UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Zinc oxide nanowires for dynamic strain sensing Tsan, Derek 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


24-ubc_2013_spring_tsan_derek.pdf [ 5.85MB ]
JSON: 24-1.0071971.json
JSON-LD: 24-1.0071971-ld.json
RDF/XML (Pretty): 24-1.0071971-rdf.xml
RDF/JSON: 24-1.0071971-rdf.json
Turtle: 24-1.0071971-turtle.txt
N-Triples: 24-1.0071971-rdf-ntriples.txt
Original Record: 24-1.0071971-source.json
Full Text

Full Text

Zinc Oxide Nanowires for Dynamic Strain Sensing by Derek Tsan B.A.Sc., The University of British Columbia, 2009 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF Master of Applied Science in THE FACULTY OF GRADUATE STUDIES (Electrical and Computer Engineering) The University Of British Columbia (Vancouver) April 2013 © Derek Tsan, 2013 Abstract A dynamic strain sensor using piezoelectric zinc oxide nanowires was demon- strated for potential application in structural health monitoring. Simulations and reviews of literature determined that strain of the nanowires by uniaxial com- pression yields the largest piezoelectric potential and that the piezoelectric coef- ficient of zinc oxide nanowires is enhanced due to nanoscale size effects. The fabrication of zinc oxide nanowires on various substrates was investigated in or- der to determine the ideal materials and seed layer deposition methods to yield high quality vertically-aligned nanowires. Nanowires were grown on indium tin oxide-coated glass slides. The tips of the nanowires were electrically connected us- ing poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate) conductive polymer, which formed a Schottky barrier with zinc oxide allowing for the separation of charge across the nanowire-electrode junction. The piezoelectric coefficients of several fabricated devices were measured by applying pressure to the top of the nanowires and measuring the charge. Variations in performance between the dif- ferent sensors were observed due to differences in the fabrication of each sensor. The highest coefficient measured was 11.5 pC/N, which is 16% higher than the bulk value for zinc oxide. The charge and voltage sensitivity to quasistatic pres- sure loading of the best performing sensor was calculated to be 1.32 pC/kPa and 16.7 mV/kPa. The response to clamped pressure stimulation from 1-90 kHz was evaluated using a piezoelectric stack actuator coupled with the zinc oxide nanowire sensor. The sensor showed excellent linearity to different amplitude vibrations at 1 kHz, and reasonably constant magnitude of charge output over the 1-90 kHz range for a constant vibration amplitude. The resonant frequency of the sensor and the response to free vibration could not be measured due to limitations in the ii available measuring equipment. The fabrication process for the nanowire sensor was found to be simple but inconsistent and could be improved by using repeat- able processes such as photolithography for precisely defining electrode and seed layer geometries. The as-fabricated nanowire sensor shows promise as a dynamic strain sensor for structural health monitoring applications or pressure sensing but requires further characterization and optimization through modeling in order to compete with commercial sensors. iii Preface Work on zinc oxide nanowire growth presented in this thesis was a collaboration between myself and Dr. Lisheng Wang. I was responsible for selecting substrates for nanowire growth, investigating different methods of seeding the substrates, and performing electrical measurements and characterization on nanowire devices af- ter growth. Dr. Wang optimized the general parameters for the nanowire growth including the concentrations of chemicals in the growth solution, growth time and temperature. Nanowire growth was performed by either Dr. Wang or by myself, following his procedures. Scanning electron microscope images of nanowires pre- sented in this thesis were taken by Dr. Wang or by myself with his assistance. iv Table of Contents Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv Table of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.1 Zinc Oxide Nanowire Applications . . . . . . . . . . . . . . . . . 5 2.2 Piezoelectricity . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.2.1 Measuring Techniques . . . . . . . . . . . . . . . . . . . 9 3 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3.1 Zinc Oxide Nanowires . . . . . . . . . . . . . . . . . . . . . . . 14 3.1.1 Enhanced Piezoelectric Effects . . . . . . . . . . . . . . . 17 3.2 Zinc Oxide Nanowire Growth . . . . . . . . . . . . . . . . . . . 19 3.2.1 Seed Layer . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.2.2 Nanowire Growth Solution . . . . . . . . . . . . . . . . . 25 3.2.3 Substrate Selection . . . . . . . . . . . . . . . . . . . . . 27 3.3 Electrodes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 v 3.3.1 Indium Tin Oxide (ITO) . . . . . . . . . . . . . . . . . . 30 3.3.2 Poly(3,4-ethylenedioxythiophene) Poly(styrenesulfonate) . 31 4 Design and Characterization . . . . . . . . . . . . . . . . . . . . . . 34 4.1 Device Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.2 Device Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.3 Non-Piezoelectric Device Characteristics . . . . . . . . . . . . . 41 4.4 Piezoelectric Coefficient . . . . . . . . . . . . . . . . . . . . . . 42 4.5 Vibration Testing . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.5.1 Clamped Sensor Response . . . . . . . . . . . . . . . . . 48 4.5.2 Resonant Frequency . . . . . . . . . . . . . . . . . . . . 52 5 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.1 Sensor Design and Performance . . . . . . . . . . . . . . . . . . 54 5.2 ZnO Nanowires Versus Other SHM sensors . . . . . . . . . . . . 56 5.3 Limitations of Current Design . . . . . . . . . . . . . . . . . . . 57 5.3.1 Repeatability of Fabrication . . . . . . . . . . . . . . . . 57 5.3.2 Materials . . . . . . . . . . . . . . . . . . . . . . . . . . 57 5.3.3 Characterization . . . . . . . . . . . . . . . . . . . . . . 58 6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 7 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 A COMSOL Models of Nanowire Strain . . . . . . . . . . . . . . . . . 70 B Matlab Code for Clamped Vibration Test . . . . . . . . . . . . . . . 72 vi List of Tables Table 2.1 Piezoelectric coefficients of common materials . . . . . . . . . 9 Table 4.1 Geometrical properties of fabricated zinc oxide nanowire strain sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Table 4.2 Electrical properties of fabricated zinc oxide nanowire strain sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Table 6.1 Properties of demonstrated zinc oxide nanowire strain sensor (Sensor A) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Table A.1 Elastic constants, ci j (GPa), for ZnO nanowire model . . . . . 70 Table A.2 Piezoelectric coupling matrix, ei j (C/m2), for nanowire model . 71 Table A.3 Relative permittivity matrix, εi j (×ε0 F/m), for nanowire model 71 vii List of Figures Figure 2.1 Piezoelectric dipole forming in quartz . . . . . . . . . . . . . 8 Figure 2.2 Equivalent circuit representations for piezoelectric devices . . 10 Figure 2.3 Schematic of charge amplifier circuit . . . . . . . . . . . . . 12 Figure 3.1 Wurtzite crystal structure of zinc oxide . . . . . . . . . . . . 15 Figure 3.2 COMSOL Multiphysics simulations for nanowire stretching and bending . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Figure 3.3 Results of ultrasonic vapour deposition of zinc acetate . . . . 22 Figure 3.4 Nanowires grown using spin-coated seed layer . . . . . . . . 23 Figure 3.5 Nanowires grown using inkjet-printed seed layer . . . . . . . 26 Figure 3.6 ZnO nanowires grown on copper foil . . . . . . . . . . . . . 28 Figure 3.7 ZnO nanowires grown on inkjet-printed silver . . . . . . . . . 29 Figure 3.8 ZnO nanowires grown on polyimide . . . . . . . . . . . . . . 30 Figure 3.9 ZnO nanowires growing on the edge of an air bubble on poly- imide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Figure 3.10 ZnO nanowires grown on ITO . . . . . . . . . . . . . . . . . 32 Figure 3.11 ZnO nanowires grown on silicon . . . . . . . . . . . . . . . . 33 Figure 3.12 Typical IV curve obtained across an ITO-ZnO-PEDOT:PSS in- terface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Figure 4.1 Schematic of nanowire sensor . . . . . . . . . . . . . . . . . 35 Figure 4.2 Fabricated nanowire sensor . . . . . . . . . . . . . . . . . . . 36 Figure 4.3 Long-exposure photograph showing poor connection across nanowire layer . . . . . . . . . . . . . . . . . . . . . . . . . 37 Figure 4.4 Grid of dots for nanowire layer . . . . . . . . . . . . . . . . . 38 viii Figure 4.5 Indentations from nanowires in PEDOT:PSS electrode layer . 39 Figure 4.6 Schematic of substrate with glass barrier for nanowire growth 40 Figure 4.7 Growth apparatus for hydrothermal growth of nanowires . . . 41 Figure 4.8 Typical charge response of nanowire sensor . . . . . . . . . . 44 Figure 4.9 Charge vs. change in compressive force with bias of 0.5 N . . 45 Figure 4.10 d33 vs. change in compressive force with bias of 0.5 N . . . . 46 Figure 4.11 Angled view of nanowire dot showing angled nanowires around outer edge . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Figure 4.12 Schematic showing bending of angled nanowires. (a) Vertical nanowires are compressed at low force. (b) Angled nanowires are bent at higher force. . . . . . . . . . . . . . . . . . . . . . 48 Figure 4.13 Schematic of clamped vibration test setup . . . . . . . . . . . 50 Figure 4.14 Sensor displaying linear response at 1 kHz vibration. x-axis shows applied voltage to the stack actuator. . . . . . . . . . . 51 Figure 4.15 Response of sensor from 1-90 kHz at constant actuator voltage of 0.75 V . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Figure 4.16 Schematic of vibration measurement using laser Doppler vi- brometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 ix Acknowledgments I would like to express my deepest gratitude to my supervisors Dr. Konrad Walus and Dr. Boris Stoeber for their guidance during my Master’s research. If it wasn’t for my positive experience as an undergraduate student working in Dr. Walus’ lab, I never would have had the forethought or opportunity to pursue a Master’s degree. Throughout my degree both Dr. Walus and Dr. Stoeber have consistently pushed me to find new ways to approach a problem and suggest new directions that I may not have considered, and they have provided invaluable encouragement in order to complete this work. I would like to thank Dr. John Madden for being a part of my defense committee, and for the use of his lab’s Bose ElectroForce instrument. I would also like to thank CMC Microsystems for loaning equipment used for characterization tests. I wish to acknowledge funding of this work from the Natural Sciences and Engineering Research Council of Canada (NSERC) and the Canadian Foundation for Innovation (CFI). My sincere thanks to Dr. Lisheng Wang who helped me take the majority of the SEM images that appear in this thesis and also introduced me to the won- derful world of zinc oxide nanowires. Together we shared in the complexities, frustrations, and ultimate triumphs of zinc oxide nanowire hydrothermal growth. My thanks to my labmates Simon Beyer, Robert Busch and John Berring for the countless hours we shared battling the inkjet printer and conquering nanotubes, nanowires, and piezoelectric polymers, and to Christoph Sielmann for his insight and knowledge of everything, especially of charge amplifiers and oscilloscopes. My heartfelt thanks to Vegas Cheung for keeping me sane