This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
 Author's personal copy

Nano Energy (2012) 1, 13–24

Available online at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/nanoenergy

REVIEW

Piezoelectric nanogenerators—Harvesting ambient

mechanical energy at the nanometer scale
Xudong Wang

Department of Materials Science and Engineering, University of Wisconsin–Madison, WI 53706, USA

Received 19 August 2011; received in revised form 16 September 2011; accepted 16 September 2011
Available online 4 October 2011

KEYWORDS Abstract
Nanogenerator; Harvesting ambient mechanical energy at the nanometer scale holds great promises for powering
Piezoelectric; small electronics and achieving self-powered electronic devices. The self-powering capability allows
ZnO; electronic device packages to exclude bulky energy storage components and makes possible forgoing
Mechanical energy the inclusion of bulky battery components. Recent development of nanogenerators (NGs) has
harvesting;
demonstrated a possible solution for the design of self-sufficient power source that directly draws
Nanowire
energy from ambient mechanical resources. Piezoelectric nanowires (NWs) are the building blocks of
NGs. In this review paper, theoretical calculations and experimental characterization methods for
predicting or determining the piezoelectric potential output of NWs are reviewed first.
Representative models of NGs are then discussed for harvesting mechanical energy from high-
frequency acoustic waves and low-frequency vibrations/frictions. A numerical calculation is also
presented to estimate the energy output from NW-based NGs. A potential practical application of NGs
for harvesting energy from respiration is shown using piezoelectric polymer thin films. At the end,
perspectives of the NG concept are discussed. The nanometer-scale piezoelectric and mechanical
properties, the piezotronic effect, and large-scale manufacturing capability are suggested to be the
essential aspects that would eventually lead the promising NG concept to a practical power source.
& 2011 Elsevier Ltd. All rights reserved.

Introduction potential for remote/wireless sensing, charging batteries, and

powering electronic devices [3–8].
Vibration-based mechanical energy is the most ubiquitous Devices to harvest the vibration-based energy have been
and accessible energy source in the surroundings. For successfully built on three different principles: piezoelectric
random vibrations with frequencies from hundreds Hz to [9], electrostatic [10], and electromagnetic [11] transducers
kHz, the available energy density is within the range of a in micro-electromechanical systems (MEMS). Among these
few hundred microwatt to milliwatts per cubic centimeter techniques, piezoelectric transducer is distinguished by its
[1,2]. Therefore, harvesting this type of energy offers a great high practical output power and relatively easier manufac-
turing. The designs are mostly based on the cantilever beam
configuration and a single operational unit is typically in the
E-mail address: [email protected] size range from hundreds of micrometers to centimeters.

2211-2855/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.nanoen.2011.09.001
 Author's personal copy

14 X.D. Wang

The relatively large size and the ceramic-thin-film-based (2) Superior mechanical properties. The lattice perfection
functional unit may restrict its compatibility to nano- or of NWs enables much larger critical strain, higher
micro-electronic systems and further improvement of the flexibility, and longer operational lifetime.
power density. (3) High sensitivity to small forces. Large aspect ratio and
In addition, with the rapid development of nanotechnol- small thickness allow the creation of significant strain in
ogy, solar, thermal, and chemical energy harvesting tech- the NWs under a force at the nano-Newton or even pico-
nologies are experiencing a magnificent improvement on Newton level.
their building blocks, designs, efficiencies, as well as costs.
Therefore, although the merits of scavenging ambient mecha- Herein, we will first review and discuss the theoretically
nical energy are unique, whether this technique can compete predicted potential–strain relationship and the flexoelectric
for entry into the mainstream of the renewable/alternative enhancement effect that are essential for predicting and
energy field would largely rely on the creation and develop- evaluating the potential of a piezoelectric NW for mechan-
ment of new materials, new designs, and even new principles ical energy harvesting. The experimental characterization
that may take the advantage of nanotechnology in shrinking the of the piezoelectric output on different piezoelectric materials
size and improving the performance [12,13]. will also be reviewed.
Recently, piezoelectric ZnO nanowires (NWs) have been
demonstrated as a promising concept to harvest micro- and Theoretical predictions
nano-scale mechanical energy from the surrounding—the
nanogenerator (NG) [14]. This device has the potential to When a piezoelectric NW is deflected, a piezoelectric potential
fundamentally improve the mechanical energy harvesting can be generated on the side surfaces due to the crystal lattice
capability with advanced NW building blocks and compact distortion. For ZnO NWs, the tensile side surface gives a
designs, which might eventually lead to an effective power positive potential; while a negative potential appears on the
source for self-powered electronic systems with higher compressive side surface. In 2007, Gao and Wang did a
energy density, higher efficiency, longer life time, as well theoretical calculation on the potential distribution in an ideal
as lower cost [15–17]. In this review article, the fundamental ZnO NW, which is a dielectric material without free charge
operation principles, energy harvesting capabilities, and carriers and subjecting no body force [19]. Perturbation
practical designs of piezoelectric NGs will be reviewed and expansion was applied to solve the potential distribution by
the perspective of this novel technique will be discussed. assuming the NW has a cylindrical shape with a uniform cross-
section. Calculation showed that the piezoelectric potential in
the NW is independent on the length position. Thus, the NW
Piezoelectric NWs for mechanical energy
would act approximately like a ‘‘parallel plate capacitor’’
harvesting (Fig. 1a). The calculated electric field distribution on the NW
cross-section is shown in Fig. 1b, where the highest potential
NW is a one dimensional (1D) nanomaterial that typically has of 70.27 can be identified. This NW has a diameter of 50 nm
a diameter less than 100 nm and a length more than 1 mm. and a length of 600 nm, and is bent 145 nm laterally at the tip
Most ceramic NWs are single crystals. Compared to the by an 80 nN lateral force. The numerical calculation showed
conventional thin-film-based piezoelectric cantilever trans- that the maximum piezoelectric potential that can be
ducers, using piezoelectric NWs for mechanical energy generated at the NW surface is directly proportional to the
scavenging offers three unique advantages: lateral displacement of the NW and inversely proportional to
the length-to-diameter aspect ratio of the NW.
(1) Enhanced piezoelectric effect. 400–500% enhancement It should be noted that the above calculation was based
of the piezoelectric effect was predicted due to the on the assumption that there is no free charge carrier
flexoelectric effect [18], when a strain gradient is inside the ZnO NW. In such a case, ZnO was considered as a
experienced by a ferroelectric NW with a thickness of perfect insulating piezoelectric material. In reality, ZnO is a
a few tens of nanometers. wide-bandgap semiconductor material and has considerably

