A flexible, gigahertz, and free-standing thin film piezoelectric MEMS resonator with
high figure of merit
Yuan Jiang, Menglun Zhang, Xuexin Duan, Hao Zhang, and Wei Pang
Citation: Appl. Phys. Lett. 111, 023505 (2017); doi: 10.1063/1.4993901
View online: http://dx.doi.org/10.1063/1.4993901
View Table of Contents: http://aip.scitation.org/toc/apl/111/2
Published by the American Institute of Physics
A flexible, gigahertz, and free-standing thin film piezoelectric MEMS
resonator with high figure of merit
Yuan Jiang, Menglun Zhang,
Xuexin Duan, Hao Zhang, and Wei Pang
State Key Laboratory of Precision Measuring Technology and Instruments, Tianjin University, Tianjin 300072,
(Received 17 April 2017; accepted 1 July 2017; published online 13 July 2017)
In this paper, a 2.6 GHz air-gap type thin ﬁlm piezoelectric MEMS resonator was fabricated on a
ﬂexible polyethylene terephthalate ﬁlm. A fabrication process combining transfer printing and
hot-embossing was adopted to form a free-standing structure. The ﬂexible radio frequency MEMS
resonator possesses a quality factor of 946 and an effective coupling coefﬁcient of 5.10%, and
retains its high performance at a substrate bending radius of 1 cm. The achieved performance is
comparable to that of conventional resonators on rigid silicon wafers. Our demonstration provides
a viable approach to realizing universal MEMS devices on ﬂexible polymer substrates, which is of
great signiﬁcance for building future fully integrated and multi-functional wireless ﬂexible
electronic systems. Published by AIP Publishing. [http://dx.doi.org/10.1063/1.4993901]
The rapid development of ﬂexible electronic systems
such as ﬂexible displays,
enabled by signiﬁcant advancements in basic ﬂexible elec-
tronic components, e.g., ﬂexible transistors,
diodes (LEDs), passive components,
Among different realization technologies, ﬁlm bulk acoustic
resonators (FBARs), traditionally used as basic building
blocks of modern RF ﬁlters,
are natural candidates for
wireless ﬂexible electronics. Moreover, as MEMS devices
that bridge the mechanical and electrical domains, FBARs
demonstrated signiﬁcant sensing
potentials. Transplanting this technology into the ﬂexible
electronic realm will enrich the functionalities of the future
ﬂexible electronic systems.
Flexible FBARs in previously reported works were real-
ized by either chemical thinning of the silicon wafer
ricating FBARs directly on polymer substrates, such as
polyimide (PI), that can withstand high processing tempera-
By thinning the silicon wafer, semi-ﬂexible
FBARs with high performance can be obtained. However,
this process is time-consuming, and further thinning of the
silicon wafer would result in very fragile devices. Using the
latter method, thin and ﬂexible FBARs can be fabricated, but
the device performance is limited. The performance of a
FBAR is typically indicated by its ﬁgure of merit (FOM),
deﬁned as the product of the quality factor (Q) and the effec-
tive coupling coefﬁcient (k2
teff ). The difﬁculty in depositing a
high quality piezoelectric thin ﬁlm on a polymer substrate
and lack of an effective acoustic reﬂection structure (air cav-
ity or Bragg reﬂector) result in a FOM below 10, compared
to a FOM of 39 in the previous method.
In this work, high-performance FBARs are achieved
using a process combining transfer printing
embossing: the composite membranes are ﬁrst fabricated on
a silicon wafer and then transfer printed on a ﬂexible
polyethylene terephthalate (PET) substrate with hot-
embossed air cavities, forming suspended FBARs. The ﬂexi-
ble resonator has a resonant frequency of 2.656 GHz, a Q
factor of 948, and a k2
teff of 5.10%. The device FOM is
around 48, which is comparable to conventional FBARs fab-
ricated on rigid silicon wafers.