and making sure I didn’t starve to death during the final weeks, days, hours, and minutes of thesis writing. Lastly but certainly not least I would like to thank my parents for their x unending love, encouragement and support from the very beginning. xi Chapter 1 Introduction The main motivation for this work is to develop dynamic strain sensors using zinc oxide nanowires, with applications for structural health monitoring (SHM). Struc- tural health monitoring is an area of research that focuses on providing realtime feedback on the integrity of a mechanical structure. Some examples include build- ings, bridges, pipelines, aircraft, and industrial motors. The aim of SHM is to identify and localize damage to a structure as it occurs so that corrective measures can be taken to ensure structural integrity. It is estimated that more than $200 billion is spent on maintenance of plant, equipment, and facilities in the United States every year [1]. The use of SHM systems could dramatically reduce these costs by providing early warnings of ma- jor failures so that less costly preventative maintenance programs can be imple- mented. In a report on the implementation of SHM systems on 40 bridges around the world, the average cost for short-term monitoring projects was found to be less than $50,000 with the installation of 20 sensors or less per structure [2]. Larger installations aimed at permanent autonomous monitoring typically cost $100,000 to $500,000, with 50 to more than 100 installed sensors. Compared with the cost of rebuilding infrastructure due to catastrophic failure, SHM implementations are quite cost effective. In 2007, Worden et al. proposed several axioms of structural health monitoring, which they formulated by reviewing the SHM literature of the past two decades [3]. The purpose of these axioms are to give new researchers in the field a common 1 basis and stimulate discussion on these axioms in order to advance the field. The axioms are provided below. Axiom I All materials have inherent flaws or defects. Axiom II The assessment of damage requires a comparison between two system states. Axiom III Identifying the existence and location of damage can be done in an unsupervised learning mode, but identifying the type of damage present and the damage severity can generally only be done in a supervised learning mode. Axiom IVa Sensors cannot measure damage. Feature extraction through signal processing and statistical classification is necessary to convert sensor data into damage information. Axiom IVb Without intelligent feature extraction, the more sensitive a measure- ment is to damage, the more sensitive it is to changing operational and envi- ronmental conditions. Axiom V The length- and time-scales associated with damage initiation and evo- lution dictate the required properties of the SHM sensing system. Axiom VI There is a trade-off between the sensitivity to damage of an algorithm and its noise rejection capability. Axiom VII The size of damage that can be detected from changes in system dy- namics is inversely proportional to the frequency range of excitation. From the axioms one can see that there are two broad considerations for SHM implementation: interpretation of the measured data (Axioms II-IVb), and mea- surement of an event (Axioms V-VII). In this work the main focus is on devel- opment of sensors for the event detection. Although not discussed here in detail, methods for interpreting the data received from SHM sensors and sensor networks are as important as development of the sensors themselves, and as such are beyond the scope of this work. 2 For event detection, networks of sensors are placed at strategic locations around a structure and are continually monitored for any anomalous data that could indi- cate a structural fault. Examples of such faults include cracks or fractures forming in concrete, bending or sagging support beams, and loose or missing bolts. The signals received by such sensors are usually mechanical vibrations, which are re- leased as a fault develops. The main challenge is to correctly identify a genuine fault from its vibrational signature, as opposed to benign vibrations from ordinary events; i.e., traffic across a bridge, or normal motor vibrations. As shown from the axioms, the type of sensor required to detect specific types of damage depends on the expected length- and time-scales of the event, sensitivity, and frequency range of excitation. There are two main types of SHM sensing systems: active and passive. Active sensing involves using a transmitter and receiver pair placed across a suspected fault or strategic location. The transmitter periodically sends out a non-destructive vibrational excitation signal, which travels through the structure to the receiver. Under normal conditions, the signal will reach the receiver through a predictable, well-characterized path, and successful receipt of this signal will indicate that the structure is healthy. The presence of a fault will cause an interaction with the exci- tation signal, altering its characteristic by reflecting some vibrational energy back to the transmitter or scatter it elsewhere in the structure, and the signal received by receiver will be different from the healthy case. The amount of signal loss or alter- ation should correspond to the severity of the fault. As an example, Ramamoorthy et al. investigated the use of ultrasound waves for detecting the depth of cracks in concrete [4]. They found that waves with frequencies from 400-600 kHz were eas- ily scattered by the presence of cracks, which made detection possible by analyz- ing the reflected wave. Active sensors are also particularly useful around structural joints which can become loose over time or after an impact. The resonant behaviour of a mechanical system depends on how tightly bonded the individual components are to each other, so that a loose connection will have a different resonant behaviour even if there is otherwise no structural damage. The resonant frequency can be de- tected by sending an excitation signal through the structure at various frequencies and determining the frequency at which maximum vibrations are detected. A shift in resonant frequency from the healthy case, or the presence of additional reso- 3 nances may indicate a loose or damaged connection. Active sensors utilized in this way enable the measurement of quasistatic strain, since a loose connection may not be moving or emitting acoustic vibrations. Passive sensing systems do not utilize an excitation signal, and instead directly monitor vibrational energy emitted from faults as they develop [5, 6]. The magnitude and frequency of the received signal can give information on the developing fault. Increases in the frequency of mea- sured events could indicate a widening crack and signal an impending structural failure, for example. Although simpler to implement than an active system since there are half as many devices, interpreting the data received from passive sensors can be challenging. In both active and passive systems, sensors need to be tuned to monitor for specific kinds of faults, which will have identifiable characteristics such as duration and frequency. The nature of SHM fault detection generally requires dynamic sensors, since developing structural faults and the excitation signals used to detect them are of- ten time-sensitive, high frequency events. Static sensors such as metal foil strain gauges can be employed in some applications, such as when an existing exterior crack has been identified and its width needs to be monitored over a long period of time. In this case, the strain gauge is placed across the crack and fixed at ei- ther end. As the crack widens, the strain gauge elongates, causing its resistance to increase proportionally to the length increase. For dynamic strain measurements, piezoelectric devices make ideal SHM sensors since they can directly convert the vibrational energy into an electrical signal to be measured. Conversely, they can be used to transmit vibrational energy as well, making them suitable for both active and passive sensing. The proposed device is a piezoelectric strain sensor using zinc oxide nano- wires as the piezoelectric material. Acoustic vibrations are coupled to the device to dynamically strain the nanowires. The strain on the nanowires generates an electric field across the nanowires, allowing separation of charge in proportion to the strain. Such a device could receive vibrations directly from acoustic emissions from cracks in a structure for passive sensing, or receive acoustic transmissions from another piezoelectric actuator in an active configuration. 4 Chapter 2 Background 2.1 Zinc Oxide Nanowire Applications For more than the past decade, researchers have been studying nanostructures of zinc oxide, which are promising for numerous applications including sensing and actuation, optoelectronics, light-emitting diodes, and biomedical devices [7]. Zinc oxide is a wide bandgap semiconductor (3.37 eV), has piezoelectric and pyroelec- tric properties, and is bio-compatible and biodegradable [7]. Nanowires of zinc oxide have become the focus for many researchers due to their one-dimensional high aspect ratio geometry, ease and repeatability of fabrication, and good piezo- electric and semiconducting properties. The most widely investigated application for zinc oxide nanowires is energy harvesting. Typically, mechanical vibrations in the environment are converted into electrical energy and stored for later use. Energy harvesting devices could, for example, be mounted on industrial machines to power wireless transmitters used to relay information from sensors in remote or hazardous locations such as in- side process tanks or underground. They could also be used as energy sources for bio-implanted devices such as pacemakers and hearing aids, as an alternative to batteries. Two forms of zinc oxide nanowire generators were described by Xu et al. [8]: a vertically integrated nanogenerator (VING); and a linearly integrated nanogen- erator (LING). The VING configuration consisted of a forest of vertically aligned 5 nanowires grown on a gold-coated silicon substrate. The wires were coated with poly(methyl methacrylate) (PMMA) for structural support. Platinum was deposited on a second piece of silicon and used as the top electrode. By pressing on the top electrode, a piezoelectric voltage of about 90 mV was generated, corresponding to an applied pressure of 6.25 MPa. By connecting three VINGs in series, the voltage was increased to 240 mV. The power output of the device was estimated at 2.7 mW/cm3. The LING was fabricated by depositing photolithographically- patterned stripes of ZnO seed material onto a polyimide (Kapton) film. Chromium was deposited on top of the seed material, exposing only the sidewalls for nano- wire growth. After growth, gold was deposited onto the ends of the nanowires, electrically connecting them in series. In total, 700 rows of nanowires were con- nected in series with each row containing approximately 20,000 nanowires. The flexible Kapton substrate was fixed at both ends and pressed in the middle to strain the nanowires. A maximum voltage of 1.26 V was generated in response to 0.19% applied strain. The authors estimated an output power density of about 70 nW/cm2 with this device. Although the voltage generated is larger, the power generated in this device is significantly less than in the VING configuration. It was proposed that the deposited gold electrodes did not make adequate contact with the nano- wires leading to high resistance and low current density. Nanowires have also been proposed to be embedded into fabrics