Figure 1 Potential distribution for a ZnO NW. (a) Side view of the potential profile. (b) Cross-sectional view of the piezoelectric
potential [19]. (c) Cross-sectional view of the piezoelectric potential when the contribution of free electrons is considered [20].
 Author's personal copy

Piezoelectric nanogenerators 15

amount of intrinsic defects that makes it n-type with a uniform cantilever beams, numerical analysis was applied to
typical dopant concentration at the level of 1017 cm3. ZnO and BaTiO3. BaTiO3 is a perovskite materials and ideally
Thus, the free charge carrier effect needs to be considered has a rectangular cross-section (Fig. 2a), which would
for predicting the piezoelectric potential output. produce uniform potential on the two side surfaces. ZnO
Fig. 1c shows the overall potential distribution on a cross has a hexagonal cross-section (Fig. 2b) and the piezoelectric
section perpendicular to the NW axis when the free charge potential was estimated on the two opposite planes of
contribution to the piezoelectric potential is considered the hexagon. Thus, the maximum piezoelectric potential
[20]. This calculation was done by using the same ZnO of rectangular and hexagonal NWs can be expressed by
constants as those used in the non-free-charge calculation Eqs. (1) and (2), respectively:
and by selecting a typical doping concentration of 1  1017
cm3. The electric potential maximum in the positive side of fy ð1þvÞb3 e15
DVmax,rNW ¼ ð1Þ
the NW is significantly reduced from 0.3 V in Fig. 1b, which 6Ixx ½2ð1þvÞe215 þ k11 E
corresponds to an insulator case, to less than 0.05 V in
pffiffiffi
Fig. 1c, which considers the moderate doping in ZnO. The 3fy ð19 þ16vÞb3 e15
potential on the negative potential side is very well preserved. DVmax,hNW ¼ ð2Þ
36Ixx ½2ð1 þvÞe215 þk11 E
The significant potential quenching at the positive side is due
to the transportation of free electrons from substrate to the Above equations were used to estimate the potential-to-
positive potential region. The negative potential side only force relations of NWs when they are deflected under a
interacts with the positively charged ionized donors. There- constant external force (fy). Both ZnO and BaTiO3 in this
fore, at very high doping level, i.e. 41018 cm3, the negative analysis were considered as a perfect dielectric medium
potential side could be completely screened and the overall with negligible free charge carrier density. The maximum
NW would exhibit no net potential output. piezoelectric potentials were plotted as functions of the
A comprehensive calculation was also presented to reveal external force and their dimensions, as shown in the insets
the size-force-potential relationship for different piezoelectric of Fig. 2c and d for BaTiO3 and ZnO NWs, respectively. In
nanomaterial morphologies [21]. By assuming the NWs are general, their maximum potential is proportional to the

fy
24 l
0.75
20
Voltage (V)

b b 18
10 Voltage (V)
1
0
l 0.8
12 20
10 150 1.5
z 0 50 0.6
2
6 3 0.4
5
0.2
o y 0
x 50 100 150 200 0.01
Width (nm)

fy
l

8
100
Voltage (V)

Voltage (V)
b
6 50 0.75

1 8
0
z l 4 9
6
3 60 6
020 1.5
2 2
4
3
o 5 2
x y 0 20 40 60 80 100
Side width (nm)

Figure 2 Schematic structures and the coordinate systems of (a) a rectangular NW and (b) a hexagonal NW. The static analysis of
the maximum allowable piezoelectric potential that can be generated by (c) a BaTiO3 NW and (d) a ZnO NW. Insets are the
corresponding full range plots of the piezoelectric potential to the force and size relationships [21].
 Author's personal copy

