MM-Wave RF-MEMS Switches in
SiGe BiCMOS Technologies
vorgelegt von
M. Sc.
Selin Tolunay Wipf
an der Fakult¨at IV – Fakult¨at Elektrotechnik und Informatik
der Technischen Universit¨at Berlin
zur Erlangung des akademischen Grades
Doktor der Ingenieurwissenschaften
- Dr.-Ing. -
genehmigte Dissertation
Promotionsausschuss:
Vorsitzender: Prof. Dr.-Ing. Friedel Gerfers
Gutachter: Prof. Dr.rer.nat. Bernd Tillack
Gutachter: Prof. Dr.-Ing. Wolfgang Heinrich
Gutachter: Prof. Dr.-Ing. Ingmar Kallfass
Tag der wissenschaftlichen Aussprache: 24. September 2019
Berlin 2020
Copyright
Selin Tolunay Wipf: MM-Wave RF-MEMS Switches in SiGe BiCMOS Technologies.
This work is licensed under the Creative Commons-NonCommercial-NoDerivatives
Attribution 4.0 International License (CC BY-NC-ND 4.0). Detailed information
about this license can be found on https://creativecommons.org/licenses/by-nc-nd/4.0/.
Figures labeled with c
YYYY IEEE, rights are reserved by the publisher of the original
paper or other parties. If interested in reprinting/republishing IEEE copyrighted ma-
terial for advertising or promotional purposes or for creating new collective works for
resale or redistribution, please go to https://www.ieee.org/publications/rights/rights-
link.html to learn how to obtain a License from RightsLink.
The similar procedure is applied for articles published on Cambridge University
Press and the European Microwave Association.
II
Abstract
In the last decade, silicon germanium (SiGe) bipolar complementary metal oxide semi-
conductor (BiCMOS) technologies opened a new cost-efficient market for the mm-wave
applications. Starting with the commercial use of automotive radars at 77 GHz, the
market now has a strong interest on radar, sensor and imaging products at mm-wave
and sub-THz frequencies. The latest developments on SiGe hetero bipolar transis-
tors (HBTs) with an fmax of beyond 700 GHz boost the research and development
effort on circuit and system areas to take share from the new market. In parallel to
the developments on SiGe HBT performances, extra functionalities to the standard
BiCMOS processes can be added by following the “More than Moore” path. With
the followed path, different modules that include photonics, microfluidics or micro-
electro- mechanical system (MEMS) can be integrated into the circuits and systems for
multi-functionality. Radio frequency micro- electro- mechanical system (RF-MEMS)
switches can offer very high RF isolation, very low insertion loss, high signal linearity,
almost zero power consumption, very large bandwidth and miniaturization. With their
superior RF performances they can be used in switching and reconfigurable matching
networks, phase arrays or active circuits.
In this thesis, mm-wave RF-MEMS switches in SiGe BiCMOS technologies are deve-
loped. Initially the process integration of the RF-MEMS switches is presented, inclu-
ding different wafer-level packaging approaches. The process integration is followed
by electromagnetic modeling and RF optimization of a thin film wafer-level encap-
sulated RF-MEMS switch for D-band (110 – 170 GHz) applications in 0.13 µm SiGe
BiCMOS technology. After the demonstration of the D-band switch, a second type of
switch in 0.13 µm SiGe BiCMOS technology is developed and demonstrated for J-band
(220 – 325 GHz) applications. The thesis continues with different RF-MEMS design ex-
amples in both 0.25 µm and 0.13 µm SiGe BiCMOS technologies. In the 0.13 µm SiGe
BiCMOS technology, an RF-MEMS based D-band single-pole double-throw (SPDT)
switch and in the 0.25 µm SiGe BiCMOS technology, wafer-level packaged K-band
(18–27 GHz) single-pole single-throw (SPST), RF-MEMS based SPDT switches and
a charge pump with the SPST switch are demonstrated. Lastly, yield analyses of the
developed examples in the 0.25 µm SiGe BiCMOS technology are given.
III
Zusammenfassung
In den vergangenen zehn Jahren sind die SiGe BiCMOS Technologien in den Markt
der mm-Wellen Anwendungen vorgedrungen. Beginnend mit dem kommerziellen Ein-
satz von Autoradarsystemen bei 77 GHz, hat der Markt nun ein großes Interesse an
Radar, Sensor und Bildgebungsprodukten bei mm-Wellen und Sub-THz Frequenzen.
Die neuesten Entwicklungen auf dem Gebiet der SiGe HBT mit einem fmax von mehr
als 700 GHz beschleunigen die Forschungs und Entwicklungsanstrengungen in den Be-
reichen Schaltkreise und Systeme, um vom neuen Markt zu profitieren. Parallel zu den
Fortschritten bei den SiGe HBT Eigenschaften werden, der “More than Moore” Stra-
tegie entsprechend, zus¨atzliche Funktionalit¨aten in den BiCMOS Prozess integriert.
Demzufolge, k¨onnen verschiedene Module wie zum Beispiel Photonik, Mikrofluidik
oder MEMS in den Schaltungen und Systemen verwendet werden, wodurch deren
Funktionalit¨at deutlich erweitert wird. RF-MEMS Schalter haben eine sehr hohe
HF-Isolation, eine sehr geringe Einf¨uged¨ampfung, eine hohe Signallinearit¨at, nahezu
keine Leistungsaufnahme, eine sehr große Bandbreite und sind sehr kompakt. Mit
ihren hervorragenden HF-Eigenschaften k¨onnen sie als HF-Schalter, in konfigurierba-
ren Anpassungsnetzwerken, in Phasenschieber Schaltungen und in aktiven Schaltun-
gen eingesetzt werden. In dieser Arbeit wurden mm-Wellen RF-MEMS Schalter in
zwei SiGe BiCMOS Technologien entwickelt. Zun¨achst wird die Prozessintegration
der RF-MEMS nebst verschiedenen Verkapselungskonzeptenen vorgestellt. Auf die
Prozessbeschreibung folgt die elektromagnetische Modellierung und HF-Optimierung
eines, mittels D¨unnfilm auf Wafer-Level, verkapselten RF-MEMS Schalters f¨ur D-Band
Anwendungen in einer 0.13 µm SiGe BiCMOS Technologie. Im folgenden Abschnitt
wird die Entwicklung eines zweiten Schalters in derselben Technologie f¨ur J-Band An-
wendungen beschreiben. Die Dissertation setzt mit RF-MEMS Designbeispielen aus
der 0.25 µm und der 0.13 µm SiGe BiCMOS Technologie fort. In der 0.13 µm SiGe Bi-
CMOS Technologie wurde ein RF-MEMS D-Band SPDT Schalter und in der 0.25 µm
Technologie ein auf Wafer-Level verkapselter K-Band SPST Schalter, ein RF-MEMS
SPDT Schalter und ein mittels integrierter Ladungspumpe gesteuerter SPST Schalter
entworfen. Abschließend wird die Ausbeute der entwickelten Demonstratorschaltun-
gen aus der 0.25 µm SiGe BiCMOS Technologie dargestellt und diskutiert.
IV
Acknowledgment
First of all, I would like to thank my ’Doktor-Vater’ Prof. Dr. Bernd Tillack for pro-
viding me the opportunity to work in IHP on my PhD thesis. I would like to express
my special gratitude to my supervisor, Dr. Mehmet Kaynak, for all his support and
technical guidance throughout all these years.
I am very thankful to my colleagues, Alexander G¨oritz and Matthias Wietstruck
who I have worked with on RF–MEMS switches. Without their support, it would
have been a much more difficult journey.
I appreciate my colleagues from Heterointegration of Devices and Technologies group
and Electrical Characterization group of IHP for their friendship and technical support.
Finally, my deepest gratitude goes to my husband Christian Wipf, my mother Zuhal
Tolunay and my father Ahmet Engin Tolunay who were always there for me with their
unconditional love and endless support. Without them I would not manage to finalize
my PhD journey successfully.
V
Contents
1 Introduction 1
1.1 Motivation and Background . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Status of the RF-MEMS switch technologies . . . . . . . . . . . . . . . 4
1.2.1 RF-MEMS switches - Principle of operation . . . . . . . . . . . 4
1.2.2 Market overview of RF-MEMS devices . . . . . . . . . . . . . . 6
1.3 Structure of the thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2 Technology 12
2.1 Introduction................................. 12
2.2 Integration of MEMS module in 0.25 µm SiGe BiCMOS Technology . 13
2.2.1 Process integration of MEMS module . . . . . . . . . . . . . . 14
2.2.2 Silicon cap packaging . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Integration of MEMS module in 0.13 µm SiGe BiCMOS Technology . 20
2.3.1 Process integration of MEMS module . . . . . . . . . . . . . . 22
2.3.2 Wafer-level encapsulation . . . . . . . . . . . . . . . . . . . . . 23
2.4 Conclusion ................................. 29
3 Modeling of RF-MEMS Switches 30
3.1 Introduction................................. 30
3.2 D-band RF-MEMS SPST Switch . . . . . . . . . . . . . . . . . . . . . 31
3.2.1 Electromagnetic modeling . . . . . . . . . . . . . . . . . . . . . 31
3.2.2 Lumped-element modeling . . . . . . . . . . . . . . . . . . . . . 53
3.2.3 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . 56
3.3 J-band RF-MEMS SPST Switch . . . . . . . . . . . . . . . . . . . . . 62
3.3.1 Electromagnetic modeling . . . . . . . . . . . . . . . . . . . . . 62
3.3.2 Lumped-element modeling . . . . . . . . . . . . . . . . . . . . . 64
3.3.3 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . 67
3.4 Conclusion ................................. 71
VI
4 RF-MEMS Design Examples 73
4.1 Introduction................................. 73
4.2 D-Band RF-MEMS SPDT Switch . . . . . . . . . . . . . . . . . . . . . 73
4.2.1 EMmodeling............................ 76
4.2.2 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . 78
4.3 K-band RF-MEMS Test Vehicles for space applications . . . . . . . . 83
4.3.1 Experimental results . . . . . . . . . . . . . . . . . . . . . . . . 84
4.3.2 Effect of silicon cap packaging . . . . . . . . . . . . . . . . . . 91
4.3.3 Yieldanalysis............................ 93
4.4 Conclusion ................................. 97
5 Conclusion and Outlook 98
5.1 Technology ................................. 98
5.2 Modeling of RF-MEMS Switches . . . . . . . . . . . . . . . . . . . . . 98
5.3 RF-MEMS Design Examples . . . . . . . . . . . . . . . . . . . . . . . 99
Bibliography 101
List of Figures 118
List of Tables 121
VII
Glossary
AlCu aluminum copper
AR aspect ratio
BEOL back-end-of-line
BiCMOS bipolar complementary metal oxide semiconductor
C-V Capacitance-Voltage
CMOS complementary metal oxide semiconductor
CMP chemical mechanical polishing
CO2carbon dioxide
CP charge pump
DC direct current
DEC dynamic evaluation circuit
DHBT double hetero bipolar transistor
EDX energy dispersive X-ray
EM electromagnetic
ESCC European Space Components Coordination
FEM finite-element-method
FEOL front-end-of-line
FET field-effect transistor
FIB focus ion beam
GSG ground signal ground
GSGSG ground signal ground signal ground
GSM global system for mobile communications
HBT hetero bipolar transistor
VIII
HDR high deposition rate
HFVPE hydrofluoric acid vapor phase etching
HV high-voltage
IC integrated circuit
IL insertion loss
InP indium phosphide
ISO isolation
ISS impedance standard substrate
LDV laser-doppler-vibrometer
LRM+ Thru-Reflect-Match plus
LRRM Thru-Reflect-Reflect-Match
LTE long-term evolution
M1 Metal1
M2 Metal2
M3 Metal3
M4 Metal4
M5 Metal5
MEMS micro- electro- mechanical system
MIM metal-insulator-metal
MIMO multiple input multiple output
PECVD plasma enhanced chemical vapor deposition
RF-MEMS Radio frequency micro- electro- mechanical system
RIC representative integrated circuit
RIE reactive ion etching
SEM scanning electron microscopy
Si silicon
Si3N4silicon nitride
SiGe silicon germanium
SiO2silicon dioxide
SL signal line
IX
SOI silicon on isolator
SOLT short open load thru
SPDT single-pole double-throw
SPST single-pole single-throw
SW#1 switch-1
SW#2 switch-2
TCV technology characterization vehicle
T-line transmission line
T/R transmit/receive
TiN titanium nitride
TM1 TopMetal1
TM2 TopMetal2
VNA vector network analyzer
WLE wafer-level encapsulated
WLP wafer-level packaging
X
List of symbols
A area of two parallel plates m2
Ccont contact air capacitance of RF-MEMS switch F
Celec1 parasitic capacitance between membrane and left side HV electrode F
Celec2 parasitic capacitance between membrane and right side HV electrode F
CencapSL1 lumped-1 capacitance between TM2 plate and M5 RF- SL F
CencapSL2 lumped-2 capacitance between TM2 plate and M5 RF- SL F
CencapSL3 lumped-3 series lumped-1 capacitance between TM2 plate and M5 RF- SL F
CencapSL4 lumped-4 series lumped-1 capacitance between TM2 plate and M5 RF- SL F
Coff off-state/up-state contact air capacitance F
Con on-state/down-state contact air capacitance F
Cparallelplate parallel-plate capacitance F
Csubst substrate capacitance between ground and RF- SL F
d distance between two parallel plates m
doff off-state distance between membrane and RF-signal line m
don on-state distance between membrane and RF-signal line m
ε0permittivity of free space F/m
εrrelative permittivity of material between two paralel plates
fmax maximum oscillation frequency Hz
fTunity current gain frequency Hz
Larm inductance of membrane arm H
Larm1 inductance of the membrane arms (left side) H
Larm2 inductance of the membrane arms (right side) H
Lencap1 lumped-1 series inductance of TM2 plate H
Lencap2 lumped-2 series inductance of TM2 plate H
Lencap3 lumped-3 series inductance of TM2 plate H
XI
Lencap4 lumped-4 series inductance of TM2 plate H
Lencap5 lumped-5 series inductance of TM2 plate H
Ltotal total inductance of membrane, arms and ground-ring of RF-MEMS switch H
Rarm1 series resistor of the membrane arms (left side) Ω
Rarm2 series resistor of the membrane arms (right side) Ω
Rencap1 lumped-1 series resistance of TM2 plate Ω
Rencap2 lumped-2 series resistance of TM2 plate Ω
Rencap3 lumped-3 series resistance of TM2 plate Ω
Rencap4 lumped-4 series resistance of TM2 plate Ω
Rencap5 lumped-5 series resistance of TM2 plate Ω
Rsubst substrate resistance between ground and RF- SL Ω
toff switch-off time of RF-MEMS switch s
ton switch-on time of RF-MEMS switch s
XII
1 Introduction
1.1 Motivation and Background
New generation communication system technologies demand for not only miniaturiza-
tion but also multifunctionality. In this point of view, MEMS technology is a promising
option in order to add functionality to the RF systems. A variety of MEMS devices
(switches [1, 2], tunable capacitors [3, 4], high-Q inductors [5, 6] and resonators [7, 8],
etc.) are under study throughout the world since they serve as fundamental building
blocks in telecommunication and radar systems for RF and mm-wave applications.
In recent years, there is a growing need and interest in RF-MEMS switches for mm-
wave application as a result of their low insertion-loss, high isolation, high linearity,
near-zero power consumption and low fabrication cost [9].
The main interest of the microelectronic industry is on the commercial applications
which operate below 40 GHz, thus; the developments of RF-MEMS switches in the li-
terature are mostly in this frequency band. With the high performance semiconductor
technologies, recently developed radar and imaging applications are moving towards
the upper frequencies of the mm-wave spectrum. Specifically, SiGe BiCMOS techno-
logy is a proven technology for the mm-wave applications such as 77 GHz automotive
radar [10], and is promising for the emerging applications such as 140 GHz radar
front-ends for active imaging systems [11] or THz spectroscopic systems at 240 GHz
and beyond [12]. For such high frequencies, complementary metal oxide semiconduc-
tor (CMOS) and BiCMOS technologies offer high speed transistors. Recently, 0.13 µm
SiGe BiCMOS processes have reached a unity current gain frequency (fT) of 505 GHz
and a maximum oscillation frequency (fmax) of 720 GHz [13]. With the high perfor-
mance HBTs, SiGe technologies have become more attractive for mm-wave frequency
applications during the last decade.
Following the “More than Moore” approach [14], monolithic or hybrid integration
techniques can provide different technology modules in the baseline technologies [15–
17]. One of the modules that can be added into the CMOS or BiCMOS technolo-
gies is the RF-MEMS switch module. The decision of the integration technique for
the RF-MEMS switch modules in CMOS/BiCMOS technologies is driven by different
1
1 Introduction
factors. Compared to the monolithic integration of MEMS devices, the hybrid inte-
gration solutions such as interposer techniques, chip-to-wafer bonding, wafer-to-wafer
bonding, or any other wafer stacking and 3D integration techniques offer more flex-
ibility in terms of combining MEMS devices with CMOS/BiCMOS chips fabricated
in different technologies. Since MEMS devices have larger sizes compared to stan-
dard CMOS/BiCMOS devices; the cost per chip including CMOS/BiCMOS circuits
and MEMS devices together are increased in case of monolithic integration. However,
against the flexibility of the hybrid integration, the monolithic integration has four
main advantages. In [18], these main advantages are defined as; “1) miniaturization,
2) single-chip integration offering low manufacturing/testing cost for large-volume pro-
duction, 3) reduction of parasitic effects due to the use of on-chip interconnects, and 4)
improvement of device sensitivity and signal-to-noise ratio due to the close proximity
of the electronics (amplifiers and read out circuits) to the MEMS device”.
In the past, RF-MEMS switches have been demonstrated with high isolation in W-
band (75 – 110 GHz) [19]. On the other side, the monolithically integrated RF-MEMS
switches have shown low insertion loss (IL) (<0.5 dB) up to 140 GHz in 2010 [20] and
recently up to 300 GHz [21]. At such high frequencies, the integration of the RF-MEMS
switches together with the other system blocks requires minimum parasitics to mini-
mize the undesired effects. Parasitic losses can be minimized with shorter connections
to the active circuits with monolithic integration of the RF-MEMS switches in SiGe
technologies. The integration of the RF-MEMS switches into a 0.25 µm SiGe BiCMOS
technology has been firstly demonstrated in [22] and showed valuable results for this
integration technique with respect to robustness and reliability [23]. The monolithic
integration of the RF-MEMS switch into 0.13 µm SiGe BiCMOS process technology
gives the possibility to use RF-MEMS components together with very high perfor-
mance HBTs [13, 24] and provides circuits with unprecedented low attenuation, to be
used as SPDT switches [25] in phased arrays [26, 27]. In [25], a transceiver integrated
circuit (IC) with novel fully integrated differential RF-MEMS SPDT switch for short-
range F-band radar systems is presented in IHP’s 0.13 µm SiGe BiCMOS technology.
Shortly, IHP’s 0.13 µm SiGe BiCMOS technology is well-fitting for the RF and mm-
wave applications with the high performance HBTs and the CMOS devices and gives
the possibility to integrate the on-chip MEMS actuation circuitry.
Although the RF-MEMS switches have shown that they can provide good RF per-
formances in all the mm-wave range, packaging is still the challenge on the way to
their commercialization [28]. Similar to the all other electronic components, a good
RF-MEMS package should not only provide the interface to the next level, but should
2
1 Introduction
also be cost and area effective. However, the packaging of the RF-MEMS devices
is more complex than the packaging of the other electronic components since there
is a need for a cavity inside the package. Moreover, the type of package has also a
significant effect on their long term reliabilities. Besides, it should be preferably fa-
bricated using the wafer-level processes to increase the throughput since thousands of
RF-MEMS devices can be fabricated on a single 8-inch wafer.
The main motivation of this thesis is to expand the operating frequency range of
high performance RF-MEMS switches for the future need of the market on the mm-
wave applications such as radar, imaging and spectroscopy.
Main objectives of this thesis are:
•Design, electromagnetic (EM) modeling and RF optimization of a thin film wafer-
level encapsulated (WLE) RF-MEMS switch for D-band (110 – 170 GHz) appli-
cations.
•Design and EM modeling of an RF-MEMS SPST switch in J-band (220 – 325 GHz)
and an RF-MEMS SPDT switch at 140 GHz targeted frequency with beyond
state-of-the-art performance figures in D-band (110 – 170 GHz).
•Yield analysis and the effect of the silicon (Si) cap packaging of K-band (18 –
27 GHz) RF-MEMS test vehicles.
3
1 Introduction
1.2 Status of the RF-MEMS switch technologies
1.2.1 RF-MEMS switches - Principle of operation
RF-MEMS switches can be divided into four groups in terms of actuation principles;
electrostatic [29–31], piezoelectric [32–34], electrothermal [35, 36] and electromagnetic
[37]. Most of the developed RF-MEMS switches are based on electrostatic actuation
and their main advantages are their simplicity in technological implementation, fast
switching and low power consumption [38]. Electrostatically actuated switches can be
divided into two basic types with a contact perspective; capacitive and ohmic swit-
ches. The top and cross sectional views of capacitive and ohmic RF-MEMS switches
are shown in Fig. 1.1.
(a) (b)
Figure 1.1: Cross-sectional and top views of (a) capacitive shunt and (b) ohmic series
types of RF-MEMS switches.
4
1 Introduction
Capacitive Switches with electrostatic actuation:
Capacitive type of RF-MEMS switches can switch the RF signal on and off with the
low and high capacitances in the contact region. When the movable membrane is in
the upper position, it provides low capacitance to the RF- signal line (SL). By applying
a high-voltage (HV) between the RF- SL and the grounded membrane, the movable
membrane moves closer to the RF- SL and provides a high capacitance. The high
contact capacitance creates a LC series resonance together with the inductance of the
movable membrane. The dielectric on top of the RF- SL provides direct current (DC)
isolation between the RF- SL and the movable membrane. Although the capacitive
type of RF-MEMS switches can be used in both shunt and series circuit configurations,
the typical circuit configuration with the capacitive type of RF-MEMS switches is the
shunt configuration.