for energy harvesting clothing, which could power or recharge personal electronics, for ex- ample. Energy would be harvested through body motions or even wind blowing on the fabric. Qin et al. [9] grew zinc oxide nanowires on Kevlar fibres using a hydrothermal method. They coated one of the nanowire-covered fibres with gold and entangled it with another uncoated nanowire-covered fibre. By fixing one fibre and pulling on the second, they observed a piezoelectric voltage generated from the bending of the zinc oxide nanowires as the two fibres brushed past each other. The peak voltage generated was about 1 mV and they estimated that a square metre of nanowire fabric could generate 4-16 mW of power. Most recently, the output from a zinc oxide nanowire device was used to electri- cally stimulate the nerve of a frog’s leg, demonstrating the potential for biomedical applications [10]. Zhu et al. created a zinc oxide nanogenerator using vertically- aligned nanowires on an ITO-covered silicon substrate. The nanowires were com- 6 pletely covered with a thin PMMA layer to electrically isolate the nanowires from a metal electrode subsequently deposited on top of the PMMA layer. This eliminated charge leakage through the nanowires as they were compressed, but still allowed induced charges to accumulate on the electrodes. The PMMA layer also served to mechanically connect all of the nanowires together, so that all of the nanowires could be compressed and contribute to the output signal. When pressed forcefully by a human palm, their device generated output signals of up to 58 V. Although no pressure data was given, this is the largest voltage signal ever reported for a zinc oxide nanowire device. The output of the device was connected to the sciatic nerve of a frog’s leg, which was shown to contract as a result of the nanogenerator output voltage. The power density of the nanogenerator was estimated at 0.78 W/cm3. Aside from energy generation applications, zinc oxide nanowires have also been investigated for sensing applications such as humidity sensing [11], glu- cose level monitoring [12], and gas sensors for hydrogen, ethanol [13] and carbon monoxide [14]. To date, the use of zinc oxide nanowires for structural health monitoring sen- sors has not been reported. The operating principle of the proposed device is sim- ilar to a piezoelectric energy harvesting device, in that vibrational energy is con- verted into an electrical output, but instead of being stored the output is analyzed for its frequency content and amplitude. Provided that the devices are sufficiently sensitive, zinc oxide nanowires could be a good candidate for structural health monitoring sensors in both active and passive configurations. 2.2 Piezoelectricity Piezoelectricity is a phenomenon that occurs in solid materials in which the ma- terial becomes polarized in response to an applied stress. The word piezoelectric is derived from the Greek word piezo, which means to squeeze. A general prop- erty of piezoelectric materials is that they lack inversion symmetry, meaning that their crystal structure is not symmetric when reflected across the centre of the unit cell, normal to the direction of applied stress. When the material is deformed, an electric dipole is formed resulting in an electric field across the piezoelectric layer. Figure 2.1 shows a simplified unit cell of quartz (SiO2, only one oxygen per sili- 7 con atom shown), demonstrating polarization under stress. In the unstrained case, both positive (Si) and negative (O) atoms are balanced so there is no net dipole. When stressed in the y-direction, the structure compresses such that the atoms are displaced from their neutral position, forming a net dipole in the x-direction. Con- versely, the application of an electric field across a piezoelectric material will result in material deformation. (a) Unstrained quartz with no dipole (b) Strained quartz with net dipole Figure 2.1: Piezoelectric dipole forming in quartz The piezoelectric effect in a material can be described by the coupled equations S = sET+dtE, (2.1) and D = dT+ εTE, (2.2) where S is the strain vector, sE is the compliance matrix at constant or zero electric field, T is the stress matrix, dt is the transposed piezoelectric coefficient matrix, E is the electric field matrix, D is the electric displacement vector and εT is the di- electric permittivity matrix at constant or zero stress. This set of coupled equations is known as the strain-charge form of the piezoelectric constitutive equations since they relate the strain of the material to the charge generated. The terms can be re- 8 arranged to yield equivalent stress-charge, strain-voltage, and stress-voltage forms as well. The piezoelectric coefficient, d, is a material-dependent constant which re- lates the charge generated to the force applied (typically given in units of Coulombs per Newton) or equivalently, the amount of displacement in response to an applied voltage (in meters per Volt). Since the piezoelectric effect depends on the crystal structure of the material, the effect is highly anisotropic. The constitutive equations are therefore expressed in tensor notation, with the matrix elements of each term representing different directions. For example, the piezoelectric coefficient, d, is commonly written as di j, where i is the direction of electric displacement and j is the direction of applied stress. Table 2.1 shows the piezoelectric coefficients of some common piezoelectric materials [15, 16]. The subscripts 1, 2, and 3 denote the x, y, and z-direction, respectively. Table 2.1: Piezoelectric coefficients of common materials Material Coefficient Magnitude (pC/N) Quartz d11 2.3 Zinc Oxide d33 9.93 Lead Zirconate Titanate (PZT) d33 117 Polyvinylidene Fluoride (PVDF) d31 28 2.2.1 Measuring Techniques The output from a piezoelectric device can be measured in a number of ways, most commonly by measuring the voltage across the device or by measuring the accumulated charge directly. Equivalent Circuit Representation Piezoelectric devices can be modeled as a current source with a parallel capacitance and resistance, or equivalently as a voltage source in series with a capacitance and resistance in parallel (see Figure 2.2). The piezoelectric capacitance, Cp = εA d , (2.3) 9 (a) Current model (b) Voltage model Figure 2.2: Equivalent circuit representations for piezoelectric devices depends on the area, A, thickness, d, and dielectric permittivity, ε , of the mate- rial. The parallel resistance, Rp, accounts for charge dissipation from the internal resistance of the piezoelectric material. An ideal piezoelectric material will be perfectly insulating and the magnitude of Rp will be infinity, representing an open circuit with no dissipation of charge over time. In reality, all piezoelectric materials are less than perfect insulators (or are semiconducting) and will dissipate charge. Direct Voltage Measurement The voltage on a piezoelectric device, Vp = Qp Cp , (2.4) is related to the charge, Qp, and the capacitance,Cp, of the piezoelectric device. As the piezoelectric material is stressed, the electric field across the material attracts free charges from metal leads connected to device, which build up on the surface of the piezoelectric material, essentially charging the piezoelectric capacitance. The charge on this capacitor results in a voltage according to Equation 2.4. This voltage can often be measured directly but its accuracy depends on certain factors. The internal resistance of the piezoelectric material will determine how quickly the charge dissipates across the piezoelectric layer according to the time constant, τ = RpCp. (2.5) 10 Dissipation of charge is caused by conduction across the piezoelectric layer, which reduces the amount of separated charge and the voltage over time. A large internal resistance will cause the charge to dissipate at a slower rate than a small internal resistance. If the time constant of the device is faster than the sampling rate of the measurement system, the measured voltage may be less than the peak generated voltage. Ideally, the capacitance, Cp, is dependent only on the piezoelectric layer capacitance. In reality, the cables connecting the device to the measurement circuit will add stray capacitance, Cc, in parallel, affecting both the time constant and the output voltage, since the piezoelectric device must charge the cable capacitance as well. Movement of the cable will change its capacitance and introduce noise into the measurement system. In order to accurately measure the charge from the output voltage, the effects of stray cable capacitance must be taken into account, leading to an effective piezoelectric capacitance, Cpe f f =Cp+Cc. (2.6) Due to charge dissipation, piezoelectric devices cannot provide a true DC volt- age output in response to static stress, and are usually used to output an AC voltage in response to a periodic stress. If a DC output is desired, it is possible to add a rectifier to convert the output from AC to DC. Charge Measurement The charge on a piezoelectric material can also be measured directly using a charge amplifier circuit as shown in Figure 2.3, adapted from [17]. As the material is strained, the electric field across the piezoelectric device is balanced by injecting charge on the reference capacitor, C f through the negative feedback loop, maintaining the inputs of the amplifier at 0 V. Since the device ter- minals are maintained at 0 V, charge is not allowed to build up on the piezoelectric capacitor, Cp, and the effects of charge dissipation through the piezoelectric layer are minimized. Another benefit is that stray capacitance from cabling between the device and the measuring instrument can be neglected in the charge measurement since there is essentially no electric field across the device. The voltage output of the charge amplifier circuit is directly proportional to the charge on the piezoelec- 11 Figure 2.3: Schematic of charge amplifier circuit, adapted from [17] tric device. The gain of the circuit is given by gain = 1 C f , (2.7) and can be changed by selecting different values for the reference capacitor. The voltage output of the charge amplifier circuit, Vo = Qp C f , (2.8) is therefore determined by the charge, Qp, injected onto the reference capacitor,C f . The total charge can be calculated by integrating the current from the piezoelectric material, Ip, with respect to time. The time constant of the circuit is dependent on the reference capacitor, C f , and reference resistor, R f . Together they set the low frequency cutoff of the circuit, fL = 1 2piR fC f . (2.9) The high frequency cutoff, fH = 1 2piRi(Cp+Cc) , (2.10) 12 depends on the resistance of the interface cables from the device to the measure- ment circuit Ri and the effective capacitance of the piezoelectric material and cable capacitance [17]. 