j ourna l h o mepage: www.elsev ier .com/ locate /sna
cale-like compliant gold electrode: Towards high strain capacitiveevices for energy harvesting
eremy Galineau, Jean-Fabien Capsal, Pierre-Jean Cottinet ∗, Mickael Lallartniversité de Lyon, INSA-Lyon, LGEF EA 682 8, rue de la Physique, 69621 Villeurbanne Cedex, France
r t i c l e i n f o
rticle history:eceived 2 July 2013eceived in revised form 25 February 2014ccepted 26 February 2014
a b s t r a c t
Highly compliant electrodes are of primary importance for high strain capacitive energy harvesting.Herein, we present a compliant gold sputtered electrode on a natural rubber substrate. Electrical con-ductivity remained remarkably good even at strains of 500%. The robustness of the electrodes has been
assessed in fatigue tests and resistivity of less than 25 � cm were observed after 1500 cycles between200% and 300% strain. These electrodes were then used in harvesting energy for large strains and exper-imental energy densities up to 3.3 mJ cm−3 cycle−1 have been recorded, showing the capabilities of suchelectrodes for efficiently ensuring electrical contact under high strain for converting mechanical energyinto electricity.
Energy harvesting using electroactive polymeric materials haseen a hot topic in the scientific community for the last decade dueo their low-cost, processability and light weight [1–8]. Harvestingnergy using dielectric elastomers can be done using electrostaticarvesters where the capacitance variation of the material sub-
ected to a mechanical force (vibrations, stretching. . .) is the mainrocess ([9a–11]). For instance, this type of system could be inte-rated into fabrics in order to harvest energy from body movementhere strain over 100% could be expected. Although it is theo-
etically possible to harvest energy by stretching elastomers, veryew can get a large capacitance difference between stretched andnstretched polymer as the maximum deformation allowed is usu-lly hindered by the compliance of the electrodes deposited ontohe dielectric material surface [12,13]. The electrodes should there-ore be made in such a way that no loss of conductivity is observedhen stretching it.
Techniques such as the use of microstructured metallic elec-rode [14,15], conducting polymers [16], ion implantation [17,18],ust or carbon grease [19] all lack from conductivity, ease of use or
oss of conductivity when stretched, and thus, are not ideal solution
or capacitive energy harvesting or, as for carbon loaded grease, areot practical to use in systems. Indeed, the techniques describedbove for metallic and conducting polymer electrodes can only
be employed where strain <100% are recorded. In addition, filmscovered with dust or carbon grease electrodes are very difficult tohandle and can result in messy system when they are implemented.
Recently, Wallace et al. reported the fabrication of compliantgold/polypyrrole electrodes showing loss of conductivity above 30%longitudinal strain [20]. Strain value for the loss of conductivity –<100% strain – has been reported by many in the microelectroniccommunity where stretchable electronics is a source of interest[21–24]. The main technique employed in these papers is to pre-stretch the material before conducting material deposition so thatwhen released a wrinkled metallic electrode is formed making itcompliant in the pre-stretching direction.
We believe that the technique described above can be exploitedfurther and that it is possible to form fish scales-like structureswhere gold scales could slide from one to another, therefore makingthe electrode compliant in the pre-strain direction even at largestrains.
In this paper, a 500% gold sputtered stretchable electrode is pre-sented using natural rubber as substrate. Its conductivity remainedbelow 25 � cm−1 in the stretching direction. Application of thiselectrode towards capacitive energy harvesting is demonstrated.
2. Elaboration
The substrate used in this study was a 50 �m thick natural rub-
ber (NR) sheets purchased from Fisher Fabrics. NR was chosen forits highly stretchable property as well as its good mechanical andchemical resistance. An elastic strain of nearly 600% can be reachedwithout any appearance of plastic behaviour. Highly stretchable
Fig. 1. Fatigue, strain and mechanical test setup. The fatigue test consisted of con-toc
gsto(u
3
3
E
3
fwd
ipirmtmptaowls
rtftuaamm
inuously stretched and unstretched for a sample of 50 mm of length and 20 mmf width until electrode failure (where failure meant that the electrode was “non-onductive at rest”).
old electrodes were deposited in a 2-step process. First, a NRheet was pre-stretched with a strain ratio (SR) ranging from 0%o 500%. Then, 50 nm thick gold electrodes were sputtered on oner both sides of the polymer using a Cressington® Sputtering Coater208HR). The samples were then carefully unstretched with a grad-al release of the pre-strain.