Ohmic Switches with electrostatic actuation:
Ohmic type of RF-MEMS switches can switch the RF signal on and off with the low
and high ohmic contacts. In the up-state, the ohmic switch creates an open between
the RF- SL and the movable membrane. On the other side, in the down-state RF-
signal is shorted to the movable membrane with the low contact resistance. Although
the ohmic type of RF-MEMS switches can be used in both shunt and series circuit
configurations, the typical circuit configuration with the ohmic type of RF-MEMS
switches is the series configuration.
The performance of the ohmic RF-MEMS switches in down-state is limited by the
contact resistance and the switch inductance that increases with higher frequencies.
On the other side, the limitation for the capacitive RF-MEMS switches is the down-
state capacitance. To be able to provide a short circuit at lower frequencies very high
capacitances need to be achieved. However to achieve a high down-state capacitance
is not preferable due to the necessity of a bulky RF-MEMS switch.
Briefly, the ohmic RF-MEMS switches are more suitable for applications below
mm-wave and the capacitive RF-MEMS switches are more suitable for applications
at mm-wave and also above. In this thesis, capacitive RF-MEMS switches are chosen
due to the targeted mm-wave frequency applications.
5
1 Introduction
Figure 1.2: “Behaviour of the hype curve [45] of RF-MEMS technology based upon
the market forecasts published through the last decade and the fluctuating
expectations for mass-market outcomes [39]”.
1.2.2 Market overview of RF-MEMS devices
In the beginning of the 21th century, market volume expectations for RF-MEMS de-
vices are reported with a focus on their expected impacts in the mass production of
mobile handsets. The reported market volumes for RF-MEMS devices are reviewed
in [39] and itemized below;
•2004–2005: “The market volume was expected to reach from 700 $M (Millions
of US Dollars) to 1 $B (Billions of US Dollars) in 2009 [40]”.
•2009: “WTC in 2006 predicted a downsized market volume of around 10 $M in
2009 and 70 $M in 2011 [41]”.
•2010: “IHS Inc. consolidated the RF-MEMS market figure to a few $M in 2009,
and predicted a volume of 225 $M in 2014 [42]”.
•2012: “IHS Inc. in 2012 shrunk down the forecast for 2014 market to less than
100 $M [43]”.
•2013: “Yole Developpement estimated a market volume of around 50 $M in 2014
and of less than 350 $M in 2018 [44]”.
6
1 Introduction
By summarizing the above given market forecasts, [39] has published the graphi-
cal presentation (Fig. 1.2) for the expected mass-market outcomes of the RF-MEMS
technology with respect to the years. It has been stated in [39, 46] that RF-MEMS
experienced two peaks for the expectations followed by disappointments. The first
disappointment phase was in around 2006-2007 which was linked with intrinsic factors
as reliability, packaging and integration. Since researchers started to address these
intrinsic problems, extrinsic factors were to be faced for the success of the technology.
These were not directly linked to the technology itself but the needs and the accep-
tance of the market. The market was not strong enough to demand high performance
and reconfigurable devices.
With the new full screen smartphones, the quality of the communication degraded
as the mobile smartphones became smaller and thinner which made it difficult to in-
corporate the performance-enhancing circuitry required to counter dropped cellphone
calls and raise voice quality [47]. Fig. 1.3 shows the degradation of connection quality
with each generation of communication, from global system for mobile communicati-
ons (GSM) to long-term evolution (LTE) [48]. Since the smartphones had low antenna
efficiency due to multi band operation, the market needed tunable antennas to over-
come this problem [49] and a change in the strategy [46]. In the recent few years within
this scenario, RF-MEMS adaptive impedance tuners started to get into the consumer
market of 4G - LTE smartphones [50]. Although the demanded reconfigurability of
the tunable antennas can be realized by integrating pin diodes [51], varactors [52, 53],
or MEMS [54–56]; the RF-MEMS devices differs with better performance (in terms of
Q, isolation, and linearity) but have the disadvantage of higher cost for the technology
[57].
The growing market interest for the RF-MEMS based products is expected to be-
come much bigger especially with the 5G mobile devices [46]. Nevertheless, before the
5G mobile devices, an RF-MEMS device in a Samsung cellphone, Focus Flash, is iden-
tified by a research firm IHS in 2012. The identified device was an RF-MEMS based
adaptable impedence tuner which was manufactured by WiSpry [58] and it became
the first known RF-MEMS based volume-shipped products [59]. Fig. 1.4 shows the
cross-section of a complete WiSpry MEMS capacitor with an integrated charge pump
(CP) [60].
In the following years, in 2014, Cavendish Kinetics [48] and Chinese ZTE Corpora-
tion announced that Nubia Z7 smartphone includes a smart antenna that is powered
by Cavendish Kinetics‘ SmarTune antenna tuning solution. “Our revolutionary Smar-
Tune antenna tuning solution controls the electrical characteristics of the antenna and
7
1 Introduction
Figure 1.3: The degradation of connection quality with each generation of communi-
cation, from GSM to LTE due to greater use of cellular communications.
This problem is addressed to be solved by RF-MEMS tuning [48].
optimizes its performance by shifting its resonating frequency, so that it is always op-
timally matched to the operating frequency,” said Paul Dal Santo, CEO of Cavendish
Kinetics [61]. Fig. 1.5 shows the MEMS array inside Cavendish Kinetics‘ SmarTune
antenna tuner [62]. In 2015, Cavendish Kinetics announced that they shipped the
SmarTune solution in five different smartphone models to optimize the performance of
LTE multiple input multiple output (MIMO) antenna; and in 2017, it announced that
ten manufacturers for 40 different smartphone models (including the Samsung Galaxy
A8), have adopted their SmarTune RF MEMS antenna tuners. Cavendish Kinetics
claims that their “RF-MEMS Tuners outperform traditional silicon on isolator (SOI)
switch based antenna tuning solutions by 2-3 dB, resulting in much higher data rates
(up to double) and improved battery life (up to 40%)” and states that the smartphone
manufacturers recognize the benefit of aperture tuning over impedance matching, and
find that the SmarTune solution allows them to design high efficiency antennas for
slimmer smartphones, without sacrificing radio performance [61].
Other examples for RF-MEMS devices that are commercially available since 2016
in the market are ADGM1304 and ADGM1004 MEMS switches for 0 Hz (DC) to
14 GHz operation frequencies from Analog Devices [63]. The introduced switches are
hermetically sealed in Si cap and contain two dies including driver circuits and MEMS
8
1 Introduction
Figure 1.4: Cross-section of the complete WiSpry MEMS capacitor with an integrated
CP [60].
Figure 1.5: “MEMS array inside Cavendish Kinetics’ antenna tuner” [62].
switch. Fig. 1.6 shows the commercially avaible MEMS switch technology of Analog
Devices. Analog Devices claims that the introduced MEMS switches are “95 percent
smaller, 30 times faster, 10 times more reliable, and use 10 times less power than con-
ventional electromechanical relays” and with their superior performances can replace
electromechanical relays (in e.g. automatic test equipments) [64].
In summary, RF-MEMS technology solutions are getting more attention as they
become more mature. Larger market volumes for RF-MEMS based products are ex-
pected in the future, especially concerning the 5G mobile devices and communication
infrastructures [46]. As a result of this demand, it is expected that the hyper curve of
RF-MEMS technology will look more pleasant in the future.
9
1 Introduction
Figure 1.6: “MEMS switch technology of Analog Devices with driver circuits (Left)
and MEMS switch (Right) mounted on and wire bonded to a metal lead
frame” [64].
1.3 Structure of the thesis
This thesis begins with the process integration chapter of RF-MEMS switches in IHP’s
high performance SiGe BiCMOS technologies. These monolithically integrated RF-
MEMS switches are real single chip solutions with the advantages of miniaturization
and minimum parasitics compared to the hybrid integrated switches. Although the
process integration of the RF-MEMS modules [65] are not in the main work of this
thesis, the thesis begins with the general process flows of the RF-MEMS modules
in Chapter 2 to give a better understanding for the electromagnetically modeled and
characterized RF-MEMS switches in the following chapters. Additional to the general
process flows of the RF-MEMS modules, the two different packaging approaches for the
RF-MEMS switches are also briefly explained in Chapter 2. In short, for the 0.25 µm
SiGe BiCMOS technology a Si cap wafer-level packaging technique and for the 0.13 µm
SiGe BiCMOS technology a thin film wafer-level encapsulation technique is selected.
The developed RF-MEMS switches target applications in mm-wave with a special
focus on above 110 GHz applications. The main work of this thesis is given in Chapter 3
and presents the D-band and J-band RF-MEMS switches in 0.13 µm SiGe BiCMOS
technology. More specifically, Chapter 3 presents initially an EM model and a small-
signal (lumped-element) model of RF-MEMS switches in D-band and J-band. The
EM model of the RF MEMS switches are developed to be able to roughly estimate
the RF performances before their fabrications. After the fabrication of the switches,
the developed EM models are used to get accurate S-parameter simulation results. In
10
1 Introduction
parallel as alternative to the EM models, the small-signal models of the fabricated RF-
MEMS switches are developed for faster simulations to predict the RF performances
of the switches from a pure electrical point of view. The experimental results of the
0.13 µm SiGe BiCMOS embedded RF-MEMS switches show beyond state of the art
RF performances in D-Band and J-band with high on- off-state contact capacitance
ratios (Con / Coff).
Chapter 4 presents different RF-MEMS design examples in the both SiGe BiCMOS
technologies. Specifically, it presents a D-band RF-MEMS SPDT switch including the
EM modeling and experimental results, and K-band RF-MEMS test vehicles including
the experimental results, yield analysis and investigation for the effect of the Si cap
packaging on RF performances.
Finally, Chapter 5 gives the conclusion of the thesis and includes the future per-
spective of this work.
11
2 Technology
2.1 Introduction
Integration of MEMS devices with CMOS electronics can be done in different ap-
proaches; pre-CMOS, intra-CMOS, post-CMOS and post processing of CMOS-BEOL
layers. These four approaches are summarized in [18]: In pre-CMOS approach MEMS
devices are fabricated before the CMOS devices; in intra-CMOS approach they are
fabricated in parallel; in post-CMOS approach MEMS devices are fabricated after the
CMOS devices and in post processing of CMOS-BEOL layers approach MEMS are
embedded in CMOS BEOL layers. Each approach has its advantages and limitations,
and should be selected accordingly.
Monolithic integrations of RF-MEMS devices into BiCMOS BEOL have advantages
in terms of stable process conditions and minimum parasitics, however it also has
risk of yield degradation for the CMOS devices. The yield degradation of CMOS
devices must be avoided by keeping the stress level of the complete stack, front-end-
of-line (FEOL) and BEOL. In 2009, process integration of RF-MEMS module into
IHP’s 0.25 µm SiGe BiCMOS technology was demonstrated [22]. For the success of
the integration, it was necessary to optimize the mechanical stress of the metalization
layer (Metal3 (M3)) used for the movable membrane. The developed M3 recipe did
not significantly affect the M3 sheet resistance and the requirement of the process
specifications were still fulfilled. In this thesis, the developed RF-MEMS devices are
monolithically integrated into IHP‘s 0.25 µm and 0.13 µm BiCMOS BEOLs [66].
This chapter provides general overviews of the process integration of RF-MEMS mo-
dules in both IHP technologies; 0.25 µm (SG25) and 0.13 µm (SG13) SiGe BiCMOS
technologies. Section 2.2 provides introduction of IHP’s 0.25 µm SiGe BiCMOS techno-
logy, detailed information about the process integration of MEMS module in 0.25 µm
node and the process flow of Si cap wafer-level packaging for the RF-MEMS switches.
In the next section, Section 2.3, introduction of IHP’s 0.13 µm SiGe BiCMOS techno-
logy, detailed information about the process integration of MEMS module in 0.13 µm
node and the process flow of thin-film wafer-level encapsulation for the RF-MEMS
switches are provided.
12
2 Technology
2.2 Integration of MEMS module in 0.25 µm SiGe BiCMOS
Technology
IHP’s 0.25 µm node has three main BiCMOS technologies; SG25H3 and SG25H4
and SGB25V. SGB25V technology includes 3 different HBTs with maximum fmax of
95 GHz. SG25H3 technology offers HBTs with fmax up to 180 GHz. The highest per-
formance of 0.25 µm IHP technologies is the SG25H4 with the fmax values for HBTs up
to 220 GHz. The main performance parameters of HBTs for different 0.25 µm techno-
logies [67] are summarized in Table 2.1.
Table 2.1: Main performance parameters of HBTs for different 0.25 µm technologies of
IHP [67].
SG25V High
Performance Medium Voltage High Voltage
fmax 95 GHz 90 GHz 70 GHz
fT75 GHz 45 GHz 25 GHz
BVCEO 2.4 V 4 V 7 V
SG25H3 High
Performance Medium Voltage High Voltage
fmax 180 GHz 140 GHz 80 GHz
fT110 GHz 45 GHz 25 GHz
BVCEO 2.3 V 5 V 7 V
SG25H4 npn1 npn2
fmax 190 GHz 220 GHz
fT190 GHz 180 GHz
BVCEO 1.9 V 1.9 V
The BEOL of the 0.25 µm SiGe BiCMOS technology offers three thin and two thick
aluminum copper (AlCu) layers, and tungsten vias in between. The cross section of the
FEOL with the detailed BEOL module is shown in Fig. 2.1. The thin metal layers are
named from bottom to top as Metal1 (M1), Metal2 (M2), M3. The two thick metal
layers are TopMetal1 (TM1) and TopMetal2 (TM2). Additionally, metal-insulator-
metal (MIM) capacitor is available between M2 and M3 in order to achieve a high
capacitance density of 1 fF/µm2. The passivation on top includes thickness of 1.5 µm
silicon dioxide (SiO2) and 0.4 µm silicon nitride (Si3N4).
13
2 Technology
Figure 2.1: The cross section of the SG25H3/H4 BEOL [68].
2.2.1 Process integration of MEMS module
The integration of the RF-MEMS switches into a BiCMOS process gives the pos-
sibility to use them together with CMOS electronics and high performance HBTs
[24]. The integration of the RF-MEMS switches into IHP‘s 0.25 µm SiGe BiCMOS
technology was firstly demonstrated in [22] and showed valuable results with respect
to robustness and reliability [23]. Fig. 2.2 summarizes the additional process steps for
RF-MEMS switch integration into IHP‘s SG25 BiCMOS process. After finalizing the
SG25 BiCMOS process, the first lithography step is done to determine the area of the
RF-MEMS switch for the passivation nitride opening. Later on, it is followed by the
reactive ion etching (RIE) process to open the passivation nitride in the determined
MEMS area. After the RIE process, the second lithography step is done to determine
the same MEMS area where the RF-MEMS switch will be wet etched from. Follo-
wing the second lithography step, RF-MEMS switch is wet etched down to M1 where
14
2 Technology
the high voltage electrodes of the switch were patterned. Before finalizing the RF-
MEMS switch integration, the wafer is cleaned and rinsed. Finally, the carbon dioxide
(CO2) critical point drying step is applied in order to prevent the unwanted stiction
of the RF-MEMS switch after the suspended membrane is released by the wet etching.
Standard SG25 BiCMOS process
Lithography step to define the
area for RF-MEMS switch with
passivation nitride opening
Opening passivation nitride
in the defined area with RIE
Lithography step to define the area
for RF-MEMS switch for wet etching
Releasing RF-MEMS switch from the
defined area until HV electrodes (M1)
CO2critical point drying step to
prevent the unwanted stiction
Figure 2.2: Additional process steps for RF-MEMS switch integration into the SG25
BiCMOS process.
Fig. 2.3 shows the cross section of the embedded SG25 RF-MEMS switches. The
RF-MEMS switches are built between the M1 and the M3. The high-voltage electrodes
are defined in M1, the RF- SL in M2 and the suspended membrane in M3. A thin
Si3N4/ titanium nitride (TiN) stack, which is part of the BiCMOS MIM capacitor,
forms the contact region and provides DC isolation between the grounded membrane
and the RF- SL. The first generation SG25 BiCMOS embedded RF-MEMS switch is
shown in Fig. 2.4 [22].
15
2 Technology
c
2017 IEEE
Figure 2.5: The generic cross section of the packaged 0.25 µm SiGe BiCMOS techno-
logy RF-MEMS test vehicles [66, 70, 71].
2.2.2 Silicon cap packaging
After the fabrication of the RF-MEMS switches in SG25 BiCMOS, the RF-MEMS
cavities are sealed with the Si caps at wafer-level in Fraunhofer IZM, Berlin [69].
The generic view of the packaged 0.25 µm BiCMOS technology RF-MEMS switches is
shown in Fig. 2.5. The applied wafer-to-wafer packaging technique is independent from
the size of the Si cap; thus the packaging of several RF-MEMS switches can be realized
at the same time. The high throughput and flexible size Si caps make it possible
to provide BiCMOS embedded RF-MEMS switch technology with a semi-hermetic
packaging. The process flow of the Si cap packaging process which is developed and
used by Fraunhofer IZM, Berlin is given in Fig. 2.6 [70].
Initially, a glass carrier wafer is bonded to a Si wafer which is coated with an adhesive
material. In subsequent process steps, the Si wafer is processed to generate the cap
structures. Processes such as back grinding, lithography and Si dry etching are used
to form Si cap structures. The structuring of the caps includes the removal of the
material in between the particular cap structures by using an additional lithography
step and dry etching of the Si. In that case, the adhesive layer acts as an etch stop
layer. In the next process steps, the created cap wafer is aligned and bonded to the
corresponding 200 mm BiCMOS wafer. The required alignment marks on the cap
wafer are created during the cap processing. The caps are structured into a thin
Si wafer (50 µm) which is created by back grinding from full thickness (725 µm) after
bonding. Each Si cap has an adhesive (polyimide) bonding frame around the rim, with
17
2 Technology
c
2017 IEEE
Figure 2.6: The process flow of the wafer-level packaging process for the 0.25 µm SiGe
BiCMOS embedded RF-MEMS switch [70].
a height of 5 µm. The bonding frames are structured by lithography and dry etching
before the Si caps are structured. The adhesive layer on the carrier wafer around the
caps is also removed by dry etching. The Si caps are transferred to the target wafer
using a wafer-to-wafer thermo compression bonding process at a temperature of less
than 300 ◦C and with a pressure of 0.1 – 0.2 MPa. After the Si caps are bonded to the
BiCMOS wafer, the carrier wafer needs to be removed from the backside of the Si caps.
In case the polyimide is only present underneath the caps, a full area laser exposure of
the carrier wafer can be performed. The microscope images of the SG25 RF-MEMS
switch before and after the wafer-level Si cap packaging are shown in Fig. 2.7.
After the Si cap packaging in Fraunhofer IZM, Berlin; the wafer-level packaged
RF-MEMS switches are initially investigated with the FIB analysis in IHP. The FIB
analysis shows (Fig. 2.8) the stable polyimide underneath the Si cap even with the
TM2 surface topology of the extended RF- SL.
18
2 Technology
2.3 Integration of MEMS module in 0.13 µm SiGe BiCMOS
Technology
IHP’s 0.13 µm node has two BiCMOS technologies; SG13S and SG13G2. SGB13S
technology includes two different HBTs with maximum fmax of 340 GHz. The highest
performance of 0.13 µm IHP technologies is the SG13G2 with the fmax values for HBTs
up to 500 GHz [24]. The main performance parameters of HBTs for different 0.13 µm
technologies [67] are summarized in Table 2.2. In 2016, IHP has demonstrated that
SiGe HBTs can go upto fT/ fmax values of 505 GHz/ 720 GHz [13]. With the high
performance HBTs and passive elements, IHP’s 0.13 µm SiGe BiCMOS technology is
a well-fitting technology for the RF and mm-wave applications.
Table 2.2: Main performance parameters of HBTs for different 0.13 µm technologies of
IHP [67].
SG13S npn13p npn13V
fmax 340 GHz 165 GHz
fT250 GHz 45 GHz
BVCEO 1.7 V 3.7 V
SG13G2 npn13G2
fmax 500 GHz
fT300 GHz
BVCEO 1.7 V
The BEOL of the 0.13 µm SiGe BiCMOS offers five thin and two thick AlCu metal
layers, and tungsten vias in between. The cross section of the FEOL with the detailed
BEOL module is shown in Fig. 2.10. The thin metal layers are named from bottom
to top as M1, M2, M3, Metal4 (M4) and Metal5 (M5). The two thick metal layers
are TM1 and TM2. Additionally, MIM capacitor is available between M5 and TM1
in order to achieve a high capacitance density of 1.5 fF/µm2. The passivation on top
includes thickness of 1.5 µm SiO2and 0.4 µm Si3N4.
20
2 Technology
Figure 2.10: The cross section of the SG13 BEOL, including the marked metal layers
and vias for the integration of the RF-MEMS switch.
2.3.1 Process integration of MEMS module
BEOL metalization of 0.13 µm BiCMOS technology has seven metal layers instead
of five and the distances between the metals are also different compared to 0.25 µm
BiCMOS technology. Indeed, it is not possible to easily transfer the 0.25 µm RF-
MEMS switch to 0.13 µm BiCMOS technology and some significant effort is necessary
in order to develop a switch at targeted bands (D-/J-band). The main changes from
0.25 µm to 0.13 µm technology are the smaller dimensions of the switch to achieve
higher frequency of operation, the thicker metal layer of the membrane and the incre-
ased distance between the RF- SL to the Si substrate. With the thicker metal layer of
the membrane and the reduced size of the switch, mechanical properties like stiffness
also changes which results in an increased pull-in voltage. Most importantly, with the
increased distance between the RF- SL to the Si substrate, the substrate coupling of
the RF-signal to the Si substrate is reduced.
With all the challenges in development of RF-MEMS switch in IHP’s 0.13 µm Bi-
CMOS technology, the capacitive RF-MEMS switch is designed between M4 and TM1
of the technology. Similar to the 0.25 µm BiCMOS technology, RF-MEMS switch
in 0.13 µm SiGe BiCMOS technology consists of HV electrodes, RF- SL and mova-
ble membrane. Fig. 2.10 shows the cross section of the SG13 BEOL, including the
RF-MEMS layers that are used during the integration of the RF-MEMS switch.
22
2 Technology
2.3.2 Wafer-level encapsulation
An huge effort has been spent on RF-MEMS technologies during the last few decades.