Bending test shows that the
device performance is sustainable for a substrate bending
radius of 1 cm. Our proposed method could be easily adopted
for the fabrication of other piezoelectric MEMS devices
on polymer substrates as well.
Figure 1illustrates the main steps used to fabricate the
ﬂexible FBAR devices and the fabricated ﬂexible devices.
The receiver ﬂexible substrate and the device donor wafer
were prepared separately. A piece of commercially available
dicing tape (UDT-1025c, Denka), which contains a 100 lm
PET layer and a 25 lm UV-curable adhesive layer, was cho-
sen as the receiver substrate. A silicon mold with pentagon-
shaped protrusions (10 lm in height) was fabricated by
photolithography and reactive ion etch (RIE). Then, the
dicing tape was laminated to the silicon mold, and together
they were put between two heated and pressurized plates
(0.2 MPa and 95 C) for 5 min. After cooling down to room
temperature, the dicing tape was peeled off from the silicon
mold, and thus, pentagon-shaped air cavities were formed in
the adhesive layer of the dicing tape, as shown in Fig. 2(a).
The fabrication of the donor device began with the
deposition of 300 nm silicon oxide as a sacriﬁcial layer on
the silicon wafer. The sacriﬁcial layer was then patterned by
a buffered oxide etchant (7:1 BOE), exposing parts of the
silicon surface underneath. The bottom electrode was formed
by deposition and patterning of 300 nm molybdenum (Mo).
Part of the bottom electrode was formed on the silicon sur-
face, thus serving as anchors during the release step. 600 nm
aluminum nitride (AlN) was deposited and patterned as the
piezoelectric layer. Another layer of 300 nm Mo was depos-
ited and patterned as the top electrode, forming the sand-
wiched thin ﬁlm structure. Removal of the sacriﬁcial layer in
a 1:10 solution of HF (hydroﬂuoric acid, 49%) for 3 h
Electronic mail: firstname.lastname@example.org
Electronic mail: email@example.com
0003-6951/2017/111(2)/023505/4/$30.00 Published by AIP Publishing.111, 023505-1
APPLIED PHYSICS LETTERS 111, 023505 (2017)
released the FBAR membranes from the substrate, while
anchors kept the membranes on the substrate, as shown in
Finally, a soft elastomer stamp made from polydime-
thylsiloxane (PDMS) was used to pick up the device from
the donor wafer. In Fig. 2(c), the anchor left on the wafer
shows a clean fracture along the designed mechanical weak
edges. It is worth mentioning that the silicon wafer can be
reused after removing the left anchors. The stamp carrying
the FBAR membrane was carefully aligned with the air cav-
ity and then brought into contact with the adhesive surface
using a microscope and a XYZ translation stage. After
retracting the stamp, the device membrane eventually
remained on the adhesive surface because of a larger adhe-
sion force at the FBAR–adhesive interface rather than the
FBAR–PDMS interface. In the last step, UV exposure from
the back side of the tape fully cured the adhesive. The adhe-
sive turned harder and became less adhesive after curing,
rendering the device more mechanically stable and less sus-
ceptible to particulates or contaminants, and made the whole
device stable. An intentional half-cut device is shown in Fig.
2(d) in order to demonstrate the suspended structure.
The transfer process is carried out at room temperature.
The applied pressure during the pick and place steps was
controlled by monitoring deformations of the device and the
stamp in the microscope, as an intimate contact must be
ensured without inducing too much stress to the device. To
optimize the transfer yield, sharp corners and narrow inter-
sections should be avoided in the geometry design to mini-
mize stress concentration during the transfer process. Round
holes at the connection edge [as shown in Fig. 2(b)] help
create stress concentration at the anchor sites, making sure
that the anchors break easier than other parts of the device.
Hole densities of 0.5 and 0.75 were ﬁnally chosen from
balancing between the easiness of retrieval and the stability
of attaching devices on the substrate during the washing and
cleaning process. Discussion about the design principle of
the device geometry and information of the design evalua-
tion experiment can be found in the supplementary material.
