(94A0=>4?D:1$099>D7A,94,(94A0=>4?D:1$099>D7A,94,
&.3:7,=7D:88:9>&.3:7,=7D:88:9>
0;,=?809?,7$,;0=>!!
0;,=?809?:1!0.3,94.,7924900=492
;;740/!0.3,94.>
$40E:070.?=4.,7@849@894?=4/09,9:070.?=:80.3,94.,7,.?@,?:=>$40E:070.?=4.,7@849@894?=4/09,9:070.?=:80.3,94.,7,.?@,?:=>
"4;@9&493,
(94A0=>4?D:1$099>D7A,94,
94;@9>0,>@;0990/@
=,3,8*,-4>E0B>64
(94A0=>4?D:1$099>D7A,94,
2=,3,8B>0,>@;0990/@
%,>30/!,3,800/
(94A0=>4?D:1$099>D7A,94,
),70=D)0780?>20=
'02,7:=;:=,?4:9
&3,B9!',990=
'02,7:=;:=,?4:9
&0090C?;,201:=,//4?4:9,7,@?3:=>
:77:B?34>,9/,//4?4:9,7B:=6>,?3??;>=0;:>4?:=D@;0990/@80,8+;,;0=>
%0.:8809/0/4?,?4:9%0.:8809/0/4?,?4:9
&493,"4;@9*,-4>E0B>64=,3,8!,3,800/%,>30/0780?>20=),70=D)',990=&3,B9!
,=;4.6%:-0=?*,9/;4,EE,4,97@.,$40E:070.?=4.,7@849@894?=4/09,9:070.?=:80.3,94.,7
,.?@,?:=>
0;,=?809?,7$,;0=>!!
3??;>=0;:>4?:=D@;0990/@80,8+;,;0=>
:;D=423?80=4.,99>?4?@?0:1$3D>4.>'34>,=?4.708,D-0/:B97:,/0/1:=;0=>:9,7@>0:97D9D:?30=@>0
=0<@4=0>;=4:=;0=84>>4:9:1?30,@?3:=,9/?3080=4.,99>?4?@?0:1$3D>4.>
%0;=49?0/1=:8
$40E:070.?=4.,7@849@894?=4/09,9:070.?=:80.3,94.,7,.?@,?:=>"4;@9&493,=,3,8*,-4>E0B>64%,>30/
!,3,800/),70=D)0780?>20=&3,B9!',990=%:-0=?*,=;4.6,9/4,97@.,$4,EE,;;7$3D> 0??
#
$@-74>30=(% 3??;7496,4;:=27496$$
'34>;,;0=4>;:>?0/,?&.3:7,=7D:88:9>3??;>=0;:>4?:=D@;0990/@80,8+;,;0=>
:=8:=0491:=8,?4:9;70,>0.:9?,.?=0;:>4?:=D;:-:C@;0990/@
$40E:070.?=4.,7@849@894?=4/09,9:070.?=:80.3,94.,7,.?@,?:=>$40E:070.?=4.,7@849@894?=4/09,9:070.?=:80.3,94.,7,.?@,?:=>
->?=,.?->?=,.?
'34>70??0==0;:=?>?3048;70809?,?4:9:1@7?=,?34998,7@849@894?=4/07";40E:070.?=4.7,D0=>1:=
?301,-=4.,?4:9:1A0=?4.,77D/0J0.?4929,9:,.?@,?:=>'30I78>0C34-4?,9,A0=,20;40E:070.?=4.