This effort has brought the RF-MEMS technologies into a point that the first real RF-
MEMS products are in the market now. However, it has also been understood that
without a good package, an RF-MEMS device cannot be pronounced as a product.
It has also been understood that the RF-MEMS device and package development
cannot be considered as different processes to find the most optimum solution. This
means, the packaging solutions need to be considered and decided during the RF-
MEMS device development, not afterwards. In the wafer bonding approaches for
the wafer-level packaging (WLP), the cost of at least one lid wafer is added to the
total production cost [73]. These additional lid wafers can also add more process
steps like wafer grinding or chemical mechanical polishing (CMP). The wafer bonding
approach does not only increase the cost but also requires additional area around the
MEMS devices to accommodate the cap. Indeed, wafer-level encapsulation [74–76]
has advantages over the WLP with wafer bonding approach in terms of cost and area.
Predictably, the wafer-level encapsulation approach is also the chosen method for the
well-known commercial packaged MEMS devices [60, 77, 78]. Therefore, as part of
the integration of the RF-MEMS switch into IHP’s 0.13 µm SiGe BiCMOS technology;
thin film wafer-level encapsulation is developed [79, 80].
Fig. 2.11 shows the process flow schemes of the embedded RF-MEMS switch in
0.13 µm SiGe BiCMOS technology. The developed switch consists of two M4 HV
electrodes, a M5 RF- SL, a TM1 movable membrane, and a TM2 plate with releasing
holes. The TM2 plate is placed on top of the switch for the wafer-level encapsulation.
Integration of the RF-MEMS switch includes supplementary process developments
additional to the standard BEOL flow of the passives which mainly includes lateral
and vertical etch stops and the planarized oxide on the TM2 plate. Initially, TM2
layer of the RF and DC pads are reached with the passivation opening (Fig. 2.11 (a)).
Afterwards, with an additional mask, the RF-MEMS switches are released through the
TM2 plate holes until the M4 HV electrodes by hydrofluoric acid vapor phase etching
(HFVPE) (Fig. 2.11 (b)). Fig. 2.12 shows the scanning electron microscopy (SEM)
image of the released RF-MEMS switch before the wafer-level encapsulation. The
releasing steps are then followed by the 4 µm thick high deposition rate (HDR) oxide
deposition on the TM2 plate to fill the holes and have encapsulated switch. Finally,
the HDR oxide covered pads are reopened by RIE to provide electrical connections for
the measurements (Fig. 2.11 (c)). Fig. 2.13 demonstrates the additional process steps
for the integration of RF-MEMS switches into the SG13 BiCMOS technology.
23
2 Technology
c
2016 IEEE
Figure 2.12: The SEM image of the released RF-MEMS switch before the wafer-level
encapsulation (including the FIB cross section) [79, 80].
SG13 BiCMOS process with
etch stop layer after M4 and
planarized oxide after TM2
Lithography step to define
the area for RF-MEMS switch
Opening passivation nitride
in the defined area with RIE
Releasing RF-MEMS switch from the
defined area until HV electrodes (M4)
Deposition of HDR ox-
ide for encapsulation
Reopening the pads by RIE
for electrical characterization
Figure 2.13: Additional process steps for RF-MEMS switch integration into the SG13
BiCMOS process.
25
2 Technology
For the success of the wafer-level encapsulation, different design modifications are
done and process parameters are optimized. In detail, the TM2 plate’s hole size was
one of the most important design parameter for both the HF vapor phase etching
and the wafer-level encapsulation. For the investigation of the hole size impact, the
quadratic hole sizes are varied with the side length (a) of 1.25 µm, 1.5 µm, 1.75 µm
and 2.0 µm. Considering a TM2 thickness of d= 3 µm results in aspect ratios (ARs)
(AR = d/a) from 2.4 down to 1.5.
Beside the hole size, the plasma enhanced chemical vapor deposition (PECVD)
process for the HDR oxide deposition is optimized to achieve the required low step
coverage. By minimizing the process temperature of the HDR oxide deposition it is
also intended to reduce its influence on the active elements of the FEOL and on the
mechanical behavior of the RF-MEMS switch. Consequently, a maximum temperature
of 200◦C for HDR oxide deposition has been developed to achieve closing of the TM2
plate holes.
As a result of the variable TM2 plate hole size investigation, successfully encapsu-
lated RF-MEMS switches are achieved with an AR of 2.4 and 2.0 (Fig. 2.14 (a) and
(b)). With an AR of 1.7, a carbon containing material inside the closing holes is
determined by energy dispersive X-ray (EDX) spectroscopy which is coming from the
spin-coating of the photoresist mask before the RIE opening of the pads (Fig. 2.14 (c)).
The photoresist can only enter from the TM2 plate holes in case of a not fully closed
TM2 plate holes. Clearly with an AR of 1.5, a fully closed encapsulation also could
not be achieved (Fig. 2.14 (d)). Indeed, with AR of 2.4 and 2.0 it became possible to
close the TM2 plate holes and prevent the unwanted HDR oxide deposition through
the holes into the cavity.
Fig. 2.15 shows the SEM image of the fabricated WLE RF-MEMS switch with its
marked region for the FIB cut. FIB analysis shows that RF-MEMS switch is released
and encapsulated successfully. The FIB cross section of the WLE RF-MEMS switch
is given in Fig. 2.16 and it shows the covered TM2 plate with the HDR oxide, the
released TM1 membrane, the M4 HV electrode and the M5 RF- SL in detail.
26
2 Technology
c
2016 IEEE
(a)
c
2016 IEEE
(b)
c
2016 IEEE
(c)
c
2016 IEEE
(d)
Figure 2.14: The focused ion beam SEM images of the wafer-level encapsulation with
varied aspect ratios (a) AR= 2.4, (b) AR= 2.0, (c) AR = 1.7 (with a
carbon containing material inside the closing holes which is determined
by EDX spectroscopy) and (d) AR= 1.5 [79, 80].
27
2 Technology
2.4 Conclusion
In Chapter 2, the process integration of MEMS modules in 0.25 µm and 0.13 µm SiGe
BiCMOS technologies have been presented. Within both technology sections, additio-
nal process steps for the MEMS module have been detailed. MEMS modules of 0.25 µm
and 0.13 µm SiGe BiCMOS technologies have presented two different packaging appro-
aches (including their process flows); namely wafer-to-wafer bonding Si cap packaging
and thin film wafer-level encapsulation. Wafer-to-wafer bonding Si cap packaging
technique as presented in Section 2.2 can be used in niche markets for low-volume
production. On the other hand, the thin-film wafer-level encapsulation technique that
has been presented in Section 2.3 has more advantages in terms of smaller area and
has already proven to be used in mass production for mobile applications [48, 58].
29
3 Modeling of RF-MEMS Switches
3.1 Introduction
Chapter 3 presents two different RF-MEMS switches, that are embedded into IHP’s
0.13 µm SiGe BiCMOS process technology. The first switch is developed and opti-
mized for D-band applications and the second switch is developed and optimized for
J-band applications. The chapter includes electromagnetic modeling, lumped-element
modeling and the experimental results of both switches. The simulation domain in
this thesis is the electromagnetic domain where a 3D EM solver is preferred to 2.5D
planar EM solvers to develop and optimize the RF-MEMS switches.
To realize an RF-MEMS switch in mm-wave operating frequencies, the first miles-
tone is developing an accurate EM model. In the development of the accurate EM
model, selection of the EM solver plays an important role. Between different EM
solvers, the 2.5D planar EM solvers are the mostly preferred ones where modeling of
passive devices is very convenient and straightforward. However, most of the 2.5D
planar EM solvers have the limitation of defining non-uniform dielectric planes in ho-
rizontal direction. Nevertheless, this is in contrast with the MEMS devices since the
mechanical parts of the MEMS devices are mostly in air but the anchors and routing
layers are in a dielectric. To be able to avoid this problem, a modeling solution with
2.5D planar EM solver is demonstrated in [65, 81] for frequencies up to 110 GHz. The
authors present accurate simulation results by implementing the measured contact
parameters of the RF-MEMS switch as lumped elements with a defined port in the
contact region. However, the presented solution do not guarantee the accuracy for
the modeling of RF-MEMS switches at above 110 GHz frequencies. Shortly, a more
careful modeling strategy is essential for EM simulation of RF-MEMS devices at high
frequencies.
In this thesis, accurate 3D EM models for BiCMOS embedded RF-MEMS switches
in both 0.25 µm and 0.13 µm technologies are built up in ANSYS HFSS 3D finite-
element-method (FEM) solver. The 3D EM solver allows defining the air cavity of the
switch and the BEOL oxide simultaneously in the same model, in contrast to 2.5D
planar EM solvers. Using the developed model, RF behavior of the MEMS device in
30
3 Modeling of RF-MEMS Switches
air and the routing layers in oxide can be simulated at the same time which is very
important considering the very high operating frequency. Furthermore, the process
effects such as bending of the membrane can also be modeled by changing the initial
gap in the simulator. The advantages of the 3D EM solver are explained in detail in
Section 3.2.1.
After the development of the RF-MEMS switches, the fabricated switches are opti-
cally and electrically characterized. For the dynamic behavior of the D-band switch,
laser-doppler-vibrometer (LDV) technique is used. For the electrical characterizations
of both switches in this chapter, Capacitance-Voltage (C-V) and S-parameter measu-
rements are performed.
3.2 D-band RF-MEMS SPST Switch
Section 3.2.1 begins with the challenges of modeling RF-MEMS switches at mm-wave
and the special methods that has been used for accurate EM simulations. EM modeling
of the RF-MEMS switch for D-band applications in SG13 BiCMOS technology starts
with usage of an inductive loaded design, similar to the SG25 switches. With the
wafer-level encapsulation approach in packaging, the design of the inductive loaded
switch is changed by removing the inductive loads which are in BEOL oxide and
increasing the length of the membrane arms which are in air. Later on, the updated
switch design is used in EM models for the optimizations of its important parameters;
such as the effect of the TM2 encapsulation plate, the arm width of the membrane,
the contact region, the RF- SL width, the membrane hole density and the ground-
ring width. Additional to the EM model, a lumped-element model is created (see
Section 3.2.2) which represents the behavior of the switch based on RLC elements and
can be used in circuits on system-level. Lastly in Section 3.2.3, the RF optimized
WLE RF-MEMS switch for D-band applications is characterized by LDV, C-V and
S-parameter measurements.
3.2.1 Electromagnetic modeling
The EM simulations of the RF-MEMS devices are commonly performed by the 2.5D
planar or 3D FEM solvers with the well-known material properties such as conductivity
and dielectric constant. Nevertheless, there are some common limitations of the EM
simulators for simulating MEMS devices. Typically, it is not possible to include the
bended structure of the suspended layers or the surface roughnesses of the materials.
31
3 Modeling of RF-MEMS Switches
Therefore, a careful design methodology is necessary to model the RF-MEMS devices
specifically for mm-wave applications.
3D vs 2.5D EM case comparisons
The main limitation of the 2.5D planar EM solvers is that all the layers are defined
infinite in horizontal directions. Since all the layers are defined infinite in horizontal
plane, the RF-devices cannot be simulated with released parts in air cavity and other
parts buried in dielectric. For this reason, the RF-MEMS switch can only be simula-
ted with all the anchors and routings in complete air or in complete another dielectric
environment. Such a case might have less influence on the results of the EM simula-
tions considering the operating frequencies less than 40 GHz. However, at frequencies
especially above 110 GHz, the anchors and the routing layers of the switch which are
normally in the BEOL oxide has a significant effect on the overall performance of the
RF-MEMS switch, considering the low resistive silicon substrate (50 Ω·cm) below the
RF-MEMS switch.
In order to compare the RF performance of the switch in two different environment
conditions, RF simulations of the switch are performed as if all parts of the switch,
including the parts located in back-end-of-line (BEOL) oxide, are in air. Fig. 3.1 shows
the created EM setups for the comparison of 2.5D planar EM solver and 3D EM
solver cases. The comparison of the RF performances of the switch for two different
environment conditions is shown in Fig. 3.2. The bold red and blue curves show the RF
performance of the switch when it‘s all in air and RF performance of the switch when
the anchors and the routing layers of the RF-MEMS switch are in the BEOL oxide,
respectively. As two environment conditions are compared in EM simulations, in the
off-state (up-state) RF-MEMS switches at high frequencies have quite different S21 and
S11 curves when in the on-state (down-state) they have quite similar isolations. In both
environment conditions contact region is simulated in air. Similar RF performances
in down-state is due to the domination of the on-state contact capacitance, Con, when
compared with the other parts of the switch. Different RF behaviors in up-states are
because of the domination of the anchors and routings of the switches, as compared
to Coff, that are simulated in oxide or in air. Comparison of the RF performance of
the switches shows that at high frequencies depending on the simulation environment
different results in up-state can be obtained. Indeed, an accurate EM model for D-
band RF-MEMS switch needs to consider the anchors and the routing layers in BEOL
oxide. Therefore, it is mandatory to simulate the mm-wave RF-MEMS devices in 3D
EM solvers.
32
3 Modeling of RF-MEMS Switches
Figure 3.1: Comparison of 2.5D planar EM solver (left) and 3D EM solver cases (right).
(a)
(b)
Figure 3.2: Comparison of the RF performances of the switch for two different en-
vironment conditions (2.5D case and 3D case): (a) S21 off-state and S21
on-state, (b) S11 off-state.
33
3 Modeling of RF-MEMS Switches
Initial approach for the switch
Although the intrinsic part of the RF-MEMS switch can be simulated in air and
the rest in oxide with the 3D EM simulations, there is another limitation coming
with process variations during fabrication. The contact capacitance of the switch can
vary with the membrane bending in both up and down states and also with surface
roughness in down state. For an accurate switch model in a 3D FEM solver, the
contact air capacitances of the switch in both states needs to be known. Therefore,
the contact air capacitances for off (up) and on (down) states, Coff and Con, should be
extracted from measured C-V data and the appropriate membrane positions should
be given into the EM model with respect to the extracted values for each state. The
mentioned above contact air capacitances are the capacitances between the membrane
and the RF- SL for each states of the RF-MEMS switch. The 3D EM simulation setup,
that is created during the development of the D-band BiCMOS embedded RF-MEMS
switch considers the air cavity inside the BEOL oxide and is shown in Fig. 3.3. Fig. 3.4
shows the contact region of an RF-MEMS switch in its initial EM model including the
inductive loading.
EM simulation results of the inductive loaded RF-MEMS switch for different contact
air capacitances that change with different distances in the contact region are shown
in Fig. 3.5. Each graph includes S-parameter results with the change of the membrane
distance by 50 nm steps over 710 nm initial distance. The bold red and blue curves
show the expected up and down state results, respectively. As it can be seen, the
electrical resonant frequency of the switch is decreasing with the reduced distance
between membrane and MIM layer of the RF- SL. With respect to the EM simulation,
IL less than 2 dB and isolation (ISO) higher than 32 dB can be achieved, including the
RF measurement pads at 140 GHz. The return loss for up-state is higher than 25 dB
at 140 GHz.
34
3 Modeling of RF-MEMS Switches
New approach in packaging and new switch design
During the initial EM modeling of the D-band switch, packaging possibilities for the
device is considered. Two different packaging approaches have already been demon-
strated with the SG25 RF-MEMS switches in [83] and [69]. With a new packaging
approach, geometrical modifications were necessary on the initial model of the D-band
RF-MEMS switch. The new design for D-band RF-MEMS switch is shown in Fig. 3.6
(a) and does no more include the extra inductive loading between the arms and the
RF ground-ring.
For an accurate EM model in the 3D FEM solver, the distances between TM1
membrane and M5 RF- SL for both off (doff) and on-states (don) of the switch are
needed to be inserted into the simulator. The mean value of the distance between the
TM1 and M5 layers is 900 nm in IHP’s 0.13 µm SiGe BiCMOS technology. In order
to estimate the possible RF performance differences due to the process variations,
parametric EM simulations with different doff and don are performed (Fig. 3.6 (b, c)).
For the insertion loss in the parametric EM simulation the membrane position is swept
from 100 nm above the specified TM1 layer until 100 nm below with a step size of 50 nm.
The maximum and minimum doff vary the calculated contact air capacitance between
10.6 fF and 13.2 fF. Fig. 3.7 (a) shows the change of the insertion loss curves with
different doff values. Varying doff by 200 nm leaded to 0.16 dB change of insertion loss
at 170 GHz. Furthermore, the parametric EM simulations are extended with different
don values for the isolation curves. The distance don is stepwise (10 nm) decreased
starting from 100 nm down to 40 nm and the simulation results are shown in Fig. 3.7
(b). The maximum and minimum don vary the calculated contact air capacitance
between 106.24 fF and 265.62 fF.
Decreasing don by 60 nm caused a resonant frequency shift from the lower end of the
D-band to the upper end. Briefly, the varying doff due to process variations did not
affect the insertion loss tremendously but the resonant frequency shifted significantly
with the varying don. For this reason, the contact air capacitances are extracted from
the C-V measurements of the RF-MEMS switch to have an accurate 3D FEM result
of the D-band RF-MEMS switch especially for the on-state. Based on the extracted
contact air capacitances (Coff and Con) and the known contact area of the switch
(Acont), doff and don can be calculated using the simple parallel plate capacitor equation
(equation 3.1) and inserted into the EM model. In equation 3.1, Cparallelplate is the
capacitance in farads which is constructed by two parallel plates of area, A, in square
meters and separated by a distance, d, in meters. εris the relative permittivity of the
37
3 Modeling of RF-MEMS Switches
c
2016 IEEE
(a)
c
2016 IEEE
(b)
Figure 3.7: The simulated S-parameters of the WLE RF-MEMS switch with varying
TM1 membrane and M5 RF- SL distances due to process variation: (a)
insertion loss, (b) isolation [79, 80].
material between the plates and ε0is the permittivity of free space, 8.854·10-12 F/m.
Cparallel plate =εr·ε0·A
d(3.1)
For the EM modeling and optimization, the important parameters of the 140 GHz
targeted encapsulated RF-MEMS switch are determined as the effect of the TM2
encapsulation plate, the arm width of the released membrane, the contact region, the
RF- SL width, the membrane hole density and the ground-ring width, as given in
Fig. 3.8. In the following subsections, the aforementioned critical parameters of the
RF-MEMS switch are modeled and optimized in D-band.
39
3 Modeling of RF-MEMS Switches
(a)
(b)
Figure 3.8: The modeled RF–MEMS switch with its key EM optimization parameters;
(a) isometric view with clip plane (b) top view with hidden TM2 plate.
40
3 Modeling of RF-MEMS Switches
Figure 3.9: Three different EM simulations in order to investigate the effect of the TM2
plate (a) without TM2 plate (b) with TM2 plate, including the releasing
holes (c) with TM2 plate, without the releasing holes.
Effect of the TM2 plate
It is crucial to consider the influence of the package from the beginning of the RF-
MEMS device development since its influence on the device performance cannot be
avoided. Therefore, the RF optimization of the D-band RF-MEMS switch is started
with a focus on the TM2 encapsulation plate to investigate the influence of the package.
The initial step to see the effect of the package is in the definition of the EM
model. Since the TM2 plate includes many small holes which significantly increases the
simulation time, three different EM models are created for the preliminary comparison:
the RF-MEMS switch without a TM2 encapsulation plate (Fig. 3.9 (a)), with a TM2
encapsulation plate including small holes (Fig. 3.9 (b)) and with a TM2 encapsulation
plate without holes (Fig. 3.9 (c)). The EM simulation comparison (Fig. 3.10) has shown
significant differences in S-parameter results if an encapsulation plate is introduced.
With the TM2 plate, the off-state performance of the RF-MEMS switch has shown
an increased loss in all D-band and the resonant frequency of the on-state has shifted
from 153 GHz to 140 GHz. However, the comparison of two models, including the
TM2 plates with and without the releasing holes has shown that there is no significant
difference in terms of RF performance between the two models. By omitting the TM2
releasing holes, the number of meshed elements and the total simulation time are
reduced by half. As a result of this, all the further simulations of this work are done
including the TM2 plates but without the releasing holes.
41
3 Modeling of RF-MEMS Switches
Figure 3.11: The generic view of the varied arm width of the RF-MEMS switches.
Optimization of the arms
After simplification of the EM model by removing the TM2 plate holes, EM simulations
are continued with an investigation on the varied arm width of the RF-MEMS switch
to show the effect of the arm inductance (Larm) on the RF performance. It is worth
to mention that the limits of the process lithography and mechanical constraints are
considered during the optimization of the arms for maximizing the RF performance.
By tuning the contact air capacitance or the arm inductance of the RF-MEMS
capacitive switches, operation at different frequency bands can be achieved [83]. The
main series resonant frequency in on-state is calculated, as follows:
f=1
2π·√Ccont ·Ltotal
(3.2)
In equation 3.2, Ccont is the contact air capacitance and Larm is the total inductance
coming from the membrane, arms and the ground-ring of the RF-MEMS switch where
arms have the most significant contribution. Fig. 3.11 shows the generic view of the
varied RF-MEMS arm width between 2 µm to 4 µm. It should be noted that the doff
and don are kept identical during the simulations with different arm widths.
The off-state S21 curves in Fig. 3.12 (a) and S11 curves in Fig. 3.12 (b) have shown
no significant differences with the varied arm width as expected. On the other hand,
the on-state S21 curves in Fig. 3.12 (c) have shown a shift in the resonance frequency as
a result of the varied inductances of the RF-MEMS switch with different arm widths
The resonance frequency of the RF-MEMS switch is shifted from 132 GHz to 136 GHz,
141 GHz and 143 GHz with the increased arm widths from 2 µm to 2.5 µm, 3 µm, and
4µm, respectively. For the optimized RF-MEMS switch, 3 µm arm width is chosen
due to the targeted operating frequency of 140 GHz.
43
3 Modeling of RF-MEMS Switches
Figure 3.13: The generic view of the varied contact region width of the RF-MEMS
switches (a) 15 µm (b) 20 µm (c) 30 µm (d) 40 µm.