. Characterization
.1. Structure characterization
The gold structure was assessed by SEM using a FEI XL30 FEGSEM.
.2. Resistance, fatigue and mechanical characterization
Three types of measurement have been realized: electrodeatigue under mechanical excitation, variation of the resistanceith strain and elastic modulus measurements using the test benchescribed below (Fig. 1).
First the variation of electrical resistance over applied mechan-cal strain was tested using a dedicated bench test (Fig. 1). Therocedure consists of applying strain with a linear motor (XM550
ronless linear motor—Newport Cop., Irvine, CA) and measuring theesistance between the two probes of the Ohmmeter. In order toeasure the electrode resistance, the sample was placed between
wo clamps, as shown in Fig. 1. The resistance of the electrode waseasured with the help of a multimeter (ITC-920) where the two
robes were separated by 10 mm (d) at all time. The structure ofest bench presented in Fig. 1 was composed of two parts: one fixednd a second that could be moved in the 1-direction with the helpf the linear motor. As a consequence, the electrodes were drivenith a given strain profile and assumed to be strained along their
ength. The generated stress was measured with a help of forceensor (ELPF-T2M-250N, Measurement Specialities, Paris).
The fatigue of the electrodes over mechanical strain is alsoequired to fully characterize the electrodes. Classical approacheso fatigue design involve the characterization of total fatigue life toailure in terms of the cyclic stress or strain range. In these methods,he number of stress or strain cycles necessary to induce fatigue fail-re in initially uncracked specimens is estimated under controlled
mplitudes of cyclic stresses and strains. The test bench describedbove was also used for this study. Fatigue resistance of the goldetallized was tested by repeated transverse strain and resistanceeasurements were made every 5 min during 15,000 cycles for a
tuators A 211 (2014) 1–7
strain varying from 200% to 300%. The measurements were alwaysrealized at the same position on the sample, with a 10 mm dis-tance between the two probes (d). The bench test described abovewas also use for mechanical characterization. The load cell wasused to measure the stress and the linear stage to apply the strain.Experiments were performed on 4 cm by 1 cm samples at 1 Hz.
3.3. Energy harvesting capabilities
Fig. 2 provides a schematic representation of the setup devel-oped for characterizing the energy harvesting by the polymer film.One end was fixed by a rigid clamp fitted with copper foil elec-trodes that was connected to electrode surface of the sample, andthe other end was mounted on the ironless linear motor whichproduced the mechanical strain. For harvesting energy an electri-cal preload is necessary, it is produced by a power amplifier (Trek20/20C) driven by a function generator (Agilent 33220A). The elec-trical responses due to geometric variation of the sample weremeasured with the help of voltage probe (Model 820, Trek, USA)and current amplifier (SR570, Stanford Research Systems Inc. Sun-nyvale, CA). The data were monitoring with the help of oscilloscope(Agilent—DSO7034A).
4. Results and discussion
4.1. Structure
Fig. 3 shows the observed structures on unstretched (A), afterreleasing a 500% pre-strain (B), and when applying a 200% strain onelectrodes after releasing 500% pre-strain (C and D). The structuresobtained after releasing were scale-like and conduction remainedgood even a large applied strains as conduction paths between goldscales existed.
On unstretched substrates, the gold layer exhibited minor cracksand the overall structure is smooth. However, when sputtering on apre-stretched substrate, gold scales are formed upon releasing theapplied pre-strain. It is believed that during the pre-strain release,wrinkles started to form and that the spatial frequency of the wrin-kles increased. At one point during the release the forces on thegold layer were large enough for it to crack, leaving chips of goldon the NR surface attached to it on one side.
When stretching the electrodes (Fig. 3C and D) gold scales areredistributed over the surface of the substrate. It is seen in Fig. 3Cand D that although gaps between gold scales begin to appear uponstretching the electrodes at 200%, a conduction paths still exists assome gold scales are still connected one to another.