13 Chapter 3 Materials 3.1 Zinc Oxide Nanowires Zinc oxide is a semiconductor with a wide bandgap of 3.37 eV. Its crystal structure is known as a Wurtzite structure, and is a hexagonal crystal composed of alternating zinc and oxygen planes. The zinc and oxygen atoms bond to form tetrahedronal cells with a single zinc atom surrounded by four oxygen atoms and vice versa. Zinc oxide is a piezoelectric and pyroelectric material since the tetrahedronal cells lack inversion symmetry. Its basal plane (0001) is a polar surface, terminated at one end by positively charged zinc ions, and by negatively charged oxygen ions at the other end. The other two most common surfaces of the zinc oxide crystal are the non-polar {2110} and {0110} [18]. The crystal structure of zinc oxide is shown in Figure 3.1 (adapted from [19]), with zinc atoms in yellow, and oxygen atoms in grey. The tetrahedronal cells are highlighted, illustrating the asymmetrical structure. The (0001), {2110} and {0110} surfaces are also highlighted. The black outline shows the rhombic unit cell of zinc oxide, three of which form the hexagonal unit cell of a zinc oxide nanowire. Zinc oxide can form a wide variety of nanostructures including nanowires and nanorods, nanobelts [18], nanoribbons, nanorings [7], whose shapes are dictated by the different growth rates along different crystal planes, and controlled to some extent via the addition of catalysts or rate inhibitors during the growth process. In general, the crystal structure forms in such a way as to maximize the surface 14 Figure 3.1: Wurtzite crystal structure of zinc oxide, adapted from [19] area of lower-energy non-polar surfaces such as {2110} and {0110}. In the case of the nanowire geometry, the area of the non-polar sidewalls are maximized and the polar end surfaces fixed by rapid growth in the [0001] direction. Nanowires are the most commonly investigated nanostructure of zinc oxide for piezoelectric applications since they are stable, easy to fabricate, and can be grown directly on a variety of surfaces in order to form useful devices. The largest piezoelectric potential is generated by applying uniaxial stress to the nanowire, leading to a strain along the nanowire and a piezoelectric potential developing along its length. Due to the high aspect ratio and small diameter of the nanowire, it can be difficult in practice to apply uniform axial stress without the wire bending or buckling during compression, or possibly fracturing upon stretch- ing. The theoretical limit for uniaxial tensile strain was determined by Agrawal et al. to be 6% [20]. Riaz et al. determined the compressive stress limit before buckling of nanowires to be 235.8 MPa, corresponding to critical strain percent- ages of 1-4%, depending on the geometry of the nanowire [21]. However, Yang 15 et al. demonstrated that a much lower strain percentage (0.05-0.1%) is sufficient to generate a measurable voltage (up to 50 mV) [22]. An advantage of actuating nanowires with uniaxial stress is that it simplifies the electrical connection require- ments across each nanowire. Since the piezoelectric potential is measured across the ends of the nanowire, electrodes can remain permanently connected to either end. Typically the nanowires are grown on a conductive substrate, forming a per- manent electrode connection at the base of each nanowire. A top electrode can be made by evaporation of metals (e.g. gold) or otherwise depositing other conductive materials (silver ink, polymers, etc.) on top of the nanowires. Having permanently connected electrodes improves the robustness of the nanowire device. Bending the nanowire is an alternative method of actuation. In this configura- tion, nanowires are typically fixed to the substrate at one end and are left free at the other end. Bending will generate a smaller piezoelectric potential than uniax- ial stress, which will develop across the thickness of the bent end of the nanowire with the sides under compression and tension having opposite polarity. In this configuration, permanently attached electrodes cannot be used since it would be almost impossible to place two separate electrodes at the free end of each individ- ual nanowire. A single electrode covering the whole nanowire end would short the electric field during bending, resulting in no generated voltage. Efforts to ac- tuate nanowires by bending have involved using sharp-tipped electrodes such as an atomic force microscope (AFM) tip [23], or an inverted pyramid-shaped grid etched in silicon [24]. In both cases the electrode is in contact with one side of the nanowire’s free end initially (the side of the nanowire under tension), forming a reverse-biased Schottky contact at which free charge from the electrode builds up but there is no current flow. Upon further bending the electrode contacts the other side of the nanowire, forming a forward-biased contact allowing built up charge to discharge through the compressed side of the nanowire, and for current to be measured. Although uniaxial compression generates a larger piezoelectric poten- tial than bending for a given strain, in practice it is easier to apply greater strain through bending, as the nanowire will be more flexible. Finite element modeling (FEM) simulations performed using COMSOL Multi- physics software show the magnitude and direction of the electric field for uniaxial and bending strain (Figure 3.2). For the stretching simulation, 85 nN of force was 16 applied to the ends of the wire, corresponding to a strain of 2×10−5. The geometry of the nanowires and values for the piezoelectric coefficients for zinc oxide used in the simulation were as published by [25] and the results were consistent. For the bending simulation, 85 nN force was applied in the x-direction, correspond- ing maximum strain in the z-direction at the base of the nanowire, with a tensile strain of 1.3×10−3 on the −x side and a compressive strain of −1.3×10−3 on the +x side. From the results it is shown that tensile/compressive strain generates a larger signal than bending, therefore compression of the nanowires was chosen as the operating principle of the demonstrated device. Details of the COMSOL simulation are provided in Appendix A. 3.1.1 Enhanced Piezoelectric Effects A number of theoretical and experimental works have shown that ZnO nanowires can exhibit a significantly higher piezoelectric response when compared to bulk zinc oxide. Lee et al. measured the piezoelectric coefficient of individual nano- wires using piezoresponse force microscopy, which involved applying a voltage to the nanowire using a conductive AFM tip [16]. For nanowires with diameters from 1.2-4 µm, the coefficient d33 was found to be from 1-45 pm/V, compared with to bulk value of 9.93 pm/V. They also found that by applying compressive stress to the nanowire, the coefficient could be further enhanced (0.32 GPa stress for 20% increase in d33), but beyond a certain value the coefficient would apparently de- crease due to buckling of the nanowire. Zhu et al. also measured the piezoelectric coefficient of individual zinc oxide nanowires by using the nanowire as an oscil- lator [26]. Each nanowire was clamped between two platinum electrodes (which they called the source and drain contacts, as in a traditional field effect transistor) suspended over a third electrode on a silicon substrate. The third electrode served as the gate electrode where an AC voltage was applied to drive the nanowire into mechanical resonance through an electrostatic effect. By adding a DC bias to the nanowire, its length change would change the resonant frequency of the nanowire and the piezoelectric coefficient d33 could be extracted from this change. Nano- wires with diameters from 200-600 nm were found to have coefficients ranging from 3-12 nm/V, significantly higher than the bulk value. The authors attribute 17 (a) Nanowire stretching simulation (b) Nanowire bending simulation Figure 3.2: COMSOL Multiphysics simulations for nanowire stretching and bending 18 the increased piezoelectric response to be due to the nanoscale structure having free outer boundaries allowing for greater volume expansion or contraction [27], as opposed to a bulk material, which would be constrained by the presence of surrounding bulk material. Additionally, the nanowires exhibit fewer structural de- fects than a bulk material, so that piezoelectric dipoles are preserved. Defects in the piezoelectric material would disrupt the orientation of aligned dipoles, reduc- ing the polarization potential. A theoretical study by Agrawal and Espinosa on the piezoelectric coefficient of very small diameter nanowires (0.6-2.5 nm) supports the explanation that the piezoelectric coefficient of nanowires is indeed enhanced compared with bulk materials [28]. They calculated the piezoelectric coefficient using first principles and found a 1-2 order of magnitude increase in piezoelectric coefficient over the bulk value. They determined that the enhanced piezoelectric effect was due to surface relaxation (due to the free boundary of the nanowire sur- face) causing a radial contraction of the nanowire structure. This would cause the volume of the nanowire to be smaller than the equivalent number of atoms in a bulk material. The piezoelectric coefficient depends on the polarization per unit volume and is therefore enhanced for the nanowire structure as compared with the bulk. Aside from the enhanced piezoelectric coefficients observed, the authors claim that conversion efficiency is also enhanced in nano-sized materials because they can tolerate larger deformations compared with their bulk counterparts, and since the piezoelectric charge is directly proportional to strain, nanowires can sep- arate more charge over a larger range. In other words, for the same given force a nanowire will deform much more than a bulk material, thus providing greater charge separation. Enhancement of the piezoelectric coefficient was not observed in the FEM simulations performed in this work because the coefficient was an ex- plicitly defined material parameter and the zinc oxide nanowire was treated as a bulk material, as opposed to the atomistic first principles model by Agrawal and Espinosa [28]. 3.2 Zinc Oxide Nanowire Growth In this work, ZnO nanowires have been grown using a simple hydrothermal method on a variety of substrates including copper, silicon, and glass slides coated with 19 indium tin oxide (ITO). The growth process begins with the deposition of seed material onto the substrate, heating of the seed material to convert it into zinc oxide, and finally growing nanowires by immersing the substrate in a heated growth solution for several hours. 3.2.1 Seed Layer In general, growth of ZnO nanowires requires a seed layer of ZnO crystals to be deposited on the substrate. During growth, these seed crystals provide nucleation sites for the nanowire growth. The precursor used for the seed layer is zinc ac- etate dihydrate (Zn(CH3COO)2•H2O). Upon heating, zinc acetate dihydrate de- composes and forms solid zinc oxide through the following reactions [29] : Zn(CH3COO)2 •H2O−→ Zn(CH3COO)2+H2O ↑; (3.1) 4Zn(CH3COO)2+H2O−→ Zn4O(CH3COO)6+2CH3COOH ↑; (3.2) Zn4O(CH3COO)6+3H2O−→ 4ZnO+6CH3COOH ↑; (3.3) Zn4O(CH3COO)6 −→ 4ZnO+3CH3COCH3 ↑+3CO2 ↑. (3.4) Initially, at around 50 ◦C, zinc acetate dihydrate begins to lose water (3.1) and be- comes anhydrous zinc acetate. As temperature is increased up to 270 ◦C, several gaseous products (carbon dioxide, acetone, acetic acid) are produced, leaving be- hind solid zinc oxide on the substrate (3.2) to (3.4). As water evaporates through heating, the mechanism for the formation of zinc oxide will shift from (3.3) to (3.4). It has been shown that simple heating of zinc acetate dihydrate in an air flow for several hours can yield nanowires with an approximate diameter of 40 nm and lengths of several micrometers [29]. However, these nanowires are randomly ori- ented and therefore cannot be used for piezoelectric devices. Hydrothermal growth has been proven to provide highly aligned vertically oriented nanowires and is the 20 preferred growth method for this work. Deposition Methods Several methods were investigated for depositing the seed layer including drop- casting, ultrasonic vapour deposition, spin coating, and inkjet printing. Drop Casting Drop casting did not yield satisfactory results because the thickness of the layer and drying pattern could not be controlled. Zinc acetate dissolved in water took several minutes to dry in