Our proposed fabrication method is quite robust and gen-
eral, as ten FBARs were transferred in succession with a yield
of 100% (Figs. S1 and S2, supplementary material). On the
other hand, since the receiver substrate experiences no harsh
conditions such as high temperature and corrosive etchant,
there are a vast number of choices for other ﬂexible substrates
such as PDMS and PI. The air cavity can be formed by
alternative methods like molding and plasma etching (Fig.
S3, supplementary material). Among those methods, hot-
embossing technique represents the most efﬁcient patterning
method and can be easily implemented within a roll-to-roll
fabrication system. Since the transfer print process is also
scalable, a high-throughput production can be envisioned
by upgrading the current manual transfer process to an auto-
mated transfer print system equipped with robotic arms, par-
allel stamp arrays, and computer vision modules.
The resonance characteristics of ﬂexible FBARs were
measured with an Agilent E8368b vector network analyzer
FIG. 1. (a) Illustration of the main fabrication process of a ﬂexible FBAR
(drawn not to scale). (b) Photo of a plastic substrate carrying ﬂexible resona-
tors and zoomed-in microscopy images of one resonator in an air cavity.
FIG. 2. Scanning electron microscopy (SEM) images of (a) the hot-
embossed air cavity on the adhesive, (b) the anchor structure that tethers the
FBAR at the silicon wafer, (c) the anchor left on the silicon wafer after the
device transfer, and (d) FBAR transferred on the adhesive (the composite
membrane was intentionally half-cut before transfer in order to show the
023505-2 Jiang et al. Appl. Phys. Lett. 111, 023505 (2017)
using a 150 lm pitch GS probe (Cascade Microtech).
Analytical results were calculated using a one-dimensional
The measured and analytical reﬂection coef-
ﬁcients (S11) are plotted on a Smith chart in Fig. 3(a). The
measured S11 circle is very close to the edge of the Smith
chart, indicating that the FBAR has a relatively high FOM.
Figure 3(b) presents the measured and analytical impedance
spectra. The two impedance peaks correspond to the series
resonant frequency (f
) at 2.656 GHz and the parallel reso-
nant frequency (f
) at 2.713 GHz, respectively. The effective
coupling coefﬁcient (k2
teff ) is calculated to be 5.10%, indicat-
ing a good electrical–mechanical conversion efﬁciency of
the piezoelectric AlN. It can be seen from Figs. 3(a) and 3(b)
that the analytical data agree well with the measured data,
showing that the Mason model can be effectively utilized to
guide the design of the ﬂexible air-gap type FBAR as well.
The device Q factor as a function of frequency can be
where sis the group delay of the measured S11. Q factors at
the series resonant frequency (Q
) and the parallel resonant
) were calculated to be 947 and 694, respec-
tively. Temperature characteristic of ﬂexible FBAR was
tested by adhering the substrate to a heated wafer chuck, and
each dataset was measured at 5 min after the chuck reached
its set temperature. Figure 3(c) shows that the thermally
induced frequency shift of the ﬂexible FBAR is quite linear
from 30 to 60 C. The corresponding coefﬁcient of frequency
(TCF) is about 24 ppm/C, which is comparable with rigid
AlN FBARs on silicon. A further increase of the temperature
will soften the adhesive below the electrodes, rendering
probe tests difﬁcult. Other polymers with a higher glass tran-
sition temperature, e.g., polyimide and polyimide adhesives,
can be used as substrates for a wider operating temperature
The mechanical ﬂexibility of FBARs was investigated
by attaching them on plastic columns of radii ranging from
3 cm to 0.5 cm, as shown in Fig. 4(a). The FBARs functioned
well at bending radii of 3 cm, 2 cm, and 1 cm, and there were
no obvious impedance spectral shifts at these bending levels.