16 X.D. Wang

elastomers [29], and even biological species [30]. Fig. 3a

illustrates a net dipole moment resulted from an inhomoge-
neous strain in a NaCl crystal. Once a strain gradient appears
due to bending of the crystal, flexoelectricity will be generated
along the gradient direction (Fig. 3b) [31].
Nevertheless, such an effect is usually negligible for most
dielectric materials due to the low flexoelectric coefficient
(f), which is of the order of e/a (or 1011–1010 C/m), where
e is the electronic charge and a is the lattice parameter. f is
also related to the dielectric susceptibility w as f =gw(e/a),
where g is a material related constant [32]. This relation
Figure 3 (a) Schematic of a NaCl unit cell under non-uniform suggests that high dielectric permittivity materials could
strain. (b) Polarization in NaCl crystal due to bending. possibly exhibit high flexoelectric effect [33,34].
Following this prediction, Ma and Cross determined the
external force and inversely proportional to the thickness or
flexoelectric coefficients of perovskite ferroelectric materials,
width but not related to their length. In real cases, the
including lead magnesium niobate (PMN) [35,36], barium
deflection of nanostructures is restricted by their mechanical
strontium titanate (BST) [37], lead zirconate titanate (PZT)
strength and flexibility, thus the highest potentials plotted in
[38,39], and BaTiO3 [40]. These materials exhibited much
the insets of Fig. 3c and d are practically impossible. In order
higher f values at the level of 10–6 C/m owing to their large
to reveal the achievable piezoelectric potentials by NWs, the
permittivities (3000–11000). The constants g are also related to
mechanical limitation boundaries were calculated and super-
the chemistry of materials due to the different ionic bonding
imposed into the potential plots.
strength. The Ba-based perovskites showed much higher
Thus, the mechanical limitation boundaries (dark lines in
g values ðgBST 9:3; gBaTiO3 12Þ; while the g value of other
the insets of Fig. 3c and d) were calculated at 7.0% strain (from
perovskites were below one (gPMN0.65, gPZT0.57). These
reported data on ZnO NWs [22]) and 7301 deflection angle,
results indicated that Ba-based perovskites could be better
whichever was reached first. The boundary lines for NWs with
candidates if the flexoelectric effect is of major interest.
various lengths were determined and the allowable potential
When the lattice size is only an order of magnitude or two
distributions inside the boundary lines are shown in Fig. 2c and
smaller than the width of the beam, as is in the case of a
d for NWs with lengths of 0.75 mm, 1 mm, 1.5 mm, 2 mm, 3 mm,
NW, the strain gradient and the resulting flexoelectric effect
and 5 mm.
would become significant [41,42]. For nanosized piezo-
Comparing BaTiO3 and ZnO NWs with the same size and
electric materials, the piezoelectricity would be further
under the same force, ZnO exhibits much higher piezoelectric
enhanced by the appreciable flexoelectricity. The mathe-
potential, although its piezoelectric coefficient is much
matic framework presented by Majdoub et al. predicted the
smaller than that of BaTiO3. The reason for BaTiO3 NWs to
size-dependent flexoelectric enhancement on the piezo-
show lower potential is attributed to their much larger
electricity of ferroelectric nanostructures [18]. A non-linear
dielectric constant, which leads to a large capacitance. It is
increase of the effective piezoelectric coefficient (deff) was
known that the piezoelectric effect directly induces charge
suggested when the beam thickness decreases. Using the
and the piezoelectric potential is derived from the amount of
flexoelectric coefficients estimated from ab initio calcula-
charge via the relation V=Q/Cp. Therefore, significantly larger
tions [43] (5.46 nC/m), a more than 400% enhancement on
dielectric constant of BaTiO3 would induce a lower effective
the effective piezoelectric coefficient (the overall measure
voltage appearing along the NW surface.
of the electromechanical coupling considering both piezo-
electric and flexoelectric effects) was predicted on tetra-
Flexoelectric enhancement at the nanometer scale gonal BaTiO3 beam when the thickness was below 1 mm [44].
The same group also estimated the energy harvesting
When a NW is applied for mechanical energy harvesting, the capability of ferroelectric beams when the flexoelectric
NWs are acting as cantilevered beams with their one end fixed effect was considered [45]. Due to the electromechanical
on the substrate and the other end free-standing. Thus, the coupling effect, the beam bending rigidity will increase with
piezoelectricity is generated by the bending induced lattice the enhancement of polarization. This is reflected by the
distortion rather than the uniform compression/extension increasing of ‘‘effective coupling coefficient’’ (xeff). Com-
along one axis. Under this circumstance, one important effect pared to the known value of PZT beam harvester [46], the
has to be considered is the flexoelectric effect [23,24], which harvested power was expected to show a 50–100% improve-
is the polarization induced by inhomogeneous strain and is ment when the beam thickness is below 50 nm in the short
defined by the relation [25] circuit resonance mode.
@Ejk
Pi ¼ fijkl ð3Þ
@xi Characterization of piezoelectric potential from
where fijkl is the fourth rank tensor (or flexoelectric coeffi- piezoelectric NWs
cient), Pi is the polarization, Ejk is the elastic strain, and xi is
the direction of the strain gradient. The flexoelectricity is a Contact AFM scanning approach
more general property than piezoelectricity since a nonuniform The first detection of strain-induced potential from piezo-
lattice deformation could induce a charge displacement even in electric NWs was performed by AFM on vertically aligned
centrosymmetric systems [26,27], liquid crystals [28], soft ZnO NW arrays [47]. The ZnO NWs were grown on GaN
 Author's personal copy

Piezoelectric nanogenerators 17

substrate and connected through a silver bottom electrode Similar experiments have also been conducted on larger
(Fig. 4a). A conductive AFM tip (Si coated with Pt) was used sized single ZnO wire, and 10 mV potential outputs was
to create deformation of the vertical NWs and collect detected each time when the Pt-coated AFM tip bent
electrical signal simultaneously. Through a contact mode, the wire and scanned across the wire and reached the
the AFM tip scanned across the NW tips and bent the NWs other negatively charged side [48]. Tens of mV output was
laterally, as shown in Fig. 4b. Bending of the ZnO NW typically observed from ZnO NWs, although the theoretical
created positive potential on the stretching side and prediction has suggested a much higher potential output.
negative potential on the compressive side. Because the The big discrepancy can be associated with the contact issue
metal–semiconductor contact between the Pt-coated AFM between the AFM tip and the NW surface, as well as the
tip and the ZnO NW is a Schottky-type contact, only when relatively high intrinsic doping concentration of ZnO
the semiconductor side has lower potential than the metal synthesized in regular vapor-based growth processes.
side, a forward-biased connection is formed, thus free
charges can flow through and neutralize the piezoelectric Three dimensional Kelvin probe microscopy (3DKPM)
potential. Little charge flow is allowed when the ZnO side approach
exhibits higher potential than the metal side (the reversely Surface potential can also be measured in a non-contact way
biased connection), which is the condition when the AFM by scanning Kelvin probe microscopy (SKPM). However, a
pushes the stretching side of the NW. Under this situation, general class of measurement errors is always introduced by
piezoelectric energy would accumulate inside the NW, and sample-topographical effects, especially the strong depen-
then be partially released in a pulsed form when the AFM tip dence of the measured surface potential on distance between
touches the other side. This condition can be considered as the AFM tip and sample, thus it is not able provide reliable
integrating a regular piezoelectric beam with a diode characterization on NWs due to the large topography
together into one system. variation. Recently, Bayerl and Wang devised a new 3DKPM
The Schottky contact between the Pt surface and ZnO NW technique to transcend the limitations of SKPM and enable
gives unipolar electrical output pulses when the AFM tip quantitative surface potential mapping of high topography
scans a ZnO NW array, as shown in Fig. 4c. The output surfaces [49].
potential was measured on an external load of resistance 3DKPM entails scanning an AFM probe configured for non-
RL =500 MO and most peaks were identified as 6–9 mV. contact SKPM operation. The probe is sequentially moved to

AFM tip

Vs>0 (Vs)+

Vs=0

Vs<0 (Vs)-

ZnO NW

Ag

Figure 4 (a) Vertically aligned ZnO NW arrays. (b) Schematic diagram showing the AFM-based measurement of the bending-induced
piezoelectric potential on ZnO NWs. (c) Electric signal collected by AFM on a ZnO NW array [47].
 Author's personal copy