Optimization of the RF-signal line contact region
An RF- SL defines the RF characteristics of an RF-MEMS switch, in terms of insertion
loss, return loss and isolation. The central part of the SL, directly underneath the
membrane, is named as the contact region. The area of the contact region determines
the contact capacitance, Ccont, of the switch and the resonance frequency together
with the inductance of the RF-MEMS switch (Ltotal). In the EM simulations, the
contact region width of the RF-MEMS switch is varied between 15 µm to 40 µm with
a fixed SL width of 10 µm. It should be noted that the contact region optimization
is performed before the SL width optimization because the main resonance at down-
state of the RF-MEMS switch is defined by the contact region capacitance. Fig. 3.13
shows the generic view of the varied contact region of the RF-MEMS switch. With
the increased width of the contact region, the loss increases (Fig. 3.14 (a)) in off-
state which can be correlated with the increased mismatch of the switch (Fig. 3.14
(b)). On the other hand, the S21 on-state curves in (Fig. 3.14 (c)) show the shift of
the resonance frequency to the lower frequencies with the increased contact region
widths, as expected. In summary, the S-parameter simulation results show a trade-off
between the minimum loss and the maximum isolation at 140 GHz. Therefore for the
next optimizations, 30 µm contact region width is chosen to obtain desired isolation
values of more than 30 dB at 140 GHz.
45
3 Modeling of RF-MEMS Switches
Figure 3.15: The generic view of the varied RF-signal line width of the RF-MEMS
switches (a) 5 µm (b) 10 µm (c) 15 µm (d) 20 µm.
Optimization of the RF-signal line width
The designed SL consists of M5 layer in the contact region and continues in each side
until the RF pads. During the simulations with the varying SL width, the contact
region width is kept at 30 µm to provide a constant contact capacitance which en-
sures the desired isolation at 140 GHz of operation. Four different signal line widths
(5 µm, 10 µm, 15 µm and 20 µm) are chosen for the comparison of the RF performances.
Fig. 3.15 shows the generic view of the varied RF- SL width of the RF-MEMS swit-
ches. The corresponding S21 and S11 curves are given in Fig. 3.16. The S-parameter
simulations have shown 0.48 dB variation in loss at 140 GHz between the switches with
a 5 µm and a 20 µm signal line widths. The minimum loss (Fig. 3.16 (a)) and the best
matching (Fig. 3.16 (b)) are achieved with the 5 µm width SL in the complete D-band.
In on-state, the isolation curves (Fig. 3.16 (c)) of the switches have shown no shift
in the resonance frequency. Despite the fact that the 5 µm width SL has shown the
best performance in off-state, 10 µm width SL is selected for the optimized RF-MEMS
switch for a small IL increase penalty to have a better stability of the process against
the under etch of the SL during the releasing process. Here, the decision is taken
considering both RF performance and process stability.
47
3 Modeling of RF-MEMS Switches
Figure 3.17: The generic view of TM1 membranes with (a) high, (b) low hole densities
and (c) no holes.
Effect and optimization of the membrane hole density
The RF-MEMS switches are released through the holes of the membrane. In order to
investigate the influence of the hole densities of membrane on the RF performance,
three cases are compared with the EM simulations: 19.6 % hole density (Fig. 3.17 (a)),
11.1 % hole density (Fig. 3.17 (b)) and without holes (Fig. 3.17 (c)). Although the
RF-MEMS switch without the releasing holes is not suitable for the releasing of the
device, it is simulated to investigate the maximum achievable contact capacitance and
corresponding RF performance without any holes. The S-parameter simulations have
shown that there are no significant differences between the three cases in the S21 and
S11 curves for the off-state (Fig. 3.18 (a), (b)) but shift in the resonance frequency for
the on-state with the increased contact capacitances (Ccontact) due to the lower hole
densities on the membrane (Fig. 3.18 (c)). With the results of the hole effect, 11.1 %
hole density membrane with the resonance frequency at 140 GHz is taken to be used
during the next optimization steps.
49
3 Modeling of RF-MEMS Switches
Figure 3.19: The generic view of the varied ground-ring width of the RF-MEMS swit-
ches (a) 3.5 µm (b) 5 µm (c) 7.5 µm (d) 10 µm and (e) 12.5 µm.
Optimization of the ground-ring width
The RF-MEMS switch needs a well-defined RF ground similar as in an mm-wave cir-
cuit to operate properly. Such effect can be ignored for lower frequency operations but
especially for mm-wave RF-MEMS devices, the performance is significantly affected
by the ground connections. Additionally, the RF ground should be connected to the
membrane through the four arms and enclose the structure. However, the designed
ring for RF ground has comparable size with the RF-MEMS switch and presumably
introduces RF coupling of the SL and influences the RF performance of the switch. For
these reasons, an investigation on the influence of the ground-ring width is essential.
Fig. 3.19 shows the generic view of the designed RF ground-ring with the variation of
the width from 3.5 to 12.5 µm by keeping the inner radius of the ring constant. The
performed EM simulations have shown an increase of the loss in off-state (Fig. 3.20
(a)) with wider RF ground-ring which can be correlated with the increased RF sig-
nal coupling from the M5 SL to the TM1 of the ground-ring. The loss at 140 GHz
is increased from 0.98 dB to 1 dB, 1.12 dB, 1.27 dB and 1.43 dB with respect to the
increased ground-ring widths. The S11 curves of the off-state (Fig. 3.20 (b)) show the
variation of the matching due to the varying ground-ring width. The return loss at
140 GHz varies between 8.2 dB and 11.3 dB. On the other hand, the ground-ring width
has a very small effect on isolation in on-state (Fig. 3.20 (c)), since the RF signal is
shorted to the ground. The RF ground-ring width value is chosen as 3.5 µm for the
EM optimized RF-MEMS switch as it has given the minimum loss.
51
3 Modeling of RF-MEMS Switches
Figure 3.21: The cross section of WLE switch with lumped-element capacitances [80].
c
Cambridge University Press and the European Microwave Association
2017
3.2.2 Lumped-element modeling
Although the 3D FEM solvers provide accurate EM simulation results with the calcu-
lated doff and don values from the extracted contact air capacitances (Coff and Con),
lumped-element models give faster simulation results. In [84], an accurate lumped-
element model of RF-MEMS switches has been presented up to 110 GHz and has
shown good matching with the S-parameter measurement results. Furthermore, a
lumped model of an RF-MEMS switch extends the usage of the devices since with the
help of a lumped-model, RF-MEMS switches can be simulated using a spice simulator
in various circuits on system-level.
The main capacitances for the lumped-element model of the WLE RF-MEMS switch
are shown on the process cross section in Fig. 3.21 [80]. Similar to the EM model,
extracted contact parameters of the switch are necessary for the lumped element model
of the RF-MEMS switch especially in on-state. Fig. 3.22 shows the lumped-element
model of the WLE RF-MEMS switch. The extracted contact air capacitances (Ccont),
Coff of 13.4 fF and Con of 149 fF are used to model the contact region. The details of
the C-V measurements are given in Section 3.2.3. Furthermore in the on-state model,
the contact resistance of 5 kΩ is also considered.
For the S-parameter simulations with the lumped-element model, both ports of the
device are terminated with 50 ohm. The M5 RF- SL is modeled using RLC elements
between the ports and the arms of the membrane are modeled as inductors (Larm1,
Larm2) with series resistors (Rarm1, Rarm2). Celec1 and Celec2 are the parasitic capa-
53
3 Modeling of RF-MEMS Switches
Table 3.1: Small-signal component values of the WLE RF-MEMS switch [80].
c
Cambridge University Press and the European Microwave Association
2017
Component Value
Lencap1, Lencap2, Lencap3, Lencap4, Lencap5 70 pH
Rencap1, Rencap2, Rencap3, Rencap4,
Rencap5, Rarm1, Rarm2, RSL1, RSL2, RSL3,
RSL4, RSL5, RSL6
0.5 Ω
CencapSL1, CencapSL4 2 fF
CencapSL2, CencapSL3, CSL1, CSL5 3 fF
Larm1, Larm2 15 pH
Cpad1, Cpad2 11 fF
LSL1, LSL6 42 pH
LSL2, LSL5 3 pH
LSL3, LSL4 1 pH
CSL2,CSL4 1.65 fF
CSL3 10 fF
Csubst 170 fF
Rsubst 400 Ω
Table 3.2: Small-signal component values of the WLE RF-MEMS switch, varied due
to states [80]. c
Cambridge University Press and the European Microwave
Association 2017
States Celec1, Celec2 Ccont Rcont
Off-state 10 fF 13.4 fF (∞)
On-state 12 fF 149 fF 5 kΩ
citances between the TM1 membrane and the M4 HV electrodes. The values of the
RLC elements of the lumped model for both states of the switch are given in Table 3.1
below. Due to the down bending of the TM1 membrane in on-state, Celec1 and Celec2
values are modeled slightly larger than the off-state values. Table 3.2 shows the varied
Celec1, Celec2 and Ccont values with respect to the states of the WLE RF-MEMS switch.
Both the 3D EM model and the lumped-element model consider the wafer-level en-
capsulation of the RF-MEMS switch. To include the WLE in the lumped model, TM2
plate is modeled with series inductances (Lencap1, Lencap2, Lencap3, Lencap4, Lencap5)
and resistances (Rencap1, Rencap2, Rencap3, Rencap4, Rencap5). Additionally for its effect
on the RF-MEMS switch, the coupling capacitances (CencapSL1, CencapSL2, CencapSL3,
CencapSL4) between the TM2 plate and the M5 RF- SL is also inserted into the lumped-
54
3 Modeling of RF-MEMS Switches
model. Lastly, the substrate effect is considered in the model and inserted as parallel
capacitance (Csubst) and resistance (Rsubst) between ground and the RF- SL. The com-
parison of the measured S-parameters of the fabricated WLE RF-MEMS switch with
simulation results of the EM and lumped-element models are given in Section 3.2.3.
3.2.3 Experimental results
This section presents the different characterization methods and their results for the fa-
bricated RF-MEMS switch. For the characterization of the D-band RF-MEMS switch,
dynamic measurements are done by LDV and the electrical characterizations are done
with C-V and S-parameter measurements.
Dynamic Measurements using Laser-Doppler-Vibrometer
A LDV (MSA-500, Polytec) [85] is used to characterize the dynamic behavior of the
RF-MEMS switch. It provides the total displacement of the moving parts, switch-on
time (ton) and switch-off (toff) time using optical methods. Therefore, the measure-
ments can only be done on RF-MEMS switches without the deposition of HDR oxide
for encapsulation. Despite the fact that the TM2 plate of the RF-MEMS switch makes
the optical observation of the membrane displacement difficult, a careful study using
the LDV is performed through the releasing holes to extract the displacement versus
time curve. The generic view of the LDV setup is given in Fig. 3.23. The displacement
versus time curve of the switch with the 60 V actuation voltage by the LDV measure-
ment is shown in Fig. 3.24. From the figure, a displacement of 0.9 µm, ton and toff of
less than 10 µs can be seen.
C-V Measurements
After the LDV measurements with the not-encapsulated RF-MEMS switch, the cha-
racterizations are continued with the encapsulated RF-MEMS switch for the extraction
of the contact capacitances. The WLE RF-MEMS switch is initially characterized by
C-V measurements.
For the C-V measurements, the high frequency impedance analyzer Agilent E4991 is
used and the capacitance values are taken at 3 GHz. An Open/Short/Load calibration
on an impedance standard substrate (ISS) from Cascade Microtech is applied to remove
the parasitics of the measurement setup. The capacitance values are measured for
the actuation voltages between -80 V and +80 V with 5 V steps. Fig. 3.25 shows the
measured C-V graph of the WLE RF-MEMS switch and the extracted Ccont versus
56
3 Modeling of RF-MEMS Switches
c
2016 IEEE
Figure 3.25: The measured C-V graph of the WLE RF-MEMS switch (red) and the
extracted contact air capacitance vs voltage graph (black) [79, 80].
actuation voltage graph. The developed switch does not show any hysteresis, since
the RF- SL distance to membrane is less than the 2/3 of the distance between the
electrodes to membrane.
Extraction of the Cconts is done with the help of an additional test structure that
contains an RF- SL without membrane and TM2 encapsulation plate. With the test
structure, the coupling capacitance (52 fF) from the signal line to ground ring and the
Si substrate is measured and the Cconts are extracted by its subtraction. The Cconts
are extracted as 13.4 fF Coff and 149 fF Con with a Con/Coff ratio of 11.1. Moreover
it is observed that the pull-in occurs after 50 V and the Con is stable after 65 V of
actuation.
S-Parameter Measurements
Two port on-wafer S-parameter measurements of the WLE RF-MEMS switch are
performed from 110 GHz to 170 GHz on semi-automated wafer probe station. The S-
parameter measurements are performed with a setup from Rohde & Schwarz, consisting
of a 4 port ZVA24 as vector network analyzer (VNA) / system controller and two
ZVA170 Millimeter-Wave Converters. The Cascade 75 µm pitch infinity(R) ground
signal ground (GSG) waveguide probes were connected via the WR6 waveguide s-
bend with the millimeter-wave modules. For the calibration, the ISS 138–356 was
placed together with an RF absorber on an auxiliary ceramic chuck and a full two
port Thru-Reflect-Reflect-Match (LRRM) calibration was performed. To actuate the
membrane of the RF-MEMS switch, Agilent Source Measurement Module E5281B was
used.
Fig. 3.26 shows the comparison between the measured S-parameters of the fabricated
58
3 Modeling of RF-MEMS Switches
WLE RF-MEMS switch with simulation results of the EM and the lumped-element
models. With the developed RF-MEMS switch, better than 0.67 dB insertion loss and
more than 16 dB isolation in all D-band is achieved. The measured S21 curve in off-
state (Fig. 3.26 (a)) shows a loss of 1.07 dB and the measured S11 curve in off-state
(Fig. 3.26 (b)) is 9.13 dB at 140 GHz. For the insertion loss curves (Fig. 3.26 (c)),
the signal reflection due to mismatch [87] is considered by equation 3.3. In the end
of the EM optimization steps, the measured S-parameter results of the RF-MEMS
switch give 0.6 dB insertion loss (Fig. 3.26 (c)) and 33.8 dB isolation (Fig. 3.26 (d)) at
140 GHz. Both EM model and lumped-element model simulation results are in good
agreement with the measured S-parameter results in the D-band. The small differences
between the measurement and simulation curves can be explained with the difficulty
of calibration in frequencies above 100 GHz.
IL = 10 ·log |S21|2
1−|S11|2(3.3)
Lastly, Fig. 3.27 shows the S21 and S11 curves of the WLE RF-MEMS switch with
the varied actuation voltages from 0 V to 80 V with 5 V steps. From the figure it can
be observed that the switch is in off-state until 50 V, in transition in 55 and 60 V and
in on-state from 65 to 80 V, which is also verified by the C-V measurements (Fig. 3.25).
59
3 Modeling of RF-MEMS Switches
(a)
(b)
Figure 3.27: The measured S-parameters of the WLE RFMEMS switch for (a) S21 and
(b) S11 with actuation voltages from 0 V to 80 V (with 5 V steps).
61
3 Modeling of RF-MEMS Switches
c
2017 IEEE
Figure 3.28: The fabricated J-band SPST RF-MEMS switch with the TM2 plate [21].
3.3 J-band RF-MEMS SPST Switch
This section presents the EM modeling, lumped-element modeling and experimental
results of an RF-MEMS switch for J-band applications that is fabricated in a 0.13 µm
SiGe BiCMOS process technology. Fig. 3.28 shows the microscope image of the fabri-
cated J-band RF-MEMS switch in 0.13 µm SiGe BiCMOS process technology.
3.3.1 Electromagnetic modeling
For the development of an 240 GHz targetted J-band RF-MEMS switch, an EM model
is built up in ANSYS HFSS 3D FEM solver (Fig. 3.29 (a)). IHP’s 0.13 µm SiGe
BiCMOS technology has a mean value of 900 nm as the distance between the TM1
and the M5 layer, which forms the distance between the membrane and the RF- SL
of the designed switch. However, the Coff and Con of the RF-MEMS switch can vary
with the process variations across the wafer such as different surface roughness, metal
and oxide thicknesses or stress behavior of the suspended TM1 membrane. As a result
of the variable contact air capacitances, it is essential to estimate the possible RF
performance variation of the RF-MEMS switch before the fabrication.
For the RF performance estimations, parametric EM simulations are performed
with different doff and don distances which simulate the aforementioned contact air
62
3 Modeling of RF-MEMS Switches
capacitance variations (Fig. 3.29(b), (c)). For the off-state EM simulation, the mem-
brane position is swept from 100 nm above the specified TM1 layer until 200 nm below
with a step size of 50 nm. The calculated maximum and minimum Ccont vary between
4.45 fF and 6.36 fF. The change of the insertion loss with different doff values is shown
in Fig. 3.30. Varying the doff by 300 nm leads to only 0.06 dB change of the insertion
loss at 240 GHz but 0.48 dB change at 325 GHz. For the on-state, the parametric EM
simulations are extended with different don values for the isolation. The simulation
results are shown in Fig. 3.30 for the decreased don, starting from 120 nm down to
60 nm with 10 nm steps. The maximum and minimum don vary the calculated contact
air capacitance between 37.1 fF and 74.2 fF. Such small gaps (don) between the moving
membrane and the RF- SL are crucial to model the down-state, resulting due to the
process variations. Decreasing the don by 60 nm causes the shift of the resonance fre-
quency from 276 GHz to below 220 GHz. Shortly, this method is useful for preliminary
simulations to estimate the possible range of the resonance frequency. However, the
extraction of the Coff and Con from the C-V measurements of the RF-MEMS switches
is essential to have accurate EM simulation results for the 220 – 325 GHz frequency
band. The values of doff and don can be calculated by using the simple parallel plate
capacitor equation with the extracted contact air capacitances and inserted into the
EM simulation model.
3.3.2 Lumped-element modeling
Since lumped-element models give faster simulation results and the RF-MEMS switch
can be simulated in various circuits, a lumped-model is also created for the J-band
RF-MEMS switch. The main capacitances considered in the lumped-element model of
the J-band RF-MEMS switch are similar compared to the lumped-element model of
the WLE D-band switch which are shown on the process cross section in Fig. 3.21 [80].
Thus, the same detailed lumped-element model of the D-band RF-MEMS switch can
be used for the J-band RF-MEMS switch which is given in Fig. 3.22. The extracted
contact air capacitances, Coff of 6.07 fF and Con of 53.3 fF are used to model the
contact region. Furthermore in the on-state model, the contact resistance of 5 kΩ is
also considered.
For the S-parameter simulation with the lumped-element model, Port1 and Port2 are
terminated with 50 ohm. The M5 RF- SL is modeled using RLC elements between the
two ports and the arms of the membrane are modeled as inductors (Larm1, Larm2) with
series resistors (Rarm1, Rarm2). Celec1 and Celec2 are the parasitic capacitances between
the TM1 membrane and the M4 HV electrodes. The values of the RLC elements of the
64
3 Modeling of RF-MEMS Switches
Table 3.3: Small-signal component values of the J-band RF-MEMS switch.
Component Value
Lencap1, Lencap2, Lencap3, Lencap4, Lencap5 70 pH
Rencap1, Rencap2, Rencap3, Rencap4,
Rencap5, Rarm1, Rarm2, RSL1, RSL2, RSL3,
RSL4, RSL5, RSL6
0.5 Ω
CencapSL1, CencapSL4 0.7 fF
CencapSL2, CencapSL3 1.1 fF
CSL1, CSL5 2 fF
Larm1, Larm2 12.5 pH
Cpad1, Cpad2 7 fF
LSL1, LSL6 27.5 pH
LSL2, LSL5 2 pH
LSL3, LSL4 0.66 pH
CSL2,CSL4 1 fF
CSL3 4.8 fF
Csubst 170 fF
Rsubst 400 Ω
Table 3.4: Small-signal component values of the J-band RF-MEMS switch, varied due
to states.
States Celec1, Celec2 Ccont Rcont
Off-state 6.5 fF 6.07 fF (∞)
On-state 7.8 fF 53.3 fF 5 kΩ
lumped model for both states of the J-band switch are given in Table 3.3 below. Due
to the down bending of the TM1 membrane in on-state, Celec1 and Celec2 values are
modeled slightly larger compared to the off-state values. Table 3.4 shows the varied
Celec1, Celec2 and Ccont values with respect to the states of the J-band RF-MEMS
switch.
Both the 3D FEM model and the lumped-element model consider the effect of the
TM2 plate on the RF-MEMS switch. The TM2 plate is modeled with series induc-
tances (Lencap1, Lencap2, Lencap3, Lencap4, Lencap5) and resistances (Rencap1, Rencap2,
Rencap3, Rencap4, Rencap5). Additionally for its effect on the RF-MEMS switch, the
coupling capacitances (CencapSL1, CencapSL2, CencapSL3, CencapSL4) between the TM2
plate and the M5 RF- SL is also inserted into the lumped-model. Lastly, the substrate
effect is considered in the model and inserted as parallel capacitance (Csubst) and
66
3 Modeling of RF-MEMS Switches
c
2017 IEEE
Figure 3.31: The measured C-V characteristic of the J-band RF-MEMS switch (red)
and the extracted contact air capacitance vs. voltage graph (black) [21].
resistance (Rsubst) between ground and the RF- SL. The comparison of the measured
S-parameters of the fabricated J-band RF-MEMS switch with simulation results of the
EM and the lumped-element models will be given in the next section, Section 3.3.3.
3.3.3 Experimental Results
C-V Measurements
The fabricated RF-MEMS switch is initially characterized by C-V measurements. The
capacitances are measured for the actuation voltages between -90 V and +90 V with
5 V steps. Fig. 3.31 shows the measured C-V characteristic and the extracted contact
air capacitance versus actuation voltage graph of the RF-MEMS switch. Extraction
of the contact air capacitances is done with the help of an additional test structure
that consists of an RF- SL without a membrane and a TM2 plate. With the test
structure, the coupling capacitance of 36.4 fF from the RF- SL to the silicon substrate
is measured and the contact air capacitances are extracted by its subtraction. The
contact air capacitances are extracted as 6.07 fF Coff and 53.3 fF Con (at 75 V) with a
Con/Coff ration of 8.78. Based on the measured Con and Coff capacitances (Fig. 3.31)
and the considered ∼500 µm2contact area, doff is calculated as ∼750 nm and don as
∼80 nm. Furthermore, it is observed from the figure that the pull-in occurs after 65 V
and Con is stable after 70 V.