A small resonant frequency shift (less than 100 kHz, about
38 ppm) was observed in the zoom-in view, which is
acceptable for most applications. This shift is much lower
than the expected value of 9 MHz in theory, as derived in the
supplementary material. The discrepancy may be caused by
alternative factors other than thickness change, including
parasitic capacitance and inductance of the electrodes, and
residual stresses left in the Mo and AlN ﬁlms during deposi-
tion. The effect of device bending on strain distribution was
studied using a FEA solver. Figure 4(b) shows that the maxi-
mum strain resides in the center of the pentagon-shaped
membrane, due to the presence of air cavity. The maximum
strain increases exponentially with the decrease of bending
radii. Bending radii of 3 cm, 2 cm, and 1 cm correspond to
strain values of 0.25%, 0.35%, and 0.7%, respectively. The
maximum strain increased signiﬁcantly to 1.3% at the radius
of 0.5 cm, causing the electrode to break. Further
FIG. 3. Measured and analytical results: (a) the Smith chart and (b) the impedance spectrum. (c) Resonant frequency drift versus temperature change.
FIG. 4. (a) Resonance curves of a ﬂexible FBAR at different bending radii,
along with a magniﬁed view of the resonance peak. (b) Correlation of the
bending radius versus maximum strain in a device membrane solved using a
023505-3 Jiang et al. Appl. Phys. Lett. 111, 023505 (2017)
improvements to the ﬂexibility of our device include using a
thinner substrate, optimization of the device shape to avoid
stress concentration, and adding a top encapsulation layer
according to the neutral plane theory.
In summary, a ﬂexible piezoelectric MEMS resonator
has been demonstrated on a polymer substrate. Transfer print-
ing and hot-embossing were used to form the free-standing
structure, which is the key to retain high device performance.
The 2.6 GHz FBAR has a Q
of 947, a Q
of 694, a k2
5.10%, and a TCF of 24 ppm/C. The device functioned
well at bending radii of 3 cm, 2 cm, and 1cm. The fabrication
process we developed could be adopted for other MEMS
devices. It greatly enriches the library of available building
blocks for a multi-functional ﬂexible electronic system.
See supplementary material for a detailed comparison
with previous works, experiments for yield estimation, the
process demonstration on alternative substrates, the design
principles of the device geometry, and the deduction of the
theoretical frequency shift of an FBAR under tensile strain.
This work was supported by Natural Science
Foundation of China (NSFC No. 51375341), the 111 Project
under Grant No. B07014, and the National High Technology
Research and Development Program of China (863 Program)
under Grant No. 2015AA042603.
S.-I. Park, Y. Xiong, R.-H. Kim, P. Elvikis, M. Meitl, D.-H. Kim, J. Wu, J.
Yoon, C.-J. Yu, Z. Liu, Y. Huang, K-c. Hwang, P. Ferreira, X. Li, K.
Choquette, and J. A. Rogers, Science 325, 977 (2009).
T. Q. Trung and N. E. Lee, Adv. Mater. 28, 4338 (2016).
M. L. Hammock, A. Chortos, B. C. K. Tee, J. B. H. Tok, and Z. Bao, Adv.
Mater. 25, 5997 (2013).
E. Menard, K. J. Lee, D. Y. Khang, R. G. Nuzzo, and J. A. Rogers, Appl.
Phys. Lett. 84, 5398 (2004).
J-H. Ahn, H-S. Kim, K. J. Lee, Z. Zhu, E. Menard, R. G. Nuzzo, and J. A.
Rogers, IEEE Electron Device Lett. 27, 460 (2006).
G. Qin, J. H. Seo, Y. Zhang, H. Zhou, W. Zhou, Y. Wang, J. Ma, and Z.
Ma, IEEE Electron Device Lett. 34, 262 (2013).
L. Sun, G. Qin, H. Huang, H. Zhou, N. Behdad, W. Zhou, and Z. Ma,
Appl. Phys. Lett. 96, 013509 (2010).
T. Rai, P. Dantes, B. Bahreyni, and W. S. Kim, IEEE Electron Device
Lett. 34, 544 (2013).
M. Koo, K. I. Park, S. H. Lee, M. Suh, D. Y. Jeon, J. W. Choi, K. Kang,
and K. J. Lee, Nano Lett. 12, 4810 (2012).