4.2. Mechanical characterisation
To ensure that the electrodes did not increase the stiffness ofthe film, mechanical characterization have been performed. Theresults are depicted in Fig. 4. The Elastic modulus was calculated tobe around 2.5 MPa for both samples (with and without electrodes).Therefore, it is demonstrated that the electrode has no effect on thestiffness of the film.
4.3. Electrodes resistivity
Firstly, the resistivity of the fabricated electrodes was assessedusing the dedicated bench test described above. Fig. 5-a shows theresitivity vs. applied strain for electrodes deposited with pre-strainsof 0%, 100%, 250% and 500%. It appeared that loss of conductivity
occurred with a strain 50% higher that the pre-strain, i.e. at 50%,150%, 300% for a pre-strain of 0%, 100% and 250%, respectively.The compliance of the electrodes is also clearly demonstrated asresistivity under 25 � cm−1 in the strain direction was recorded
J. Galineau et al. / Sensors and Actuators A 211 (2014) 1–7 3
Fig. 2. A schematic illustration of the experimental setup for the energy harvesting measurements.
Fig. 3. SEM micrograph of the gold compliant electrode (A), on unstretched substrate (magnification: 500×, scale bar: 20 �m), (B) after releasing a 500% pre-strain (magnifi-cation: 500×, scale bar: 20 �m), (C and D) after releasing a 500% pre-strain followed by a 200% applied strain ((C), magnification: 500×, scale bar: 20 �m, (D) magnification:1500×, scale bar: 10 �m).
ig. 4. Stress–strain graph for natural rubber without electrodes (red) and withlectrode (black) (For interpretation of the references to color in this figure legend,he reader is referred to the web version of this article.).
or a 500% strain for electrodes sputtered onto NR pre-stretchedt 500%. In addition, it can be seen on Fig. 5-a that, for electrodes
puttered onto pre-stretched substrates, the resistivity increasedt low strain. This is certainly due to the fact that when notrain is applied, the conduction path measured more than 1 cmetween the two electrodes. In other words, the gold scales made by
ig. 5. (a) Measured resistance as a function of the strain for pre-strains of 0%, 100%,50% and 500% and (b) fatigue test under strains ranging from 100% to 300%. Linesre drawn to guide the eye.
tuators A 211 (2014) 1–7
releasing the pre-strain make the conduction path quite long com-pared to the resistance measurement probes distance of 1 cm(Fig. 3). Therefore, the prestrain applied before gold depositionshould be tuned to meet the working strain in order to get thelowest possible value of the electrode resistance.
Furthermore, Fig. 6 shows the possibility of structuring thedeposited electrode and therefore demonstrates the potential useof such electrodes where complexe electrode designs are required.In this instance, the compliant gold electrode was sputteredthrough a mask made by photolithography into a AgNi alloy. It canbe seen in Fig. 6-B–D that the LED is on, i.e. that a current is passingthrough the electrode, upon stretching. Fig. 6-A depicts the systemwithout connection as a control, the diode did not emit any light.
Secondly, a fatigue test was conducted for strains ranging from200% to 300% and results are depicted in Fig. 5-b. The resistivity didnot increased dramatically over 1500 cycles, thus, the electrodespresented here are robust and can potentially be used for energyharvesting from large strain deformations for a long period of time.
4.4. Modeling of the electrode behaviour
The linear resistivity behaviour of the electrode may be decom-posed into two resistances according to the applied strain vs. thepre-strain during electrode deposition. When the strain is less thatthe deposition pre-strain, the initial length of the electrode does notchange when the strain is released, and therefore, assuming no con-tact between two distinct points, the linear resistivity is inverselyproportional to the strain:
�release = �0
(S0 + 1S + 1
)(1)
with S0 the deposition pre-strain value, S the strain and �0 the initiallinear resistivity. If the strain value is greater to the deposition pre-strain, then the gold electrode is lightly expanding, but also somecracks appear that limit the conductivity. Such a phenomenon canbe described by an exponential law as:
�stretch = �0 exp[
˛(
S − S0
Smax − S
)], (2)
where Smax refers to the maximal strain before breakdown and ̨ isa constant coefficient. Therefore the total linear resistivity � yields:
� ={
�0(
S0 + 1/S + 1)
for S ≤ S0
�0 exp[˛
(S − S0/Smax − S
)]for S > S0
(3)
Experimental data, along with fitted curve from theoretical anal-ysis (Eq. (3)) with a exponential coefficient chosen as ̨ = 2 aredepicted in figure. This figure clearly demonstrates the advantageof pre-stretching as well as using very flexible materials to disposeof very compliant electrodes (Fig. 7).