67
3 Modeling of RF-MEMS Switches
S-Parameter Measurements
After the C-V analysis of the RF-MEMS switch, two port on-wafer S-parameter me-
asurements of the device are performed from 220 to 325 GHz at room temperature
on the semi-automated probe station. For the S-parameter measurements, a setup
from Rhode & Schwarz, consisting of a 4 port ZVA24 as VNA / system controller
and two ZVA325 Millimeter-Wave Converters, are used. The 50 µm pitch GSG wa-
veguide probes from Picoprobe are connected via a WR3 waveguide s-bend with the
millimeter-wave modules. For calibration, the ISS, CS-15, is placed together with an
RF absorber on an auxiliary ceramic chuck and a full two port Thru-Reflect-Match
plus (LRM+) calibration is performed. To actuate the membrane of the RF-MEMS
switch a 100 V Agilent Source Measurement Module E5281B is used.
The comparison of the measured and simulated S-parameter results of the RF-
MEMS switch is given in Fig. 3.32. The switch shows maximum isolation of 24.67 dB
at 238 GHz with the insertion loss of 0.49 dB. The return loss is better than 9.6 dB over
the J-band. The calculated doff and don values from the C-V measurements are then
inserted into the EM model for the comparison with the S-parameter measurements.
The 3D FEM simulation and lumped-element model results are in good agreement
with the measured S-parameters in the 220 – 325 GHz frequency band. It should be
noted that the differences between the measurement and simulation curves can be
explained by the calibration challenges in these high frequencies.
Lastly, Fig. 3.33 shows the S21 and S11 curves of the RF-MEMS switch with the
varied actuation voltages between 0 V and 90 V with 5 V steps. It can be observed
from the measured data that the switch is in off-state until 55 V, in transition at 60 V
and 65 V and in on-state between 70 V to 90 V of actuation.
68
3 Modeling of RF-MEMS Switches
3.4 Conclusion
With 3D FEM solvers, it is possible to estimate the RF performances of RF-MEMS
switches. However accurate EM models need a careful modeling approach as one of
the key elements, the contact air capacitances, in both on and off-states of the switches
are not known precisely before their fabrication. With the known contact areas of the
switches and the measured contact air capacitances, the distances between the RF- SL
and the released membrane can be calculated by the simple parallel plate capacitance
formula. Afterwards the calculated distances can be given into the EM models for the
accurate EM simulations.
Beside the accurate EM models, lumped-element models of the RF-MEMS switches
are also necessary especially for the circuit designers to predict the RF-behavior of the
designed circuits using the switches. The lumped-element models of the RF-MEMS
switches increase the simulation speed remarkably compared to the EM models and
give the possibility to simulate the switches with various circuits in a system level.
Section 3.2 has presented the EM modeling, EM optimizations, lumped-element mo-
deling and the electrical characterizations of a WLE RF-MEMS switch embedded in a
0.13 µm SiGe BiCMOS process technology for D-band applications. A 140 GHz targe-
ted operating frequency WLE RF-MEMS switch with its measured 0.6 dB of insertion
loss and 33.8 dB of isolation at 140 GHz, has been successfully demonstrated. To the
best of my knowledge, the 0.13 µm BiCMOS embedded thin film wafer-level encap-
sulated RF-MEMS switch is the first RF-MEMS switch in literature which operates
with state of the art RF performances in D-band. The results of the both EM model
and lumped-element model of the switch are in good agreement with the S-parameter
measurements in D-band [80] which proves the efficiency of the models.
Section 3.3 has presented a J-band RF-MEMS switch embedded in a 0.13 µm SiGe
BiCMOS process technology. To the best of my knowledge, the demonstrated SPST
switch is the first RF-MEMS switch presented in the complete J-band. This work
shows that the BiCMOS embedded RF-MEMS switches can provide low insertion loss
and high isolation at sub-THz frequencies [21].
Table 3.5 shows the measured performance of the designed D-band [80] and J-band
[21] SPST switches in comparison to the other published and above 110 GHz SPST
switches in the literature, based on the indium phosphide (InP)-double hetero bipolar
transistor (DHBT), SOI-CMOS and BiCMOS technologies. In comparison with the
published SPST performances in literature, it is clearly shown that the presented
D-band and J-band RF-MEMS SPST switches have the minimum insertion losses.
71
3 Modeling of RF-MEMS Switches
Table 3.5: Measured Performance Comparison of mm-wave SPST Switches.
This work:
D-band switch
[80]
[88] [89]
This work:
J-band switch
[21]
Technology 0.13 µm SiGe
BiCMOS
45 nm CMOS
SOI
250 nm InP
DHBT
0.13 µm SiGe
BiCMOS
Topology RF-MEMS Double Shunt
FET
Double-shunt
transistor RF-MEMS
Frequency (GHz) 110 – 170 140 – 220 220 – 320 220 – 325
IL (dB) 0.6 @ 140 GHz <1.5 1.2 – 2.9 0.44*– 1.7
ISO (dB) 33.8 @ 140 GHz 15 – 20 13.1 – 15.7 13.24 – 25.1
IP1dB (dBm) — 7 – 8
@ 50 – 65 GHz
16 @ 300 GHz
(simulated) —
Chip Area (mm2) — 0.2 0.12 0.1
* at 240 GHz
72
4 RF-MEMS Design Examples
4.1 Introduction
In this chapter, RF-MEMS design examples in both 0.13 µm and 0.25 µm SiGe Bi-
CMOS technologies are described. In Section 4.2, firstly a D-band SPST switch in
SG13 BiCMOS technology is presented which is used as a key element to design a
D-band SPDT switch in the same technology. The developed D-band RF-MEMS ba-
sed SPDT switch is presented with its EM modeling and S-parameter measurement
results. In the next section, Section 4.3, Si cap packaged K-band RF-MEMS test vehi-
cles for space applications are presented in SG25 BiCMOS technology. Firstly the
C-V and S-parameter measurement results of the two test vehicles; RF-MEMS based
SPST and SPDT switches are presented for both before and after Si cap packaging
cases. Later on, the yield analyses of all the test vehicles; RF-MEMS based SPST,
SPDT switches and a on-chip CP integrated with the RF-MEMS based SPST switch
are given. Lastly, the effect of Si cap packaging is studied in more detail with 3D
EM simulations in HFSS which shows the RF performance differences with different
resistivity values of the Si caps.
4.2 D-Band RF-MEMS SPDT Switch
SPDT switches are essential components for many different RF and mm-wave system
applications; such as in 94 GHz passive imaging systems as Dicke SPDT switch [90, 91]
and in 120 – 140 GHz radar and sensor transceivers as transmit/receive (T/R) switch.
In most of these applications, SPDT is the component coming right after the antenna;
thus adding its noise figure on top of the overall noise figure of the system. Therefore,
a very low loss is required by the SPDTs in order to achieve a high overall system
performance.
73
4 RF-MEMS Design Examples
c
2016 IEEE
(a)
c
2016 IEEE
(b)
Figure 4.1: Fabricated D-band RF-MEMS SPST switch with the TM2 plate and its
top generic view with hidden TM2 plate [86].
Key element of the RF-MEMS SPDT Switch: RF-MEMS SPST Switch
A D-band RF-MEMS SPST switch is used as the key element of the RF-MEMS SPDT
switch. The main differences between this key element RF-MEMS SPST switch and
the developed switch in Section 3.2 are the design of the RF pads and the encapsulation.
Similar to the key-element RF-MEMS SPST switch, the designed RF-MEMS SPDT
switch is also fabricated as shown in Section 2.3. Both switches are not encapsulated
but include the TM2 plate on top for their protection. Additionally, both of the
switches include different RF pads compared to the RF pads of the presented WLE
switch in Section 3.2. Fig. 2.11 (b) shows the process integration scheme of the RF-
MEMS SPST and SPDT switches in the 0.13 µm SiGe BiCMOS technology without
WLE and the micrograph of the fabricated RF-MEMS SPST switch is shown in Fig. 4.1
(a). The top view of the embedded RF-MEMS SPST switch with hidden TM2 plate
is shown in Fig. 4.1 (b).
The two port S-parameter measurements of the fabricated RF-MEMS SPST switch
are performed from 110 to 170 GHz with the measurement setup explained in Section
3.2.3. Although the fabricated RF-MEMS SPST switch has pull-in after 45 V; 60 V
is applied for down-state to obtain a stable down-state capacitance. The measured
and simulated S-parameter results of the SPST switch are given in Fig. 4.2, both for
the up and down states. The results are in very good agreement in the 110 – 170 GHz
frequency band with respect to the EM simulations, performed by the 3D FEM solver.
After the S-parameter measurements of the RF-MEMS switch, the pad capacitances
74
4 RF-MEMS Design Examples
c
2016 IEEE
(a)
c
2016 IEEE
(b)
Figure 4.2: Comparison of the measured open-deembedded S-parameters with the EM
simulation results for the insertion loss in the up-state (a) and the isolation
in the down-state (b) of the D-band RF-MEMS SPST switch [86].
are removed by the open-deembedding and an insertion loss of 0.46 dB and an isolation
of 36 dB at 140 GHz are extracted.
RF-MEMS switches can be used in different mm-wave circuits under different RF
and DC bias conditions. During the design phase of the circuits, the bias conditions
should already be considered as RF-MEMS switch has connections to the other com-
ponents of the circuit. Due to non-accurate bias conditions on the RF-MEMS switch
two different failures can occur; namely the self-actuation and latching.
75
4 RF-MEMS Design Examples
Self actuation: Self actuation occurs in a high DC or RF level difference applied be-
tween the RF- SL and membrane of the RF-MEMS switch. The applied bias difference
generate a force to pull the membrane down without any additional voltage applied
to the electrodes. The self-actuation voltage of the D-band is measured as 21 V. This
bias difference between the RF- SL and membrane is quite high as its compared to the
typical biasing conditions of mm-wave circuits. It should be also noted that typical RF
power levels in mm-wave circuits are also much less than the power values which can
cause the self-actuation. Indeed, self-actuation does not create a significant problem
for the circuits that include the presented D-band RF-MEMS switch.
Latching: Latching also occurs in a high DC or RF level difference applied between
the RF- SL and membrane of the RF-MEMS switch which generates a force to keep
the membrane in down state. To investigate the latching of the D-band RF-MEMS
switch, the switch is actuated by applying 60 V to the electrodes and the voltage
applied to the RF- SL is increased simultaneously to catch the latching voltage. The
latching voltage of the D-band switch is measured as 3.9 V when 60 V is applied to the
electrodes. Shortly, the DC potential difference between RF- SL and the membrane
should be below 3.9 V for the safe operation in mm-wave circuits.
4.2.1 EM modeling
By using the above presented RF-MEMS SPST switch, a 140 GHz center frequency D-
band SPDT switch design is targeted. The RF-MEMS based SPDT switch is designed
with a tee junction that is connected to the two RF-MEMS SPST switches with each
side λ/4 microstrip lines. The schematic of the designed SPDT switch and its EM
model in HFSS are shown in Fig. 4.3 and Fig. 4.4, respectively.
The designed SPDT switch behaves like a thru between Port1 and Port2 (transmis-
sion mode) when the switch-1 (SW#1) is open (up-state) and the switch-2 (SW#2)
shorted (down-state). The SW#2 together with the transmission line (T-line) in the
right side branch provide an open circuit since the λ/4 T-lines is capacitively shorted
to ground by the SW#2. Although the right side branch is supposed to provide open
circuit, due to the non-optimized T-line lengths, a specific loss always occurs in this
branch. In the second state, the switch has a high attenuation between Port1 and
Port2 (isolation mode) when the SW#1 is in short (down-state) and the SW#2 is in
open (up-state). In this state, the Port1 and the Port2 are isolated by the help of λ/4
T-lines and the SW#1 in the left side branch.
76
4 RF-MEMS Design Examples
c
2016 IEEE
Figure 4.5: Micrograph of the D-band RF-MEMS based SPDT switch [86].
Since the comparison of the S-parameter measurements and the EM simulations of
the RF-MEMS SPST switch has given a very good match, same distances in SPST
switch’s EM model are also given to the SPDT switch’s EM model between the TM1
membrane and M5 RF- SL. For the EM simulations of the insertion loss between Port1
and Port2 of the SPDT switch, the SW#1 SPST switch’s TM1 membrane is kept in
up-state position while SW#2 SPST switch’s TM1 membrane is shifted into the down-
state position. On the other hand for the EM simulations of the isolation between Port1
and Port2 of the SPDT switch, the SW#1 SPST switch’s TM1 membrane is shifted
into the down-state position while SW#2 SPST switch’s TM1 membrane is kept in
up-state position. During the development of the RF-MEMS based SPDT switch, the
tee junction is fine tuned with comparing different simulation results to provide the
minimum insertion loss and maximum isolation at 140 GHz targetted frequency. After
the RF optimizations of the RF-MEMS based SPDT switch in HFSS, the developed
SPDT switch is fabricated in IHP’s SG13 BEOL, excluding the WLE process. Fig. 4.5
shows the micrograph of the fabricated D-band RF-MEMS based SPDT switch.
4.2.2 Experimental results
The RF measurements of the fabricated RF-MEMS based SPDT switch were done by
only two port S-parameter measurements because of the lack of ground signal ground
signal ground (GSGSG) probes in 110 – 170 GHz frequency band. With the actuation
of the SW#2 membrane (Fig. 4.5) with 60 V while SW#1 was in off-state (no voltage
at the electrode), the measurement results have shown an insertion loss of 1.42 dB at
78
4 RF-MEMS Design Examples
140 GHz. In the case of an actuated SW#1 (60 V actuation voltage) while SW#2 is in
off-state (no voltage at the electrode) an isolation of 54.5 dB is measured at 140 GHz.
Fig. 4.6 shows the performance of the RF-MEMS based SPDT switch for the measured
and the EM simulated cases. For the 2-port simulations, the non-contacted third port
is also left as open circuit, like in the case of the measurements. To check the effect of
the not connected third port, an additional setup is created where all the three ports
are terminated with 50 Ω. The 3-port EM simulations show that the insertion loss and
the isolation do not change significantly in the interested frequency band (Fig. 4.6).
Additional to the D-band simulations, Fig. 4.7 shows the comparison of the 2-port and
3-port EM simulation results in a wider frequency band - from DC to 170 GHz. This
comparison has shown that 3-port EM simulations give similar results with 2-port
simulation in D-band for insertion loss and isolation graphs, however show differences
below 110 GHz in insertion loss and isolation. Since the length of the λ/4 T-lines is
optimized for the targetted 140 GHz, for higher or lower frequencies the length of the
T-lines differs from λ/4. This causes differences in below 110 GHz frequencies due to
the traveling of the RF-signal to both of the T-lines.
In literature, although there are no RF-MEMS based SPDT switches beyond 110 GHz,
there are many studies on transistor based SPDT switches. Unlike the transistor ba-
sed SPDTs, the RF-MEMS based SPDTs have much lower power consumption, can
handle higher RF power and provide excellent linearity. Table 4.1 shows the measured
performance of the designed RF-MEMS based SPDT switch in comparison to the ot-
her published millimeter-wave SPDT switches, based on the BiCMOS and SOI-CMOS
based technologies. With a state of the art 0.13 µm SiGe BiCMOS technology, the de-
signed SiGe HBT based SPDT in [92] has an insertion loss of 2.6 – 3 dB and a maximum
isolation of 29 dB between 96 – 163 GHz. In [93], a single-shunt SPDT switch based
on a 32 nm SOI CMOS technology with an insertion loss of 2.6 – 4 dB and an isolation
of 22 dB in 110 – 170 GHz frequency band is published. Another doubled-shunt SPDT
switch is demonstrated based on 45 nm SOI CMOS technology with insertion loss of
3 – 4.5 dB and isolation of 20 – 30 dB in 140 – 220 GHz frequency band. To the best of
my knowledge, the results achieved in this study are the lowest insertion loss and the
highest isolation of a SPDT reported in D-band [86].
79
4 RF-MEMS Design Examples
Table 4.1: Measured Performance Comparison of mm-wave SPDT Switches [86].
[92] [93] [88] This Work [86]
Technology 0.13 µm SiGe
BiCMOS
32 nm CMOS
SOI
45 nm CMOS
SOI
0.13 µm SiGe
BiCMOS
Topology Double Shunt
Sat. HBT
Double Shunt
FET
Double Shunt
FET RF-MEMS
Frequency (GHz) 96 – 163 110 – 170 140 – 220 110 – 170
IL (dB) 2.6 – 3 2.6 – 4 3 – 4.5 1.23*– 1.86
ISO (dB) 23.5 – 29 22 20 – 30 18.25 – 54.5
Switching Time 0.4 ns (simulated) — — <10 µs
c
2016 IEEE * at 157 GHz
82
4 RF-MEMS Design Examples
c
2017 IEEE
Figure 4.8: The microscope images of a designed, fabricated and packaged TCV (left),
DEC (middle) and RIC (right) [71].
4.3 K-band RF-MEMS Test Vehicles for space applications
In the previous chapters, it has already been pointed to the importance of packaging
for commercialization. In this section, RF-MEMS test vehicles for K-band space ap-
plications are demonstrated. The demonstrated test vehicles are embedded in IHP’s
0.25 µm SiGe BiCMOS process and packaged with Si cap wafer-level packaging in
Fraunhofer IZM, Berlin [69–71]. As required by the European Space Components
Coordination (ESCC) specification (n◦2269010), three RF-MEMS test vehicles for
space applications have been designed, fabricated and packaged (Fig. 4.8); namely the
technology characterization vehicle (TCV), the dynamic evaluation circuit (DEC) and
the representative integrated circuit (RIC).
•TCV is an RF-MEMS SPST switch.
•DEC is an RF-MEMS SPDT switch including the standard building block, TCV.
•RIC is an RF-MEMS SPST switch integrated together with an on-chip high-
voltage CP.
All the fabricated RF-MEMS test vehicles are packaged by the Si cap wafer-to-wafer
packaging technique explained in Section 2.2.2.
83
4 RF-MEMS Design Examples
4.3.1 Experimental results
The designed and fabricated TCV, DEC and RIC are electrically characterized before
and after the WLP to investigate the effect of the packages and the packaging processes
on the RF performances and the mechanical behaviors (pull-in voltages). Since the
performance of the RIC is identical to the TCV, the electrical characterization of the
RIC only focuses on the DC measurements of the integrated CP.
For the capacitance measurements [79] of the TCV and DEC, the RF-MEMS swit-
ches are actuated between -45 V to +45 V with 5 V steps and the contact capacitances
are extracted using the high frequency impedance analyzer (Agilent E4991). The
two port S-parameters of the TCV and DEC are measured on wafer-level from DC to
67 GHz. For the measurement, a 67 GHz Agilent PNA E8361A is used. For the calibra-
tion, the impedance standard substrate is placed on an auxiliary chuck and a full two
port short open load thru (SOLT) calibration is performed. A 200 V Agilent Source
Measurement Module 41420A is used to actuate the membrane of the RF-MEMS test
vehicles.
Technology Characterization Vehicle (TCV: an RF-MEMS SPST switch)
Microscope images of the TCV before packaging are shown in Fig. 2.7. After the com-
parison of the C-V curves of the TCV before and after packaging (Fig. 4.9), approx-
imately 35 fF increase in the capacitance values is observed which is correlated with
the additional parasitic capacitance between the RF- SL and the Si cap. It is worth to
mention that the pull-in voltage is stable and not influenced by the packaging process
of the TCVs proving the mechanical stability of the membrane.
The comparison result of the measured S-parameters of a TCV before and after the
packaging process is shown in Fig. 4.10. The S-parameter measurements have shown
that the loss of the TCV has slightly increased after the packaging. The packaged
TCVs have shown an ∼0.2 dB increase of the insertion loss in off-state until 30 GHz and
a 1 dB decrease of the isolation in the resonance frequency in on-state after packaging.
On the other hand, the matching (S11) is still better than 10 dB up to 67 GHz. The
reason for the increase of the loss is the capacitive coupling of the RF-signal to the
package. It should be noted that all the given results are the not de-embedded results
which include the additional loss due to the RF-pads, considering the real use scenario
with bond-wires.
84
4 RF-MEMS Design Examples
c
2017 IEEE
Figure 4.9: The measured C-V graphs of a TCV before (blue) and after (red) the
packaging [70, 71].
Dynamic Evaluation Circuit (DEC: an RF-MEMS SPDT switch)
As expected from the TCV, the capacitance values of the DEC show a significant
increase after the packaging process (Fig. 4.11). The reason of the ∼300 fF capaci-
tance increase is again the parasitic capacitance between the RF- SL and the Si cap.
Microscope image of the DEC before packaging is shown in Fig. 4.12. Since the RIC
includes significantly longer RF- SLs compared to the TCV, the coupling to the Si cap
is much stronger. The pull-in voltage of the DEC is stable as in the TCV and the
mechanical behavior of the membrane is not influenced by the packaging process.
In Fig. 4.13, the comparison results of the measured S-parameter results of a DEC
before and after the packaging process are shown. The comparison shows clearly the
effect of the package on the measured DEC with an increase of ∼1.5 dB loss in the
up-state and a 1.5 dB decrease of isolation in the down-state at 23 GHz.
85
4 RF-MEMS Design Examples
Figure 4.14: Microscope image of a RIC before the Si cap packaging.
Representative Integrated Circuit (RIC: an RF-MEMS SPST switch integrated
with an on-chip high-voltage CP)
The RIC includes an RF-MEMS SPST switch integrated with an on-chip high-voltage
CP in the 0.25 µm SiGe BiCMOS technology. The necessary actuation voltage for the
RICs SPST RF-MEMS switch is generated by the on-chip CP. A CP is designed in
IHP’s 0.13 µm SiGe BiCMOS technology [94], using stacked BEOL capacitors to over-
come the output voltage limitations of charge pumps with standard MIM capacitors.
Microscope image of the RIC including the RF-MEMS SPST switch integrated with
the on-chip high-voltage CP is shown in Fig. 4.14, before the Si cap packaging of the
RF-MEMS SPST switch. Fig. 4.15 shows the microscope image of the RICs during
electrical characterization after the Si cap packaging of the RF-MEMS SPST switch.