.:0K.409?/FH;"B34.34>.:8;,=,-70?:4?>84.=:>.,70.:@9?0=;,=?'34>,77:B>A0=?4.,7
/0J0.?4:9>,>7,=20,>981=:8G87:92,9/98?34.68@7?47,D0=.,9?470A0=-48:=;3-0,8>
B4?3),.?@,?4:9@=?30=8:=049;7,90>?=0>>,9/>?=0>>2=,/409?>3,A0-009>48@7?,90:@>7D
.:9?=:770/'30I78>0C34-4?70,6,20.@==09?>7:B0=?3,99.8,?),9/3,A0,9,A0=,20=07,?4A0
/4070.?=4..:9>?,9?:1,;;=:C48,?07D,>49?34.60=I78>'30>08,?0=4,7.3,=,.?0=4>?4.>,9/,.?@,?4:9
=0>@7?>8,60?307"9,9:I78>4/0,7.,9/4/,?0>1:=?30=0,74E,?4:9:19,9:070.?=:80.3,94.,7>B4?.30>
1:=7:B;:B0=7:24.,;;74.,?4:9>
0DB:=/>0DB:=/>
,7@8494@8.:8;:@9/>.,9?470A0=>)>084.:9/@.?:=>70,6,20.@==09?>9,9:070.?=:80.3,94.,7
/0A4.0>9,9:>?=@.?@=0/8,?0=4,7>;0=84??4A4?D;40E:070.?=4.,.?@,?:=>;40E:070.?=4.8,?0=4,7>
;40E:070.?=4.?349I78>;40E:070.?=4.4?D>084.:9/@.?:=?349I78>>?=0>>0110.?>B4/0-,9/2,;
>084.:9/@.?:=>
:8809?>:8809?>
:;D=423?80=4.,99>?4?@?0:1$3D>4.>'34>,=?4.708,D-0/:B97:,/0/1:=;0=>:9,7@>0:97D9D
:?30=@>0=0<@4=0>;=4:=;0=84>>4:9:1?30,@?3:=,9/?3080=4.,99>?4?@?0:1$3D>4.>
%0;=49?0/1=:8
$40E:070.?=4.,7@849@894?=4/09,9:070.?=:80.3,94.,7,.?@,?:=>"4;@9&493,=,3,8*,-4>E0B>64
%,>30/!,3,800/),70=D)0780?>20=&3,B9!',990=%:-0=?*,=;4.6,9/4,97@.,$4,EE,;;7
$3D> 0??#
$@-74>30=(% 3??;7496,4;:=27496$$
@?3:=>@?3:=>
"4;@9&493,=,3,8*,-4>E0B>64%,>30/!,3,800/),70=D)0780?>20=&3,B9!',990=%:-0=?
*,=;4.6,9/4,97@.,;4,EE,
'34>5:@=9,7,=?4.704>,A,47,-70,?&.3:7,=7D:88:9>3??;>=0;:>4?:=D@;0990/@80,8+;,;0=>
Piezoelectric aluminum nitride nanoelectromechanical actuators
Nipun Sinha,
1,a兲
Graham E. Wabiszewski,
1
Rashed Mahameed,
2
Valery V. Felmetsger,
3
Shawn M. Tanner,
3
Robert W. Carpick,
1
and Gianluca Piazza
1,2
1
Department of Mechanical Engineering and Applied Mechanics, University of Pennsylvania,
Philadelphia, Pennsylvania 19104, USA
2
Department of Electrical and Systems Engineering, University of Pennsylvania, Philadelphia,
Pennsylvania 19104, USA
3
Tegal Corporation, San Jose, California 95134, USA
共Received 27 April 2009; accepted 14 July 2009; published online 4 August 2009兲
This letter reports the implementation of ultrathin 共100 nm兲 aluminum nitride 共AlN兲 piezoelectric
layers for the fabrication of vertically deflecting nanoactuators. The films exhibit an average
piezoelectric coefficient 共d
31
⬃−1.9 pC/ N兲, which is comparable to its microscale counterpart. This
allows vertical deflections as large as 40 nm from 18
m long and 350 nm thick multilayer
cantilever bimorph beams with 2 V actuation. Furthermore, in-plane stress and stress gradients have
been simultaneously controlled. The films exhibit leakage currents lower than 2 nA/ cm
2
at 1 V, and
have an average relative dielectric constant of approximately 9.2 共as in thicker films兲. These material
characteristics and actuation results make the AlN nanofilms ideal candidates for the realization of
nanoelectromechanical switches for low power logic applications. © 2009 American Institute of
Physics. 关DOI: 10.1063/1.3194148兴
The growth of the semiconductor industry has been
made possible by the continuous scaling of the complemen-
tary metal-oxide semiconductor 共CMOS兲 technology. An im-
portant aspect that hinders further reduction in the channel
lengths 共the length of a channel between the source and
drain in a transistor兲 and supply voltages in metal-oxide-
semiconductor field-effect transistors 共MOSFET兲 is the pas-
sive power dissipation in the device. This power loss is ex-
perienced due to subthreshold conduction that occurs due to
the presence of a physical channel in the semiconductor ma-
terial for the transfer of carriers between the source and
drain. Purely nanomechanical structures for implementing
