HBAR’s impedance.
Figure 4.16: Imaginary part of the measured HBAR’s impedance.
56
Figure 4.17: Quality factor of HBAR at ~ 7.55 GHz.
4.4. Summary
This chapter introduces Bulk Acoustic Wave (BAW) resonators, beginning with a background and
the fundamental principles. Two types of BAW resonators, the Film Bulk Acoustic Resonator
(FBAR) and the High Overtone Bulk Acoustic Resonator (HBAR), are described. A focal point of
the chapter is the exploration of reducing ZnO film thickness to reduce the ZnO breakdown voltage
for the sake of the tamper detector. The design and fabrication processes of FBARs and HBAR,
along with experimental results, are described.
This chapter shows that, for the specific application of tamper detection, HBARs offer a
distinct advantage over FBARs due to HBARs' reliance on the acoustic quality of the nonpiezoelectric high-Q substrate, e.g., Sapphire.
57
Chapter 5: Tamper Detection with PREAT Detector
In this chapter, the integration of LiNbO3 PEH (accumulating charge in response to de-soldering),
HBAR (working as a single RFID tag), mechanical switch (delivering PEH charges to HBAR),
and RFID antenna is discussed for both an initial bulky proof-of-concept tamper detector and a
miniature Passive RFID Embedded Anti-Tamper (PREAT) chip prototype.
Dielectric breakdown experiments are conducted to elucidate the breakdown
characteristics and requirements of the ZnO film within the HBAR. Additionally, for the PREAT
chip prototype, a MEMS switch is explored to transmit PEH charges to the RFID tag when
subjected to substantial mechanical shocks. The exploration extends to the implementation and
characterization of RFID antennas. This includes utilizing a microstrip patch antenna (MSPA) for
the proof-of-concept bulky design and a miniature chip antenna for the PREAT chip. Also, the
results show successful wireless authentication of the HBAR through the discernible
differentiation in the backscattered signals from intact HBARs and broken HBARs.
5.1. Overview of Tamper Detector System
The initial prototype is a bulky tamper detection system [53] (Fig. 5.1) that can (1) detect
and record (without battery) a semiconductor-chip tamper activity (i.e., de-soldering followed by
mechanical banging), which a counterfeiter does to scavenge semiconductor chips from a printed
circuit board and (2) be wirelessly interrogated without need to open semiconductor packages. The
58
sensor is based on an HBAR working as an RFID tag, which can be permanently broken down by
the voltage and charge generated by a PEH.
Figure 5.1: Proof-of-concept bulky wireless tamper detection system with a wireless interrogator
transmitting a signal and measuring the backscattered signal from the tamper detector. The strength of the
backscattered signal depends on the HBAR RFID tag that is connected (1) directly to a microstrip patch
antenna and (2) to a pyroelectric energy converter (PEC) via a cantilever-based acceleration switch. The
PEC generates a large voltage (and charge) in response to temperature change due to a de-soldering
process, while the cantilever switch is turned on by mechanical banging (which follows the de-soldering)
to let a large voltage and charge flow from the PEC to HBAR to break down the HBAR permanently and
thus to alter the characteristic of the backscattered signal.
The experimental results presented in this chapter indicate that the tamper detector will
offer the benefits of markedly (1) reduced cost and time to authenticate semiconductor chips, (2)
improved robustness and accuracy of such authentication, and (3) increased costs on the
counterfeiter’s end to scavenge semiconductor chips from PCB.
After the initial prototype, a Passive RFID-Based Embedded Anti-Tamper (PREAT) chip
is designed to have 3.72.62.5 mm³ dimensions, significantly reducing the volume of the previous
un-integrated detector, to be fit into IC packages.
59
Figure 5.2: Schematic of the front and back sides of a PREAT chip along with a top view of a highovertone bulk acoustic resonator (HBAR) used in the PREAT.
Figure 5.2 illustrates the detector chip composed of a LiNbO3 PEH, an HBAR RFID tag, a
40 – 50 g MEMS shock switch, and an ultra-wideband (UWB) chip antenna. These elements are
mounted on an adhesive-less double-sided copper-clad laminate flexible PCB (ThinFlex-W22, W1005ED-N4, ThinFlex Corporation) made of polyimide sandwiched between two copper layers.
60
Figure 5.3: Schematic showing the components on the top and bottom sides of PREAT PCB made of an
adhesive-less polyimide-based substrate. The shock switch (HT-Micro AT-50-T) and the antenna
(Kyocera A1001312) share the same side of the PCB, while the HBAR and the PEH are on opposite
sides. The top and bottom pads are connected through vias.
The PCB (Fig. 5.3) is built on a 3.72.6 mm2
copper/polyimide/copper substrate without
any adhesive, and the PREAT components are soldered on the PCB with solder paste in a reflow
process. A very thin conductive epoxy is applied to the Al solder pads to mount the components
on the pads on the PCB. For demo purposes, the plastic package of sample IC with through-hole
pins is engraved with a laser machine to carve a hole in the cap and embed PREAT inside the
package (Fig. 5.4).
61
Figure 5.4: Photo of a PREAT chip inside a laser-engraved IC package.
The experimental results presented in this chapter indicate that PREAT offers a reliable and lowcost authentication of semiconductor chips.
5.2. HBAR’s Breakdown Characteristics
5.2.1. Breakdown Characterization Setup
The breakdown behavior of HBAR in response to a gradual increase in the applied voltage is
characterized in a setup illustrated in Fig. 5.5 with a programmable power supply (Rigol DP832A)
controlled by Python scripts for a voltage sweep along with a programmable multimeter (Rigol
DM 3058) to measure the current through HBAR due to the voltage.
62
Figure 5.5: Experimental setup for characterizing HBAR’s electrical breakdown with a programmable
power supply; the voltage across the top and bottom electrodes of HBAR is increased from 0 to 4.2 volts
while the current through the HBAR is measured with a digital multimeter. The power supply and
multimeter are controlled via the PyVisa library.
Virtual Instrument Software Architecture (VISA) Application Programming Interface
(API) is used for communication between a computer and the electrical units, particularly Pythonbased PyVISA.
5.2.2. Breakdown Characterization Results
The voltage across HBAR is swept gradually from 0 to 4.2 V, back and forth multiple times
with 0.05 V step resolution. During the first sweep from zero to 1 V, the measured I-V curves (Fig.
5.6) indicate that pinholes in the 0.35 m thick ZnO film (in HBAR) become non-negligible as the
HBAR area grows.
63
Figure 5.6: Measured I-V characteristics for seven HBARs of various areas under DC voltage applied
between the top and bottom electrodes of HBAR. The measurements show no sign of breakdown for
HBARs up to 1 VDC.
As the voltage increases (Figs. 5.7 and 5.8), the HBARs experience the initial phase of
permanent dielectric breakdown at 2.25 – 3.25 V, which decreases the DC resistance from tens of
MΩs to several kΩs. When the voltage exceeds 3.5 V, the second permanent breakdown occurs
where the current surges to tens of mA, and the HBARs’ resistances drop further to hundreds of
Ωs.
64
Figure 5.7: Measured I-V characteristics for three HBARs with areas of (Top) 15,200 µm2
, (Middle)
8,550 µm2
, (Bottom) 3,800 µm2
, a DC voltage over 0 - 4.2 V is applied between the top and bottom
electrodes three times. During the first sweeps, the initial and second permanent breakdowns occur
around 2 – 3 V, making the HBAR’s DC resistance from several MΩs down to about 1 kΩ.
The breakdowns are measured to be permanent through multiple sweeps of the applied
voltage, which shows the I-V characteristics of a resistor with several hundreds of Ωs resistance.
65
Figure 5.8: A closer look at the first sweeps on the three HBARs shows that pre-breakdown DC
resistances depend on the HBAR’s top-view areas, while the post-breakdown resistances are not related to
the area, as the dielectric breakdown happens over weak spots that are random. However, due to the
breakdown, the quality factors (Qs) of the HBARs drop by several factors.
For this low voltage breakdown, the HBAR’s appearance looks identical before and after the
breakdown. Increasing the applied voltage to more than 8 V makes the previously broken film
explode and makes the device open-circuited (Table 5.1).
Table 5.1: Measured electrical properties of HBAR at each breakdown phase.
As we want to break down the HBAR with a limited amount of current (or charge) from the PEC,
Phase Resistivity (Ω.cm) Dielectric Constant
Initial 107
- 108 ~ 25 1
Initial Breakdown 80 – 1,200 ~ 100-1,000 ~ 10-6
- 10-5
Ultimate Breakdown ∞ Unmeasurable ∞
66
reaching the initial phase of the permanent breakdown (which we have observed with 2.25-3.25 V
DC voltage) is our initial target.
When the large voltage and charges are applied, the HBAR goes through the initial dielectric
breakdown due to the voltage and charge from the PEH, which has experienced a 250 ˚C
temperature change when the mechanical switch is connected due to the mechanical shocks that
emulate the tamper activity. The measured Q’s of the HBAR before and after the breakdown are
shown in Fig. 5.9. The Q can vary from 20% to 80% through the initial breakdown phase.
Figure 5.9: Measured HBAR Q’s before and after the first breakdown phase over (a) a wide 2.5 GHz
range with 1.9 MHz resolution and (b) a narrow 10 MHz range with 19 kHz resolution. Note that the
actual peak for Q before the breakdown is more than 2,000 (Fig. 10b), which is not shown in Fig. 10a
67
over the wide range due to the network analyzer's limited number of data acquisition points. The peak Q
is less than 500 after the Phase I breakdown.