room temperature, during which the contact line of the droplet would recede, resulting in a much smaller seeded area. Heating the sub- strate accelerated the drying time but the resulting pattern was still uneven and unpredictable. Using methanol instead of water greatly reduced the drying time, but the seed material dried in a coffee-ring pattern with more zinc acetate at the edge of the droplet than in the centre. Ultrasonic Vapour Deposition For ultrasonic vapour deposition, a piezoelectric disk was used to vapourize a so- lution of zinc acetate in water, which was then forced by air into an enclosed con- tainer containing the substrate. The substrate was heated to dry the condensed vapour on the substrate. This method yielded macroscopically uniform coverage on the substrate, but it was found to be composed of small (approximately 2 µm) particles of zinc acetate, which did not merge to form a uniform film (Figure 3.3a). The resulting nanowires grew in flower-like formations instead of having good ver- tical alignment (Figure 3.3b). Spin-Coating Spin-coating was performed using a seed solution of zinc acetate in methanol. The resulting film was very uniform in the centre of the substrate, but thicker along the edges of the substrate geometry. The nanowires grown were very well vertically aligned, as shown in Figure 3.4. Very high concentrations of zinc acetate in the seed solution for spin-coating 21 (a) Ultrasonically deposited seed layer (b) Flower-like nanowires after growth Figure 3.3: Results of ultrasonic vapour deposition of zinc acetate 22 (a) Nanowires grown on ITO covered glass (b) Closeup of nanowires Figure 3.4: Nanowires grown using spin-coated seed layer 23 produced a seed layer which would crack and flake off the substrate during heat- ing. This is likely due to the difference in thermal expansion between zinc oxide and the substrate creating stress at the interface as the seed layer is oxidized, es- pecially if the heating is rapid. This was observed for both copper substrates and on ITO-coated glass substrates. Lower concentration seed layers are less prone to flaking off since the resulting zinc oxide layers are less dense and not completely connected, leading to lower internal stress and greater relaxation, while maintain- ing adhesion to the substrate. During growth, the bases of the nanowires merge, leading to a seed layer that is completely connected. Drop-on-Demand Inkjet Printing The seed layer was also patterned using a drop-on-demand inkjet printer system. The printer system uses piezoelectrically actuated nozzles (MicroFab Technolo- gies) to dispense picolitre-volume droplets of low viscosity fluid. It is coupled with a two-dimensional stage with a lateral resolution of 2.5 µm. By controlling the movement of the stage and droplet dispensing frequency using a computer, precise patterns can be printed on any flat substrate. The seed solution was prepared by dissolving zinc acetate dihydrate in water to form a 0.1 M solution. Ethylene glycol (5 wt.%) was added to the solution to reduce the evaporation rate of the solution and prevent clogging at the inkjet printer nozzle. It was found that without the addition of ethylene glycol, the zinc acetate would crystalize at the nozzle tip and eventually prevent droplets from being deposited. After the addition of ethylene glycol, the solution remained stable for several hours before clogging occurred. The presence of ethylene glycol in the seed solution was not found to affect the growth of the nanowires. Ethylene glycol has typically been used with zinc acetate to form zinc oxide thin films or nanoparticles through sol-gel synthesis, which requires heating of the solution [30, 31]. It has not been reported elsewhere as being used as an additive for seed solution preparation in inkjet printing. Inkjet printing proved to produce the highest quality vertically aligned nanowires, and was the most control- lable, as each droplet had an even spacing and similar volume. The degree of crys- tallinity of the nanowires has been shown to increase with seed layer thickness [32]. Inkjet-printing the seed material should yield much thicker seed layers than spin 24 coating, as the discrete droplets do not spread out on the substrate. Although not quantified, the printed seed layers were visibly thicker than low-concentration spin- coated seed layers. The printed seed layers were also able to maintain adhesion to the substrate during heating compared with high-concentration spin-coated seed layers which flaked off the substrate during heating. This is likely due to lower interfacial tension between the substrate and seed layer since the surface area of each seed droplet is small. The as-grown nanowires are shown in Figure 3.5. 3.2.2 Nanowire Growth Solution Nanowire growth occurs in an aqueous solution containing zinc nitrate hexahy- drate, hexamethylenetetramine (HMTA), polyethyleneimine (PEI), and ammonium hydroxide. The role of each component of the solution is as follows. Zinc nitrate hexahydrate provides a source of Zn2+ ions for ZnO formation. HMTA in its most basic role serves as a source of OH− ions for ZnO formation. Upon heating, it decomposes through the following reactions [18] : (CH2)6N4+6H2O←→ 6HCHO+4NH3; (3.5) NH3+H2O←→ NH+4 +OH−. (3.6) Zn2+ ions react with OH− to form ZnO through precipitation : 2OH−+Zn2+ −→ ZnO(s)+H2O. (3.7) There is some debate about additional roles of HMTA in the growth reaction. It is thought that because HMTA is nonpolar, it will preferentially bind to the non- polar sidewalls of the ZnO nanowires, leading to more favoured growth along the nanowire length as opposed to width [18]. Additionally it is suggested that HMTA serves as a pH buffer, maintaining the solution pH favourable for nanowire growth since the rate of HMTA decomposition is independent of the reaction yielding ZnO (3.7) [33]. PEI is thought to perform a similar role as HMTA in binding to the sidewalls of the nanowires, perhaps to a greater extent, allowing nanowire aspect ratios greater 25 (a) Nanowires on inkjet-printed droplet (b) Closeup of nanowires Figure 3.5: Nanowires grown using inkjet-printed seed layer 26 than 125 [34]. Ammonium hydroxide is used to suppress homogenous nucleation of ZnO in the growth solution by forming complexes of zinc ions [35] : Zn2++nNH3←→ Zn(NH3)n2+, (3.8) where n= 1,2,3 or 4. The presence of these complexes lowers the degree of Zn2+ saturation in the solution, encouraging ZnO growth only at the seeded substrate. The addition of PEI further discourages homogenous nucleation in the solution. Indeed, it has been seen in this work that without the addition of ammonium hy- droxide, the growth solution quickly becomes cloudy due to formation of ZnO par- ticles in the growth solution, while with ammonium hydroxide the growth solution remains clear. 3.2.3 Substrate Selection Substrate selection was found to be an important factor in the quality of the as- grown nanowires. High quality nanowires are characterized as being very well aligned vertically with uniform hexagonal cross section. The major factor influ- encing nanowire quality appears to be surface roughness, with extremely smooth surfaces yielding the highest quality nanowires. Rough surfaces produced nano- wires with random orientation, forming grasslike nanowire forests. Substrates in- vestigated from lowest to highest quality included unpolished copper foil, inkjet- deposited silver ink, polyimide tape, ITO-coated glass, and silicon. For unpolished copper foil (Figure 3.6), although macroscopically smooth, the foil had many sur- face scratches resulting in nanowires with poor alignment. Nanowires grown on inkjet-printed silver ink were thin and very densely packed and grasslike with tapered cross-section (Figure 3.7). The silver ink is composed of nanoparticles of silver dispersed in ethylene glycol and ethanol. After deposition the ink is heated in order to sinter the nanoparticles together to form a continu- ous conductive layer. The fused nanoparticles might provide many small islands for nanowires to grow from, which would explain why they grow thin and dense. Nanowires were found to grow directly on silver without the need for seed material. Polyimide was investigated as a substrate potentially for use in a flexible piezo- electric device. Patterns of zinc acetate seed material were inkjet-printed on poly- 27 Figure 3.6: ZnO nanowires grown on copper foil imide sheets. The nanowires were found to grow on polyimide with good selectiv- ity for growth only on seeded areas. The resulting growth was dense and reason- ably well aligned (Figure 3.8). Unfortunately, the nanowires could not be used in a device since polyimide is not conductive. The nanowires were also found to grow spontaneously on inhomogeneities in the substrate such as along surface scratches, defects, and debris. Nanowires were observed to grow along the edges of air bub- bles formed in polyimide tape that was heated (in order to convert zinc acetate into zinc oxide), even though these areas had not been seeded (Figure 3.9). Nanowires grown on ITO were found to be very well aligned, as shown in Figure 3.10. Silicon wafers proved to be very good substrates for nanowire growth since they are highly polished and optically smooth. The best aligned nanowires were grown on silicon, however the available silicon substrates were not as conductive as ITO. Ultimately, ITO-coated glass was chosen as the device substrate due to its conductivity and smoothness. For structural health monitoring applications, an ideal substrate should be stiff in order to effectively couple vibrational energy from the structure to the substrate 28 Figure 3.7: ZnO nanowires grown on inkjet-printed silver and nanowires. Flexible substrates may have the advantage of being able to con- form to the surface of a curved structure but may dampen vibrations, reducing the amount vibrational energy reaching the nanowires. 3.3 Electrodes The selection of proper electrodes for connection with zinc oxide nanowires is important for sensing and energy harvesting applications. Zinc oxide has an elec- tron affinity of 4.5 eV [22]. For adequate charge separation during applied stress, charges must not be allowed to travel through the nanowire, otherwise the gener- ated electric field across the nanowire will be neutralized. Therefore, at least one of the nanowire-electrode junctions should form a Schottky barrier [22]. A Schot- tky barrier is formed when the work function of the electrode is higher than the electron affinity of the semiconductor. To equalize the fermi levels between elec- trode and semiconductor upon connection, the conduction and valence bands of the semiconductor bend downwards, but remain pinned at the electrode-semiconductor interface. This forms a potential barrier preventing current from flowing through 29 Figure 3.8: ZnO nanowires grown on polyimide the semiconductor, allowing an electric field to be generated across the semicon- ductor, which can be measured using an external circuit. This process is illustrated and explained in detail in [22]. Several electrode materials were initially investigated including copper, iron, silver, gold, indium tin oxide (ITO), and PEDOT:PSS. The conductive polymer PEDOT:PSS was found to form the largest Schottky barrier with ZnO and has been chosen for the final device. 3.3.1 Indium Tin Oxide (ITO) ITO is a transparent conductor most often used for solar cells. It has a work func- tion typically ranging from 4.3-4.7 eV [36, 37]. ITO is usually deposited onto a substrate by sputtering, which yields a uniform thin film on the substrate surface. For this work, 0.7 mm thick ITO-coated glass slides (Structure Probe, Inc., USA) were used as substrates. The coated ITO layer is 700 nm thick, with a sheet re- sistance of 8-12 Ω/square. ITO deposited on glass is easily etched by strong acids such as sulfuric acid, where the glass substrate serves as an etch stop. This is also useful for patterning different electrode areas on the same substrate. 