logic and memories
1–5
have been proposed to resolve this
specific issue. The current leakage due to presence of a solid
channel can be greatly reduced by replacing the channel with
an air gap. This air gap can be closed by mechanical actua-
tors, therefore significantly reducing the subthreshold slope
of a conventional MOSFET.
Conventional mechanical actuation mechanisms, such as
capacitive, magnetomotive, and thermoelastic, which have
been used to drive nanoscale devices
6,7
have the drawback of
either being nonlinear in nature or requiring high power for
operation. For example, capacitive actuation, whose sensitiv-
ity depends directly on electrode area, does not scale well to
submicron dimensions,
3
magnetomotive actuation needs the
presence of high magnetic fields to function,
3
and thermal
actuation is still comparatively inefficient at this scale.
7
On
the other hand, the well-established piezoelectric actuation
mechanism offers the advantages of extremely low power
consumption and linear actuation.
8
In addition, the energy
density per input unit voltage of piezoelectric materials is
inversely proportional to the square of the film thickness.
However, piezoelectric film technology scaled to nanoscale
thicknesses is only in its infancy, and raises a challenge be-
cause it is hard to deposit ultrathin films that retain the prop-
erties of their thicker microscale counterparts. Studying the
electromechanical properties of piezoelectric nanofilms is
therefore the first step toward the realization of a structure
that can be used to implement nanomechanical actuators.
Furthermore, the demonstration of nanomechanical devices
for logic applications dictates having simultaneously fast
switching times and large deflection characteristics. The only
way to preserve substantial mechanical responsiveness along
with high operating frequencies is to scale both the thickness
and the length of the device, i.e., the device would have to be
extremely small and thin at the same time. This letter reports
on the implementation of ultrathin 共100 nm兲 aluminum
nitride 共AlN兲 piezoelectric films for the fabrication of
vertically deflecting piezoelectric nanoelectromechanical
共P-NEMS兲 actuators.
Lead zirconate titanate and AlN are two commonly used
piezoelectric materials that have already been used for the
fabrication of actuators and microelectromechanical
共MEMS兲 switches.
9–12
Out of these two materials, AlN
stands out for its high dielectric strength, ease of deposition,
and processing 共involving low temperatures and nontoxic
precursors兲, and its potential for integration with CMOS de-
vices. AlN has previously been used for the fabrication of
MEMS contour mode filters,
13
film bulk-wave acoustic reso-
nator 共FBAR兲 filters
14
共AlN being the material of choice for
commercial FBAR production兲, high frequency resonators,
15
and switches.
9,10
Scaling of piezoelectric transduction to the nanoscale
has failed in the past because of degradation of piezoelectric
properties due to limited orientation in thin films and in-
creases in internal stresses 共cracking and excessive deforma-
tions in released structures兲. The class of P-NEMS devices
presented in this letter has been made possible by precise
control of film quality 共orientation and stress兲 at 100 nm
thickness. Owing to optimization of the actuator design and
AlN sputter technology, not only was good piezoelectric re-
sponse in 100 nm thick films achieved, but bimorph actua-
tors were formed by stacking two of these thin AlN films
a兲
Electronic mail: nipun@seas.upenn.edu. Tel.: ⫹1-215-573-3276.