K-Y. Hashimoto, RF Bulk Acoustic Wave Filters for Communications
(Artech House, 2009).
G. Wingqvist, J. Bjurstr€
om, L. Liljeholm, V. Yantchev, and I. Katardjiev,
Sens. Actuators, B 123, 466 (2007).
I. Voiculescu and A. N. Nordin, Biosens. Bioelectron. 33, 1 (2012).
Y. Chang, N. Tang, H. Qu, J. Liu, D. Zhang, H. Zhang, W. Pang, and X.
Duan, Sci. Rep. 6, 23970 (2016).
E. S. Kim, in 2010 IEEE International Frequency Control Symposium
(Institute of Electrical and Electronics Engineers (IEEE), 2010).
J. Friend and L. Y. Yeo, Rev. Mod. Phys. 83, 647 (2011).
H. Qu, Y. Yang, Y. Chang, Z. Tang, W. Pang, Y. Wang, H. Zhang, and X.
Duan, Sens. Actuators, B 248, 280 (2017).
Y. R. Kang, S. C. Kang, K. K. Paek, Y. K. Kim, S. W. Kim, and B. K. Ju,
Sens. Actuators, A 117, 62 (2005).
R. Wright, G. Hakemi, and P. Kirby, Microelectron. Eng. 88, 1006
N. N. Hu, X. L. He, X. L. Bian, G. H. Chen, S. R. Dong, and J. K. Luo, in
2014 IEEE International Conference on Electron Devices Solid-State
Circuits, EDSSC 2014 (2014), Vol. 2,p.4.
C. Zhou, Y. Shu, Y. Yang, H. Jin, S.-R. Dong, M. Chan, and T.-L. Ren,
J. Micromech. Microeng. 25, 055003 (2015).
G. Chen, X. Zhao, X. Wang, H. Jin, S. Li, S. Dong, A. J. Flewitt, W. I.
Milne, and J. K. Luo, Sci. Rep. 5, 9510 (2015).
E. Smecca, F. Maita, G. Pellegrino, V. Vinciguerra, L. L. Magna, L.
Maiolo, G. Fortunato, G. G. Condorelli, A. Alberti, E. Smecca, F. Maita,
G. Pellegrino, V. Vinciguerra, L. L. Magna, S. Mirabella, L. Maiolo, G.
Fortunato, G. G. Condorelli, and A. Alberti, Appl. Phys. Lett. 106, 232903
A. Carlson, A. M. Bowen, Y. Huang, R. G. Nuzzo, and J. A. Rogers, Adv.
Mater. 24, 5284 (2012).
R. C. Ruby, P. Bradley, Y. Oshmyansky, A. Chien, and J. D. Larson, in
2001 IEEE Ultrasonics Symposium (IEEE, 2001), Vol. 1, pp. 813–821.
Y. Lu and D. A. Horsley, J. Microelectromech. Syst. 24, 1142 (2015).
J. Zou, C.-M. Lin, C. Lam, and A. P. Pisano, J. Appl. Phys. 121, 154502
J. Zou, C.-M. Lin, Y.-Y. Chen, and A. P. Pisano, J. Appl. Phys. 115,
W. P. Mason, Electromechanical Transducers and Wave Filters (D. Van
Nostrand Co., 1948).
D. A. Feld, R. Parker, R. Ruby, P. Bradley, and D. Shim, Ultrasonics
Symposium 2008, IUS 2008 (IEEE, 2008), p. 431.
J. Kim, M. Lee, H. J. Shim, R. Ghaffari, H. R. Cho, D. Son, Y. H. Jung,
M. Soh, C. Choi, S. Jung, K. Chu, D. Jeon, S.-T. Lee, J. H. Kim, S. H.
Choi, T. Hyeon, and D.-H. Kim, Nat. Commun. 5, 5747 (2014).
023505-4 Jiang et al. Appl. Phys. Lett. 111, 023505 (2017)