Electrical characterization of the CP has shown that the necessary high voltage (45 V)
for the actuation of the RF-MEMS SPST can be successfully generated on-chip. In
order to understand the effect of the control voltage (Vctrl) on the output voltage
(Vout), Vctrl is swept between 0 to 1.1 V. Fig. 4.16 shows the Vctrl versus Vout graph
of the CP when the supply voltage (Vdd) is fixed at 3 V and input voltage for first CP
stage (Vin) is fixed at 0 V. It shows that 45 V Vout is possible with minimum 1.1 V
Vctrl.
89
4 RF-MEMS Design Examples
c
2017 IEEE
(a)
c
2017 IEEE
(b)
Figure 4.17: The 3D EM simulation model of the inductive loaded M3 membrane RF-
MEMS switch, (a) without Si cap and (b) with Si cap [70].
4.3.2 Effect of silicon cap packaging
In this section, the effect of Si cap packaging on the SG25 BiCMOS embedded RF-
MEMS switch performance is studied [70]. To show the effect of the Si caps on the
RF performance of the RF-MEMS switches, two EM models are built up in ANSYS
HFSS and shown in Fig. 4.17.
The first model of the RF-MEMS switch is created without the Si cap; instead the
second model is created with the Si cap. The model including the Si cap is simula-
ted with a parametric sweep for different Si resistivity values: 10 kΩ·cm (0.01 s/m),
50 Ω·cm (2 s/m) and 1 Ω·cm (100 s/m). Fig. 4.18 shows the EM simulated insertion
loss results of the RF-MEMS switch without the Si cap and with the different resisti-
91
4 RF-MEMS Design Examples
c
2017 IEEE
Figure 4.18: The extracted insertion loss results of the EM simulated RF-MEMS
switch without Si cap and with Si caps of different resistivity values
(10 kΩ·cm, 50 Ω·cm, and 1 Ω·cm) [70].
Figure 4.19: The extracted insertion loss comparison of the measured and simulated
RF-MEMS switches in off-state without Si cap and with 1 Ω·cm resistivity
Si cap.
vity Si caps. The IL curves are similar in case of without the Si cap and with the Si
caps of 10 kΩ·cm and 50 Ω·cm resistivity. However, the IL degrades with the usage of
a 1 Ω·cm resistivity Si cap.
The insertion loss comparison of the RF-MEMS switch before and after Si cap; and
the EM simulations without and with the 1 Ω·cm Si cap shows very good agreement
(Fig. 4.19). Indeed, the EM simulations clearly show that the ∼0.2 dB increase of the
loss occurs due to the low resistive Si cap (1 Ω·cm). It is worth taking into consideration
that the RF performance degradation can be minimized with a medium resistive Si
cap (50 Ω·cm) which makes the packaging cost-effective compared to the high resistive
Si substrates.
92
4 RF-MEMS Design Examples
4.3.3 Yield analysis
In recent years, numerous outstanding RF performance results [95–97] have been achie-
ved with RF-MEMS switches. However, the repeatability and the yield of the RF-
MEMS processes are still in the main challenges before their commercialization.
With respect to the C-V measurement results of the TCVs and DECs, three different
regions are defined on wafer-level to show the process uniformity. The divisions into
regions are done with respect to the mechanical functionalities and the pull-in voltages
of the test vehicles over the 8-inch wafer. Additionally, the CPoutput voltage of the
RIC over an 8-inch wafer is also given.
Technology Characterization Vehicle (TCV: an RF-MEMS SPST switch)
Although 83.5 % of the measured TCVs are functional on wafer-level (Fig. 4.20), the
functional TCVs are also divided into two different regions with respect to their pull-in
voltages. The TCVs, for the further reliability tests, are selected from the safe opera-
tion region because of their higher pull-in voltages (stiffer membrane) [98]. The three
wafer regions for the TCVs (Fig. 4.21) are decided as the following: (1) not functional
region with less than 230 fF on-state capacitance (Con), (2) the safe operation region
between 230 and 320 fF Con, (3) the low pull-in voltage region with more than 320 fF
Con at 45 V. The yield of the TCVs is distributed into these three wafer regions as
63.5 % functional from the safe operation region, 20 % functional but with low pull-in
voltage and 16.5 % not functional region. The different performances of the TCVs are
due to the non-uniform deposition and etching processes over the wafer during their
fabrication.
Dynamic Evaluation Circuit (DEC: an RF-MEMS SPDT switch)
The 68 % of the measured DECs are functional on wafer-level (Fig. 4.22). Similar as
in the TCV case, the functional DECs are divided into two different regions with re-
spect to their pull-in voltages. The selected DECs, for the further reliability tests,
are also taken from the safe operation region with the higher pull-in voltages (stiffer
membrane). The three wafer regions for the DECs (Fig. 4.23) are defined as the fol-
lowing: (1) the not functional region with less than 1010 fF Con, (2) the safe operation
region between 1010 and 1100 fF Con, (3) the low pull-in voltage region with more
than 1100 fF Con at 45 V. The yield of the DECs is distributed into these three wafer
regions as 46 % functional from the safe operation region, 22 % functional but with
low pull-in voltage and 32 % not functional region.
93
4 RF-MEMS Design Examples
c
2017 IEEE
Figure 4.24: The wafer-level measured Vout values of RIC’s charge pump, showing the
yield and uniformity of the BiCMOS process [71].
Charge pump (CP) of the Representative Integrated Circuit (RIC)
The on-wafer yield analysis of the integrated CP Vout is performed applying a Vdd of
3 V and a Vctrl of 1.1 V. The wafer-level measured Vout values are shown in Fig. 4.24.
The CPs of the RICs are 100 % functional on wafer-level. The wafer map shows
that the 13 % of the chips have generated between 40 V and 44 V Vout. The rest of
the wafer, 87 % of the chips has generated more than 44 V Vout which is sufficient
actuation voltage for the RF-MEMS test vehicles.
96
4 RF-MEMS Design Examples
4.4 Conclusion
Section 4.2 section has presented an RF-MEMS based SPDT switch embedded in IHP’s
0.13 µm SiGe BiCMOS process technology. A 140 GHz targeted operating frequency
RF-MEMS based SPDT switch, with its measured 1.42 dB insertion loss and 54.5 dB
isolation at 140 GHz, has been successfully demonstrated in 110 – 170 GHz frequency
band. To the best of my knowledge, the demonstrated SPDT switch is the first RF-
MEMS based SPDT switch presented in D-band and provides the lowest insertion
loss and the highest isolation values compared to the field-effect transistor (FET) and
HBT based SPDTs.
Section 4.3 has presented the performance and yield analyses of packaged RF-MEMS
test vehicles; SPST, SPDT and high-voltage CP integrated with SPST switch that are
embedded in IHP’s 0.25 µm SiGe BiCMOS process technology for space applications at
K-Band. Since the S-parameter and C-V measurement results of the SPST and SPDT
switches have shown changes after packaging of the test vehicles, the influence of the
Si cap packaging is investigated more in detail by extending the EM simulations. As
a result of this investigation, the S-parameter measurements and the EM simulations
have shown a similar increase of ∼0.2 dB dB for the loss in SPST switch which is
introduced by the low resistive Si cap packages, i.e. 1 Ω·cm. To conclude, the presented
wafer-level packaging technique has shown a good compromise between low cost and
good RF performance for the RF-MEMS switches at K-band.
In Chapter 4, RF-MEMS design examples from both 0.13 µm and 0.25 µm SiGe Bi-
CMOS technologies have been given. In 0.13 µm SiGe BiCMOS technology, a D-band
SPST has been firstly developed to be used as the key element during the development
of a D-band SPDT switch. Both of the developed switches have shown state-of-the-art
RF performances and their EM simulation results have matched very good with the
S-parameter measurement results. Later on the design examples of RF-MEMS devices
in 0.25 µm SiGe BiCMOS technology have been presented. The presented devices have
included K-band RF-MEMS based SPST and SPDT switches and an on-chip CP with
an integrated RF-MEMS based SPST switch. The presented switches of the 0.25 µm
SiGe BiCMOS technology have shown more than 68 % yield in terms of functionality
on 200 mm wafers together with their wafer-to-wafer bonded Si cap packages. The
differences in the C-V characteristics between different regions of the wafer can be
explained with process variations overall the wafer. With non-uniform deposition and
etching processes, membranes of the switches have different initial bending due to
stress which causes variations on Coff,Con and pull-in voltages.
97
5 Conclusion and Outlook
This chapter summarizes the main challenges, remarks and achievements of this thesis.
5.1 Technology
Monolithical integration of RF-MEMS switches in SiGe BiCMOS technologies has
the big advantage of stable and well-controlled fabrication environment. However,
this integration also brings challenges during the design and fabrication phase of the
devices. More specifically; the devices should use the available metals and the standard
metal thicknesses of the BEOL modules, should be full-filling the minimum design
rule requirements of the BEOL modules. Additionally, the additional process steps
for the fabrication of the devices should not decrease the yield and performance of the
standard CMOS devices.
5.2 Modeling of RF-MEMS Switches
For the realization of the high performance mm-wave RF-MEMS switches that are
presented in this thesis, it was essential to have accurate electromagnetic models addi-
tional to the necessities of well developed process flows and stable process conditions.
Therefore, the accurate electromagnetic modeling of mm-wave RF-MEMS switches
has been selected as the main focus of this thesis. These accurate EM models made
the realization of RF-MEMS switches with state-of-the-art RF performances possible,
even up to 320 GHz. The research work done in this thesis has been devoted to the
design, electromagnetic modeling and RF optimizations of the mm-wave RF-MEMS
switches that are embedded in IHP’s BiCMOS technologies.
Packaging of the RF-MEMS devices is not only critical due to the development of
an entire process flow but also due to additional RF performances losses that can
be caused by the package. The additional loss can become more significant with the
increasing operating frequencies. Therefore, the electromagnetic models of mm-wave
RF-MEMS devices should include the effect of the packages from the beginning to
98
5 Conclusion and Outlook
foresee the differences in RF performances. Foreseeable effects can be avoided with
different packaging strategies or design/process optimizations. An example for avoi-
ding the RF performance loss due to the packaging can be given as the decision of the
packaging material. The right selected packaging material would avoid the additional
loss in RF performance of the RF-MEMS device, however the cost of the material
should be also taken into account before the final decision about the package.
Development of RF-MEMS devices with very good RF performances take years,
starting from the process development up to electrical characterization. An accurate
model for the EM simulations reduces the final cost and development time of the RF-
MEMS devices since the optimizations can be done with the help of the simulators.
However, a good match between the simulation and measurement results still takes
time until an accurate model is created. Modeling and electrical characterization
supplies information to each other as a feedback loop during the optimizations. With
the back and forth iterations, an accurate model can be created. The developed
feedback methodology can be also used for development of other RF-MEMS devices
in the future.
5.3 RF-MEMS Design Examples
The presented wafer-to-wafer Si cap packaging process for the K-band test vehicles
has shown successful results in terms of high yield and quality of the bonding process.
However, for the Si cap packaging process an additional ∼100 µm distance between
the pads and the switch is required for the bonding. This distance brings extra cost
because of the necessity of a larger area and an additional loss with longer transmis-
sion lines. On the other hand, the presented second wafer-level packaging technique,
thin film wafer-level encapsulation, has shown more promising results for the future
in terms of less number of process steps and lower cost as a result of no additional
wafer for bonding and has required no extra distance between the switch and the pads
which can cause to additional loss and area.
The necessary voltage for the actuation of the BiCMOS embedded RF-MEMS swit-
ches can be provided by on-chip charge pumps [71, 94] with an additional cost due to
increased chip area.
99
5 Conclusion and Outlook
The main achievements of this thesis are as follows:
•For the first time, successful demonstration of an RF-MEMS switch in D-band.
The switch shows maximum isolation of 51.6 dB at 142.8 GHz with the inser-
tion loss of 0.65 dB. With the developed RF-MEMS switch, better than 0.67 dB
insertion loss and more than 16 dB isolation in all D-band are achieved.
•For the first time, successful demonstration of an RF-MEMS switch in J-band.
With the developed RF-MEMS switch, better than 1.22 dB insertion loss and
more than 13.24 dB isolation in all J-band are achieved. At 240 GHz, an insertion
loss of 0.44 dB and an isolation of 24.6 dB are demonstrated.
•For the first time, successful demonstration of an RF-MEMS based SPDT switch
in D-band.
The demonstrated RF-MEMS SPDT switch shows less than 1.86 dB insertion
loss and more than 18.25 dB isolation in all D-band.
•Successful demonstration of high yield RF-MEMS test vehicles with Si cap pack-
aging.
The presented K-band RF-MEMS test vehicles for space applications show high
yield of packaged devices; thus promising for prototyping and volume production.
Despite the state-of-the-art performances of the presented RF-MEMS switches in
0.13 µm SiGe BiCMOS technology, there are additional points that need to be addres-
sed in the near future:
•Yield analysis of the thin-film wafer-level encapsulated RF-MEMS switches has
to be done in order to reveal the process related problems that can occur during
the wafer-level packaging.
•Investigations on the 0.13 µm SiGe BiCMOS technology RF-MEMS switch re-
liability have to be performed in order to reveal the failure modes due to the
design or process integration.
100
Bibliography
References
[1] C. L. Goldsmith, Zhimin Yao, S. Eshelman, and D. Denniston. “Performance of
low-loss RF MEMS capacitive switches”. In: IEEE Microwave and Guided Wave
Letters 8.8 (Aug. 1998), pp. 269–271. issn: 1051-8207. doi:10.1109/75.704410.
[2] Y. Q. Zhu, L. Han, L. F. Wang, J. Y. Tang, and Q. A. Huang. “A Novel Three-
State RF MEMS Switch for Ultrabroadband (DC-40 GHz) Applications”. In:
IEEE Electron Device Letters 34.8 (Aug. 2013), pp. 1062–1064. issn: 0741-3106.
doi:10.1109/LED.2013.2269993.
[3] A. Dec and K. Suyama. “Micromachined electro-mechanically tunable capacitors
and their applications to RF IC’s”. In: IEEE Transactions on Microwave Theory
and Techniques 46.12 (Dec. 1998), pp. 2587–2596. issn: 0018-9480. doi:10.
1109/22.739251.
[4] R. Mahameed and G. M. Rebeiz. “Electrostatic RF MEMS tunable capacitors
with analog tunability and low temperature sensitivity”. In: 2010 IEEE MTT-S
International Microwave Symposium. May 2010, pp. 1254–1257. doi:10.1109/
MWSYM.2010.5514843.
[5] S. Pinel, F. Cros, S. Nuttinck, S. W. Yoon, M. G. Allen, and J. Laskar. “Very
high-Q inductors using RF-MEMS technology for System-On-Package wireless
communication integrated module”. In: IEEE MTT-S International Microwave
Symposium Digest, 2003. Vol. 3. June 2003, 1497–1500 vol.3. doi:10.1109/
MWSYM.2003.1210420.
[6] N. Khalid, J. Singh, H. P. Le, J. Devlin, and Z. Sauli. “A very high Q-factor
inductor using MEMS technology”. In: 2009 Asia Pacific Conference on Postgra-
duate Research in Microelectronics Electronics (PrimeAsia). Jan. 2009, pp. 77–
80. doi:10.1109/PRIMEASIA.2009.5397444.
[7] S. S. Li. “CMOS-MEMS resonators and their applications”. In: 2013 Joint Eu-
ropean Frequency and Time Forum International Frequency Control Symposium
(EFTF/IFC). July 2013, pp. 915–921. doi:10.1109/EFTF-IFC.2013.6702098.
101
References
[8] A. Uranga, J. Verd, and N. Barniol. “CMOS-MEMS resonators: From devices
to applications”. In: Microelectronic Engineering 132 (2015). Micro and Nano-
fabrication Breakthroughs for Electronics, MEMS and Life Sciences, pp. 58–73.
issn: 0167-9317. doi:https://doi.org/10.1016/j.mee.2014.08.015.
[9] G.M. Rebeiz. RF MEMS: Theory, Design, and Technology. Wiley, 2004. isbn:
9780471462880.
[10] M. SeyyedEsfahlan, E. ¨
Ozt¨urk, M. Kaynak, and I. Tekin. “77-GHz Four-Element
Phased-Array Radar Receiver Front End”. In: IEEE Transactions on Compo-
nents, Packaging and Manufacturing Technology 6.8 (Aug. 2016), pp. 1162–1173.
issn: 2156-3950. doi:10.1109/TCPMT.2016.2571742.
[11] Project NANOTEC. 2017-08-11. url:http://project-nanotec.com/.
[12] K. Schmalz, N. Rothbart, P. F. X. Neumaier, J. Borngr¨aber, H. W. H¨ubers, and
D. Kissinger. “Gas Spectroscopy System for Breath Analysis at mm-wave/THz
Using SiGe BiCMOS Circuits”. In: IEEE Transactions on Microwave Theory
and Techniques 65.5 (May 2017), pp. 1807–1818. issn: 0018-9480. doi:10.1109/
TMTT.2017.2650915.
[13] B. Heinemann, H. R¨ucker, R. Barth, F. B´arwolf, J. Drews, G. G. Fischer, A. Fox,
O. Fursenko, T. Grabolla, F. Herzel, J. Katzer, J. Korn, A. Kr´uger, P. Kulse,
T. Lenke, M. Lisker, S. Marschmeyer, A. Scheit, D. Schmidt, J. Schmidt, M. A.
Schubert, A. Trusch, C. Wipf, and D. Wolansky. “SiGe HBT with fT/fmax of 505
GHz/720 GHz”. In: 2016 IEEE International Electron Devices Meeting (IEDM).
Dec. 2016, pp. 3.1.1–3.1.4. doi:10.1109/IEDM.2016.7838335.
[14] G. Q. Zhang. More than Moore. Creating High Value Micro/Nanoelectronics
Systems. New York Springer, 2009. isbn: 9780387755922.
[15] M. Kaynak, M. Wietstruck, C. B. Kaynak, A. G¨oritz, S. Tolunay, and B. Tillack.
“MEMS - BiCMOS monolithic integration”. In: 2014 XXXIth URSI General
Assembly and Scientific Symposium (URSI GASS). Aug. 2014, pp. 1–4. doi:
10.1109/URSIGASS.2014.6929473.
[16] M. Kaynak, M. Wietstruck, C. B. Kaynak, S. Tolunay, A. G¨oritz, and B. Til-
lack. “Modular extension of high performance SiGe BiCMOS technologies - Fol-
lowing the More-than-Moore path”. In: 2015 Asia-Pacific Microwave Conference
(APMC). Vol. 1. Dec. 2015, pp. 1–3. doi:10.1109/APMC.2015.7411742.
102
References
[17] M. Kaynak, M. Wietstruck, A. G¨oritz, S. Tolunay Wipf, M. Inac, B. Cetindo-
gan, C. Wipf, C. B. Kaynak, M. W¨ohrmann, S. Voges, and T. Braun. “0.13 µm
SiGe BiCMOS technology with More-than-Moore modules”. In: 2017 IEEE Bi-
polar/BiCMOS Circuits and Technology Meeting (BCTM). Oct. 2017, pp. 62–65.
doi:10.1109/BCTM.2017.8112912.
[18] R. R. Mansour. “RF MEMS-CMOS Device Integration: An Overview of the
Potential for RF Researchers”. In: IEEE Microwave Magazine 14.1 (Jan. 2013),
pp. 39–56. issn: 1527-3342. doi:10.1109/MMM.2012.2226539.
[19] J. Rizk, G.-L. Tan, J. B. Muldavin, and G. M. Rebeiz. “High-isolation W-band
MEMS switches”. In: IEEE Microwave and Wireless Components Letters 11.1
(Jan. 2001), pp. 10–12. issn: 1531-1309. doi:10.1109/7260.905952.
[20] M. Kaynak, M. Wietstruck, R. Scholz, J. Drews, R. Barth, K.E. Ehwald, A. Fox,
U. Haak, D. Knoll, F. Korndorfer, S. Marschmeyer, K. Schulz, C. Wipf, D. Wol-
ansky, B. Tillack, K. Zoschke, T. Fischer, Y.S. Kim, J.S. Kim, W.-G. Lee, and
J.W. Kim. “BiCMOS embedded RF-MEMS switch for above 90 GHz applicati-
ons using backside integration technique”. In: IEEE International Electron De-
vices Meeting. Dec. 2010, pp. 36.5.1–36.5.4. doi:10.1109/IEDM.2010.5703488.
[21] S. Tolunay Wipf, A. G¨oritz, C. Wipf, M. Wietstruck, A. Burak, E. T¨urkmen,
Y. G¨urb¨uz, and M. Kaynak. “240 GHz RF-MEMS switch in a 0.13 µm SiGe
BiCMOS Technology”. In: 2017 IEEE Bipolar/BiCMOS Circuits and Technology
Meeting (BCTM). Oct. 2017, pp. 54–57. doi:10.1109/BCTM.2017.8112910.
[22] M. Kaynak, K. E. Ehwald, J. Drews, R. Scholz, F. Kornd¨orfer, D. Knoll, B.
Tillack, R. Barth, M. Birkholz, K. Schulz, Y. M. Sun, D. Wolansky, S. Leidich,
S. Kurth, and Y. G¨urb¨uz. “BEOL embedded RF-MEMS switch for mm-wave
applications”. In: 2009 IEEE International Electron Devices Meeting (IEDM).
Dec. 2009, pp. 1–4. doi:10.1109/IEDM.2009.5424219.
[23] M. Kaynak, F. Kornd¨orfer, M. Wietstruck, D. Knoll, R. Scholz, C. Wipf, C.
Krause, and B. Tillack. “Robustness and reliability of BiCMOS embedded RF-
MEMS switch”. In: 2011 IEEE 11th Topical Meeting on Silicon Monolithic In-
tegrated Circuits in RF Systems. Jan. 2011, pp. 177–180. doi:10.1109/SIRF.
2011.5719336.
[24] H. R¨ucker, B. Heinemann, and A. Fox. “Half-Terahertz SiGe BiCMOS techno-
logy”. In: 2012 IEEE 12th Topical Meeting on Silicon Monolithic Integrated Cir-
cuits in RF Systems. Jan. 2012, pp. 133–136. doi:10.1109/SiRF.2012.6160164.