APPLIED PHYSICS LETTERS 95, 053106 共2009兲
0003-6951/2009/95共5兲/053106/3/$25.00 © 2009 American Institute of Physics95, 053106-1
Downloaded 14 Sep 2009 to 130.91.117.41. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
sandwiched between three layers of thin 共⬃50 nm兲 platinum
共Pt兲. The nanoactuator shown in Fig. 1 was fabricated as a
beam clamped at both ends and successively cut by focused
ion beam to free one end and test different length structures
as bending cantilevers. Platinum electrodes were used to ap-
ply an electric field across the piezoelectric film, which
causes the thin film to strain by the inverse piezoelectric
effect. The basic mechanism of actuation of the beam is iden-
tical to what was demonstrated for MEMS devices:
9,10
the
strains in either or both AlN layers of the structure generates
a transverse moment about the neutral axis of the whole de-
vice and causes the beam to bend. The actuation voltage can
be applied to each of the two piezoelectric layers separately
and consequently permits the nanomechanical actuator to op-
erate as a unimorph 共single piezoelectric layer actuated兲 or as
a bimorph 共two layers actuated in opposite directions兲 struc-
ture. This lead to the demonstration of bimorph actuation at
the nanoscale using AlN piezoelectric films that have pre-
served the same stress-free state and high piezoelectric coef-
ficients of their macroscopic counterparts.
One of the biggest hurdles in the fabrication of
multilayer structures is the presence of residual stress 共and
stress gradients兲 in the composite released structure. These
unwanted stresses arise from both the thermal expansion
mismatch between adjacent layers in the stack and intrinsic
residual stresses, which can be affected by the parameters
used to control the deposition of each layer. In this effort,
highly c-axis oriented and low stress AlN thin films were
deposited by a dual cathode S-Gun magnetron using ac re-
active sputtering technology of Tegal Corporation.
16
It was
experimentally found that both the in-plane stress and the
stress gradient of AlN can be controlled by preheating the
substrate before and during a fraction of the deposition pro-
cess.
X-ray diffraction analysis of these thin AlN films has
shown that the full width at half maximum 共FWHM兲 varies
from 2.1° to 4.4° for AlN films of thickness 100–200 nm.
According to rocking curve measurements performed on
films whose thickness ranged from 100 nm to 1
m, it can
be concluded that the FWHM depends on the substrate used
for deposition. For example, the FWHM on bare silicon was
found to be consistently lower than on patterned platinum.
However, the results show that the change in FWHM does
not significantly affect the piezoelectric coefficient of AlN.
Further studies were carried out to analyze the break-
down characteristics and relative dielectric constant
r
of the
thin AlN films using a capacitance test structure formed by
two Pt layers 共50 nm thick兲 sandwiching a 100 nm thick AlN
layer. The I-V characteristics of these films, measured using
a Keithley 6517A electrometer, show a low leakage current
of ⬃2nA/ cm
2
at 1 V over a 13⫻13
m area. The approxi-
mate breakdown voltage of 100 nm thick films is measured
to be 12 V.
Thin film AlN offers a multitude of advantages for elec-
trical devices operating at high frequencies, such as NEMS
switches, as it possesses a relatively high dielectric strength
共thus resisting breakdown兲 and low dielectric constant 共thus
requiring low power to actuate兲. Since AlN does not exhibit
a Curie temperature, it is an ideal candidate for high tem-
perature and high frequency applications that need low
power to function. For the proposed ultrahigh frequency 共i.e.,
300 MHz to 3 GHz兲 and super high frequency 共i.e., 3 GHz to
30 GHz兲 operations it becomes imperative to study the rela-
tive dielectric constant of AlN at high frequencies. Prelimi-
nary two-port capacitance measurements on the same struc-
tures used for the leakage measurements were performed in a
Lakeshore rf probe station with an Agilent PNA-L N5230
network analyzer. An identical open structure was de-
embedded from the direct measurements and the resultant
admittance was fitted to an equivalent series LC-circuit. An
average value of 9.2 was measured for the dielectric con-
stant. This value is similar to what has been recorded for
thicker AlN films.