18 X.D. Wang

each designed testing position where it is lifted to a uniform left (compressed) and right (stretched) sides is apparent near
height above the surface and slowly lowered until the probe the surface of the bent NW (Fig. 6a); while the straight NW
achieves contact with the surface (Fig. 5a). The potential gave a fairly symmetric potential distribution (Fig. 6b). By
recorded during the approaching corresponds to the potential extracting the potential difference on the two side surfaces,
distribution in the column of space above each testing position. the piezoelectric potential was identified to be 400–600 mV
Such spatially resolved potential data can then be recon- from different bending conditions (Fig. 6c), which matches
structed into a 3D image throughout a volumetric scan. Fig. 5b well with theoretical calculations [19]. 3DKPM methods
shows the spatial potential distribution in a plane bisecting a therefore have the potential to become an invaluable tool
single pair of Au electrodes (left one grounded and right one for ongoing characterization of size-enhanced piezoelectric
biased by 100 mV), which is qualitatively identical to the finite polarization effects in piezoelectric micro/nanostructures of
element simulation (not considering the potential scaling high topography and low dimension.
factor) evidencing the accurate spatial potential mapping
capability of 3DKPM. The surface potential identified from the
electrode surface scaled linearly with the applied bias (Fig. 5c) MEMS-based approaches
showing that the 3DKPM technique can accurately quantify the MEMS-based devices have also been developed for char-
potential of charged surfaces. 3DKPM can accurately map the acterizing the piezoelectric potential of NWs. By integrating
electric potential in a 3D space with an electric potential a BaTiO3 NW with a mobile base that can provide a precise
gradient resolution of at least 5 mV per 50 nm. linear motion (resolutiono1 nm), the piezoelectric re-
With the capability for high-resolution surface potential sponses were determined under axial strains [50]. The maxi-
quantification established, 3DKPM was applied to characteriz- mum piezoelectric potential was found to be 25 mV and the
ing the piezoelectric potential of ZnO NWs. Spatial potential output potential was linearly related to the driving ampli-
mapping was obtained on a ZnO NW when it was bent and tude. Such output voltage amplitude was comparable to
straight. Asymmetry in the potential distribution between the those determined from ZnO NW, although the piezoelectric

3DKPM
Potential (mV)

Simulation
Potential (mV)

Vapplied
Au electrode
standard sample

Figure 5 (a) Schematic of the 3DKPM approach. (b) 3DKPM measured and simulated potential distribution above a pair of
electrodes. (c) Plot of measured potential versus applied potential.

Figure 6 3DKPM potential maps of (a) bent and (b) straight ZnO NW. (c) Piezoelectric potential quantification of ZnO NWs with
multiple bending configurations (insets).
 Author's personal copy

Piezoelectric nanogenerators 19

coefficient of BaTiO3 is higher. As discussed earlier, this is Ultrasonic wave

due to the higher dielectric constant of BaTiO3. From the Silicon zigzag electrode
number of charge produced by the BaTiO3 NW, the output coated with Pt
electrical energy per straining cycle was calculated to be
0.3 aJ at driving amplitude of 5 V.
A bridge-type device was made by placing a poly(vinyli-
dene fluoride) (PVDF) fiber across a pair of electrode on a ZnO NWs
flat flexible substrate [51]. The voltage and current signal
was measured from the PVDF nanofiber by stretching and Bottom conductive electrode
releasing the flexible substrate at a constant rate (2 Hz).
The typical voltage was measured to be 5–30 mV and
current was 0.5–3 nA. Integrating the product of voltage
and current with respect to time gives the overall electric
energy output. Comparing this energy to the input mechan-
ical energy that was needed to create the strain, the energy
conversion efficiency was estimated to be as high as 21.8%. 300 nm
Such a value is much higher than those can be achieved by
PVDF thin films (typically 0.5–4%). It was suggested that the
higher energy conversion efficient of PVDF nanofibers was
due to the much smaller domain wall motion barrier in PVDF
nanofibers (ca. 0.01% strain) compared to that of PVDF thin
films (ca. 0.3% strain). Thus, higher polarization could be
produced by the nanofibers under the same strain.

Prototypes of piezoelectric NGs

The discovery of appreciable piezoelectricity from ZnO

NWs has led to the development of a series of NW-based
piezoelectric NGs. Most recent models have showed the
capability of powering small electronic devices. Represen-
tative models of piezoelectric NGs will be reviewed in this Voltage (V)
section.

Ultrasonic wave-driven NGs—the first prototype

VOC = 0.01V
JNG (µA/cm2)

Integrating a Pt coated serrated electrode with vertically

aligned ZnO NWs created the first NG prototype that
converted ultrasonic waves into electricity [14]. As sche-
matically shown in Fig. 7a, an array of aligned ZnO NWs was
covered by a serrated Si electrode coated with Pt. The Pt
coating not only enhanced the conductivity of the elec-
trode, but also created a Schottky contact at the interface ISC = 1.1Acm-2
with ZnO. The NWs were grown on GaN substrates and the
top electrode was composed of parallel serrated trenches Figure 7 ZnO NW-based piezoelectric NGs. (a) Schematic
fabricated on a (0 0 1) orientated Si wafer and coated with a diagram showing the design and structure of the NG. (b) Cross-
thin layer of Pt. A cross-sectional image of a packaged NG is section of the NG showing the integration of aligned NWs and
shown in Fig. 7b. Some NWs were in direct contact with the the top electrode. Inset shows a NW that is forced by the
top electrode, but some were located between the teeth of electrode to bend [14]. (c) A typical DC current output of an
the electrode. Under the application of ultrasonic waves, ultrasonic-driven NG [52]. (d) J–V curve of the NG during normal
the top electrode would move downward and push the NW operation, where the short-circuit current and open-circuit
leading to a lateral bending. The bending-induced piezo- voltage can be identified [53].
electric potential would then be collected by the Pt film when
the Schottky junction between the NW and Pt electrode was
forward biased. Schottky barrier. When the connection of electrical mea-
Due to the in situ rectifying effect of the Schottky surement system was switched between the two electrodes
contact, the detected output electric signal exhibited a of the NG, the polarization of the output signal would be
direct current (DC) characteristic. Although each NW would reversed accordingly but the signal amplitude was unaf-
still produce electric pulses, accumulation of the output fected, as shown by the black curve in Fig. 7c [52]. This
from all NW generated a constant DC current signal with phenomenon further proved the Schottky effect to the NG
large noises, as shown in Fig. 7c. The polarization of the DC output and it is considered as a necessary condition to
signal was consistent with the bias direction of the Pt–ZnO distinguish valid signals from artifacts.
 Author's personal copy