5. Application to mechanical energy harvesting
The availability of very flexible materials and electrodes opensa wide range of applications, including mechanical to electricalenergy conversion from large vibration magnitudes (e.g. humanmotions). The purpose of this section is to provide a practicalexample of the use of the proposed system featuring longitudinalcompliant electrodes for energy harvesting using the change in thedielectric capacitance of the substrate.
Considering an unstretched initial position, the capacitance ofthe device is given by:
C0 = εA
e(4)
with ε the permittivity of the material, A the surface and e the thick-ness. As the system is stretched, the length varies proportionally
J. Galineau et al. / Sensors and Actuators A 211 (2014) 1–7 5
Fig. 6. Conduction through a structured compliant electrode. (A) Control (no contact), (B–D) electrode stretched at 75%, 250%, 400%, respectively.
Fig. 7. Experimental and fitted linear resistivities as a function of the strain.
6 J. Galineau et al. / Sensors and Actuators A 211 (2014) 1–7
s for
ws
C
tI(ci
((
((
Fig. 8. Principles and associated cycle
ith the strain while the width and thickness decrease with theame ratio. Therefore the capacitance is changed to:
S = εA
e(1 + S) (5)
Hence, if an electrical preload is applied to the system, the elec-rical energy on the capacitance is changed because of its variation.t is then possible to harvest this electrical energy for a further usesensor powering, . . .). Several energy conversion cycles may beonsidered ([9a–11]), but one of the simplest approaches consistsn performing an Ericsson cycle (Fig. 8) that consists in:
1) Stretching the polymer
2) Applying a voltage when the capacitance is maximal (stretched
state)3) Releasing the mechanical stress (resulting in zero strain)4) Removing the voltage.
-0.4 -0.2 0 0.2 0.40
50100
a
Time (s)
Stra
in (%
)
-0.4 -0.2 0 0.2 0.40
5001000
Time (s)
Vol
tage
(V)
-0.4 -0.2 0 0.2 0.4-101
x 10-6
Time (s)
Cur
rent
(A)
Fig. 9. (a) Experimental time-domain waveforms for a 1000 V applied vo
the Ericsson energy harvesting cycle.
Using such an approach, the total harvested energy W is givenby ([9a,11])
W = 12
V20 (CS − C0) = εA
2eV2
0 S (6)
where V0 is the applied voltage.Experimental waveforms using Ericsson cycle-based energy
harvesting are depicted in Fig. 9 for a pre-load voltage of 1000 V,along with the obtained theoretical and experimental harvestedenergies for a strain of approximately 100% and initial voltagesof 100, 500 and 1000 V. As expected from Eq. (6), the harvestedenergy follows a quadratic behaviour with the initial voltage, andexperimental results show good agreement with the predictedenergy. From the experimental measurements, it is found that theharvested energy density equals to 3.3 mJ cm−3 cycle−1 when the
pre-load voltage is 1000 V. Hence, considering a 1 Hz vibrationwith a strain magnitude of 100% (which can typically be obtainedfrom human motion), such an approach would permits harvest-ing 3.3 mW. Although such a value seems to be much below values
b
0 200 40 0 60 0 800 100 0 12 000
10
20
30
40
Applied voltage (V)
Har
vest
ed e
nerg
y (
J) ExperimentalTheore tical
ltage and (b) harvested energy as a function of the applied voltage.
and Ac
rigstst8t
ict6h3e
6
nac5pettoTmaiaof
p
R
[
[[[
[
[
[
[
[
[
[
[
[[
[[[
[
J. Galineau et al. / Sensors
eported in the literature (which are around 1 J cm−3 cycle−1—[25]),t has to be noted that here, because the harvesting system isiven as an application example, the operating conditions are cho-en far below the maximal operating conditions. Hence, in ordero compare, the harvested energy density per cycle, per squaredtrain and squared initial electric field, which can be defined ashe material figure of merit for energy harvesting [26], equals to.3 �J cm−3 cycle−1 (m/m)−2 (V/�m)−2, which is more similar toypical values, and even slightly higher [27].