5.2.3. MEMS Switch and Shock Tests
A customized shock table is built to characterize the miniature shock switch for charge
delivery from PEH to HBAR (Fig. 5.10) with a programmable linear actuator (Aerotech
ACT115DL). The applied shock acceleration is done with an accelerometer (Analog Devices 3-
axis accelerometer ADXL 372) placed on the metal plate that houses the switch. The metal plate
is lifted up by an electric magnetic gripper attached to the linear actuator and dropped at a certain
height for a shock when the metal plate hits the base stage. An Arduino Zero connected to the
accelerometer records the 3-axis data.
Figure 5.10: Illustration of a mechanical shock table with (1) an electrically controlled magnetic gripper
that grips a metallic plate holding a miniature shock switch (HT-Micro AT-50) and a 3-axis high-g
accelerometer (Analog Devices ADXL372) and (2) a programmable linear actuator (Aerotech
ACT115DL) which lifts up the plate. At an elevated point, the gripper is turned off; the plate falls onto
the table’s base, and a shock, which makes the shock switch be on. An oscilloscope reads the signal
passed through the switch, while the shock acceleration is recorded through an Arduino Zero.
68
According to Fig. 5.11, a minimum height of 6 mm is needed to reach a shock acceleration
of 40 g. Also, the results from the switch response to various applied accelerations (Fig. 5.12)
show that the average duration of the connection period is more than 5 ms in all cases when the
peak acceleration is more than 40g, indicating a sufficient time for PEH charges to be delivered to
HBAR.
Figure 5.11: Measured Z-axis shock accelerations due to the impact of the metallic plate on the base when
the plate is released from various heights. Shocks beyond 40g are observed from a free-fall distance
greater than 6 mm.
Banging the same metal plate carrying the switch and the accelerometer to the stage by
hand produces relatively strong shocks in the directions perpendicular to the main direction (here,
along the Z-axis), indicating that a single-axis (not three-axis) shock switch may be sufficient to
capture banging of PCB by a counterfeiter (Fig. 5.13).
69
Figure 5.12: Transient waveforms of the signal passing through the shock switch due to the free falls
producing more than 40g shock. Higher acceleration makes a more extended transient switch on time. In
all cases, the average on-time is beyond 5 ms, sufficiently long for delivering all the generated charge
from the PEH into the HBAR according to the waveforms shown in Fig. 3.17.
Figure 5.13: Measured 3-axis accelerations due to hand banging of the metallic plate onto a hard object
twenty different times, showing substantial accelerations directed towards the other two axes (X-axis and
Y-axis) when the banging is mainly toward the Z-axis.
70
5.3. RFID Antennas and Wireless Experiments
5.3.1. Antenna Design and Modeling
For the proof-of-concept bulky sensor, in order to interrogate the RFID tag wirelessly, a
pair of microstrip patch antennas (MSPAs) is designed, fabricated, and characterized on 22 cm2
Kapton substrates (made of polyimide) that can withstand temperatures up to 400 ˚C [54]. One of
these antennas transmits an interrogation signal towards another antenna (connected to HBAR),
which re-transmits a backscattered signal. The transmitting antenna reads the backscattered signal
and detects if the HBAR is broken.
As a rule of thumb, the thickness of an antenna substrate is recommended to be about 0.05
of the target electromagnetic (EM) wavelength [55]. Hence, for a 7.5 GHz HBAR, the substrate
thickness (h) is chosen to be about 2 mm, which Cirlex® [56] offers. The main operating frequency
of MSPA is:
≈
2√
(5.1)
where c is the velocity of EM waves in free space; l is the length of the top patch electrode; and
is the dielectric constant of the substrate material (polyimide). According to the simulation, our
designed MSPA shows a -10 dB bandwidth of 420 MHz around 7.5 GHz (Fig. 5.15). MSPAs are
fabricated on 2 mm thick Cirlex® substrates by depositing 20 nm thick evaporated titanium and 1
µm thick sputtered copper (in that order) and then patterning the copper and titanium with copper
etchant (CE-100) and buffered oxide etchant, respectively.
71
The reflected power Pr from the passive RFID tag at the interrogator’s end can be obtained with
= − 2 + + 4 (5.2)
where
is the transmitted power; L is the free-space path loss; K is the backscattered power
coefficient [57] from the tag; and is the gain of the MSPAs. The backscattered coefficient
is
= 10 log (
4
2
| + |
2
)
(5.3)
where and are the impedance and the real part of the impedance of the MSPA,
respectively, while is the impedance of the HBAR RFID tag.
Figure 5.14: Perspective view of the micro-strip patch antenna, (b) top-view design of the antenna with
key dimensions.
72
Figure 5.14: Perspective view of the micro-strip patch antenna, (b) top-view design of the antenna with key
dimensions.
.
Figure 5.15: Simulated S11 parameter of the antenna showing the maximum absorption of -30 dB at the
target operating frequency and -10 dB bandwidth of 420 MHz.
For the miniature PREAT chip, a 6 - 8.5 GHz UWB chip antenna (Kyocera 1005194) is
selected to initially send and then receive the backscattered interrogation signal from PREAT
antenna (Kyocera A1001312 chip antenna) through a vector network analyzer (VNA).
5.3.2. Antenna Characterization
With a fabricated MSPA connected to a Sub-Miniature version A (SMA) connector, we
measure S11 with 37347a Anritsu network analyzer after calibrating the network analyzer up to the
SMA connector, using a standard open load, short load, and 50 Ω load on an SMA calibration kit.
73
Figure 5.16: Photo of E-shaped microstrip patch antenna on Cirlex® substrate next to a U.S. quarter coin.
The measured S11 shows a maximum absorption of -18 dB at ~7.5 GHz with a -10 dB
bandwidth of 280 MHz (Fig. 5.17).
Figure 5.17: (Left) Measured S11 parameter of the antenna showing the maximum absorption of -18 dB at
the target operating frequency and 280 MHz -10 dB bandwidth. (Right) Calculated S12 of the antenna into
air, based on the measured S11.
To measure the RF performance of the PREAT chip antenna (soldered on the PCB flex)
and the interrogator antenna are characterized individually (Figs. 5.18a and 5.18b) using a VNA,
showing a decent performance for both antennas (sub minus 10 dB S11 at the frequency range of
interest).
74
Figure 5.18: Measured S11’s (a) of the interrogator’s antenna (Kyocera 1005194 as a transmitting antenna)
and PREATS antenna (Kyocera A1001312 chip antenna as a receiving antenna) mounted on the PCB. (b)
of the two antennas on the Smith chart.
5.3.3. PREAT Wireless Characterization
The wireless RFID test is performed by placing Tx and Rx antennas in a way that they face each
other.
For the bulky proof-of-concept design, the bottom MSPA is connected to an HBAR RFID
tag using an SMA connector and a Cascade microprobe, while the top MSPA is connected to
37347a Anritsu network analyzer through a high-frequency RF coaxial cable (Fig. 5.19).
75
Figure 5.19: (a) Schematic of the measurement set-up showing two MSPAs facing each other with the top
MSPA connected to a Vector Network Analyzer and the bottom MSPA connected to an HBAR RFID tag
through a Cascade microprobe. (b) Actual photo of two MSPAs separated by an air gap (and facing each
other) during the wireless test.
The measured S11 shows a dip in its magnitude and a rapid phase change at the target
resonance (Fig. 5.20a and b), exhibiting the exact characteristics of an undamaged HBAR. A Q of
76
~2,326 obtained with this wireless setup is almost the same as the one obtained through a wired
setup.
Figure 5.20: (a) Measured S11 magnitude and (b) phase of HBAR obtained through the wireless test over
1 MHz range with 14 kHz resolution. (c) and (d) Imaginary and real parts of the wirelessly measured
HBAR’s impedance, respectively.
On the other hand, with a broken HBAR, we do not observe any noticeable dip in the S11
magnitude (Fig 21a). A small peak in the reactance curve (Fig. 21c) is likely due to the dielectric
breakdown.
77
Figure 5.21: (a) Measured S11 magnitude and (b) phase of a broken HBAR obtained through the wireless
test over a 30 MHz with 19 kHz resolution. (c) Imaginary and (d) real parts of the wirelessly measured
impedance of the broken HBAR, respectively.
Also, for the PREAT chip, the S11 of an HBAR is measured through wired and wireless
setups before and after its dielectric breakdown (Fig. 5.22), showing a distinct pattern from the
intact HBAR compared to a broken tag, with a very high frequency resolution.
78
Figure 5.22: Measured S11 parameter’s magnitude and phase of the HBAR across a zoomed window of
1.6 MHz through wired and wireless setups before the breakdown while having an fQ >1013 compared to
the dead signal after HBAR’s breakdown.
5.4. Summary
A bulky battery-less and wireless tamper detector is first designed, fabricated, and characterized
for a proof-of-concept demonstration, as (1) a PEH based on LiNbO3 is shown to be able to break
HBAR when the PEH experiences a temperature rise of 250 – 300 ˚C; (2) the resonant frequencies
of HBAR are shown to be wirelessly interrogable through a pair of microstrip patch antennas; and
(3) the exact sequence of temperature rise (250 – 300 ˚C) followed by mechanical banging (50 g),
not just one of the two nor in reverse order, is shown to be the necessary condition for the HBAR
to be permanently damaged.
79
In the second phase of the project, the PREAT chip, a miniature version of the proof-ofconcept bulky tamper detector, is developed. The HBAR-based RFID tag will be very difficult to
replace or replicate once the tamper activity breaks it because the HBAR’s Q is close to several
thousand at several GHz, leading to precise resonant frequencies with 3-4 significant decimal
digits. The experimental results presented in this paper indicate that PREAT offers a reliable and
low-cost authentication of semiconductor chips.