20 X.D. Wang

The DC output of ultrasonic wave-driven NGs allows the nanowires rooted at the other fiber. Thus, when there was
performance characterization by connecting a variable DC a relative sliding/deflection between them, the bending of
voltage source as the load, which is a condition similar to the uncoated ZnO NWs produced a piezoelectric potential
that of solar cell characterization. A typical J–V curve of a across their width, and the Au-coated NWs acted as the
NG is shown in Fig. 7d demonstrating an open circuit voltage electrode for collecting and transporting the charges.
of 10 mV and a short circuit current density of 1 mA/cm2, The performance of the fiber NGs was characterized by
which suggested an optimum output power of 10 nW/cm2. measuring the short-circuit current (Isc) and open-circuit
From the shape of the curve, the fill factor (FF) was voltage (Voc) when the gold-coated fiber was pulled and
determined to be 25%. Thus the operational power of that released at a controlled frequency. 5 pA current pulses
particular NG was 2.5 nW/cm2 [53]. The low FF was likely were detected at each pulling-releasing cycle (blue curve in
associated with the large internal resistance due to a large Fig. 8d). Negative current pulses with the same amplitude
number of branching NW and the poor contact between NWs were received (pink curve in Fig. 8d) when the current
and the Pt electrode. In an ideal case, if each NW were meter was reverse-connected. The small output current is
involved in the energy conversion process, the output power attributed mainly to the large loss in the fiber due to an
would be raised to the mW/cm2 level or higher [54]. extremely large inner resistance (RiE250 MO). The open-
circuit voltage was measured to be 1–3 mV from the same
Power fibers fiber NG. The piezoelectric output can be enlarged by
integrating multiple fibers together. In a small yarn contain-
ing 6 fibers (three covered with Au-coated NWs, and three
In addition to the ultrasonic wave driven NG, a textile fiber
covered with untreated ZnO NWs), an average current of
based NG has been developed for harvesting low-frequency
0.2 nA was achieved, which was 30–50 times larger than
vibration/friction energies [55]. The fibers used in the
the output signal from a double-fiber NG. This is due to the
fabrication were Kevlar 129 fibers, on which ZnO NWs were
enhanced surface contact area among those fibers. Reducing
grown radially (Fig. 8a). Along the entire length of the fiber,
the inner resistance of the fiber and the NWs was also found
ZnO NWs exhibited a very uniform coverage and well
effective for enhancing the output current. By depositing a
preserved cylindrical shape (Fig. 8b). The spaces between
conductive layer directly onto the fiber before growing the ZnO
the NWs are large enough for them to be bent to generate
NWs, the output current Isc of a double-fiber NG was increased
the piezoelectric potential.
from 4 pA to 4 nA. Isc was approximately inversely propor-
A double-fiber system was designed to demonstrate the
tional to the inner resistance of the NG. Based on the current
capability of harvesting mechanical energy. As shown in
and voltage data received, calculation showed that weaving a
Fig. 8c, two fibers, one coated with a 300-nm-thick gold
large number of such fibers together would bring the output
layer and the other as-grown, were entangled to form the
power up to 4–16 mW/m2 if the consumption from the device
core for power generation. Once the two fibers were firmly
resistance and capacitance were neglected.
entangled together, some of the gold-coated NWs pene-
trated slightly into the spaces between the uncoated
Prediction of the power output from piezoelectric
NWs

The first independently operated NG prototype was designed

to be driven by ultrasonic waves. In fact, stimulating a
piezoelectric beam into vibration has been broadly used for
With Au coating harvesting mechanical energy in most MEMS devices. There-
fore, to quantitatively understand the dynamic responses of
Without Au coating
a piezoelectric NW under vibration is essential for the
development of NGs. The system used for dynamic analysis
is schematically shown in Fig. 9a. The nanostructure was
assumed to have its bottom end fixed on the substrate and
the other end to be free. An external mechanical energy in a
wave form is applied through the substrate and agitates the
vibration of the NW at the same frequency. The external
mechanical energy is also assumed to be continuous,
constant and sufficiently large. Such vibrations will induce
alternating electrical charges along the two side surfaces of
the NW.
Considering each NW as a single degree-of-freedom (DOF)
system and using the strain–potential relationship derived
for a bending NW, the power and efficiency were calculated
Figure 8 ZnO NW-based fiber NGs. (a) A Kevlar fiber uniformly at different lengths and thicknesses by assuming that the
coated with ZnO NW arrays. (b) A cross-section image of the NWs are agitated into oscillation at their resonant frequen-
fiber. (c) Optical image of the NG fabricated by two ZnO NW- cies with a fixed mechanical damping ratio of 0.005 [21].
coated fibers, one with gold coating and one without. (d) The The plots of these relationships for ZnO and BaTiO3 NWs are
current output measured from a double-fiber NG [55]. shown in Figs. 9b and c, respectively. Two general conclu-
 Author's personal copy

Piezoelectric nanogenerators 21

ZnO NW 3.5 BaTiO3 NW

0.8
35

Power (nW)

Efficiency (%)
16 2.5

Efficiency (%)
Power (nW)
0.6
25
0.4 12 1.5
nanowire
Em 8 15
0.2 0.5
4 5
100 150
70 50 4 1
40 4 1 7
10 7

Figure 9 (a) Schematic of a dynamic analysis model, where the NW is agitated into vibration by an external vibration mechanical
energy applied through the substrate. The output power and energy conversion efficiency of (b) ZnO and (c) BaTiO3 NWs at their
resonant frequencies as functions of their length and thickness [21].