30 Figure 3.9: ZnO nanowires growing on the edge of an air bubble on poly- imide 3.3.2 Poly(3,4-ethylenedioxythiophene) Poly(styrenesulfonate) Poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate) (PEDOT:PSS) is a con- ductive polymer with a work function ranging from 4.8-5.3 eV [38]. The polymer is easily dispersed in aqueous solutions and therefore is ideal for a variety of de- position methods including drop-casting, spin coating, and inkjet-printing. The PEDOT:PSS used in this work is Clevios 1000 (Heraeus) and was prepared as an ink for drop-on-demand inkjet printing. Dimethylsulfoxide (DMSO) from Sigma Aldrich (5 wt%) was added to the PEDOT:PSS as a conductivity enhancer, as de- scribed in [39], giving a conductivity of approximately 1000 S/cm. Triton X-100 (Sigma Aldrich) surfactant (0.1 wt%) was added to increase adhesion to the sub- strate. Since the work function of PEDOT:PSS is higher than the electron affinity of the zinc oxide nanowires, a Schottky barrier is formed at the PEDOT:PSS-nanowire interface, allowing charge to separate across the nanowires. The typical IV curve obtained across an ITO-ZnO-PEDOT:PSS interface is shown in Figure 3.12, which was measured using a Keithley 2635a sourcemeter. The voltage applied across the nanowires was swept from −3 V to 3 V and back to −3 V for five cycles, and the 31 Figure 3.10: ZnO nanowires grown on ITO corresponding current through the nanowire was measured. For negative voltages the Schottky contact prevents current from flowing through the nanowires. Some hysteresis was observed when the device was forward-biased. Hysteresis in the IV characteristics of zinc oxide nanowire devices has been observed by Song et al. [40]. This hysteresis was proposed to be caused by the externally applied elec- tric field’s ability to draw free electrons away from the nanowire-electrode interface above a threshold voltage, resulting in a change in the local Schottky barrier height, and the observed hysteresis. This phenomenon was not investigated further in this work since it is unlikely to be encountered during operation as a strain sensor, and the test was only used to confirm the presence of a Schottky contact. The IV characteristics presented in Figure 3.12 were consistent in trend and rectifiying be- haviour among all ITO-ZnO-PEDOT:PSS devices made, although the magnitude of the current varied depending on the number of electrically connected nanowires in each device. 32 Figure 3.11: ZnO nanowires grown on silicon Figure 3.12: Typical IV curve obtained across an ITO-ZnO-PEDOT:PSS in- terface 33 Chapter 4 Design and Characterization 4.1 Device Design The zinc oxide nanowire dynamic strain sensor is intended to be bonded to the surface of a structure so that mechanical vibrations in the structure will strain the nanowires, allowing charge to be measured that is proportional to the magnitude of the vibration. When subject to vibrations from the bottom of the device, a mass on top of the nanowire layer will oppose the motion of the bottom substrate, applying tensile and compressive stress to the nanowires as it oscillates up and down. Alter- natively, the nanowires can be strained directly by applying pressure to the top of the device, thus acting as a pressure sensor. A detailed cross-section schematic of the sensor design is shown in Figure 4.1. The sensor was designed such that the wire leads to the measuring instrument are bonded in the same plane on the bottom substrate, rather than directly across the nanowires. This was accomplished by using the top ITO/glass surface as a bridge connecting the nanowire tips to the outer electrodes on the bottom substrate. This configuration allows stress to be distributed evenly across the top surface of the device (and the nanowires) since there are no wires or other materials present. The top ITO/glass slide also acts as the mass on top of the nanowires. Glass was chosen as the substrate since it is fairly rigid (Young’s modulus of 72 GPa), which aids in transferring mechanical energy to the nanowires. A polymer or other flexible substrate would likely dampen any vibrations significantly, i.e. if the PEDOT:PSS 34 Figure 4.1: Schematic of nanowire sensor electrode layer alone was used as a mass. The packaged device with components identified is shown in Figure 4.2. The total substrate area is 25 mm x 25 mm. The top electrode (outlined in blue in Figure 4.2) is approximately 7.5 mm wide and 25 mm long. The active piezoelectric region of the device is defined by the overlapping areas of the nanowire grid and the top electrode. The nanowire grid is 4 mm wide and 15 mm long, though only half of its length is in contact with the top electrode. Therefore the active area of the device is approximately 7.5 mm x 4 mm (outlined in red in Figure 4.2). Initially, the nanowire layer was prepared by spin-coating seed solution onto the glass slide. The nanowires would grow in a dense forest as shown in Figure 3.4. It was found that after assembly of the device, the roughness of the nanowire forest (caused by different length nanowires or debris stuck on top of the nanowire forest) would prevent the top electrode from contacting the majority of nanowires 35 (a) Top view (b) Angled view Figure 4.2: Fabricated nanowire sensor 36 below. To confirm the poor connection, a long-exposure photograph of the sensor was taken in a dark room, while a forward bias of 20 V was applied to the sensor leads, as shown in Figure 4.3. The current flowing through the nanowires generated a small amount of visible light, indicating areas where there was a good connection across the nanowire layer. These areas were found to be scattered sparsely within the overlapping area of the top and bottom electrode, showing that the majority of nanowires were not connected to both electrodes. No light was emitted in areas outside the overlapping electrodes, as expected. Figure 4.3: Long-exposure photograph showing poor connection across nanowire layer To improve the electrode connection, the nanowire layer was modified to a grid of discrete dots (approximately 40 µm diameter each), with each dot contain- ing many vertically aligned nanowires, as shown in Figure 4.4. An 8 µm thick layer of PEDOT:PSS was used as the top electrode. Since PEDOT:PSS is a flexible polymer, when pressed against the nanowire dots it will wrap around the nano- wires, ensuring a solid electrical connection. To prevent accidental shorting of the PEDOT:PSS layer to the bottom electrode, a layer of PMMA was used to insulate the space between the nanowire dots from the top electrode. The quality of the electrode connection was confirmed by disassembling a device and imaging the PEDOT:PSS layer using an optical profilometer (Wyko 37 Figure 4.4: Grid of dots for nanowire layer NT1100 Optical Profiling System). The nanowires were found to create inden- tations in the PEDOT:PSS layer corresponding to the size of the nanowire dots, approximately 40 µm in diameter and 1.5 µm deep. This indicates that the polymer layer is wrapping around the nanowire dots, forming a solid physical connection. The profilometer scan confirming the indentations is shown in Figure 4.5. 4.2 Device Fabrication The sensor was fabricated on an ITO covered glass slide. The ITO layer was etched to form three channels—the centre channel serving as the bottom electrode and the two outer channels providing connections to the top electrode. The areas to be etched were covered using polyimide tape, and the slide was covered with spin- coated PMMA 495 photoresist (Sigma Aldrich), which serves as an etch barrier preventing the ITO underneath from being removed. The substrate was heated at 100 ◦C for 10 minutes to evaporate any solvent in the PMMA layer. The tape was then removed and the slide was immersed in sulfuric acid for several minutes to etch away the exposed ITO. After etching, the substrate was rinsed in methanol 38 Figure 4.5: Indentations from nanowires in PEDOT:PSS electrode layer and acetone to remove the PMMA. The seed layer was deposited using a drop-on-demand inkjet printer system. A grid of seed solution was printed on top of the centre electrode of the ITO/glass substrate using an 80 µm diameter nozzle, with a 150 µm droplet spacing. After depositing the seed layer, the substrate was heated on a hotplate at 300 ◦C for 10 minutes to decompose the zinc acetate dihydrate into zinc oxide, as in equations (3.1) to (3.4). The seeded substrate was placed into a covered beaker containing the growth solution, maintained at 88 ◦C by a water bath. The substrate was placed on a glass rack in the beaker with the seeded side facing downwards to prevent any particles from settling on the surface and disrupting the nanowire growth. A glass barrier was placed 2 mm below the substrate surface in order to block convective currents within the growth solution from dislodging the seed material as the solution is heated [41], which can otherwise lead to striations in the nanowire pattern and uneven nanowire growth. The barrier was clamped to the substrate using binder 39 clips. A schematic of the substrate with glass barrier assembled is shown in Figure 4.6. The complete growth apparatus is shown in Figure 4.7. After three hours in the growth solution, the substrate was removed, rinsed with distilled water, and dried under a nitrogen flow. A layer of PMMA was spin-coated over the substrate to provide structural support between the nanowires, as well as insulate the space between nanowires from the top electrode. Figure 4.6: Schematic of substrate with glass barrier for nanowire growth The top electrode was fabricated on a small section of an ITO-coated glass slide. PEDOT:PSS conductive polymer containing 5 wt% DMSO and 0.1 wt% Triton X-100 surfactant was drop-casted onto the centre and outer edges of the ITO layer and dried at 17 ◦C to suppress the coffee-ring effect [42]. At higher tem- peratures the solvent evaporates at a faster rate, driving more solid particles to the edge of the droplet where they deposit in a ring pattern. The presence of a coffee ring prevents nanowires in the centre from contacting the top electrode since the centre of the PEDOT:PSS droplet would be thinner than the edge. The substrate 40 Figure 4.7: Growth apparatus for hydrothermal growth of nanowires was then annealed at 140 ◦C for 10 minutes. A second drop was deposited onto the centre of the electrode, then dried and annealed as before, to increase the thick- ness of the PEDOT:PSS layer and therefore increase the contact with the nanowire tips. PEDOT:PSS was deposited on the outer electrodes of the bottom nanowire substrate, then dried and annealed. The top electrode was clamped onto the bottom substrate and the edges were bonded together using cyanoacrylate glue. Once dry, wire leads were bonded to the top and bottom electrodes using conductive carbon glue (Wire Glue, Anders Products). 