80
Chapter 6: Piezoelectric MEMS Vibrational Energy Harvesting
from Mechanical Shocks
This chapter describes the development of a sub-kHz cantilever-based piezoelectric vibrational
energy harvester (VEH) based on bimorph Lead zirconate titanate (PZT) structure and Zinc Oxide
(ZnO) thin films. Harvesting electrical energy from low-frequency vibrations and mechanical
shock events with piezoelectric materials offers a means for powering electronic devices, sensors,
and other low-power applications, extending the battery lifespan of microelectronic devices or
leading to self-powered systems, and thus, reducing dependency on external power sources and
minimizing the environmental footprint associated with energy production and consumption. The
ZnO-based VEHs are based on a stress-compensated silicon nitride cantilever with piezoelectric
ZnO film. The experiments evaluate the performance of the abovementioned VEHs, showing their
capabilities to produce significant voltage and noticeable charges when exited by mechanical
vibrations.
6.1. Background
Piezoelectric energy harvesting can be used as a potential self-power source for wireless sensor
networks, as piezoelectric materials transform mechanical vibrations into electrical energy [58].
Piezoelectricity is a characteristic of specific crystalline substances like quartz, PZT, Aluminum
Nitride (AlN), and ZnO, which generate electrical charges under applied pressure. Many
researchers have investigated the applications of piezoelectric vibrational energy harvesters [59 -
81
63]. This chapter explores specific VEHs for voltage and charge generation from mechanical
shocks.
While electromagnetic energy harvesters generally can produce larger currents, their lower
output voltage needs to be increased for applications requiring a high voltage. On the other hand,
piezoelectric energy harvesters can produce a large voltage [64] but with a relatively low current.
Two Cantilever-based VEH designs are explored based on bimorph PZT and thin film ZnO.
6.2. MEMS VEH with Sub-kHz Resonant Frequencies
6.2.1. Design and Modeling
By utilizing an off-the-shelf (OTS) bimorph PZT substrate made of a metal/PZT/Brass/PZT/metal
combination, a 1-D cantilever-based VEH is made.
Figures 6.1 and 6.2 illustrate the cross-sectional views of proposed cantilever-based VEHs based
on bimorph PZT and unimorph ZnO thin film.
82
Figure 6.1: Illustration of the PZT bimorph structure, composed of two PZT layers covered by metal
electrodes, sandwiching a brass layer.
Figure 6.2: Illustration of the ZnO-based unimorph cantilever structure. The ZnO thickness is set at 0.3
µm while the cantilever is 1.5 µm thick.
For such VEHs [50], the moment of the inertia In is
=
2 [(1/2)
3 + (1/2 + 2
)
3 − (1/2)
3
]
12
(6.1)
where W is the width of the cantilever; t1 is the thickness of the brass layer; t2 is the thickness of
each PZT layer; and n is the ratio of PZT’s Young’s modulus over that of the brass.
83
The cross-section area (A) of the composite beam is
= (1 + 2 ∗ 2) (6.2)
For a 1-D cantilever
1 = 1.875 (6.3)
where L1 is the cantilever length and with
= (
4
2
2
)
0.25
(6.4)
where fn, brass, and Ebrass are resonant frequency, brass’ mass density, and brass’ Young’s
modulus, respectively. The length of the beam is finetuned o have a higher power generation at
lower frequencies, fn is set to sub-kHz values.
The capacitance of the cantilever is
=
1
22
(6.5)
At frequencies much lower than the resonance frequencies mentioned in Eq. 6.4 (where significant
spectral power of mechanical shocks exist), the normal in-plane stress is
11() =
1
2
(6.6)
11avg =
(
1
2
+
2
2
)(1)
2
(6.7)
84
where F is the cantilever mass times the applied acceleration.
The piezoelectrically induced electrical field in the z-direction is
=
13
(6.8)
Hence, the induced voltage for the bimorph is
= 2
2 (6.9)
The generated charge (Q) can be obtained by multiplying the voltage (Eq. 6.9) and capacitance
(Eq. 6.5). Also, by adding a proof mass (Fig. 6.3), the total length of the cantilever can be reduced
for a particular resonant frequency.
Figure 6.3: Illustration of the same PZT bimorph structure with an added proof mass at the tip of the
cantilever.
Table 6.1 illuminates the piezoelectric properties of various materials [59].
85
Table 6.1: Piezoelectric materials properties [59].
6.2.2. Fabrication
For the bimorph PZT structure, the VEH is fabricated by mechanically dicing the commercial.
PZT bimorph. The diced piece is attached to an acrylic grip structure to form a cantilever.
In the case of the ZnO-based VEH, the relatively large cantilever needs stress
compensation along the thickness direction for a cantilever with minimal warping. As illustrated
in Fig. 6.4, first, a 0.5 µm thick LPCVD low-stress silicon nitride (SiN) with a 6:1 flow ratio of
dichlorosilane (SiH2C12) to ammonia (NH3) is deposited at 835°C, followed by forming a
diaphragm through silicon bulk micromachining. A second 0.5 µm thick silicon nitride layer is
deposited on both sides of the diaphragm (with the reactant gas ratio of 4:1 at 835 °C) to
compensate for the stress gradient in the first SiN, as shown in Fig. 6.5 [65].
86
Figure 6.4: (a) Brief fabrication steps and (b) top view schematic of the vibration-energy harvester
(VEH).
Then, a 0.2 m thick bottom Al electrode and a 0.3 µm thick piezoelectric ZnO layer are deposited
(and patterned), followed by the deposition of 0.1 µm thick PECVD silicon nitride to prevent
charge transfer between the top and bottom electrodes.
Figure 6.5: Stress compensation through another deposition of low-stress silicon nitride after
micromachining a diaphragm [65].
The cantilever is released after a 0.2 µm thick top Al electrode is deposited and patterned along
with PECVD SiN and ZnO. The completed VEHs are flat (Fig. 10), though they have greater than
1,000 aspect ratios between the side and the thickness. The thicknesses of the layers are designed
87
to have the neutral plane below the bottom surface of the ZnO layer. The microfabricated photo of
ZnO-based cantilever VEHs is shown in Fig 6.6.
Figure 6.6: (a) Top view photo of the fabricated VEH and (b) photo of two diced VEHs with a US quarter
coin.
6.3. Experimental Results
The bimorph PZT VEH and the unimorph ZnO-based VEH are tested on a B&K shaker with the
input signal generated by a signal generator and amplified by a Bruel & Kjaer (B&K) power
amplifier (Fig 6.8). An oscilloscope reads out the voltage signals generated from the VEH, which
is connected to a high input-impedance pre-amplifier.
88
Figure 6.7: Photograph a bimorph cantilever, clamping system, and shaker. The cantilever has a total
dimension of 11 mm 1 mm 0.6 mm.
The vibration of the shaker is characterized by a Laser Doppler Displacement Meter
(LDDM) unit, which converts the time-varying displacement of the shaker into an electrical signal
[66].
Figure 6.8: Setup for measuring output voltage from VEHs with vibration applied by a shaker [66].
89
Figure 6.9: Generated voltage from bimorph PZT cantilever while applying 2.6 g acceleration.
The PZT VEH produces an open circuit peak-to-peak voltage of 2.1 V when a 2.6 g
acceleration is applied (Fig.6.9). Adding a 0.2g mass to the tip of the cantilever (Fig. 6.10)
increases the generated voltage by a factor of three on average. For instance, the maximum
amplitude of the generated voltage while applying a 0.91g acceleration changed from 250 to 740
mV after adding a 0.2g mass (Fig. 6.11).
90
Figure 6.10: Photograph the bimorph cantilever with an added 200 mg proof mass at the tip.
Figure 6.11: Generated voltage from bimorph PZT cantilever (Left) without and (Right) with a 200 mg
proof mass while applying a 0.91 g acceleration. After adding the proof mass, the generated voltage
amplitude changed from 250 mv to 740 mv.
The unimorph ZnO-based 2 mm 2 mm VEH (Fig. 6.12) shows a resonant frequency of
880 Hz, which drops to 850 Hz as the applied acceleration increases due to the softening of the
spring constant. The quality factor at the resonant frequency is about 17, which is unexpectedly
low, possibly due to squeeze film damping.
91
Figure 6.12: Output voltage of 2 mm 2 mm VEH vs. frequency as a function of applied acceleration
showing the softening of the spring with increasing vibrational amplitude.
Table 6.2: Dimension and measured properties of two fabricated VEHs.
1 mm 1 mm VEH 2 mm 2 mm VEH
Resonant Frequency 3.77 kHz
(simulation)
880 Hz
ZnO Thickness ~ 240 nm ~ 240 nm
Resistance ~ 2 GΩ ~ 530 MΩ
Capacitance ~ 80 pF ~ 310 pF
Rigidity Survives after more than 200
shocks with 50 - 500 g
Survives after more than
200 shocks with 50 - 500 g
Fabrication Yield 100% ~ 65%
Output Voltage at fo ~ 35 mV/g ~ 50 mV/g
Output Per Device Area ~ 35 mV/(g·2
) ~ 12.5 mV/(g·2
)
92
The voltage output per area of the 2 mm 2 mm VEH is about 64% less than that of the 1
mm 1 mm VEH (Table 6.2), likely because the 2 mm 2 mm VEH has more warping (and, thus,
higher stiffness) after the fabrication due to the larger size of the cantilever.
Figure 6.13: Output voltage vs time with 60 g shock applied periodically.
Using the same programmable shock table discussed in Chapter 5, the applied shock
acceleration is calibrated with a chip accelerometer (attached to the platform holding the sample).
The ZnO-based VEH is shown to produce 3 V from 60 g shock (Fig. 6.13).
93
Figure 6.14: Output voltage vs. applied acceleration as a function of frequency.
Figure 6.15: Schematic of the circuit designed to build up charges on a capacitor, as the VEH generates
voltage and charge in response to applied acceleration.