103
References
[25] V. Valenta, H. Schumacher, S. Tolunay Wipf, M. Wietstruck, A. G¨oritz, M.
Kaynak, and W. Winkler. “Single-chip transmit-receive module with a fully in-
tegrated differential RF-MEMS antenna switch and a high-voltage generator for
F-band radars”. In: IEEE Bipolar/BiCMOS Circuits and Technology Meeting.
Oct. 2015, pp. 40–43. doi:10.1109/BCTM.2015.7340564.
[26] T. Chaloun, W. Menzel, F. Tabarani, T. Purtova, H. Schumacher, M. Kaynak,
Q. Luo, S. Gao, R. Starec, and V. Ziegler. “Wide-angle scanning active trans-
mit/receive reflectarray”. In: IET Microwaves, Antennas Propagation 8.11 (Aug.
2014), pp. 811–818. issn: 1751-8725. doi:10.1049/iet-map.2013.0704.
[27] T. Chaloun, F. Tabarani, S. Tolunay Wipf, M. Kaynak, H. Schumacher, and W.
Menzel. “A modular phased array transceiver with RF-MEMS SPDT switches
in a 0.25 µm SiGe BiCMOS technology”. In: 12th European Conference on An-
tennas and Propagation (EuCAP 2018). Apr. 2018, pp. 1–5. doi:10.1049/cp.
2018.0615.
[28] H.A.C. Tilmans, J. De Coster, P. Helin, V. Cherman, A. Jourdain, P. De Moor,
B. Vandevelde, N.P. Pham, J. Zekry, A. Witvrouw, and I. De Wolf. “MEMS
packaging and reliability: An undividable couple”. In: Microelectronics Reliability
52.9 (2012). Special issue 23rd European Symposium on the reliability of electron
devices, failure pyhsics and analysis, pp. 2228–2234. issn: 0026-2714. doi:https:
//doi.org/10.1016/j.microrel.2012.06.029.
[29] G. M. Rebeiz and J. B. Muldavin. “RF MEMS switches and switch circuits”.
In: IEEE Microwave Magazine 2.4 (Dec. 2001), pp. 59–71. issn: 1527-3342. doi:
10.1109/6668.969936.
[30] S. Fouladi and R. R. Mansour. “Capacitive RF MEMS Switches Fabricated in
Standard 0.35 µm CMOS Technology”. In: IEEE Transactions on Microwave
Theory and Techniques 58.2 (Feb. 2010), pp. 478–486. issn: 0018-9480. doi:
10.1109/TMTT.2009.2038446.
[31] Y. Feng and N. S. Barker. “Electrostatic RF MEMS switch working on 500-750
GHz”. In: 2016 IEEE 17th Annual Wireless and Microwave Technology Confe-
rence (WAMICON). Apr. 2016, pp. 1–5. doi:10.1109/WAMICON.2016.7483849.
[32] H.C. Lee, J.Y. Park, and J.U. Bu. “Piezoelectrically actuated RF MEMS DC
contact switches with low voltage operation”. In: IEEE Microwave and Wireless
Components Letters 15.4 (Apr. 2005), pp. 202–204. issn: 1531-1309. doi:10.
1109/LMWC.2005.845689.
104
References
[33] R. G. Polcawich, D. Judy, J. S. Pulskamp, S. Trolier-McKinstry, and M. Dubey.
“Advances in Piezoelectrically Actuated RF MEMS Switches and Phase Shif-
ters”. In: 2007 IEEE/MTT-S International Microwave Symposium. June 2007,
pp. 2083–2086. doi:10.1109/MWSYM.2007.380297.
[34] D. J. Chung, R. G. Polcawich, D. Judy, J. Pulskamp, and J. Papapolymerou.
“A SP2T and a SP4T switch using low loss piezoelectric MEMS”. In: 2008 IEEE
MTT-S International Microwave Symposium Digest. June 2008, pp. 21–24. doi:
10.1109/MWSYM.2008.4633093.
[35] D. Girbau, L. Pradell, A. Lazaro, and A. Nebot. “Electrothermally Actuated
RF MEMS Switches Suspended on a Low-Resistivity Substrate”. In: Journal
of Microelectromechanical Systems 16.5 (Oct. 2007), pp. 1061–1070. issn: 1057-
7157. doi:10.1109/JMEMS.2007.904744.
[36] M. Bakri-Kassem and R. R. Mansour. “High Power Latching RF MEMS Swit-
ches”. In: IEEE Transactions on Microwave Theory and Techniques 63.1 (Jan.
2015), pp. 222–232. issn: 0018-9480. doi:10.1109/TMTT.2014.2376932.
[37] Y. Zhang, G. Ding, X. Shun, X. Li, and B. Cai. “Design and analysis of the mi-
cromechanical structure for an electromagnetic bistable RF MEMS switch”. In:
2005 Asia Pacific Microwave Conference Proceedings. Vol. 1. Dec. 2005, pp. 4–.
doi:10.1109/APMC.2005.1606197.
[38] S. Lucyszyn. Advanced RF MEMS. 1st ed. New York, NY, USA: Cambridge
University Press, 2010. isbn: 9780521897716.
[39] J. Iannacci. “RF-MEMS: an enabling technology for modern wireless systems
bearing a market potential still not fully displayed”. In: Microsystem Technolo-
gies 21.10 (Oct. 2015), pp. 2039–2052. issn: 1432-1858. doi:10.1007/s00542-
015-2665-6.
[40] J. Bouchaud and H. Wicht. “RF MEMS: status of the industry and roadmaps”.
In: 2005 IEEE Radio Frequency integrated Circuits (RFIC) Symposium - Digest
of Papers. June 2005, pp. 379–384. doi:10.1109/RFIC.2005.1489818.
[41] J. Bouchaud and B. Knoblich. RF MEMS switches deliver on early promise.
2018-07-18. 2007. url:http://www.memsjournal.com/2007/10/rf- mems-
switche.html.
[42] MEMS Journal. Cell phone antenna troubles RF MEMS come to the rescue.
2018-07-18. 2010. url:http://www.memsjournal.com/2010/09/cell-phone-
antenna-troubles-rf-mems-come-to-the-rescue.html.
105
References
[43] D. McGrath. Teardown finds RF MEMS in Samsung handset. 2018-07-18. 2012.
url:https://www.eetimes.com/document.asp?doc_id=1260917.
[44] J.J. DeLisle. The Future Of Connectivity: Mobile & Automobiles. 2018-07-18.
2014. url:http://www.mwrf.com/systems/future-connectivity-mobile-
automobiles.
[45] Gartner. Gartner Hype Cycle. 2018-03-02. url:http://www.gartner.com/
technology/research/methodologies/hype-cycle.jsp.
[46] J. Iannacci. “Surfing the hype curve of RF-MEMS passive components: towards
the 5th generation (5G) of mobile networks”. In: Microsystem Technologies (Jan.
2018). issn: 1432-1858. doi:10.1007/s00542-018-3718-4.
[47] R. Allan. RF MEMS Switches Are Primed For Mass-Market Applications. 2018-
04-18. 2013. url:http://www.mwrf .com /active - components/ rf- mems -
switches-are-primed-mass-market-applications.
[48] Cavendish Kinetics. 2018-03-19. url:http://www.cavendish-kinetics.com.
[49] J. R. De Luis, A. Morris, Q. Gu, and F. de Flaviis. “Tunable Duplexing An-
tenna System for Wireless Transceivers”. In: IEEE Transactions on Antennas
and Propagation 60.11 (Nov. 2012), pp. 5484–5487. issn: 0018-926X. doi:10.
1109/TAP.2012.2207869.
[50] J. Iannacci. “RF-MEMS technology: An enabling solution in the transition from
4G-LTE to 5G mobile applications”. In: 2017 IEEE SENSORS. Oct. 2017, pp. 1–
3. doi:10.1109/ICSENS.2017.8234190.
[51] Y. K. Park and Y. Sung. “A Reconfigurable Antenna for Quad-Band Mobile
Handset Applications”. In: IEEE Transactions on Antennas and Propagation
60.6 (June 2012), pp. 3003–3006. issn: 0018-926X.
[52] N. Behdad and K. Sarabandi. “A varactor-tuned dual-band slot antenna”. In:
IEEE Transactions on Antennas and Propagation 54.2 (Feb. 2006), pp. 401–408.
issn: 0018-926X. doi:10.1109/TAP.2005.863373.
[53] V. A. Nguyen, R. A. Bhatti, and S. O. Park. “A Simple PIFA-Based Tunable
Internal Antenna for Personal Communication Handsets”. In: IEEE Antennas
and Wireless Propagation Letters 7 (2008), pp. 130–133. issn: 1536-1225. doi:
10.1109/LAWP.2008.921332.
[54] K. R. Boyle and P. G. Steeneken. “A Five-Band Reconfigurable PIFA for Mo-
bile Phones”. In: IEEE Transactions on Antennas and Propagation 55.11 (Nov.
2007), pp. 3300–3309. issn: 0018-926X. doi:10.1109/TAP.2007.908822.
106
References
[55] P. Bahramzy, O. Jagielski, S. Svendsen, and G. F. Pedersen. “Compact Agile
Antenna Concept Utilizing Reconfigurable Front End for Wireless Communica-
tions”. In: IEEE Transactions on Antennas and Propagation 62.9 (Sept. 2014),
pp. 4554–4563. issn: 0018-926X. doi:10.1109/TAP.2014.2325946.
[56] P. Bahramzy, O. Jagielski, S. Svendsen, and G. F. Pedersen. “Self-matched high-
Q reconfigurable antenna concept for mobile terminals”. In: IET Science, Mea-
surement Technology 8.6 (2014), pp. 479–486. issn: 1751-8822.
[57] P. Bahramzy, O. Jagielski, S. Svendsen, P. Olesen, and G. F. Pedersen. “Aspects
of High-Q Tunable Antennas and Their Deployment for 4G Mobile Communi-
cations [Antenna Applications Corner]”. In: IEEE Antennas and Propagation
Magazine 58.4 (Aug. 2016), pp. 70–81. issn: 1045-9243. doi:10.1109/MAP.
2016.2569434.
[58] WiSpry. 2018-07-16. url:http://wispry.com/.
[59] IHS. IHS iSuppli Teardown Analysis Service Identifies First Use of RF MEMS
Part, Set to be Next Big Thing in Cellphone Radios. 2018-03-19. url:http://
news.ihsmarkit.com/press-release/design-supply-chain/ihs-isuppli-
teardown-analysis-service-identifies-first-use-rf-mems.
[60] A. S. Morris, S. P. Natarajan, Q. Gu, and V. Steel. “Impedance tuners for
handsets utilizing high-volume RF-MEMS”. In: 2012 42nd European Microwave
Conference. Oct. 2012, pp. 193–196. doi:10.23919/EuMC.2012.6459116.
[61] Cavendish Kinetics. News Releases. 2018-03-19. url:http://www.cavendish-
kinetics.com/news/news-releases/.
[62] J. Morra. For Tuning Antennas and Switching Signals, RF MEMS Make Their
Mark. 2018-21-03. 2017. url:http://www.mwrf.com/analog-semiconductors/
tuning-antennas-and-switching-signals-rf-mems-make-their-mark.
[63] Analog Devices. 2018-03-21. url:http://www.analog.com/en/index.html.
[64] E. Carty, P. Fitzgerald, and P. McDaid. The Fundamentals of Analog Devi-
ces‘ Revolutionary MEMS Switch Technology. 2018-03-21. url:http://www.
analog.com/en/technical-articles/fundamentals-adi-revolutionary-
mems-switch-technology.html.
107
References
[65] M. Kaynak, M. Wietstruck, W. Zhang, J. Drews, R. Scholz, D. Knoll, F. Kornd¨orfer,
C. Wipf, K. Schulz, M. Elkhouly, K. Kaletta, M. V. Suchodoletz, K. Zoschke,
M. Wilke, O. Ehrmann, V. M¨uhlhaus, G. Liu, T. Purtova, A. C. Ulusoy, H.
Schumacher, and B. Tillack. “RF-MEMS switch module in a 0.25 µm BiCMOS
technology”. In: IEEE Topical Meeting on Silicon Monolithic Integrated Circuits
in RF Systems. Jan. 2012, pp. 25–28. doi:10.1109/SiRF.2012.6160150.
[66] S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, C. Wipf, and M. Kaynak. “BiCMOS
Embedded RF-MEMS Technologies”. In: MikroSystemTechnik 2017; Congress.
Oct. 2017, pp. 1–3.
[67] IHP. Low-Volume & Multi-Project Service. 2018-04-09. url:https://www.ihp-
microelectronics.com/en/services/ mpw - prototyping / sigec - bicmos -
technologies.html.
[68] IHP. SG25H3 Process Specifications Document. IHP, Frankfurt Oder, 2018.
[69] K. Zoschke, M. Wilke, M. Wegner, K. Kaletta, C. A. Manier, H. Oppermann,
M. Wietstruck, B. Tillack, M. Kaynak, and K. D. Lang. “Capping technologies
for wafer level MEMS packaging based on permanent and temporary wafer bon-
ding”. In: 2014 IEEE 64th Electronic Components and Technology Conference
(ECTC). May 2014, pp. 1204–1211. doi:10.1109/ECTC.2014.6897444.
[70] S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, M. Cirillo, C. Wipf, K. Zoschke, and
M. Kaynak. “Effect of wafer-level silicon cap packaging on BiCMOS embedded
RF-MEMS switch performance”. In: 2017 IMAPS Nordic Conference on Microe-
lectronics Packaging (NordPac). June 2017, pp. 31–34. doi:10.1109/NORDPAC.
2017.7993158.
[71] S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, M. Cirillo, C. Wipf, W. Winkler,
and M. Kaynak. “Packaged BiCMOS embedded RF-MEMS test vehicles for
space applications”. In: 2017 47th European Microwave Conference (EuMC).
Oct. 2017, pp. 320–323. doi:10.23919/EuMC.2017.8230859.
[72] IHP. SG25G2/S Process Specifications Document. IHP, Frankfurt Oder, 2018.
[73] A. Morris and S. Cunningham. “Challenges and Solutions for Cost-effective RF-
MEMS Packaging”. In: 2007 32nd IEEE/CPMT International Electronic Manu-
facturing Technology Symposium. Oct. 2007, pp. 278–285. doi:10.1109/IEMT.
2007.4417077.
108
References
[74] R. N. Candler, W.-T. Park, H. Li, G. Yama, A. Partridge, M. Lutz, and T. W.
Kenny. “Single wafer encapsulation of MEMS devices”. In: IEEE Transactions
on Advanced Packaging 26.3 (Aug. 2003), pp. 227–232. issn: 1521-3323. doi:
10.1109/TADVP.2003.818062.
[75] K. D. Leedy, R. E. Strawser, R. Cortez, and J. L. Ebel. “Thin-Film Encapsulated
RF MEMS Switches”. In: Journal of Microelectromechanical Systems 16.2 (Apr.
2007), pp. 304–309. issn: 1057-7157. doi:10.1109/JMEMS.2007.892915.
[76] F. Barriere, A. Crunteanu, A. Bessaudou, A. Pothier, F. Cosset, D. Mardivirin,
and P. Blondy. “Zero level metal thin film package for RF MEMS”. In: 2010
Topical Meeting on Silicon Monolithic Integrated Circuits in RF Systems (SiRF).
Jan. 2010, pp. 148–151. doi:10.1109/SMIC.2010.5422957.
[77] R. Gaddi, R. Van Kampen, A. Unamuno, V. Joshi, D. Lacey, M. Renault, C.
Smith, R. Knipe, and D. Yost. “MEMS technology integrated in the CMOS back
end”. In: Microelectronics Reliability 50.9 (2010). 21st European Symposium on
the Reliability of Electron Devices, Failure Physics and Analysis, pp. 1593–1598.
issn: 0026-2714. doi:http://dx.doi.org/10.1016/j.microrel.2010.07.113.
[78] S. P. Natarajan, S. J. Cunningham, A. S. Morris, and D. R. Dereus. “CMOS
integrated digital RF MEMS capacitors”. In: 2011 IEEE 11th Topical Meeting
on Silicon Monolithic Integrated Circuits in RF Systems. Jan. 2011, pp. 173–176.
doi:10.1109/SIRF.2011.5719327.
[79] S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, C. Wipf, B. Tillack, A. Mai, and M.
Kaynak. “Thin film wafer level encapsulated RF-MEMS switch for D-band ap-
plications”. In: 2016 46th European Microwave Conference (EuMC). Oct. 2016,
pp. 1381–1384. doi:10.1109/EuMC.2016.7824610.
[80] S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, C. Wipf, B. Tillack, A. Mai, and
M. Kaynak. “Electromagnetic and small-signal modeling of an encapsulated
RF-MEMS switch for D-band applications”. In: International Journal of Mi-
crowave and Wireless Technologies 9.6 (2017), pp. 1271–1278. doi:10.1017/
S1759078717000137.
[81] W. Zhang, M. Kaynak, M. Wietstruck, V. M¨uhlhaus, and B. Tillack. “EM and
lumped-element model of BiCMOS embedded capacitive RF-MEMS switch”. In:
2012 The 7th German Microwave Conference. Mar. 2012, pp. 1–4.
109
References
[82] S. Tolunay, M. Wietstruck, A. G¨oritz, M. Kaynak, and B. Tillack. “Accurate
3D EM modeling of 140 GHz BiCMOS embedded RF-MEMS switch”. In: 14th
Symposium on RF-MEMS and RF-Microsystems (MEMSWAVE). 2013.
[83] M. Kaynak, M. Wietstruck, W. Zhang, J. Drews, R. Barth, D. Knoll, F. Kornd¨orfer,
R. Scholz, K. Schulz, C. Wipf, B. Tillack, K. Kaletta, M. v. Suchodoletz, K.
Zoschke, M. Wilke, O. Ehrmann, A. C. Ulusoy, T. Purtova, G. Liu, and H.
Schumacher. “Packaged BiCMOS embedded RF-MEMS switches with integra-
ted inductive loads”. In: 2012 IEEE/MTT-S International Microwave Sympo-
sium Digest. June 2012, pp. 1–3. doi:10.1109/MWSYM.2012.6259417.
[84] N. Torres Matabosch, F. Coccetti, M. Kaynak, W. Zhang, B. Tillack, R. Plana,
and J. L. Cazaux. “An accurate and versatile equivalent circuit model for RF-
MEMS circuit optimization in BiCMOS technology”. In: 2012 7th European Mi-
crowave Integrated Circuit Conference. Oct. 2012, pp. 143–146.
[85] Polytec. 2018-05-08. url:https : / / www . polytec . com / us / vibrometry /
products/microscope-based-vibrometers/msa-500-micro-system-analyzer/.
[86] S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, C. Wipf, B. Tillack, and M. Kaynak.
“D-Band RF-MEMS SPDT Switch in a 0.13 µm SiGe BiCMOS Technology”. In:
IEEE Microwave and Wireless Components Letters 26.12 (Dec. 2016), pp. 1002–
1004. issn: 1531-1309. doi:10.1109/LMWC.2016.2623245.
[87] C. Nguyen. Radio-Frequency Integrated-Circuit Engineering. Wiley Series in Mi-
crowave and Optical Engineering. Wiley, 2015. isbn: 9781118936481.
[88] M. Uzunkol and G. M. Rebeiz. “140-220 GHz SPST and SPDT Switches in 45
nm CMOS SOI”. In: IEEE Microwave and Wireless Components Letters 22.8
(Aug. 2012), pp. 412–414. issn: 1531-1309. doi:10.1109/LMWC.2012.2206017.
[89] Y. Kim, H. Lee, and S. Jeon. “A 220 - 320 GHz single-pole single-throw switch”.
In: 2016 IEEE International Symposium on Radio-Frequency Integration Techno-
logy (RFIT). Aug. 2016, pp. 1–3. doi:10.1109/RFIT.2016.7578184.
[90] J. W. May and G. M. Rebeiz. “Design and Characterization of W-Band SiGe
RFICs for Passive Millimeter-Wave Imaging”. In: IEEE Transactions on Micro-
wave Theory and Techniques 58.5 (May 2010), pp. 1420–1430. issn: 0018-9480.
doi:10.1109/TMTT.2010.2042857.
110
References
[91] A. Strodl, S. Tolunay, V. Valenta, M. Kaynak, S. Reyaz, R. Johnson, W. Wink-
ler, R. Malmqvist, and H. Schumacher. “RF-MEMS Switched W-Band mm-Wave
Passive Imaging System in a 0.13 µm SiGe BiCMOS Technology”. In: 16th Sym-
posium on RF-MEMS and RF-Microsystems (MEMSWAVE). 2015.
[92] A. C. Ulusoy, P. Song, R. L. Schmid, W. T. Khan, M. Kaynak, B. Tillack, J.
Papapolymerou, and J. D. Cressler. “A Low-Loss and High Isolation D-Band
SPDT Switch Utilizing Deep-Saturated SiGe HBTs”. In: IEEE Microwave and
Wireless Components Letters 24.6 (June 2014), pp. 400–402. issn: 1531-1309.
doi:10.1109/LMWC.2014.2313529.
[93] W. T. Khan, A. C. Ulusoy, R. Schmid, T. Chi, J. D. Cressler, H. Wang, and
J. Papapolymerou. “A D-band (110 to 170 GHz) SPDT switch in 32 nm CMOS
SOI”. In: 2015 IEEE MTT-S International Microwave Symposium. May 2015,
pp. 1–3. doi:10.1109/MWSYM.2015.7167061.
[94] M. Wietstruck, W. Winkler, A. G¨oritz, S. Tolunay Wipf, C. Wipf, D. Schmidt,
A. Mai, and M. Kaynak. “0.13 µm BiCMOS embedded On-Chip High-Voltage
Charge Pump with Stacked BEOL Capacitors for RF-MEMS Applications”. In:
17th Symposium on RF-MEMS and RF-Microsystems (MEMSWAVE). 2016.
[95] J. B. Muldavin and G. M. Rebeiz. “High-isolation CPW MEMS shunt switches.
1. Modeling”. In: IEEE Transactions on Microwave Theory and Techniques 48.6
(June 2000), pp. 1045–1052. issn: 0018-9480. doi:10.1109/22.904743.