4.3 Non-Piezoelectric Device Characteristics The geometrical properties of each layer of the fabricated devices were obtained using an optical profilometer (Wyko NT1100). These are summarized in Table 4.1. The following basic electrical properties of the fabricated devices were mea- sured using a Fluke PM6303A RCL meter, at 1 kHz, and are presented in Table 41 Table 4.1: Geometrical properties of fabricated zinc oxide nanowire strain sensors Layer Thickness Nanowires 11 µm PEDOT:PSS 8 µm PMMA 3 µm Glass substrate 0.7 mm ITO 700 nm 4.2. Table 4.2: Electrical properties of fabricated zinc oxide nanowire strain sen- sors Quantity Measured Value Cp 79 pF (at 1 kHz) Rp 500 kΩ(at1V) Rp 2 MΩ(at−1V) τp 39.5 µs - 0.158 ms 4.4 Piezoelectric Coefficient The direct piezoelectric coefficient was measured by applying force to the sensor and measuring the output charge. Compressive force was applied to the sensor using a Bose ElectroForce 3100 test instrument, and the charge was measured using a Kistler 5015A Charge Meter. A preloading force of 0.5 N was applied to the sensor in order to maintain contact between the nanowire tips and the top electrode at all times. The force was applied as a square wave pulse in 0.5 N steps from 0.5-3 N, with a period of 5 s per pulse. Each step was repeated four times. The time constant of the charge meter was set to 0.1 s so that the signal would decay before the application of the next pulse. The actual time constant of the nanowire device is not significant in this configuration since the charge meter maintains 0 V across the device terminals, negating the effects of the nanowire capacitance. Upon 42 compression of the sensor, a positive voltage (corresponding to negative charge) was measured by the charge meter. Upon relaxation of the compressive force, an approximately equal magnitude but negative voltage was measured, indicating reversal of the electric field and charge moving in the opposite direction. A typical measured response is shown in Figure 4.8 for 0.5 N and 3 N pulses. The measured charge versus change in force for three devices is shown in Fig- ure 4.9. The maximum compression force varied from a 0.5 N bias force to 3.5 N (denoted as a 3 N change in force in the figure) and the maximum relaxation force was from 3.5 N returning to the bias force of 0.5 N (denoted as −3 N). Although the three devices were fabricated the same way, they show slightly different re- sponses. Sensors A and C displayed similar charge vs. force profiles, with Sensor A having the largest output. Sensor B showed the most linear response, but the magnitude of charge output was the smallest. The effective piezoelectric coefficient d33 of each sensor was calculated by di- viding the peak charge generated by the peak applied force for each step (Figure 4.10). For increasing force applied to Sensors A and C, the effective coefficient was observed to decrease. For Sensor A the piezoelectric coefficient ranged from a maximum of 11.5 pC/N at 0.5 N compressive force to a minimum of 5 pC/N at 3 N compressive force. Sensor C did not display maximum piezoelectric coefficient at the lowest applied force but in general the coefficient was observed to decrease for higher force. For an ideal piezoelectric device, the measured charge should increase linearly with increasing force. For all measured force steps, the test in- strument consistently reached the desired force level within 70 ms, and the charge meter’s time constant was fixed at 0.1 s. Since the time constant of the meter is on the same order as the time taken by the instrument to reach the desired force level, the measured signal will begin to decay before the maximum force is reached. This will result in a constant underestimation of the maximum charge at each force step, but does not account for the nonlinear response. The different responses for the three sensors could be attributed to variations in the assembly of each device. Vari- ations in the thickness and distribution of the polymer electrode layer may cause different stress concentrations on the nanowires as the top electrodes are stressed. For thicker electrode layers, a larger proportion of applied stress may be absorbed by the polymer layer instead of directly stressing the nanowires, leading to less 43 (a) 0.5 N applied force (b) 3 N applied force Figure 4.8: Typical charge response of nanowire sensor 44 Figure 4.9: Charge vs. change in compressive force with bias of 0.5 N charge at lower applied force (potentially the case for Sensor C). For thinner elec- trode layers the nanowires may already be in good contact with the top electrode and are stressed uniformly as the top electrode is compressed (potentially the case for Sensor B). One possible explanation for the nonlinear response seen in Sensors A and C is that the compression of the nanowire is nonlinear depending on its angle on the substrate. As shown in Figure 4.11, the nanowires around the edge of the nano- wire dot grow outwards at an angle. As the top electrode is compressed onto the nanowires, these angled nanowires will bend rather than be uniaxially compressed. Since for bending strain the electric field appears across the lateral direction of the bent end and is likely shorted by contact with the top electrode, these angled nanowires will not contribute to the charge measured. However, they will absorb some of the applied force resulting in less stress on the active, vertically aligned nanowires. A schematic depicting this process is provided in Figure 4.12. In all cases it was also observed that the reverse charge generated during re- laxation of the applied force (denoted as negative change in force in Figures 4.9 and 4.10) was consistently smaller in magnitude than during compression. This is 45 Figure 4.10: d33 vs. change in compressive force with bias of 0.5 N likely due to the PEDOT:PSS layer remaining slightly compressed upon relaxation of the force (since it is a soft polymer) and not completely returning to its origi- nal thickness before the next cycle of applied force. As a result, the sensor does not undergo as much strain upon relaxation and less charge is accumulated. Re- peated actuations showed consistent results. Subsequent measurements performed over several weeks also showed similar results indicating no significant degrada- tion over time. It is possible that excessive force, such as greater than 100 N, will damage the nanowires, the PEDOT:PSS-nanowire connection or the glass substrate but such forces are not expected to occur during normal operation as a strain sensor. In addition to the direct piezoelectric coefficient, the converse piezoelectric co- efficient (measuring the amount of strain for a given applied voltage) was initially intended to be measured by applying a voltage to the sensor and measuring the dis- placement of the top electrode using a laser Doppler vibrometer. However, it was observed that applying a voltage greater than 5 V DC for a few seconds caused the PEDOT:PSS electrode to permanently degrade, resulting in a lowering of the work function and decreased leakage resistance. As a result this rendered the de- vice unusable for subsequent charge measurements. It is possible that too much 46 Figure 4.11: Angled view of nanowire dot showing angled nanowires around outer edge current flowed through the PEDOT:PSS and nanowires causing damage through Joule heating, but this was not investigated further since large applied voltages are unlikely to be encountered when used as a dynamic strain sensor. The converse piezoelectric performance is not important in this application since the device is not intended to be used as a piezoelectric actuator. For actuator applications using ZnO nanowires, a more robust electrode material may be required. 4.5 Vibration Testing The nanowire sensor’s response to vibration was evaluated by using a piezoelectric ceramic stack actuator (Physik Instrumente Model P885.21) to vibrate the sensor over a range of frequencies. Sensor A was measured for the following tests as it showed the highest piezoelectric response. 47 (a) (b) Figure 4.12: Schematic showing bending of angled nanowires. (a) Vertical nanowires are compressed at low force. (b) Angled nanowires are bent at higher force. 4.5.1 Clamped Sensor Response Direct loading of the nanowire sensor was achieved by clamping the sensor and stack actuator together, allowing the actuator to directly strain the nanowires. A schematic of the test setup is shown in Figure 4.13. The test setup consists of a micrometer which holds the sensor, stack actuator, and load cell, which is then clamped in a vise to provide a rigid, level structure. This configuration ensures that vibrations are transferred directly to the nanowire sensor and are not dampened by the test fixture. The load cell (Transducer Techniques Model THA-250-Q) is used to measure the amount of preloading force applied to the sensor and actuator assembly, and it has a maximum load value of 250 lb (1100 N), and a stiffness of 22 N/µm. The response of the load cell was measured using the Bose Electro- 48 Force 3100 test instrument by applying a range of compressive force and recording the output voltage. Using a fixed input voltage of 10 V, the sensitivity of the load cell was determined to be 16.2 µV/N. The preloading force holds the sensor and actuator in the test fixture, and also ensures that the nanowires are in contact with the top electrode for the duration of the test. The operating frequency of the stack actuator was controlled using a Stanford Research Systems DS360 Low Distortion Function Generator. The maximum actuation frequency was limited by the reso- nant frequency of the stack actuator, which is specified as 135 kHz ±20%. 90 kHz was chosen as the upper frequency limit for this test. The AC RMS voltage at the actuator was measured using an Agilent 34401a multimeter. Due to the high capacitance of the stack actuator (600 nF) the actual voltage delivered to the ac- tuator was severely attenuated at higher frequencies, since the parallel capacitance of the actuator in series with the impedance of the instrument (50Ω) and device leads acts as a low pass filter with a cutoff frequency of approximately 5.3 kHz. A MATLAB calibration script (provided in Appendix B) was used to determine the maximum possible actuation voltage at the maximum frequency of 90 kHz, and to adjust the generator voltage at each lower frequency to maintain a constant actu- ation voltage over the entire frequency range. At the maximum generator output voltage of 14.14 V RMS, the voltage delivered to the actuator was approximately 0.75 V at 90 kHz. This value was used as the constant actuating voltage for the entire frequency range. The output of the sensor was connected to a Kistler 5015A Charge Meter, which was then connected to an Agilent DSO6034a oscilloscope to record the generated charge waveform at each frequency. From Equations 2.1 and 2.2, the charge of a piezoelectric device is linearly proportional to the applied stress. Similarly, the strain is linearly proportional to the applied electric field. To verify that the nanowire sensor response is linear, the actuation voltage of the stack actuator was varied at a fixed frequency and the output charge of the sensor was measured. Since the stack actuator is also a piezoelectric device, its output force is also expected to be linearly proportional to the applied actuation voltage. Therefore, a linear change in actuation voltage of the stack actuator will result in a linear output of charge from the nanowire sensor. The measured response at 1 kHz is shown in Figure 4.14. It can clearly be seen that the sensor output is linear with the actuation voltage from 1-10 V. The nanowire 49 Figure 4.13: Schematic of clamped vibration test setup sensor response was found to be generally linear as well at higher frequencies, but due to attenuation of the actuation voltage from the capacitance of the stack actuator, the voltage range was limited for higher frequencies. This test did not show decreasing charge sensitivity at higher actuation forces as in the quasistatic case. For this test, a preloading force of 16.5 N was applied to the