To build up the charges to provide sufficient current (in addition to sufficient voltage), as
the VEH produces voltage and charge, a rectifier circuit based on a diode bridge is used (Fig. 6.15).
94
The measured voltage on a 100 nF capacitor, with the 11 mm2
and 22 mm2 VEHs in
parallel, shows that about 0.13 V and 13 nC are built up on the capacitor after about 25 shocks of
30g - 50g (Fig 6.16). Both the 11 mm2
and 22 mm2 VEHs survive after more than 200 shocks
with 50 - 500 g.
Figure 6.16: Measured voltage on the capacitor (b) as the VEH generates voltage and charge due to 30-
50g shocks applied to the VEH at 2 Hz (a).
95
6.4. Summary
Low resonant frequency VEHs with macro and microcantilever structures are designed, fabricated,
and characterized.
A PZT-based bimorph structure sandwiching a brass layer in the middle with dimensions
11 mm 1 mm 0.6 mm is measured to produce 250 mV (without) and 740 mV (with) a 200 mg
proof mass attached to the tip under 0.91 g acceleration.
The microcantilever of a stress-compensated 1.5 µm SiN layer is used to support a 0.3 µm
thick ZnO film for a unimorph VEH, resulting in a fundamental resonant frequency of 880 Hz with
22 mm2
lateral dimension. The VEH is shown to be capable of charging a 100 nF capacitor up to
140 mV when 30 – 50 g mechanical shocks are repeated at 1 Hz.
96
Chapter 7: Non-resonant Vibration Energy Harvesting from
Human Walking Motion
This chapter describes a wrist-wearable non-resonant vibrational energy harvester, 1.4 cc in
volume and 3.2 grams in weight, with two arrays of wound copper coils adjacent to a movable
array of magnets suspended by ferrofluid bearing, to generate power from a human’s walking
motion with a target frequency falling in the range of 2 – 4 Hz. Thousand-turn coils are wound
with a customized coil winding machine. Two sets of such coils are mounted on the top and bottom
of a movable magnet array to obtain a 20% improvement compared to the earlier version based on
an electroplated coil array [67] on the figure of merit (FOM) defined to be the power (delivered to
a matched load) divided by the device’s volume for a given acceleration of 1 g at 2 Hz.
The optimization of the design and fabrication is investigated, and the experimental results
evaluate the device’s performance, showing a potential for power generation from human walking
motion for wearable devices.
7.1. Background
Batteries for cell wearable and hand-held devices need recharging or replacement during or for
which the operation is interrupted. Wearable vibration energy harvesters (VEHs) have been
explored to convert the mechanical energy of human motion into electrical energy for powering
wearable devices. Some are based on piezoelectricity [68], while others use triboelectricity [69],
97
pyroelectricity [70], or thermoelectricity [71]. These approaches, though, inherently result in a
high source impedance, requiring the load to be of high impedance and being able to deliver only
a limited current.
Figure 7.1: A 3D-printed harvester based on a V-shaped spring [72].
On the other hand, an electromagnetic VEH based on magnets and coil presents a very low
source impedance (easily down to several ohms) and can deliver a large current [72]. The voltage
generated by electromagnetic VEH can be increased by increasing the number of turns for the coil
(at the cost of increased source impedance and bulkiness of the device). Thus, electromagnetic
VEHs can effectively recharge batteries and have been explored mostly with resonant structures
that produce enhanced relative displacement (between magnet and coil) at the specific resonance
frequencies. However, most of the vibration energy in a human’s walking motion is at 1 - 4 Hz
[73], [74], which presents a significant challenge for VEH based on a resonant structure [75], [76]
since the structure (which needs to be compliant for the sake of such a low resonance frequency)
presents a substantial static displacement by gravity.
Previous literature has investigated electromagnetic energy harvesting based on both
98
resonant (Fig. 7.2) [77] and non-resonant [67] structures by utilizing electroplated coil arrays (Fig.
7.3) [76], PCB coil arrays [78], and bulky wound coils [79].
Figure 7.2: Schematic and photo of a low resonant frequency VEH based on a ferrofluid liquid spring
with a cylindrical magnet array and a flexible coil. The volume and weight are 1.8 cc and 5 g, respectively
[77].
This work presents a non-resonant electromagnetic VEH with an array of micro-coils
(wound by a programmable winding machine customized for producing subminiature coils with
hundreds - thousands of turns) adjacent to an array of magnets that are moveable and suspended
by ferrofluid bearings.
99
Figure 7.3: Microfabricated planar coils for electromagnetic energy harvesting [67].
The ferrofluid bearing allows the magnet array to move with very little friction. It leads to
a large relative displacement between the magnet array and the coil array in response to an applied
vibration.
7.2. Design and Fabrication
Since the in-plane magnetic field’s spatial gradient is the largest at the boundary of abutting
magnets (Fig. 7.5), the diameter of each wound coil is designed to match the magnet width [80].
100
Figure 7.4: Illustrations of the non-resonant vibrational energy harvester (VEH) composed of five coils on
the top and bottom sides of an acrylic chamber containing four rectangular (left) and square (right)
magnets.
The acrylic chamber for housing the magnet array inside and holding coil arrays outside is
made out of an acrylic plate by first making line grooves on the plate with an LG-500 Jamieson
Laser machine and then folding the plate (aided by the grooves) into a 3D chamber.
Figure 7.5: Simulated magnetic field around the four magnets inside the acrylic chamber of VEH.
101
The side walls of the acrylic chamber (Fig. 7.8) are initially sealed with ultraviolet-lightsensitive resin, which solidifies when exposed to UV light for 2 minutes under 36-watt UV light.
The sealing is completed by gluing all contact points with super glue (Krazy cyanoacrylate glue)
immediately after placing a magnet array with ferrofluid (self-assembled along the boundaries
between the magnets) inside the chamber to avoid any evaporation of ferrofluid.
Figure 7.6: Photo of the VEH next to Samsung Galaxy Watch 4.
All internal surfaces of the acrylic chamber are coated with a super-hydrophobic layer to
make ferrofluidic bearing as spherical as possible. Evaporated silane (from a silane solution at
room temperature) is used to treat the surfaces for hydrophobicity while adding negligible
thickness (Fig. 7.7).
102
Figure 7.7: (Top) Photo showing the hydrophobicity of a coil plate with a ferrofluid and water droplet on
the surface with contact angles of more than 90˚.
Two designs are explored: one based on oval µ-coils (OMC) and the other based on circular
µ-coils (CMC). Acrylic cylindrical spools of 1.8 mm in height with 4.1 mm and 1.3 mm in diameter
are used for OMC, while acrylic cylindrical spools of 1.8 mm in height and 1.3 mm in diameter
are used for the CMC to wind 43 AWG (60 µm in diameter) self-bonding copper wire with a coil
winding machine.
Figure 7.8: Top-view photo of the VEH without the top plate.
The ACME’s AEX-01 programmable coil winding machine is customized for the first
103
time, specifically for fabricating VEH’s µ-coils. Ethanol is used to chemically bond the insulating
layers of the copper wires during the winding process. After the winding is completed, the spool
is carefully taken out from the center of the µ-coil, and the µ-coil is compressed gently with a
clamp while being heated to 320 ˚C with a hot gun to reduce the µ-coil height to ~1 mm from 1.8
mm (Fig. 7.10) and firm the coil structure.
Figure 7.9: Cross-sectional-view photo of the VEH with five-coil arrays at the top and bottom of the
acrylic chamber, along with four rectangular magnets inside the chamber.
A set of five µ-coils are placed over the top plate (as well as under the bottom plate) of the
acrylic chamber such that the rotating directions of the coils match (all clockwise or counterclockwise when viewed from the top face of the coil) for a total of 1,500 turns for OMC and 1,000
turns for CMC per each side (top or bottom). It is critical to ensure the coils are connected so that
the coil's voltages add up. Thus, the identical terminals (inner to inner or outer to outer) must be
connected for adjacent coils. Two sets of the five coils in series are attached to the top and bottom
of the acrylic chamber to generate double the number of turns and increase the power.
104
Figure 7.10: Photos of (a) ACME AEX01 coil winding machine customized for subminiature coils, (b)
customized tooling to hold a cylindrical spool (1.8 mm in height and 1.3 mm in diameter), (c) coil
winding over the spool, and fabricated (d) OMC and (e) CMC along with the acrylic spools (used for the
coil winding) and the magnets (to be used with the fabricated coils).
The optimized number of coil turns is calculated by first setting an array of single-turn
circular (or oval) coils as the first ring (with 3.2 mm in diameter and 60 μm in thickness) above an
array of square (or rectangular) magnets with 0.3 mm gap (equal to the height of the acrylic plate
and the ferrofluid bearing). When the number of turns for the coil is increased, the coil height
105
increases.
Consequently, the marginal increase of the induced voltage becomes less as the number of
turns is increased since the distance between the coil and the magnet increases, leading to an
increasingly smaller in-plane magnetic field (and its field gradient) that the added coil experiences,
as shown in the top of Fig. 7.11.
Figure 7.11: Calculated parametric and unit-less voltage (top) and power to a matched load (bottom) by
VEHs vs the number of turns in the coil; a larger number of turns means a larger average distance
between the coil (as the coil becomes thicker/taller with higher turns) and the magnets. The CMC’s peak
power is ~25% of the OMC peak power for an applied acceleration of 1 g at 4 Hz, as the CMC and OMC
have 200 and 300 turns for the coils, respectively.