[96] Hyun-Ho Yang, A. Yahiaoui, H. Zareie, P. Blondy, and G. M. Rebeiz. “A compact
high-isolation DC-50 GHz SP4T RF MEMS switch”. In: 2014 IEEE MTT-S
International Microwave Symposium (IMS2014). June 2014, pp. 1–4. doi:10.
1109/MWSYM.2014.6848359.
[97] C. Chu, X. Liao, and H. Yan. “Ka-band RF MEMS capacitive switch with low
loss, high isolation, long-term reliability and high power handling based on GaAs
MMIC technology”. In: IET Microwaves, Antennas Propagation 11.6 (2017),
pp. 942–948. issn: 1751-8725. doi:10.1049/iet-map.2016.0595.
[98] N. Torres Matabosch, F. Coccetti, M. Kaynak, B., B. Tillack, and J.L. Cazaux.
“Equivalent circuit model of reliable RF-MEMS switches for component synthe-
sis, fabrication process characterization and failure analysis”. In: International
Journal of Microwave and Wireless Technologies 6.1 (Feb. 2014), pp. 73–81. doi:
10.1017/S1759078713000871.
111
Publications
Publications
1. S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, C. Wipf, and M. Kaynak. “BiCMOS
Embedded RF-MEMS Technologies”. In: MikroSystemTechnik 2017; Congress.
Oct. 2017, pp. 1–3.
2. S. Tolunay Wipf, A. G¨oritz, C. Wipf, M. Wietstruck, A. Burak, E. T¨urkmen,
Y. G¨urb¨uz, and M. Kaynak. “240 GHz RF-MEMS switch in a 0.13 µm SiGe
BiCMOS Technology”. In: 2017 IEEE Bipolar/BiCMOS Circuits and Technology
Meeting (BCTM). Oct. 2017, pp. 54–57. doi:10.1109/BCTM.2017.8112910.
3. S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, M. Cirillo, C. Wipf, W. Winkler,
and M. Kaynak. “Packaged BiCMOS embedded RF-MEMS test vehicles for
space applications”. In: 2017 47th European Microwave Conference (EuMC).
Oct. 2017, pp. 320–323. doi:10.23919/EuMC.2017.8230859.
4. S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, M. Cirillo, C. Wipf, K. Zoschke, and
M. Kaynak. “Effect of wafer-level silicon cap packaging on BiCMOS embedded
RF-MEMS switch performance”. In: 2017 IMAPS Nordic Conference on Microe-
lectronics Packaging (NordPac). June 2017, pp. 31–34. doi:10.1109/NORDPAC.
2017.7993158.
5. S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, C. Wipf, B. Tillack, A. Mai, and
M. Kaynak. “Electromagnetic and small-signal modeling of an encapsulated
RF-MEMS switch for D-band applications”. In: International Journal of Mi-
crowave and Wireless Technologies 9.6 (2017), pp. 1271–1278. doi:10.1017/
S1759078717000137.
6. S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, C. Wipf, B. Tillack, and M. Kaynak.
“D-Band RF-MEMS SPDT Switch in a 0.13 µm SiGe BiCMOS Technology”. In:
IEEE Microwave and Wireless Components Letters 26.12 (Dec. 2016), pp. 1002–
1004. issn: 1531-1309. doi:10.1109/LMWC.2016.2623245.
7. S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, C. Wipf, B. Tillack, A. Mai, and M.
Kaynak. “Thin film wafer level encapsulated RF-MEMS switch for D-band ap-
plications”. In: 2016 46th European Microwave Conference (EuMC). Oct. 2016,
pp. 1381–1384. doi:10.1109/EuMC.2016.7824610.
8. S. Tolunay Wipf, A. G¨oritz, M. Wietstruck, C. Wipf, and M. Kaynak. “RF
Pad Optimization for a 140 GHz RF-MEMS Switch”. In: 17th Symposium on
RF-MEMS and RF-Microsystems (MEMSWAVE). 2016.
112
Publications
9. S. Tolunay, A. G¨oritz, M. Wietstruck, C. Wipf, B. Tillack, and M. Kaynak. “94
GHz RF-MEMS SPDT Switch in a 0.13 µm SiGe BiCMOS Technology”. In:
16th Symposium on RF-MEMS and RF-Microsystems (MEMSWAVE). 2015.
10. S. Tolunay, M. Wietstruck, A. G¨oritz, M. Kaynak, and B. Tillack. “Accurate
3D EM modeling of 140 GHz BiCMOS embedded RF-MEMS switch”. In: 14th
Symposium on RF-MEMS and RF-Microsystems (MEMSWAVE). 2013.
113
Co-authored publications
Co-authored publications
1. M. Wietstruck, S. Marschmeyer, S. Tolunay Wipf, C. Wipf, T. Voss, M. Ber-
trand, E. Pistono, G. Acri, F. Podevin, P. Ferrari, and M. Kaynak. “Design
Optimization of Through-Silicon Vias for Substrate-Integrated Waveguides em-
bedded in High-Resistive Silicon Interposer”. In: 2018 IEEE 20th Electronics
Packaging Technology Conference (EPTC). Dec. 2018, pp. 195–200.
2. M. Bertrand, E. Pistone, G. Acri, D. Kaddour, F. Podevin, V. Puyal, S. Tolu-
nay Wipf, C. Wipf, M. Wietstruck, M. Kaynak, and P. Ferrari. “Substrate In-
tegrated Waveguides for mm-wave Functionalized Silicon Interposer”. In: 2018
IEEE/MTT-S International Microwave Symposium - IMS. June 2018, pp. 875–
878. doi:10.1109/MWSYM.2018.8439287.
3. T. Chaloun, F. Tabarani, S. Tolunay Wipf, M. Kaynak, H. Schumacher, and W.
Menzel. “A modular phased array transceiver with RF-MEMS SPDT switches
in a 0.25 µm SiGe BiCMOS technology”. In: 12th European Conference on An-
tennas and Propagation (EuCAP 2018). Apr. 2018, pp. 1–5. doi:10.1049/cp.
2018.0615.
4. C. Wipf, R. Sorge, A. G¨oritz, S. Tolunay Wipf, A. Scheit, D. Kissinger, and M.
Kaynak. “High voltage LDMOS inverter for on-chip RF-MEMS actuation”. In:
2018 IEEE 18th Topical Meeting on Silicon Monolithic Integrated Circuits in RF
Systems (SiRF). Jan. 2018, pp. 48–50. doi:10.1109/SIRF.2018.8304226.
5. P. Rynkiewicz, A.-L. Franc, F. Coccetti, M. Wietstruck, C. Wipf, S. Tolunay
Wipf, M. Kaynak, and G. Prigent. “Ring filter synthesis and its BiCMOS 60 GHz
implementation”. In: International Journal of Microwave and Wireless Techno-
logies 10.3 (2018), pp. 291–300. doi:10.1017/S1759078718000156.
6. P. Rynkiewicz, A.-L. Franc, F. Coccetti, S. Tolunay Wipf, M. Wietstruck, M.
Kaynak, and G. Prigent. “Tunable dual-mode ring filter based on BiCMOS em-
bedded MEMS in V-band”. In: 2017 IEEE Asia Pacific Microwave Conference
(APMC). Nov. 2017, pp. 124–127. doi:10.1109/APMC.2017.8251393.
7. P. Rynkiewicz, A.-L. Franc, F. Coccetti, S. Tolunay Wipf, M. Wietstruck, M.
Kaynak, and G. Prigent. “Millimeter-wave three-state tunable stopband resona-
tor based on integrated MEMS”. In: 2017 IEEE Asia Pacific Microwave Confe-
rence (APMC). Nov. 2017, pp. 128–131. doi:10.1109/APMC.2017.8251394.
114
Co-authored publications
8. A. G¨oritz, S. Tolunay Wipf, M. Wietstruck, M. Kaynak, and A. Mai. “Pro-
zesskontrolle f¨ur die D¨unnschichtverkapselung von BiCMOS HF-MEMS-Schaltern
mittels BEOL-integriertem Pirani-Element”. In: MikroSystemTechnik 2017; Con-
gress. Oct. 2017, pp. 1–3.
9. M. Kaynak, M. Wietstruck, A. G¨oritz, S. Tolunay Wipf, M. Inac, B. Cetindo-
gan, C. Wipf, C. B. Kaynak, M. W¨ohrmann, S. Voges, and T. Braun. “0.13 µm
SiGe BiCMOS technology with More-than-Moore modules”. In: 2017 IEEE Bi-
polar/BiCMOS Circuits and Technology Meeting (BCTM). Oct. 2017, pp. 62–65.
doi:10.1109/BCTM.2017.8112912.
10. P. Rynkiewicz, A.-L. Franc, F. Coccetti, S. Tolunay Wipf, M. Wietstruck, M.
Kaynak, and G. Prigent. “R´esonateur Bi-Mode Accordable Int´egr´e en Bande
Millim´etrique”. In: Journ´ees Nationales Micro-Ondes (JNM). 2017.
11. S. Lischke, D. Knoll, S. Tolunay-Wipf, C. Wipf, C. Mai, A. Fox, F. Herzel, and
M. Kaynak. “Side-use of a Ge p-i-n photo diode for electrical application in a
photonic BiCMOS technology”. In: 2016 IEEE Bipolar/BiCMOS Circuits and
Technology Meeting (BCTM). Sept. 2016, pp. 126–129. doi:10.1109/BCTM.
2016.7738970.
12. M. Wietstruck, W. Winkler, A. G¨oritz, S. Tolunay Wipf, C. Wipf, D. Schmidt,
A. Mai, and M. Kaynak. “0.13 µm BiCMOS embedded On-Chip High-Voltage
Charge Pump with Stacked BEOL Capacitors for RF-MEMS Applications”. In:
17th Symposium on RF-MEMS and RF-Microsystems (MEMSWAVE). 2016.
13. Y. J. Du, W. Su, Y. Li, S. Tolunay, M. Kaynak, R. Scholz, and Y. Z. Xiong.
“D-band MEMS switch in standard BiCMOS technology”. In: 2015 Asia-Pacific
Microwave Conference (APMC). Vol. 3. Dec. 2015, pp. 1–3. doi:10.1109/APMC.
2015.7413319.
14. M. Kaynak, M. Wietstruck, C. B. Kaynak, S. Tolunay, A. G¨oritz, and B. Til-
lack. “Modular extension of high performance SiGe BiCMOS technologies - Fol-
lowing the More-than-Moore path”. In: 2015 Asia-Pacific Microwave Conference
(APMC). Vol. 1. Dec. 2015, pp. 1–3. doi:10.1109/APMC.2015.7411742.
15. Y. J. Du, W. Su, S. Tolunay, L. Zhang, M. Kaynak, R. Scholz, and Y. Z. Xiong.
“220GHz wide-band MEMS switch in standard BiCMOS technology”. In: 2015
IEEE Asian Solid-State Circuits Conference (A-SSCC). Nov. 2015, pp. 1–4. doi:
10.1109/ASSCC.2015.7387512.
115
Co-authored publications
16. V. Valenta, H. Schumacher, S. Tolunay Wipf, M. Wietstruck, A. G¨oritz, M.
Kaynak, and W. Winkler. “Single-chip transmit-receive module with a fully in-
tegrated differential RF-MEMS antenna switch and a high-voltage generator for
F-band radars”. In: IEEE Bipolar/BiCMOS Circuits and Technology Meeting.
Oct. 2015, pp. 40–43. doi:10.1109/BCTM.2015.7340564.
17. A. G¨oritz, M. Lisker, S. Tolunay, M. Wietstruck, and M. Kaynak. “Wafer-Level
Encapsulation for BiCMOS embedded RF-MEMS Application”. In: Internatio-
nal Conference on Micro and Nano Engineering (MNE). 2015.
18. A. Strodl, S. Tolunay, V. Valenta, M. Kaynak, S. Reyaz, R. Johnson, W. Wink-
ler, R. Malmqvist, and H. Schumacher. “RF-MEMS Switched W-Band mm-Wave
Passive Imaging System in a 0.13 µm SiGe BiCMOS Technology”. In: 16th Sym-
posium on RF-MEMS and RF-Microsystems (MEMSWAVE). 2015.
19. M. Wietstruck, G. Kahmen, A. G¨oritz, S. Tolunay, B. Tillack, and M. Kaynak.
“Reliability of BiCMOS embedded MEMS Varactors for Wideband RF VCO
Applications”. In: 16th Symposium on RF-MEMS and RF-Microsystems (MEM-
SWAVE). 2015.
20. M. Kaynak, M. Wietstruck, C. B. Kaynak, A. G¨oritz, S. Tolunay, and B. Tillack.
“MEMS - BiCMOS monolithic integration”. In: 2014 XXXIth URSI General
Assembly and Scientific Symposium (URSI GASS). Aug. 2014, pp. 1–4. doi:
10.1109/URSIGASS.2014.6929473.
21. M. Wietstruck, M. Kaynak, A. G¨oritz, S. Tolunay, and B. Tillack. “Accurate
Contact Analysis of a BiCMOS Embedded Capacitive-Type RF-MEMS Switch”.
In: 15th Symposium on RF-MEMS and RF-Microsystems (MEMSWAVE). 2014.
22. M. Wietstruck, M. Kaynak, St. Marschmeyer, A. G¨oritz, S. Tolunay, and B.
Tillack. “Monolithisch-integrierte Through-Silicon Vias f¨ur Grounding, Power
Distribution Networks und System-on-Chip Anwendungen”. In: MikroSystem-
Technik 2013; Congress. Oct. 2013, pp. 1–4.
23. M. Wietstruck, M. Kaynak, St. Marschmeyer, A. G¨oritz, S. Tolunay, S. Kurth,
and B. Tillack. “High-Frequency Optimization of BiCMOS Embedded Through-
Silicon Vias for Backside-Integrated MEMS”. In: 14th Symposium on RF-MEMS
and RF-Microsystems (MEMSWAVE). 2013.
24. A. G¨oritz, M. Fraschke, J. Drews, M. Wietstruck, S. Tolunay, M. Kaynak, and B.
Tillack. “Anwendung eines HF-Gasphasen¨atzprozesses zum Freilegen von mono-
116
Co-authored publications
lithisch integrierten RF-MEMS-Schaltern”. In: MikroSystemTechnik 2013; Con-
gress. Oct. 2013, pp. 1–3.
117
List of Figures
1.1 Capacitive shunt and ohmic series RF-MEMS switches . . . . . . . . . 4
1.2 Hype curve of RF-MEMS technology . . . . . . . . . . . . . . . . . . . 6
1.3 Decreasing voice signal quality with smart mobile phones . . . . . . . 8
1.4 Cross-section of complete WiSpry MEMS capacitor with CP . . . . . . 9
1.5 MEMS array inside Cavendish Kinetics’ antenna tuners . . . . . . . . 9
1.6 MEMS switch technology of Analog Devices . . . . . . . . . . . . . . . 10
2.1 Cross section of the SG25H3/H4 BEOL . . . . . . . . . . . . . . . . . 14
2.2 Additional process steps of RF-MEMS switch in SG25 BiCMOS . . . 15
2.3 Cross section of embedded SG25 RF-MEMS switch . . . . . . . . . . . 16
2.4 SEM image of the SG25 RF-MEMS switch . . . . . . . . . . . . . . . 16
2.5 The generic cross section of packaged SG25 BiCMOS RF-MEMS devices 17
2.6 Process flow of the wafer-level packaging for SG25 RF-MEMS switch . 18
2.7 RF-MEMS switches before and after Si cap packaging . . . . . . . . . 19
2.8 The FIB analysis of the Si cap packaged RF-MEMS switch . . . . . . 19
2.9 The cross section of the SG13S/G2 BEOL . . . . . . . . . . . . . . . . 21
2.10 The cross section of the SG13 BEOL, including MEMS layers . . . . . 22
2.11 The process flow of the RF-MEMS switch in schematics . . . . . . . . 24
2.12 SEM image of the RF-MEMS switch before encapsulation . . . . . . . 25
2.13 Additional process steps of RF-MEMS switch in SG13 BiCMOS . . . 25
2.14 FIB SEM images of wafer-level encapsulation on aspect ratios . . . . . 27
2.15 SEM image of the WLE RF-MEMS switch . . . . . . . . . . . . . . . 28
2.16 The FIB cross section of the WLE RF-MEMS switch . . . . . . . . . . 28
3.1 Comparison of 2.5D and 3D EM solvers . . . . . . . . . . . . . . . . . 33
3.2 RF performances of switch in 2.5D and 3D case . . . . . . . . . . . . . 33
3.3 Simulation setup for the initial RF-MEMS switch . . . . . . . . . . . . 35
3.4 Contact region of the RF-MEMS switch . . . . . . . . . . . . . . . . . 35
3.5 Simulated RF performance of initial RF-MEMS Switch for different
distances................................... 36
3.6 EM model ifor D-band BiCMOS embedded WLE RF-MEMS switch . 38
118
List of Figures
3.7 EM simulation results of WLE RF-MEMS switch with varying distances 39
3.8 RF–MEMS switch model with its key EM optimization parameters . . 40
3.9 EM models to investigate the effect of TM2 plate . . . . . . . . . . . . 41
3.10 Simulated S-parameters for TM2 plate investigation . . . . . . . . . . 42
3.11 Generic view of varied arm width . . . . . . . . . . . . . . . . . . . . . 43
3.12 Simulated S-parameters for varied arm widths . . . . . . . . . . . . . . 44
3.13 Generic view of varied contact region width . . . . . . . . . . . . . . . 45
3.14 Simulated S-parameters for varied contact region widths . . . . . . . . 46
3.15 Generic view of varied RF-signal line width . . . . . . . . . . . . . . . 47
3.16 Simulated S-parameters for varied RF-signal line widths . . . . . . . . 48
3.17 Generic view of varied TM1 membrane hole density . . . . . . . . . . 49
3.18 Simulated S-parameters for varied TM1 membrane hole density . . . . 50
3.19 Generic view of varied ground-ring width . . . . . . . . . . . . . . . . 51
3.20 Simulated S-parameters for varied ground-ring . . . . . . . . . . . . . 52
3.21 The cross section of WLE switch with lumped-element capacitances . 53
3.22 The lumped-element model of the WLE RF-MEMS switch . . . . . . . 55
3.23 Generic schematics of LDV setup . . . . . . . . . . . . . . . . . . . . . 57
3.24 Displacement vs time curve . . . . . . . . . . . . . . . . . . . . . . . . 57
3.25 C-V measurement results of WLE RF-MEMS switch . . . . . . . . . . 58
3.26 Comparison of measured and simulated S-parameters of WLE switch . 60
3.27 Measured S-parameters of D-band switch with actuation voltages . . . 61
3.28 The fabricated J-band SPST RF-MEMS switch with the TM2 plate . 62
3.29 EM simulation model of the J-band RF-MEMS switch . . . . . . . . . 63
3.30 Simulated S-parameters of J-band switch with varying contact distance 65
3.31 C-V measurement results of J-band RF-MEMS switch . . . . . . . . . 67
3.32 Comparison of measured and simulated S-parameters of J-band switch 69
3.33 Measured S-parameters of J-band switch with actuation voltages . . . 70
4.1 Fabricated D-band RF-MEMS SPST switch and its top generic view . 74
4.2 Measured S-parameters and EM simulation results of SPST switch . . 75
4.3 Schematic of the SPDT switch in shunt configuration . . . . . . . . . . 77
4.4 EM Simulation model of D-band SPDT switch . . . . . . . . . . . . . 77
4.5 Micrograph of the D-band RF-MEMS based SPDT switch. . . . . . . 78
4.6 Measured and EM simulated S-parameters of D-band SPDT switch . . 80
4.7 Simulated 2–port and 3-port S–parameters of D-band SPDT switch . 81
4.8 Microscope images of packaged K-band RF-MEMS test vehicles . . . . 83
119
List of Figures
4.9 Measured C-V graphs of a TCV before and after packaging . . . . . . 85
4.10 Measured S-parameters of a TCV before and after packaging . . . . . 86
4.11 Measured C-V graphs of a DEC before and after packaging . . . . . . 87
4.12 Microscope image of a DEC before packaging. . . . . . . . . . . . . . . 87
4.13 Measured S-parameters of a DEC before and after packaging . . . . . 88
4.14 Microscope image of a RICs before Si cap packaging . . . . . . . . . . 89
4.15 Microscope image of the RICs during electrical characterization . . . . 90
4.16 Vctrl vs. Vout curve of the charge pump with Vdd of 3 V and Vin of 0 V 90
4.17 EM simulation model of TCV, with and without Si cap . . . . . . . . 91
4.18 Insertion losses of EM simulated switch with and without Si caps . . . 92
4.19 ILs of measured and simulated switch with and without Si cap . . . . 92
4.20 Wafer-level Con values of the TCVs . . . . . . . . . . . . . . . . . . . . 94
4.21 C-V behaviors of TCVs on different regions of wafer . . . . . . . . . . 94
4.22 Wafer-level Con valuesofDECs...................... 95
4.23 C-V behaviors of the DECs on different regions of wafer . . . . . . . . 95
4.24 Wafer-level measured Vout values of the RIC’s charge pump . . . . . . 96
120
List of Tables
2.1 Main performance parameters of HBTs for 0.25 µm technologies . . . . 13
2.2 Main performance parameters of HBTs for 0.13 µm technologies . . . . 20
3.1 Small-signal component values of WLE RF-MEMS switch . . . . . . . 54
3.2 Small-signal component values of WLE switch with varied states . . . 54
3.3 Small-signal component values of J-band RF-MEMS switch . . . . . . 66
3.4 Small-signal component values of J-band switch, with varied states . . 66
3.5 Measured Performance Comparison of mm-wave SPST Switches . . . . 72
4.1 Measured Performance Comparison of mm-wave SPDT Switches . . . 82
121
Erkl¨arung
Ich erkl¨are hiermit, dass die vorliegende Arbeit von mir selbst und ohne fremde Hilfe
verfasst wurde. Alle benutzten Quellen sind im Literaturverzeichnis angegeben. Die
Arbeit hat in gleicher oder ¨ahnlicher Form noch keiner Pr¨ufungsbeh¨orde vorgelegen.
Frankfurt(Oder), den
122