sions can be drawn from the plots regardless of the ment of NGs. Its polymeric nature allows it to be easily
morphology and material. First, the output power increases fabricated into a 2D film morphology with a good potential for
with the increasing of the nanostructures’ volume. This scaling up. Sun et al. recently demonstrated an application of
agrees with the property of regular piezoelectric bulk PVDF thin films for harvesting energy from low-speed air flow
materials—more volume brings more piezoelectric power. as well as from simulated human respiration—an important
Second, the energy conversion efficiency at the resonant biological energy source for powering implantable electronic
frequency increases rapidly with the reducing of the devices [64].
nanostructure’s aspect ratio. This suggests that high energy The design for characterizing the capability of harvesting
conversion efficiency would be found in a thin-film-like energy from air flow is schematically shown in the inset of
morphology, which is resulted from the faster decay of the Fig. 10a. A PVDF microbelt (20 mm  2 mm  26 mm) was
mechanical damping energy compared to the electric output tightly suspended across the front edge of a trench which
energy when the aspect ratio decreases. The different was 1 cm deep. The air flow was provided by a nozzle
tracks of the output energy and efficiency indicate the placed 3 cm in front of the trench and the flow speed was
existence of an optimal size range, where the NWs will measured at the position where the microbelt was located.
output a reasonably high power with a reasonably good Direct measurement of the voltage difference between
efficiency. Such a size range can be identified from the plots the top and bottom electrodes of the PVDF microbelt
where the power surface intersects with the efficiency demonstrated that appreciable piezoelectric potential was
surface. For BaTiO3 NWs, their lengths should be 4–5 mm and produced by low-speed air flow. Fig. 10a shows the open-
width should be larger than 100 nm. For ZnO NWs, their circuit voltage (VOC) of the PVDF microbelt as a function of
lengths should be 3–4 mm and side widths should be larger air flow speed. Peak VOC of 76 V was detected at an air
than 50 nm. It should be noted that these predictions were flow speed of 5 m/s. The voltage amplitude monotonically
established based on the assumption that the ambient decreased following the reduction of air flow speed. At the
mechanical energy is always sufficient. Although the plots air flow speed of 1 m/s, the output voltage dropped to less
show a continuous increasing tendency of both the power than 0.1 V due to the very small oscillation amplitude.
and efficiency with the increase of radius or thickness, too By connecting the PVDF microbelt to a capacitor (1 mF)
large size would need a significantly high external mechan- through a bridge circuit, a practical power source was
ical energy to drive the vibration. Under this condition, the demonstrated. Based on the saturation voltage that can be
assumption of unlimited mechanical energy will not apply eventually maintained on the capacitor under different air
and the solution is beyond the capability of this method. flow speeds, the usable energy stored in the capacitor from
microbelt oscillation was determined as a function of the air
Piezoelectric polymer microbelts for harvesting flow speed (Fig. 10b). The maximum stored energy increased
energy from respiration from 8 mJ at 2 m/s to 1.8 mJ at 4 m/s. This energy level is
sufficient to power small electronic devices, which was
It has been clearly shown that increasing the number of NWs demonstrated using a digital stopwatch. As shown in the inset
while maintaining the optimal geometry could substantially of Fig. 10b, the stopwatch was connected as a load on the
improve the output power. In addition to the vertical NW capacitor. By applying constant air flow at a speed of 3.5 m/s,
arrays that were used in the first NG prototypes, laterally the stopwatch was operated normally. Sustained production of
aligning NWs are considered as a good strategy for useful-level electrical power by PVDF microbelts under low-
effectively integrating a large number of NW over a large speed air flow was thus demonstrated. Reducing the thickness
area, thus improve the output of NGs. Many current designs of the microbelts allows this device to harvest energy from
of NGs have such a lateral film-type configuration, which even lower air flow speeds (o1 m/s).
raised the output electric energy to the mW level [56–59]. The small size of PVDF microbelts and their capability
The capability of powering small electronic devices has also of producing mW-level power under low-speed air flow
been successfully demonstrated [58,60–63]. make feasible harnessing energy from respiratory processes.
As we briefly discussed in Section 2.3, PVDF is a piezo- A typical respiration air flow rate follows a sine curve with a
electric polymer that shows good promises for the develop- peak rate of 0.4 l/s (corresponding to 2 m/s considering
 Author's personal copy

22 X.D. Wang

Figure 10 (a) Open-circuit voltage measured under different air

flow speed. Inset is the schematic design of a PVDF microbelt for
harvesting energy from air flow. (b) Energy stored by the capacitor
under different wind speeds. Inset shows an operating stopwatch
powered by a microbelt under air flow speed of 3.5 m/s. Figure 11 Energy harvesting from simulated respiration by
PVDF microbelts. (a) Schematic design for simulated respiration
the average area of adult human nostrils) and a periodicity energy harvesting. (b) Air flow rate through a tubular channel
of 5–6 s [65]. Certainly, lower cut-in speed would allow with a matching pattern to the normal respiration of an adult.
more energy to be harvested by a microbelt. A 17 mm-thick (c) Corresponding VOC measured on a PVDF microbelt fixed at
PVDF microbelt was used for fabricating such a device. The one end of the tubular channel. (d) Voltage increase on a
microbelt was fixed at the center of a cylindrical channel capacitor that was connected to the PVDF microbelt through a
with a cross section area of 1 cm2. One end of the channel bridge circuit. Inset is the long-time voltage increase trend on
was connected to a compressible plastic chamber that the capacitor.
simulated the function of a lung and the other end was open
(Fig. 11a). The plastic chamber was squeezed and released tion’’ flow rate of 0.5 l/s (Fig. 11c). Lower output voltage
every 6 s producing air flow in and out through the tunnel was observed during ‘‘inhalation’’ cycles. This was likely due
with a peak rate of 2.3 m/s (Fig. 11b, calculated from the to asymmetry of the air flow profile between inhalation and
volume change of the plastic chamber per unit time 0.23 l/s). exhalation. In order to demonstrate the feasibility of
This operation provided a close simulation of typical adult’s converting energy from respiration to useful electricity, a
respiration. 1 mF capacitor was connected to the microbelt through a
Resonant oscillation of the microbelt was stimulated by bridge circuit and the increasing voltage curve is shown in
the air flow through the channel and its VOC output was Fig. 11d. It reveals that one respiration cycle could increase
continuously monitored. The voltage output was clearly the capacitor voltage by 5 mV. A majority of voltage
correlated to the instantaneous flow rate and peak voltage increase was registered with the ‘‘exhalation’’ cycles owing
output of 0.4–0.5 V was detected at the highest ‘‘exhala- to the higher voltage output of the microbelt. A continuous
 Author's personal copy

Piezoelectric nanogenerators 23

12 minutes of simulated respiration produced a capacitor piezoelectric potential [69]. Understanding of how electron
voltage output of 200 mV, corresponding to 20 mJ of distribution and semiconductor band structures are inter-
electrical energy storage. This experiment showed that acted with piezoelectric potential would offer new insights
microbelts made from PVDF can oscillate under low-speed on the capability of mechanical energy harvesting.
air flow and generate electrical energy sufficient for Last, fabrication and manufacturing: In addition to above
powering small electronic devices. This capability is ideal scientific issues, the engineering challenges are mostly lying
for harvesting energy from respiration, and thus has the on the piezoelectric nanomaterial fabrication and device
potential for powering implantable biomedical devices. integration. Designing a NG system that can be adapted to
large-scale roll-to-roll processing would substantially reduce
the fabrication cost and eventually lead this exciting
Challenges and opportunities concept toward a practical nanoscale power source.