106
Though the induced voltage increases monotonically as the number of turns increases, the
increasing rate of the induced voltage is lower than the increasing rate of the coil resistance (which
increases linearly as the length of the wire increases). Thus, the power delivered to a matched load
(i.e., the load with its resistance being the same as the coil resistance) peaks at a particular number
of coil turns, as seen at the bottom of Fig. 7.11. According to our calculation, the optimal number
of turns for a maximum delivered power to a matched load is higher than that (200 – 300 per coil),
which our current winding technique allows.
Table 7.1: Key parameters of the non-resonant VEH based on rectangular magnets (OMC design).
Table 7.2: Key parameters of the non-resonant VEH based on square magnets (CMC design).
Total Volume 1.4 cc
Total Weight 3.2 g
Magnet Size (mm3
) 6.4 ´ 3.2 ´ 3.2
Movable Range of Magnet Array 6 mm
Spool Size (mm3
) 4.1 ´ 1.3 ´ 1.8
Coil Size (mm3
) 6.4 ´ 3.2 ´ 1
Number of Coils 10
Total Coil Turns 3,000
Total Resistance (Ω) 250 Ω
Total Volume 0.75 cc
Total Weight 1.55 g
Magnet Size (mm3
) 3.2 ´ 3.2 ´ 3.2
Movable Range of Magnet Array 6 mm
Spool Size (mm3
) 1.3 ´ 1.3 ´ 1.8
Coil Size (mm3
) 3.2 ´ 3.2 ´ 0.8
Number of Coils 10
Total Coil Turns 2,000
Total Resistance 120 Ω
107
7.3. Experimental Setup and Results
The performance of the VEHs is characterized by an in-plane linear actuator (Aerotech
ACT115DL), the operating frequency and acceleration of which can be controlled with the Soloist
Motion Composer. The linear actuator is operated over 2 – 4 Hz while varying the acceleration
from 0.5 to 2g.
The VEH is connected to a matched load, and an oscilloscope with a sampling rate of 10
kHz is connected in parallel to the device in order to eliminate high-frequency noise coming from
the linear actuator (Fig. 7.12).
Figure 7.12: Experimental setup for VEH characterization.
Typical open-circuit voltages (Fig. 7.13) produced by the OMC VEH for a 2g, single-cycle
2 Hz sinusoidal acceleration (applied intermittently) show that the bottom coil plate produces a
108
slightly higher voltage than the top coil plate, likely due to the magnet coil being closer to the
bottom coil due to gravity. When the top and bottom coil plates are connected in series, the induced
voltage is the sum of the voltages produced by the top and bottom coil plates, as expected. As can
be seen in Fig. 7.13, two voltage spikes (with opposite signs) occur during each period, each by a
one-way trip of the magnet array.
Figure 7.13: Measured voltages from the bottom, top, and combined coil arrays when VEH is driven with
2g acceleration at 2 Hz.
With the OMC VEH connected to a matched load of 250 Ω, the voltage induced in the
wound coils is measured with an oscilloscope. The power delivered to the matched load vs. applied
acceleration as a function of vibration frequency over 2 - 4 Hz is shown in Fig. 7.14.
109
Figure 7.14: Measured power vs. acceleration as a function of frequency for the VEH with a 250 Ω source
resistance.
Taking the power delivered to a matched load per the VEH volume as the figure of merit
(FOM), the VEH based on wound µ-coils presented here outperforms the previous device [67]
based on electroplated planar coils by 15-20% due to the lower resistance of copper wires used in
the wound coil than that of the electroplated copper electrode which has substantial contact
resistances when the coil plates are stacked [67].
The power output of the OMC VEH is about five times larger than that of the CMC VEH
(Fig. 7.15), and thus, the OMC VEH's FOM is more than twice that of the CMC VEH. This
observation matches the simulated results shown in Fig. 7.11, which shows the peak power of the
CMC design with the number of turns limited to 200 turns per coil due to fabrication difficulties
is about 20% of that of OMC-based VEH with 300 turns per coil.
110
Figure 7.15: Power vs. frequency for the two different designs (Fig. 7.4).
7.4. Summary
This chapter presents a non-resonant electromagnetic vibration energy harvester (VEH) (1.4 cc
and 3.2 grams) for harvesting power from humans' walking motion (of which the energy is mostly
below 4Hz) without loading the person.
This VEH is an improvement from the previous non-resonant VEH (1.1cc and 2.6gram)
[67]. The improvement has been obtained by (1) using a programmable coil winding machine that
enables a very large number of turns with extremely thin wire (60 m in diameter, AWG 43) and
(2) mounting coil arrays on both top and bottom sides of the magnet array, ensuring minimum
spacing between the top coil array and the magnet array through optimized acrylic chamber height.
The VEH is measured to deliver 16 - 34 μW power to a matched load (250 Ω) from 2g acceleration
at 2 - 4 Hz.
111
Chapter 8: Conclusion and Future Direction
This thesis presents a set of devices and concepts for battery-less wireless sensing and systems
specifically designed for applications in tamper detection (enhancing the security of the
semiconductor supply chain against certain forms of integrated circuit tampering), along with
vibration energy harvesting from gait movements.
The devices showcased include LiNbO3 pyroelectric energy harvesters, film bulk acoustic
resonators, high-overtone bulk acoustic resonators, microstrip patch antennas, MEMS shock
switches, Passive RFID-based Embedded Anti-Tamper (PREAT) chips, piezoelectric cantileverbased vibration energy harvesters, and wearable non-resonant electromagnetic vibration energy
harvesters. These devices and systems are explored from conceptualization and design to
modeling, fabrication, and experimental validation.
8.1. Tamper Detector Chip
A novel wireless and battery-free approach is introduced to detect tampered ICs non-destructively
and wirelessly with tamper detection and recording done during the supply chain without requiring
a power source or battery.
During the tampering activity, a high Q high-overtone bulk acoustic resonator (working as
a single-bit RFID tag) with an fQ of 1013 is broken by a pyroelectric energy harvester, making the
tag non-interrogatable for the authenticator RFID reader at the legit vendor’s facility in the late
112
stages of the semiconductor supply chain. The LiNbO3 pyroelectric energy harvester, when heated
to 250 ˚C, delivers its voltage and charge to the RFID tag through a MEMS shock switch.
This approach, though, detects only a particular tampering technique often used for cheap
and easy detachment of IC chips from their printed circuit boards (PCB) through de-soldering and
mechanical banging of the PCBs. Counterfeiters may resort to more costly alternative recycling
techniques, such as chemical or laser detachment of ICs, that could introduce visual imperfections
to the IC package and pins, necessitating additional post-processing efforts to mimic an authentic
package. This not only reduces the profit margins for counterfeiters but also diminishes their
motivation for engaging in tampering activities.
The RFID interrogator can be custom-designed to transmit and receive signals within the
desired frequency bandwidth to enhance the wireless authentication process. Optimizing the RF
matching between the HBAR and the chip antenna can extend the detectable distance between the
interrogator's antenna and the PREATS.
The voltage and charge needed to break the HBAR may be reduced by introducing sharp
tips in the electrodes sandwiching the piezoelectric film for the HBAR. This can be achieved
through timed isotropic etching of the bottom electrode, and if successful, may allow thicker
piezoelectric film to be used for HBAR that will lead the resonant frequencies of the HBAR to be
1 – 2 GHz, where the RF interrogation can be done over a longer distance and with commercially
available RFID interrogators.
The entire detector chip needs to be reduced in size further by developing a fabrication
process that will allow the monolithic integration of a MEMS shock switch, PEC, HBAR, and
113
antenna on a single chip or two chips. For example, MEMS switch, HBAR, and antenna can be
built on a double-sided-polished (DSP) silicon substrate, while PEC is built on a LiNbO3 substrate.
Alternative substrates such as Lithium Tantalate (LiTaO3) with several factors and a larger
pyroelectric coefficient [81] can be used for PEC, which can enable a smaller required size while
enabling the monolithic integration of the whole sensor on a single substrate, as the literature
suggests the feasibility of fabricating HBAR on LiTaO3.
An incubation on the feasibility of fabricating HBAR on a LiNbO3 substrate showed a subpar electromechanical coupling coefficient at the resonances, making them less apparent in the S11
curve. However, even the HBAR performance can be improved by switching to LiTaO3 as the
HBAR’s substrate [82].
For mass-scale production of PREAT chips, required design parameters such as the size of
the PEH should be set by statistical analysis of results from tests mimicking tampering activities
by counterfeiters, and each set of design parameters must be assigned to specific packaging
geometry.
8.2. Vibration Energy Harvesting
This dissertation describes vibration energy harvesters based on piezoelectric and electromagnetic
modalities.
Bimorph (PZT-based) and unimorph (ZnO-based) piezoelectric VEHs are designed and
tested. With a 200 mg proof mass, the bimorph cantilever produces a peak voltage of 740 mV
under 0.91 g vibrations, while the unimorph cantilever is shown to charge a 100 nF capacitor up
to a steady state voltage of 140 mV when actuated by repetitive mechanical shocks within the 30
– 50 g range at 1 Hz.
114
To improve the performance of the unimorph piezoelectric VEH, having a thinner substrate
and adding a proof mass can bring the fundamental resonant frequency to lower values, which
enhances DC and low-frequency response at the cost of facing a more challenging fabrication
process with a potentially less yield. It is important to note that the stress compensation techniques
discussed in Chapter 6 require a thickness equal to the summation of three SiN layers, making it
extremely difficult to reduce the thickness to below 1 µm. Controlling and neutralizing the overall
residual stress in the diaphragm before release is the key to success. This will result in a flat
achieved cantilever after the release.
The non-resonant electromagnetic energy harvester contains two arrays of wound microcoils with hundreds of turns and a self-assembled levitating magnet array with ferrofluid as the
liquid bearing. The VEH with oval-shaped micro-coils is measured to deliver 16 - 34 μW power
to a matched load (250 Ω) from 2g acceleration at 2 - 4 Hz.