The energy density may however be further increased byncreasing the pre-load voltage. Considering a breakdown electri-al field of 136 V �m−1 obtained from experimental measurements,he maximum voltage that can be applied to the sample is800 V. Hence, taking a security factor of 2, a voltage 3.4 timesigher may be used, allowing achieving an energy density of8 mJ cm−3 cycle−1 because of the quadratic dependence of thenergy with the voltage.
. Conclusion
In this paper, we report on the harvesting capabilities of aatural rubber based on a highly stretchable device. The stretch-bility of the gold electrodes is of prior importance. To overcomeracks formation of the electrode layers during film stretching, a0 nanometer thick gold layer was deposited onto a pre-stretchedolymer film. This process allows a very low linear resistivity of thelectrode under large strains of up to 500%. Although, These elec-rodes performances are close to that of carbon grease electrodes,hey are not messy to handle and therefore make them electrodesf choices where high strain are observed for energy harvesting.he as-prepared polymer film was then used as an electro-activeaterial in a capacitive harvesting system. It is then theoretically
nd experimentally shown that this system driven under a 1000 Vnitial voltage and a strain magnitude of 100% allows an electro-ctive conversion from mechanical stretching to electrical energyf 3.3 mJ cm−3 cycle−1, although this energy can be multiplied by aactor of 11.5 considering the dielectric strength of the material.
Evaluation of energy harvesting performance of electrostrictiveolymer and carbon-filled terpolymer composites.
eferences
[1] P.J. Cottinet, D. Guyomar, B. Guiffard, L. Lebrun, C. Putson, Journal of PolymerEngineering 31 (2011) 133.
[2] K.M. Farinholt, N.A. Pedrazas, D.M. Schluneker, D.W. Burt, C.R. Farrar, Journalof Intelligent Material Systems and Structures 20 (2009) 633.
[3] D. Guyomar, L. Lebrun, C. Putson, P.J. Cottinet, B. Guiffard, S. Muensit, Journalof Applied Physics (2009) 106.
[4] M. Lallart, P.J. Cottinet, D. Guyomar, L. Lebrun, Journal of Polymer Science PartB-Polymer Physics 50 (2012) 523.
[5] K. Ren, Y.M. Liu, H. Hofmann, Q.M. Zhang, J. Blottman, Applied Physics Letters(2007) 91.
[6] R. Tiwari, K.J. Kim, S.M. Kim, Smart Structures and Systems 4 (2008) 549.[7] Y.L. Zhu, H.M. Wang, D.B. Zhao, J. Zhao, Smart Materials & Structures (2011) 20.[8] P.J. Cottinet, D. Guyomar, M. Lallart, Sensors and Actuators A-Physical 172
(2011) 497.[9] (a) S. Meninger, J.O. Mur-Miranda, R. Amirtharajah, A.P. Chandrakasan, J.H.
Lang, IEEE Transactions on Very Large Scale Integration (VLSI) Systems 9 (1)(2001) 64;(b) C. Lee, Y.M. Lim, B. Yang, R.K. Kotlanka, C.-H. Heng, J.H. He, M. Tang, J. Xie,H. Feng, Sensors and Actuators A: Physical 156 (2009) 208.
11] M. Lallart, S. Pruvost, D. Guyomar, Physics Letters A 375 (2011) 3921.12] Z.G. Suo, MRS Bulletin 37 (2012) 218.13] M. Meddad, A. Eddialal, D. Guyomar, S. Belkhiat, A. Hajjaji, A. Cherif, Y.
Boughaleb, Journal of Optoelectronics and Advanced Materials 14 (2012)
55.
14] T. Adrega, S.P. Lacour, Journal of Micromechanics and Microengineering 20(2010) 055025.
15] A. Pimpin, Y. Suzuki, N. Kasagi, Journal of Microelectromechanical Systems 16(2007) 753.
tuators A 211 (2014) 1–7 7
16] D.J. Lipomi, J.A. Lee, M. Vosgueritchian, B.C.K. Tee, J.A. Bolander, Z. Bao, Chem-istry of Materials 24 (2012) 373.