Using piezoelectric nanomaterials for harvesting mechanical

energy in the micro- or nano-scale is a very promising concept
Conclusion
for realizing self-powered nanoelectronics. To date, various
designs and piezoelectric nanomaterials have been applied to This paper reviews the development of NGs from the funda-
NGs. In order to pave the road toward integratable NGs for mental principles to theoretical predictions to practical
nanoelectronics, the essential goal is to improve the power devices. Compared to conventional piezoelectric material-
density and energy conversion efficiency thus to shrink the based mechanical energy harvesters, using NWs would poten-
device size down to the micrometer level while maintaining a tially improve the energy conversion efficiency, enhance the
reasonably high output power. Despite current research efforts sensitivity to low-level mechanical energy sources, and extend
on material processing and NG design optimizing, there are the device lifetime. ZnO NW is the first nanomaterial that has
several fundamental aspects that must be addressed. been applied for NG development. Leading by the research on
First, piezoelectric effect at the nanometer scale: The ZnO NWs, NWs made from other piezoelectric materials,
piezoelectric nanomaterials used in current NGs are mostly including BaTiO3, PZT, and PVDF were also studied to reveal
in the size range of 4100 nm. Therefore, their piezoelectric their mechanical energy conversation capability. Different
properties are basically the same as their bulk form. designs of NG were implemented mostly based on ZnO NW
Theoretical predictions have shown that within the nan- arrays. Integration of a large number of ZnO NWs was
ometer regime, flexoelectric effect may play a significant demonstrated as an effective pathway for improving the output
role and enhance the electromechanical coupling. Cur- power. The power output from NGs is already sufficient to
rently, experiment-based analysis of the piezoelectric and support the operation of small electronic devices, such as LEDs,
flexoelectric properties at the nanometer scale is very LCDs and NW-based sensors. Addressing several fundamental
limited. Fundamental understandings of the piezoelectric, scientific issues, including the nanometer-scale piezoelectric
ferroelectric, and flexoelectric effects on single-crystalline and mechanical properties and the piezotronic effect would
NWs is of great importance for predicting the piezoelectric substantially enrich our knowledge of NG designing and
potential, quantifying the mechanical-to-electric energy operation principles. Combination of in-depth scientific under-
conversion efficiency, and optimizing the size selection of standing and advanced engineering fabrication approaches
different types of nanomaterials. would eventually lead the promising NG concept to a practical
Second, mechanical property at the nanometer scale: power source for the realization of self-powered nanoelectronic
Number of research has been conducted to reveal the devices.
mechanical property (the Young’s modulus and critical
strain) of ceramic materials at the nanometer scale. Largely Acknowledgment
enhanced fracture strain was observed on the NW mor-
phologies [22,66–68]. In order to optimize the material We thank C. Sun, J. Shi, and D. Bayerl for their contributions to
selection, a systematic understanding of the mechanical the work reviewed in this article. We also thank the financial
properties as a function of the dimensions of different support from National Science Foundation under grant Nos.
piezoelectric materials is necessary. Specifically, for NWs DMR-0905914 and CMMI-0926245, DARPA under grant No.
with their sizes o100 nm, rectification might be needed for N66001-11-1-4139, and UW-Madison graduate school.
re-defining the elastic constant matrix. With this informa-
tion, the mechanical energy that is necessary to create certain
level of strain can be quantified thereby the corresponding References
energy conversion efficiency.
Third, coupling between the piezoelectric and semicon- [1] R. Shad, S.L. Eli, B. Jessy, C. Eric, R. Elizabeth, L. Elaine, O. Brian,
ductor properties (the piezotronic effect): Most piezo- M.R. Jan, V. Sundararajan, K.W. Paul, IEEE Pervasive Computing
electric materials are actually semiconductors as well. They 4 (2005) 28–36.
can be doped intrinsically or extrinsically to show appreci- [2] S. Roundy, P.K. Wright, J. Rabaey, A study of low level vibrations as
a power source for wireless sensor nodes, Computer Communica-
able free charge concentration. The piezoelectric effect
tions 26 (2003) 1131–1144.
would interact with the charge transportation and induce [3] J.A. Paradiso, T. Starner, IEEE 4 (2005) 18–27.
new phenomena or operation principles. For example, the [4] J.M. Donelan, Q. Li, V. Naing, J.A. Hoffer, D.J. Weber, A.D. Kuo,
Schottky barrier formed between ZnO and metal electrodes Science 319 (2008) 807–810.
can in situ rectify the potential output [47]. The conductiv- [5] G.K. Ottman, H.F. Hofmann, A.C. Bhatt, G.A. Lesieutre, IEEE
ity of a ZnO NW can be regulated by the appearance of Transactions on Power Electronics 17 (2002) 669–676.
 Author's personal copy