To advance the electromagnetic energy harvester, the number of turns should be
maximized to produce higher voltage values at the target frequency bandwidth. By surpassing 400
mv RMS voltage, a rectifier with a Schottky diode can enable charging a capacitor using the energy
harvester. This can be achieved by fabricating micro-coils with thinner copper wires with 40+
AWG. Also, the amount of applied ferrofluid can be finetuned to optimize the trade-off between
the smooth motion of the magnet array vs. the higher equivalent spring constant of the system and
electromagnetic field blockage by the ferrofluid.
In general, piezoelectric VEHs can produce relatively high voltages from vibrations but
typically generate limited currents. On the other hand, electromagnetic VEHs inherently result in
large currents, albeit at relatively low voltages, but tend to be bulky due to the coil and magnet. In
some applications, a hybrid approach combining piezoelectric and electromagnetic energy
115
harvesting may offer a solution by harnessing the high-voltage generation capability of
piezoelectric materials and the high-current generation capability of electromagnetic transduction.
As mentioned, the electrical outputs from piezoelectric and electromagnetic mechanisms can have
different characteristics in terms of voltage and current, and integrating these outputs into a single,
stable electrical output that can be effectively stored or used is a technical challenge. Also,
combining these mechanisms might introduce losses or inefficiencies if not designed carefully, as
the energy conversion processes are inherently different for these two approaches.
116
Bibliography
[1] J. Stradley and D. Karraker, “The Electronic Part Supply Chain and Risks of Counterfeit Parts in
Defense Applications,” IEEE Trans. Compon. Packag. Technol., vol. 29, no. 3, pp. 703–705, Sep.
2006, doi: 10.1109/TCAPT.2006.882451.
[2] G. Garcia-Vazquez, “The Electronics Industry’s Counterfeit Parts Problem,” MacroFab Blog.
Accessed: Dec. 31, 2023. [Online]. Available: https://www.macrofab.com/blog/electronics-industrycounterfeit-parts-problem/
[3] H. Livingston, “Avoiding Counterfeit Electronic Components,” IEEE Trans. Compon. Packag.
Technol., vol. 30, no. 1, pp. 187–189, Mar. 2007, doi: 10.1109/TCAPT.2007.893682.
[4] C. Gorman, “Counterfeit chips on the rise,” IEEE Spectr., vol. 49, no. 6, pp. 16–17, Jun. 2012, doi:
10.1109/MSPEC.2012.6203952.
[5] F. Koushanfar et al., “Can EDA combat the rise of electronic counterfeiting?,” in Proceedings of the
49th Annual Design Automation Conference, San Francisco California: ACM, Jun. 2012, pp. 133–138.
doi: 10.1145/2228360.2228386.
[6] M. Pecht and S. Tiku, “Bogus: electronic manufacturing and consumers confront a rising tide of
counterfeit electronics,” IEEE Spectr., vol. 43, no. 5, pp. 37–46, May 2006, doi:
10.1109/MSPEC.2006.1628506.
[7] EUIPO, “INTELLECTUAL PROPERTY CRIME THREAT ASSESSMENT 2022,” EUIPO, Mar.
2022. Accessed: Feb. 10, 2024. [Online]. Available: https://euipo.europa.eu/tunnelweb/secure/webdav/guest/document_library/observatory/documents/reports/2022_IP_Crime_Threat_
Assessment/IP_Crime_Threat_Assessment_2022_ExSum_en.pdf
[8] C. Gorman, “The Financial Risks of Counterfeit.” Accessed: Dec. 31, 2023. [Online]. Available:
https://spectrum.ieee.org/the-financial-risks-of-counterfeit-semiconductors
[9] U. Guin, D. DiMase, and M. Tehranipoor, “Counterfeit Integrated Circuits: Detection, Avoidance, and
the Challenges Ahead,” J. Electron. Test., vol. 30, no. 1, pp. 9–23, Feb. 2014, doi: 10.1007/s10836-
013-5430-8.
117
[10]K. Xiao, D. Forte, Y. Jin, R. Karri, S. Bhunia, and M. Tehranipoor, “Hardware Trojans: Lessons
Learned after One Decade of Research,” ACM Trans. Des. Autom. Electron. Syst., vol. 22, no. 1, pp.
1–23, Jan. 2017, doi: 10.1145/2906147.
[11]K. Huang, J. M. Carulli, and Y. Makris, “Counterfeit electronics: A rising threat in the semiconductor
manufacturing industry,” in 2013 IEEE International Test Conference (ITC), Anaheim, CA: IEEE,
Sep. 2013, pp. 1–4. doi: 10.1109/TEST.2013.6651880.
[12]S. Hook and SMT Corp., “The Growing Threat of Counterfeit Electronic Parts in the Critical
Infrastructure Supply Chain of the United States and Allies,” May 2023. Accessed: Jan. 01, 2024.
[Online]. Available: https://smtcorp.com/wp-content/uploads/2023/05/Growing-Threat-ofCounterfeit-Electronic-Parts-in-the-Critical-Infrastructure-Supply-Chain-Releasable-1.pdf
[13]E. Oriero and S. R. Hasan, “Survey on recent counterfeit IC detection techniques and future research
directions,” Integration, vol. 66, pp. 135–152, May 2019, doi: 10.1016/j.vlsi.2019.02.006.
[14]T. Mosavirik, F. Ganji, P. Schaumont, and S. Tajik, “ScatterVerif: Verification of Electronic Boards
Using Reflection Response of Power Distribution Network,” ACM J. Emerg. Technol. Comput. Syst.,
vol. 18, no. 4, pp. 1–24, Oct. 2022, doi: 10.1145/3513087.
[15]T. Mosavirik, P. Schaumont, and S. Tajik, “ImpedanceVerif: On-Chip Impedance Sensing for SystemLevel Tampering Detection,” IACR Trans. Cryptogr. Hardw. Embed. Syst., pp. 301–325, Nov. 2022,
doi: 10.46586/tches.v2023.i1.301-325.
[16]D.-C. Vasile and P. Svasta, “Protecting the Secrets: Advanced Technique for Active Tamper Detection
Systems,” in 2019 IEEE 25th International Symposium for Design and Technology in Electronic
Packaging (SIITME), Cluj-Napoca, Romania: IEEE, Oct. 2019, pp. 212–215. doi:
10.1109/SIITME47687.2019.8990877.
[17]D. Shahrjerdi, J. Rajendran, S. Garg, F. Koushanfar, and R. Karri, “Shielding and securing integrated
circuits with sensors,” in 2014 IEEE/ACM International Conference on Computer-Aided Design
(ICCAD), San Jose, CA, USA: IEEE, Nov. 2014, pp. 170–174. doi: 10.1109/ICCAD.2014.7001348.
[18]M. Alam, S. Chowdhury, M. M. Tehranipoor, and U. Guin, “Robust, low-cost, and accurate detection
of recycled ICs using digital signatures,” in 2018 IEEE International Symposium on Hardware
Oriented Security and Trust (HOST), Washington, DC: IEEE, Apr. 2018, pp. 209–214. doi:
10.1109/HST.2018.8383917.
[19]M. Alam, H. Shen, N. Asadizanjani, M. Tehranipoor, and D. Forte, “Impact of X-Ray Tomography on
the Reliability of Integrated Circuits,” IEEE Trans. Device Mater. Reliab., vol. 17, no. 1, pp. 59–68,
Mar. 2017, doi: 10.1109/TDMR.2017.2656839.
118
[20]SMT Corp., “Dynasolve 750 Testing.” Accessed: Feb. 15, 2024. [Online]. Available:
https://www.erai.com/customuploads/ca/sc/dynasolve-750.pdf
[21]J. English, D. Coe, R. Gaede, D. Hyde, and J. Kulick, “MEMS-Assisted Cryptography for CPI
Protection,” IEEE Secur. Priv. Mag., vol. 5, no. 4, pp. 14–21, Jul. 2007, doi: 10.1109/MSP.2007.90.
[22]J. Villasenor and M. Tehranipoor, “Chop shop electronics,” IEEE Spectr., vol. 50, no. 10, pp. 41–45,
Oct. 2013, doi: 10.1109/MSPEC.2013.6607015.
[23]H. J. Goldsmid, “Principles of thermoelectric devices,” Br. J. Appl. Phys., vol. 11, no. 6, pp. 209–217,
Jun. 1960, doi: 10.1088/0508-3443/11/6/301.
[24]S. H. Price, “The Peltier Effect and Thermoelectric Cooling.” Accessed: Feb. 17, 2024. [Online].
Available: http://ffden2.phys.uaf.edu/212_spring2007.web.dir/sedona_price/phys_212_webproj_peltier.html
[25]R. Carter, “Pyroelectrics: The Dark Side of Piezos.” Accessed: Feb. 17, 2024. [Online]. Available:
https://blog.piezo.com/pyroelectrics-the-dark-side-ofpiezos#:~:text=The%20pyroelectric%20effect%20is%20the,stretching%20that%20you%20hear%20a
bout.
[26]S. B. Lang and D. K. Das-Gupta, “Pyroelectricity,” in Handbook of Advanced Electronic and Photonic
Materials and Devices, Elsevier, 2001, pp. 1–55. doi: 10.1016/B978-012513745-4/50036-6.
[27]R. W. Whatmore, “Pyroelectric devices and materials,” Rep. Prog. Phys., vol. 49, no. 12, pp. 1335–
1386, Dec. 1986, doi: 10.1088/0034-4885/49/12/002.
[28]A. Thakre, A. Kumar, H.-C. Song, D.-Y. Jeong, and J. Ryu, “Pyroelectric Energy Conversion and Its
Applications—Flexible Energy Harvesters and Sensors,” Sensors, vol. 19, no. 9, p. 2170, May 2019,
doi: 10.3390/s19092170.