17] S. Rosset, M. Niklaus, P. Dubois, H.R. Shea, Journal of MicroelectromechanicalSystems 18 (2009) 1300.
18] S. Rosset, M. Niklaus, P. Dubois, H.R. Shea, Advanced Functional Materials 19(2009) 470.
19] P. Khodaparast, S.R. Ghaffarian, M.R. Khosrosham, N. Yousefimehr, D.Zamani, Journal of Optoelectronics and Advanced Materials 9 (2007)3585.
20] C. Wang, W. Zheng, Z. Yue, C.O. Too, G.G. Wallace, Advanced Materials 23 (2011)3580.
22] J.-H. Ahn, J.H. Je, Journal of Physics D-Applied Physics (2012) 45.23] D.-H. Kim, J. Xiao, J. Song, Y. Huang, J.A. Rogers, Advanced Materials 22 (2010)
2108.24] J.A. Rogers, T. Someya, Y. Huang, Science 327 (2010) 1603.25] S.J.A. Koh, X. Zhao, Z. Suo, Applied Physics Letters 94 (2009) 262902.26] M. Lallart, P.J. Cottinet, L. Lebrun, B. Guiffard, D. Guyomar, Journal of Applied
Physics 108 (2010) 034901.27] M. Lallart, P.J. Cottinet, D. Guyomar, L. Lebrun, Journal of Polymer Science Part
B: Polymer Physics 50 (8) (2012) 523–535.
Biographies
Dr. Jérémy Galineau graduated from Ecole NationaleSupérieure de Chimie de Rennes (ENSCR) in 2007. Hereceived is Ph.D. in 2011 from Dublin City University(DCU) in the field of analytical chemistry and conductingpolymers. He is currently a Post-doctoral Fellow in Lab-oratoire de Génie Electrique et Ferroélectricité (LGEF) atINSA de Lyon. His current interests include electroactivematerial synthesis and their integration into miniaturizedsystems.
Dr. Jean-Fabien Capsal received a M.Sc. degree in nano-technology and nanophysics in 2005 and Ph.D. degreein Material Science (Polymer Physics) in 2008 fromthe University of Toulouse. After working 2 years as aPost-doctoral Fellow in the Centre Inter-Universitaire deRecherche et d’Ingénierie des Matériaux (CIRIMAT) inToulouse (France), he joined Piezotech S.A.S. (ArkemaGroup) as a Research Engineer. Since 2011, he has beenhired as an associate professor in the Laboratoire de GénieElectrique et Ferroélectricité (LGEF, INSA Lyon, France).His current field of interest focuses on the modeling ofelectrostrictive properties of polymers, material charac-terizations and the improvement of the electromechanical
coupling of organic materials.
Dr. Pierre-Jean Cottinet graduated from the InstitutNational des Sciences Appliquées de Lyon (INSA Lyon),Lyon, France, in 2008. He received a Ph.D. degree inAcoustics in 2008 from the Institut National des Sci-ences Appliquées de Lyon (INSA) France for his thesis onelectrostrictive polymer for energy harvesting and actu-ation. During 2011, he was at Florida State Universityas a post-doctoral and working on buckypaper in HPMI(High-Performance Materials Institute). Currently, he is anassociate professor at INSA de Lyon, with research inter-ests concerning electroactive materials (polymers, CNT)and smart structures.
Dr. Mickaël Lallart graduated from Institut National desSciences Appliquées de Lyon (INSA Lyon), Lyon, France,in electrical engineering in 2006, and received his Ph.D.in electronics, electrotechnics, and automatics from thesame university in 2008, where he worked for the Lab-oratoire de Génie Électrique et Ferroélectricité (LGEF).After working as a Post-doctoral Fellow in the Center forIntelligent Material Systems and Structures (CIMSS) inVirginia Tech, Blacksburg, VA, USA in 2009, he has beenhired as an associate professor in the Laboratoire de Génie
Électrique et Ferroélectricité. His current field of inter-est focuses on vibration damping, energy harvesting andstructural health monitoring using piezoelectric, pyro-
electric or electrostrictive devices, as well as autonomous, self-powered wirelesssystems.