24 X.D. Wang

[6] G. Jonathan, J. Feenstra, H.A. Sodano, K. Farinholt, Smart [46] Y.C. Shu, I.C. Lien, Smart Materials and Structures 15 (2006)
Materials and Structures 16 (2007) 1810. 1499–1512.
[7] S. Priya, Journal of Electroceramics 19 (2007) 167–184. [47] Z.L. Wang, J.H. Song, Science 312 (2006) 242–246.
[8] A.P. Chandrakasan, N. Verma, D.C. Daly, Annual Review of [48] J.H. Song, J. Zhou, Z.L. Wang, Nano Letters 6 (2006) 1656–1662.
Biomedical Engineering 10 (2008) 247–274. [49] D. Bayerl, X.D. Wang, submitted for publication.
[9] H.A. Sodano, D.J. Inman, G. Park, The Shock and Vibration Digest [50] Z.Y. Wang, J. Hu, A.P. Suryavanshi, K. Yum, M.F. Yu, Nano Letters
36 (2004) 197–205. 7 (2007) 2966–2969.
[10] S. Meninger, J.O. Mur-Miranda, R. Amirtharajah, A. Chandrakasan, [51] C. Chang, V.H. Tran, J. Wang, Y.-K. Fuh, L. Lin, Nano Letters 10
J.H. Lang, IEEE Transactions on Very Large Scale Integration (VLSI) (2010) 726–731.
Systems 9 (2001) 64–76. [52] J. Liu, P. Fei, J. Zhou, R. Tummala, Z.L. Wang, Applied Physics
[11] C.B. Williams, R.B. Yates, Sensors and Actuators A: Physical 52 Letters. 92 (2008).
(1996) 8–11. [53] C. Xu, X.D. Wang, Z.L. Wang, Journal of the American Chemical
[12] G.M. Whitesides, G.W. Crabtree, Science 315 (2007) 796–798. Society 131 (2009) 5866–5872.
[13] V.S. Arunachalam, E.L. Fleischer, MRS Bulletin 33 (2008) 264–276. [54] Z.L. Wang, X.D. Wang, J.H. Song, J. Liu, Y.F. Gao, IEEE Pervasive
[14] X.D. Wang, J.H. Song, J. Liu, Z.L. Wang, Science 316 (2007) Computing 7 (2008) 49–55.
102–105. [55] Y. Qin, X.D. Wang, Z.L. Wang, Nature 451 (2008) 809–813.
[15] Z.L. Wang, Scientific American 298 (2008) 82–87. [56] X. Chen, S. Xu, N. Yao, Y. Shi, Nano Letters 10 (2010) 2133–2137.
[16] Z.L. Wang, Advanced Functional Materials 18 (2008) 3553–3567. [57] Y. Qi, M.C. McAlpine, Energy & Environmental Science 3 (2010)
[17] R. Yang, Y. Qin, C. Li, G. Zhu, Z.L. Wang, Nano Letters 9 (2009) 1275–1285.
1201–1205. [58] G. Zhu, R. Yang, S. Wang, Z.L. Wang, Nano Letters 10 (2010)
[18] M.S. Majdoub, P. Sharma, T. Cagin, Physical Review B 77 (2008). 3151–3155.
[19] Y. Gao, Z.L. Wang, Nano Letters 7 (2007) 2499–2505. [59] R.S. Yang, Y. Qin, L.M. Dai, Z.L. Wang, Nat. Nanotech 4 (2009)
[20] Y. Gao, Z.L. Wang, Nano Letters 9 (2009) 1103–1110. 34–39.
[21] C. Sun, J. Shi, X. Wang, J. Appl. Phys. 108 (2010) 034309–034311. [60] S. Xu, Y. Qin, C. Xu, Y. Wei, R. Yang, Z.L. Wang, Nature
[22] S. Hoffmann, F. Ostlund, J. Michler, H.J. Fan, M. Zacharias, Nanotechnology 5 (2010) 366–373.
S.H. Christiansen, C. Ballif, Nanotechnology 18 (2007) 205503. [61] S. Xu, B.J. Hansen, Z.L. Wang, Nature Communications 1 (2010) 93.
[23] V.S. Mashkevich, K.B. Tolpygo, Soviet Physics JETP 5 (1957) 435–439. [62] M.-Y. Choi, D. Choi, M.-J. Jin, I. Kim, S.-H. Kim, J.-Y. Choi, S.Y. Lee,
[24] K.B. Tolpygo, Soviet Physics Solid State 4 (1963) 1297–1305. J.M. Kim, S.-W. Kim, Advanced Materials 21 (2009) 2185–2189.
[25] S.L. Kogan, Soviet Physics Solid State 5 (1964) 2069–2070. [63] Y. Hu, Y. Zhang, C. Xu, G. Zhu, Z.L. Wang, Nano Letters 10 (2010)
[26] E.V. Bursian, N.N. Tranov, Soviet Physics Solid State 16 (1974) 5025–5031.
760–762. [64] C. Sun, J. Shi, D. Bayerl, X.D. Wang, Energy Environmental
[27] R. Maranganti, N.D. Sharma, P. Sharma, Physical Review B 74 Science, doi:10.1039/C1EE02241E, in press.
(2006) 014110. [65] J.K. Gupta, C.-H. Lin, Q. Chen, Indoor Air 20 (2010) 31–39.
[28] R.B. Meyer, Physical Review Letters 22 (1969) 918. [66] M.J. Gordon, T. Baron, F. Dhalluin, P. Gentile, P. Ferret, Nano
[29] M. Marvan, A. Havra nek, Progress in Colloid Polymer Science 78 Letters 9 (2009) 525–529.
(1988) 33–36. [67] A.V. Desai, M.A. Haque, Sensors and Actuators A: Physical 134
[30] R. Lakes, IEEE Transactions in Biomedical Engineering BME-27 (2007) 169–176.
(1980) 282–283. [68] B. Wu, A. Heidelberg, J.J. Boland, Nature Materials 4 (2005)
[31] E.V. Bursian, O.I. Zaikovskii, Soviet Physics Solid State 10 (1968) 525–529.
1121–1124. [69] J. Zhou, Y.D. Gu, P. Fei, W.J. Mai, Y.F. Gao, R.S. Yang, G. Bao,
[32] A.K. Tagantsev, Soviet Physics JETP 61 (1985) 1246–1254. Z.L. Wang, Nano Letters 8 (2008) 3035–3040.
[33] A.K. Tagantsev, Phase Transitions 35 (1991) 119–203.
[34] M. Marvan, A. Havrdnek, Solid State Communications 101 (1997) Dr. Xudong Wang is an assistant professor
493–496. in the department of Materials Science
[35] W. Ma, L.E. Cross, Applied Physics Letters 78 (2001) 2920–2921. and Engineering at University of Wisconsin–
[36] W. Ma, L.E. Cross, Applied Physics Letters 79 (2001) 4420–4422. Madison. His research interests include
[37] W. Ma, L.E. Cross, Applied Physics Letters 81 (2002) 3440–3442. studying the growth and assembly of oxide
[38] W. Ma, L.E. Cross, Applied Physics Letters 82 (2003) 3293–3295. nanowire arrays, understanding the cou-
[39] W. Ma, L.E. Cross, Applied Physics Letters. 86 (2005) 072903–072905. pling effect of semiconductor properties
[40] W. Ma, L.E. Cross, Applied Physics Letters. 88 (2006) and piezoelectric charge displacement, and
232902–232903. developing nanogenerator that uses piezo-
[41] C.L. Bauera, W.A. Brantley, Material Science and Engineering A electric nanomaterials to convert lowlevel
5 (1970) 295–297. mechanical energy into electricity. He is the recipient of DARPA Young
[42] J. Fousek, L.E. Cross, D.B. Litvin, Materials Letters 39 (1999) Faculty Award, 3M Non-Tenured Faculty Award, Ross Coffin Purdy
287–291. Award, Young Innovators Under 35 Award, and KAUST research fellow.
[43] R. Maranganti, P. Sharma, Physical Review B 80 (2009) 054109. He has published 49 papers in peer reviewed scientific journals,
[44] M.S. Majdoub, P. Sharma, Ccedil, T. Agbrevein, Physical Review B contributed 7 book chapters in his research field, and holds
77 (2008) 125424; 5 patents/provisional patents on oxide nanostructures and nanoma-
M.S. Majdoub, P. Sharma, Ccedil, T. Agbrevein, Physical Review B
79 (2009) 119904. terial-enhanced energy harvesting. His publications have been cited
- agin, Physical Review B 78 (2008) 121407.
[45] M.S. Majdoub, P. Sharma, T. C over 3,600 times by peers and his current h-index is 27.
˘