[29]G. Sebald, E. Lefeuvre, and D. Guyomar, “Pyroelectric energy conversion: Optimization principles,”
IEEE Trans. Ultrason. Ferroelectr. Freq. Control, vol. 55, no. 3, pp. 538–551, Mar. 2008, doi:
10.1109/TUFFC.2008.680.
[30]D. Guyomar, G. Sebald, E. Lefeuvre, and A. Khodayari, “Toward Heat Energy Harvesting using
Pyroelectric Material,” J. Intell. Mater. Syst. Struct., vol. 20, no. 3, pp. 265–271, Feb. 2009, doi:
10.1177/1045389X08093564.
119
[31]R. Mondal, M. A. M. Hasan, J. M. Baik, and Y. Yang, “Advanced pyroelectric materials for energy
harvesting and sensing applications,” Mater. Today, vol. 66, pp. 273–301, Jun. 2023, doi:
10.1016/j.mattod.2023.03.023.
[32]M. R. H. Sarker, J. L. Silva, M. Castañeda, B. Wilburn, Y. Lin, and N. Love, “Characterization of the
pyroelectric coefficient of a high-temperature sensor,” J. Intell. Mater. Syst. Struct., vol. 29, no. 5, pp.
938–943, Mar. 2018, doi: 10.1177/1045389X17721376.
[33]A. J. Antolak, A. X. Chen, K.-N. Leung, D. H. Morse, and T. N. Raber, “Radiatively heated high
voltage pyroelectric crystal pulser,” Nucl. Instrum. Methods Phys. Res. Sect. Accel. Spectrometers
Detect. Assoc. Equip., vol. 735, pp. 379–381, Jan. 2014, doi: 10.1016/j.nima.2013.09.040.
[34]S. Patel and N. Novak, “The pyroelectric energy harvesting and storage performance around the
ferroelectric/antiferroelectric transition in PNZST,” J. Mater. Sci., vol. 56, no. 2, pp. 1133–1146, Jan.
2021, doi: 10.1007/s10853-020-05353-4.
[35]C. R. Bowen, J. Taylor, E. LeBoulbar, D. Zabek, A. Chauhan, and R. Vaish, “Pyroelectric materials
and devices for energy harvesting applications,” Energy Env. Sci, vol. 7, no. 12, pp. 3836–3856, 2014,
doi: 10.1039/C4EE01759E.
[36]Z. Liu, H. Wu, W. Ren, and Z.-G. Ye, “Piezoelectric and ferroelectric materials: Fundamentals, recent
progress, and applications,” in Comprehensive Inorganic Chemistry III, Elsevier, 2023, pp. 135–171.
doi: 10.1016/B978-0-12-823144-9.00069-8.
[37]R. W. Whatmore and S. J. Ward, “Pyroelectric infrared detectors and materials—A critical
perspective,” J. Appl. Phys., vol. 133, no. 8, p. 080902, Feb. 2023, doi: 10.1063/5.0141044.
[38]J.-P. Cao et al., “High thermal conductivity and high electrical resistivity of poly(vinylidene
fluoride)/polystyrene blends by controlling the localization of hybrid fillers,” Compos. Sci. Technol.,
vol. 89, pp. 142–148, Dec. 2013, doi: 10.1016/j.compscitech.2013.09.024.
[39]H. Chen et al., “Thermal conductivity of polymer-based composites: Fundamentals and applications,”
Prog. Polym. Sci., vol. 59, pp. 41–85, Aug. 2016, doi: 10.1016/j.progpolymsci.2016.03.001.
[40]G. A. Smolenskii, N. N. Krainik, N. P. Khuchua, V. V. Zhdanova, and I. E. Mylnikova, “The Curie
Temperature of LiNbO 3,” Phys. Status Solidi B, vol. 13, no. 2, pp. 309–314, Jan. 1966, doi:
10.1002/pssb.19660130202.
[41]Boston Piezo Optics, “Lithium Niobate.” [Online]. Available: [14]
https://www.bostonpiezooptics.com/lithium-niobate
120
[42]Roditi, “LiNbO3 Properties.” [Online]. Available: [15]
http://www.roditi.com/SingleCrystal/LiNbO3/liNBO3-Properties.html
[43]J. F. Rosenbaum, Bulk acoustic wave theory and devices. in The Artech House acoustics library.
Boston: Artech House, 1988.
[44]Y. Liu, Y. Cai, Y. Zhang, A. Tovstopyat, S. Liu, and C. Sun, “Materials, Design, and Characteristics
of Bulk Acoustic Wave Resonator: A Review,” Micromachines, vol. 11, no. 7, p. 630, Jun. 2020, doi:
10.3390/mi11070630.
[45]M. Z. Koohi, S. Lee, and A. Mortazawi, “Fabrication of a Low insertion loss intrinsically switchable
BAW filter based on BST FBARs,” in 2017 IEEE MTT-S International Microwave Symposium (IMS),
Honololu, HI, USA: IEEE, Jun. 2017, pp. 1480–1483. doi: 10.1109/MWSYM.2017.8058900.
[46]W. Pang, “Micromachined Bulk Acoustic Resonators,” University of Southern California, 2006.
[47]H. Zhang, “Micromachined bulk acoustic resonators for radio-frequency systems and biochem mass
sensing,” University of Southern California, 2006.
[48]Hao Zhang, Jongjin Kim, Wei Pang, Hongyu Yu, and Eun Sok Kim, “5GHz low-phase-noise oscillator
based on FBAR with low TCF,” in The 13th International Conference on Solid-State Sensors,
Actuators and Microsystems, 2005. Digest of Technical Papers. TRANSDUCERS ’05., Seoul, Korea:
IEEE, 2005, pp. 1100–1101. doi: 10.1109/SENSOR.2005.1496648.
[49]Engineering Materials and Processes. [Online]. Available:
https://link.springer.com/book/10.1007/978-1-4471-4944-6
[50]E. S. Kim, Fundamentals of microelectromechanical systems (MEMS). New York: McGraw Hill, 2021.
[51]H. Zhang, W. Pang, H. Yu, and E. S. Kim, “High-tone bulk acoustic resonators on sapphire, crystal
quartz, fused silica, and silicon substrates,” J. Appl. Phys., vol. 99, no. 12, p. 124911, Jun. 2006, doi:
10.1063/1.2209029.
[52]L. Baumgartel and E. S. Kim, “Experimental optimization of electrodes for high Q, high frequency
HBAR,” in 2009 IEEE International Ultrasonics Symposium, Rome, Italy: IEEE, Sep. 2009, pp. 2107–
2110. doi: 10.1109/ULTSYM.2009.5441814.
121
[53]M. Barekatain, H. Liu, and E. S. Kim, “Wireless and Battery-Less Tamper Detection With Pyroelectric
Energy Converter and High-Overtone Bulk Acoustic Resonator,” IEEE Sens. J., vol. 22, no. 14, pp.
14639–14646, Jul. 2022, doi: 10.1109/JSEN.2022.3182940.
[54]; Du Pont Company, High Performance Materials Division, “Kapton Polyimide Films Summary of
Properties,” no. H-38492-2, 1997.
[55]“Microstrip (Patch) Antennas.” Accessed: Feb. 21, 2024. [Online]. Available: https://www.antennatheory.com/antennas/patches/antenna.php
[56]“Advanced Materials for Demanding Applications.” [Online]. Available: https://www.cirlex.com/
[57]A. Bletsas, A. G. Dimitriou, and J. N. Sahalos, “Improving Backscatter Radio Tag Efficiency,” IEEE
Trans. Microw. Theory Tech., vol. 58, no. 6, pp. 1502–1509, Jun. 2010, doi:
10.1109/TMTT.2010.2047916.
[58]H. A. Sodano, D. J. Inman, and G. Park, “A Review of Power Harvesting from Vibration Using
Piezoelectric Materials,” Shock Vib. Dig., vol. 36, no. 3, pp. 197–205, May 2004, doi:
10.1177/0583102404043275.
[59]H. Elahi, M. Eugeni, and P. Gaudenzi, “A Review on Mechanisms for Piezoelectric-Based Energy
Harvesters,” Energies, vol. 11, no. 7, p. 1850, Jul. 2018, doi: 10.3390/en11071850.
[60]H. S. Kim, J.-H. Kim, and J. Kim, “A review of piezoelectric energy harvesting based on vibration,”
Int. J. Precis. Eng. Manuf., vol. 12, no. 6, pp. 1129–1141, Dec. 2011, doi: 10.1007/s12541-011-0151-
3.
[61]C. Liu, W. Zhang, K. Yu, B. Liao, R. Zhao, and T. Liu, “Gravity-induced bistable 2DOF piezoelectric
vibration energy harvester for broadband low-frequency operation,” Arch. Civ. Mech. Eng., vol. 23,
no. 3, p. 208, Jul. 2023, doi: 10.1007/s43452-023-00739-y.
[62]F. Wang, M. Zhou, P. Wu, L. Gao, X. Chen, and X. Mu, “Self-powered transformer intelligent wireless
temperature monitoring system based on an ultra-low acceleration piezoelectric vibration energy
harvester,” Nano Energy, vol. 114, p. 108662, Sep. 2023, doi: 10.1016/j.nanoen.2023.108662.
[63]H. Zhu, T. Tang, H. Yang, J. Wang, J. Song, and G. Peng, “The State-of-the-Art Brief Review on
Piezoelectric Energy Harvesting from Flow-Induced Vibration,” Shock Vib., vol. 2021, pp. 1–19, Apr.
2021, doi: 10.1155/2021/8861821.