17
Finally, dc voltage-induced nanobeam deflections were
measured by a Zygo optical profilometer in air. Figure 2
shows the excellent agreement between the predicted and
measured deflection data for an 18
m long composite
nanobeam 共350 nm thick with two 100 nm AlN layers, as
discussed above兲 when operated as a bimorph. The beam
exhibits linear voltage-deflection behavior, showing a deflec-
tion of ⬃116 nm upon the application of 6 V as compared to
⬃118 nm predicted by finite element method 共FEM兲 simu-
FIG. 1. 共Color online兲 The false-colored scanning electron micrograph
共SEM兲 was taken with the sample tilted at 52° to the horizontal. The sample
was tilted to provide a three-dimensional view of the nanoactuator and show
its constituent layers 共350 nm thick stack formed by two 100 nm thick AlN
layers and three 50 nm thick Pt layers兲. The released cantilevered beam also
shows very small out-of-plane deflections. This is indicative of low levels of
stress gradients in the nano-AlN films. The inset schematic illustrates a
cross-sectional view of the stack of layers used to make the device and the
operating principle of the bimorph nanoactuator.
FIG. 2. 共Color online兲 Excellent agreement exists between experimental
data and FEM analysis of the vertical deflection for an 18
m long, 4
m
wide, and 350 nm thick composite beam operated as a bimorph.
053106-2 Sinha et al. Appl. Phys. Lett. 95, 053106 共2009兲
Downloaded 14 Sep 2009 to 130.91.117.41. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
lations 共for a rigidly fixed beam of the same dimensions兲
using
COMSOL. Both layers of AlN have also been individu-
ally actuated and the deflection measurement for each indi-
vidual layer closely matches the FEM predictions. The
agreement between simulation and experimental data allows
accurate prediction of the deflection behavior of AlN nano-
structures by FEM simulations.
These experimental deflection measurements along with
FEM simulation results and analytical solutions for the de-
flection of piezoelectric unimorph and bimorph beams
18
were used for extracting the experimental d
31
piezoelectric
coefficient of the ultrathin AlN films. The extracted experi-
mental d
31
is 1.92⫾ 0.14 pC/ N for the composite beam,
1.98⫾0.16 pC/ N for the top layer, and 1.89⫾0.16 pC/ N
for the bottom layer. It is important to note that AlN nano-
actuators offer a highly linear and reversible stroke that will
greatly benefit active mechanisms for opening and closing
contacts such as in piezoelectric contact switches.
Figure 3 summarizes the results for the testing of nano-
beams of different lengths 共5–18
m兲 under unimorph and
bimorph actuation at 4 V. Their deflections match well with
FEM analysis and verify the dependence of deflection on the
square of the beam length as predicted by conventional Euler
beam-bending equations. This dependence on the square of
the length shows that the piezoelectric layer generates a con-
stant moment throughout the length of the beam.
In summary, this letter reports the implementation of ul-
trathin 共100 nm兲 AlN piezoelectric films for the fabrication
of vertically deflecting NEMS devices. The optical interfero-
metric measurements of the deflection of the nanoactuators
match closely with the FEM simulations. Electrical proper-
ties of the films were measured and found to be similar to
microscale devices. The demonstrated electromechanical
properties make this thin film AlN technology amenable for
applications such as nanomechanical low power logic,
atomic force microscopy tip transduction, nanoresonator-
based sensing, and energy harvesting.