122
[64]A. A. Shkel, M. Barekatain, and E. S. Kim, “FBAR-BASED SENSOR FOR WIRELESS RFID
AUTHENTICATION OF INTEGRATED CIRCUITS,” in 2018 Solid-State, Actuators, and
Microsystems Workshop Technical Digest, Hilton Head, South Carolina, USA: Transducer Research
Foundation, May 2018, pp. 190–193. doi: 10.31438/trf.hh2018.53.
[65]Seung S. Lee, R. P. Ried, and R. M. White, “Piezoelectric cantilever microphone and microspeaker,”
J. Microelectromechanical Syst., vol. 5, no. 4, pp. 238–242, Dec. 1996, doi: 10.1109/84.546403.
[66]Q. Zhang and E. S. Kim, “Microfabricated Electromagnetic Energy Harvesters With Magnet and Coil
Arrays Suspended by Silicon Springs,” IEEE Sens. J., vol. 16, no. 3, pp. 634–641, Feb. 2016, doi:
10.1109/JSEN.2015.2487994.
[67]Y. Wang, Q. Zhang, L. Zhao, and E. S. Kim, “Non-Resonant Electromagnetic Broad-Band VibrationEnergy Harvester Based on Self-Assembled Ferrofluid Liquid Bearing,” J. Microelectromechanical
Syst., vol. 26, no. 4, pp. 809–819, Aug. 2017, doi: 10.1109/JMEMS.2017.2669978.
[68]N. Sezer and M. Koç, “A comprehensive review on the state-of-the-art of piezoelectric energy
harvesting,” Nano Energy, vol. 80, p. 105567, Feb. 2021, doi: 10.1016/j.nanoen.2020.105567.
[69]Y. Zou, V. Raveendran, and J. Chen, “Wearable triboelectric nanogenerators for biomechanical energy
harvesting,” Nano Energy, vol. 77, p. 105303, Nov. 2020, doi: 10.1016/j.nanoen.2020.105303.
[70]Q. Li, S. Li, D. Pisignano, L. Persano, Y. Yang, and Y. Su, “On the evaluation of output voltages for
quantifying the performance of pyroelectric energy harvesters,” Nano Energy, vol. 86, p. 106045, Aug.
2021, doi: 10.1016/j.nanoen.2021.106045.
[71]M. F. Sanad, A. E. Shalan, S. O. Abdellatif, E. S. A. Serea, M. S. Adly, and Md. A. Ahsan,
“Thermoelectric Energy Harvesters: A Review of Recent Developments in Materials and Devices for
Different Potential Applications,” Top. Curr. Chem., vol. 378, no. 6, p. 48, Dec. 2020, doi:
10.1007/s41061-020-00310-w.
[72]P. Constantinou and S. Roy, “A non-linear 3D printed electromagnetic vibration energy harvester,” J.
Phys. Conf. Ser., vol. 660, p. 012092, Dec. 2015, doi: 10.1088/1742-6596/660/1/012092.
[73]J. E. A. Bertram and A. Ruina, “Multiple Walking Speed–frequency Relations are Predicted by
Constrained Optimization,” J. Theor. Biol., vol. 209, no. 4, pp. 445–453, Apr. 2001, doi:
10.1006/jtbi.2001.2279.
123
[74]F. Danion, E. Varraine, M. Bonnard, and J. Pailhous, “Stride variability in human gait: the effect of
stride frequency and stride length,” Gait Posture, vol. 18, no. 1, pp. 69–77, Aug. 2003, doi:
10.1016/S0966-6362(03)00030-4.
[75]S. Wu, P. C. K. Luk, C. Li, X. Zhao, Z. Jiao, and Y. Shang, “An electromagnetic wearable 3-DoF
resonance human body motion energy harvester using ferrofluid as a lubricant,” Appl. Energy, vol.
197, pp. 364–374, Jul. 2017, doi: 10.1016/j.apenergy.2017.04.006.
[76]C. Li, S. Wu, P. C. K. Luk, M. Gu, and Z. Jiao, “Enhanced Bandwidth Nonlinear Resonance
Electromagnetic Human Motion Energy Harvester Using Magnetic Springs and Ferrofluid,”
IEEEASME Trans. Mechatron., vol. 24, no. 2, pp. 710–717, Apr. 2019, doi:
10.1109/TMECH.2019.2898405.
[77]Y. Wang, Q. Zhang, L. Zhao, Y. Tang, A. Shkel, and E. S. Kim, “Vibration energy harvester with low
resonant frequency based on flexible coil and liquid spring,” Appl. Phys. Lett., vol. 109, no. 20, p.
203901, Nov. 2016, doi: 10.1063/1.4967498.
[78]B. Yang, “Hybrid energy harvester based on piezoelectric and electromagnetic mechanisms,” J.
MicroNanolithography MEMS MOEMS, vol. 9, no. 2, p. 023002, Apr. 2010, doi: 10.1117/1.3373516.
[79]S. P. Beeby and T. O’Donnell, “Electromagnetic Energy Harvesting,” in Energy Harvesting
Technologies, S. Priya and D. J. Inman, Eds., Boston, MA: Springer US, 2009, pp. 129–161. doi:
10.1007/978-0-387-76464-1_5.
[80]Q. Zhang and E. S. Kim, “Vibration Energy Harvesting Based on Magnet and Coil Arrays for WattLevel Handheld Power Source,” Proc. IEEE, vol. 102, no. 11, pp. 1747–1761, Nov. 2014, doi:
10.1109/JPROC.2014.2358995.
[81]X. Xiao et al., “Performance of LiTaO3 Crystals and Thin Films and Their Application,” Crystals, vol.
13, no. 8, p. 1233, Aug. 2023, doi: 10.3390/cryst13081233.
[82]B. Thomas et al., “High-Q and low TCF HBAR based on LiTaO3 substrate,” in 2014 IEEE
International Frequency Control Symposium (FCS), Taipei, Taiwan: IEEE, May 2014, pp. 1–4. doi:
10.1109/FCS.2014.6859983.
Abstract (if available)
Abstract
The research elaborated in this thesis revolves around the design and application of multiple micro-electromechanical systems (MEMS) that employ piezoelectric and electromagnetic sensors and actuators, which are essential in developing highly efficient sensing systems for applications with limited power resources.
The study features the design, simulation, and experimental analysis of a zero-power wireless authentication system. This system utilizes a High-Overtone Bulk Acoustic Resonator (HBAR) as an RFID tag for passive detection of target tampering activity, i.e., temperature elevation for de-soldering followed by mechanical shocks for detaching integrated circuits (ICs) from printed circuit boards (PCBs). The novel system operates at a frequency of 7.56 GHz with an fQ product of more than 10^13 and includes an energy harvester that generates a 6V pulse capable of permanently changing the RFID tag's RF spectral properties.
Various energy harvesters have been developed using piezoelectric and pyroelectric properties on multiple substrates, including bulk ceramics and bimorph structures, and through thin films. These energy harvesters have been evaluated thoroughly to ascertain their effectiveness in converting extreme thermal and mechanical excitations into electrical energy.
Further, the dissertation explores a compact wearable energy harvester that utilizes a non-resonant electromagnetic energy harvesting modality. This device, composed of wound micro-coils and a magnet array suspended in ferrofluid within an acrylic chamber, has been fine-tuned to generate power from low-frequency movements, such as human walking, despite its minimal form factor.
In summary, this thesis presents a suite of innovative low-power solutions enabled by MEMS resonators and piezoelectric thin films, suitable for various applications including but not limited to secure wireless authentication of ICs and health monitoring using wearables.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Zero-power sensing and processing with piezoelectric resonators
Asset Metadata
Creator
Barekatain, Matin
(author)
Core Title
Battery-less detection and recording of tamper activity along with wireless interrogation
School
Viterbi School of Engineering
Degree
Doctor of Philosophy
Degree Program
Electrical Engineering
Degree Conferral Date
2024-05
Publication Date
03/20/2024
Defense Date
03/15/2024
Publisher
Los Angeles, California
(original),
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
battery-less sensing,counterfeit semiconductor,electromagnetic energy harvesting,ferrofluid,high-overtone bulk acoustic resonator (HBAR),lithium niobate (LiNbO3),low-power sensing,MEMS,microactuators,OAI-PMH Harvest,passive tags,piezoelectric energy harvesting,piezoelectricity,pyroelectric energy harvesting,pyroelectricity,PZT,radio-frequency identification (RFID),RF MEMS,secure semiconductor supply chain,semiconductor tamper detection,sensors,shock switch,smart sensors,wearable devices,wireless interrogation,zero-power sensing,ZnO
Format
theses
(aat)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Kim, Eun Sok (
committee chair
), Wu, Wei (
committee member
), Zhou, Qifa (
committee member
)
Creator Email
barekata@usc.edu,barekatainsut@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC113857445
Unique identifier
UC113857445
Identifier
etd-Barekatain-12703.pdf (filename)
Legacy Identifier
etd-Barekatain-12703
Document Type
Thesis
Format
theses (aat)
Rights
Barekatain, Matin
Internet Media Type
application/pdf
Type
texts
Source
20240325-usctheses-batch-1130
(batch),
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Repository Email
cisadmin@lib.usc.edu
Tags
battery-less sensing
counterfeit semiconductor
electromagnetic energy harvesting
ferrofluid
high-overtone bulk acoustic resonator (HBAR)
lithium niobate (LiNbO3)
low-power sensing
MEMS
microactuators
passive tags
piezoelectric energy harvesting
piezoelectricity
pyroelectric energy harvesting
pyroelectricity
PZT
radio-frequency identification (RFID)
RF MEMS
secure semiconductor supply chain
semiconductor tamper detection
sensors
shock switch
smart sensors
wearable devices
wireless interrogation
zero-power sensing
ZnO