The authors would like to acknowledge the staff of the
Wolf Nanofabrication Facility at the University of Pennsyl-
vania for their help.
1
D. C. Judy, R. G. Polcawich, and J. S. Pulskamp, Technical Digest of the
Solid-State Sensors, Actuators, and Microsystems Workshop, 2008, pp.
328–331.
2
D. C. Judy, J. S. Pulskamp, R. G. Polcawich, and L. Currano, Proceedings
of the 22nd IEEE International Conference on Micro Electro Mechanical
Systems, 2009, pp. 591–594.
3
M. L. Roukes, Tech. Dig. - Int. Electron Devices Meet. 2004, 539.
4
K. E. Moselund, D. Bouvet, M. H. Ben Jamaa, D. Atienza, Y. Leblebici,
G. De Micheli, and A. M. Ionescu, Microelectron. Eng. 85, 1406 共2008兲.
5
W. W. Jang, J. O. Lee, J. Yoon, M. Kim, J. Lee, S. Kim, K. Cho, D. Kim,
D. Park, and W. Lee, Appl. Phys. Lett. 92, 103110 共2008兲.
6
H. G. Craighead, Science 290, 1532 共2000兲.
7
I. Bargatin, I. Kozinsky, and M. L. Roukes, Appl. Phys. Lett. 90, 093116
共2007兲.
8
K. Uchino, Piezoelectric Actuators and Ultrasonic Motors 共Springer, New
York, 1996兲.
9
N. Sinha, R. Mahamameed, C. Zuo, M. B. Pisani, C. R. Perez, and G.
Piazza, Technical Digest of the Solid-State Sensors, Actuators, and Micro-
systems Workshop, 2008, pp. 22–25.
10
R. Mahameed, N. Sinha, M. B. Pisani, and G. Piazza, J. Micromech.
Microeng. 18, 105011 共2008兲.
11
H.-C. Lee, J.-H. Park, J.-Y. Park, H.-J. Nam, and J.-U. Bu, J. Micromech.
Microeng. 15, 2098 共2005兲.
12
R. G. Polcawich, J. S. Pulskamp, D. Judy, P. Ranade, S. Trolier-
McKinstry, and M. Dubey, IEEE Trans. Microwave Theory Tech. 55,
2642 共2007兲.
13
C. Zuo, N. Sinha, C. R. Perez, R. Mahamameed, M. B. Pisani, and G.
Piazza, Technical Digest of the Solid-State Sensors, Actuators, and Micro-
systems Workshop, 2008, pp. 324–327.
14
R. Ruby, P. Bradley, J. Larson III, Y. Oshmyansky, and D. Figueredo, Dig.
Tech. Pap.-IEEE Int. Solid-State Circuits Conf. 438, 120 共2001兲.
15
M. Rinaldi, C. Zuniga, and G. Piazza, Proceedings of the 22nd IEEE
International Conference on Micro Electro Mechanical Systems, 2009, pp.
916–919.
16
V. Felmetsger, P. Laptev, and S. Tanner, Proceedings of the 2008 IEEE
International Ultrasonics Symposium, 2008, pp. 2146–2149.
17
J. S. Thorp, D. Evans, M. Al-Naief, and M. Akhtaruzzaman, J. Mater. Sci.
25, 4965 共1990兲.
18
D. L. DeVoe and A. P. Pisano, J. Microelectromech. Syst. 6,266共1997兲.
FIG. 3. 共Color online兲 Results of deflection testing of beams of different
lengths with unimorph actuation 共single layer兲 and bimorph actuation
共double layer兲 at 4 V and their agreement with FEM analysis. The dashed
line shows the square dependence that exists analytically between tip dis-
placement and beam length.
053106-3 Sinha et al. Appl. Phys. Lett. 95, 053106 共2009兲
Downloaded 14 Sep 2009 to 130.91.117.41. Redistribution subject to AIP license or copyright; see http://apl.aip.org/apl/copyright.jsp