sensor using the micrometer, to hold the sensor and stack actuator assembly in place. A large preloading force was necessary since neither the stack actuator or the nanowire sensor were bonded to the test apparatus (or to each other). For a smaller preloading force, the assembly would vibrate loose or shift during the test. It is possible that due to the large preloading force, the effect of bending angled nanowires no longer dominates, allowing for a linear output dominated by compression of the vertical nanowires. The effective piezoelectric coefficient could not be calculated in this test, as it was not possible to measure the force generated by the stack actuator using the load cell due to its limited sensitivity for small forces at high frequency. According to the actuator data sheet, the actuator has a blocking force of 800 N 50 at 120 V, a stiffness of 100 N/µm, and a nominal travel of 6.5 µm at 100 V. The applied voltage of 10 V corresponds to a blocking force of approximately 67 N. Since the test fixture, load cell, and sensor are not infinitely stiff (the PEDOT:PSS electrode layer being especially soft), the actual applied force will be much less. Referring to the results in Figure 4.9, the measured results suggest an estimated actuator force of roughly 0.5 N/V. Figure 4.14: Sensor displaying linear response at 1 kHz vibration. x-axis shows applied voltage to the stack actuator. The frequency response from 1-90 kHz was obtained for the nanowire sen- sor. The output charge waveform was recorded using the oscilloscope at each frequency, and the RMS values of the output charge were calculated from these waveforms. Figure 4.15 shows the measured charge vs. actuation frequency. The magnitude of the output charge is consistent with that measured in Figure 4.14 (2.1 pC at 0.75 V, 1 kHz actuation, average 2.2 pC at 0.75 V, 1-90 kHz actua- tion). Over the measured frequency range the charge output is fairly constant, al- though some deviation is observed from approximately 30-70 kHz. The deviations could be due to resonances within the test apparatus since the elements were only clamped together by the micrometer and not rigidly bonded together. The origins 51 of these resonances could not be identified using the current setup. Figure 4.15: Response of sensor from 1-90 kHz at constant actuator voltage of 0.75 V 4.5.2 Resonant Frequency The resonant frequency of the device was intended to be measured by mounting the sensor on top of the stack actuator and measuring the displacement of the de- vice using a laser Doppler vibrometer (Polytec MSA-500). The vibrometer would be configured in a differential mode, where the difference in velocity between the vibrating top electrode and bottom substrate could be measured in order to deter- mine the resonant frequency of the sensor. A schematic of the proposed test setup is shown in Figure 4.16. A periodic chirp signal would be applied to the stack actuator and an FFT of the differential velocity at each frequency would be ob- tained. By taking the differential measurement of velocity, the resonant frequency of the stack actuator would not limit the frequency range of measurement since the measurements would be referenced to the motion of the bottom substrate. Unfortu- nately due to equipment malfunctions and time constraints this measurement could not be performed successfully. 52 Figure 4.16: Schematic of vibration measurement using laser Doppler vi- brometer Once the resonant frequency of the sensor was determined, the frequency could be tuned by changing the amount of mass on top of the nanowires. 53 Chapter 5 Discussion 5.1 Sensor Design and Performance The as-demonstrated zinc oxide nanowire device shows potential as a dynamic strain sensor, though further characterization is required. For low loading (i.e. 0.5 N) the piezoelectric coefficient of 11.5 pC/N exceeds the bulk value for zinc oxide, though not significantly. This value is within the range of values reported by Lee [16] of 1-45 pm/V, which is directly equivalent to 1-45 pC/N. It is possible that enhanced piezoelectric effects are observed due to the size of the nanowires, but since many nanowires are being stressed in parallel, each nanowire is stressed less and the enhancement may be diminished, especially if it is stress-dependent. Single nanowires have shown to provide the largest piezoelectric response due to enhanced piezoelectric coefficients and greater strain tolerance, but reliably fabri- cating a sensor using a single nanowire would prove challenging, simply due to the small scale. Similarly, reading the output of a single nanowire device would be challenging, requiring complex equipment. It would be impractical, for example, to integrate a piezoresponse force microscope AFM tip into the measurement cir- cuitry (as in [16]), especially if several sensors are to be used at a time. In theory, the amount of charge on a piezoelectric material depends only on the amount of force applied. The charge output can be increased by increasing the total force on the nanowire. A single nanowire will be very easily strained and can output a large amount of charge. Since the capacitance is fixed, the voltage will be corre- 54 spondingly large. A large voltage across the nanowire can cause it to overcome the Schottky barrier, leading to conduction through the nanowire and loss of current flowing through the external circuit. A large force may also cause the single nano- wire to fracture if the stress is too high. The same force applied to a parallel number of nanowires would output the same amount of charge as the single nanowire, how- ever, since the nanowires have more surface area the piezoelectric capacitance is increased and the voltage is kept correspondingly low. Using many nanowires in parallel allows greater force to be applied to the device without allowing too large a force or voltage to build up across the individual nanowires. Having a large area of active nanowires also allows a macroscopic device to be assembled, which is much simpler than assembling a microscopic device designed to strain a single nanowire. Hydrothermal growth allows the fabrication of many vertically aligned nanowires without the need for single nanowire manipulation. The robustness of the device is also increased with many parallel nanowires, as more nanowires will be in contact with the electrode, and the device can still function if some of those connections are damaged over time. As a pressure sensor, the fabricated device shows very good sensitivity com- pared with similar devices. Given an effective area of 114.4 mm2 (based on the area of the probe used to apply the force that was in contact with the top elec- trode), the pressure on the device per unit of force is calculated to be approximately 8.7 kPa/N. For the maximum piezoelectric coefficient of 11.5 pC/N, the maximum estimated charge sensitivity of the sensor is 1.32 pC/kPa. The voltage sensitivity can be estimated by dividing the charge sensitivity by the device capacitance. With a measured capacitance of 79 pF (measured at 1 kHz using a Fluke PM6303A RCL meter), the maximum sensitivity is 16.7 mV/kPa. This sensitivity is three orders of magnitude higher than the value of 8 mV/MPa (approximately 20 mV output at 2.5 MPa pressure) reported by Xu et al. for their vertical integrated nanogenerator (VING), which is also based on zinc oxide nanowires [8]. From the limited vibration tests performed the fabricated device shows excel- lent linearity to vibrations of different amplitudes (Figure 4.14), though this may depend on the frequency of the vibration or the preloading force on the nanowires. 55 5.2 ZnO Nanowires Versus Other SHM sensors Piezoelectric sensors for structural health monitoring have most commonly been fabricated using PZT as the piezoelectric material [43–46]. Compared with other sensors based on piezoelectric materials, ZnO nanowires have the advantage of be- ing intrinsically piezoelectric, whereas other piezoelectric materials such as PZT require poling in a large electric field and at high temperature in order to become piezoelectric. These materials have the potential to become depoled in the presence of high temperatures or electric fields, which would severely affect the piezoelec- tric performance. ZnO nanowires are also more mechanically robust due to their high elasticity, having been demonstrated to tolerate more than 70◦ of bending [8]. ZnO nanowires are also very easy to fabricate, only requiring simple apparatus as described in Section 3.2. Hydrothermal growth of nanowires has the potential for batch fabrication of many sensors in parallel by using large temperature-controlled tanks for the growth solution. Since nanowires can be easily grown on many differ- ent materials with selective growth on areas with pre-seeded patterns, it might be possible to integrate nanowire-based sensors directly on the surface of structures with complex geometry. Deposition of other piezoelectric materials would be lim- ited by constraints such as size, complex equipment or process conditions (i.e. vacuum chambers for chemical vapour deposition or high temperature furnaces). Fibre optic sensors (FOS) are another technology of interest for SHM sensors. Embedded fibre optic cables in structures have been shown to be able to detect 0.1 mm cracks in concrete [47]. The formation of a crack at any point along the ca- ble’s length will cause the cable to bend, obstructing the flow of light. This method can therefore obtain quasistatic evaluations of strain from slow forming cracks in concrete. However, for this method to work, the cable must be free to move within the structure which can prove challenging for embedding in a structure. FOS using fibre bragg gratings (FBG) have also been shown to detect dynamic events such as acoustic vibrations from PZT actuators [48], and the propagation of Lamb waves on a structure’s surface [49]. An FBG is a fibre optic cable that can reflect certain wavelengths of light while transmitting others. Such sensors can be used in a sim- ilar scheme as active piezoelectric sensors, using an actuator to generate acoustic vibrations and receiving the signal using an FBG. Compared with the ZnO nano- 56 wire sensor, FOS have the advantage of continuous long range sensing since light can travel very quickly for long distances, while the nanowire sensor is localized to the area around the device. However, implementation of an SHM system using ZnO nanowire sensors may be more straightforward as they can be simply mounted to the outer surface of an existing structure whereas FOS cables would need to be embedded inside a structure during construction (as in [47]). 5.3 Limitations of Current Design 5.3.1 Repeatability of Fabrication Although the fabrication process for the nanowire sensor was fairly simple and inexpensive, it was difficult to make devices with consistent performance due to variations in the packaging process. The etching of the ITO electrodes was not lithographically controlled, resulting in slightly different active areas in each de- vice. Alignment of the nanowire grid in the inkjet printing process was also per- formed manually, resulting in variations in the grid placement. The deposition of PEDOT:PSS was achieved using drop-casting of polymer solution via pipette onto the top electrode. This resulted in variations in the size and thickness of the poly- mer layer. In order to improve the consistency of each device, the etching of the ITO channels and deposition of the PEDOT:PSS electrode could be defined using photolithography to precisely mask off areas to be etched or to cover areas that should be insulated. 5.3.2 Materials The use of ITO-coated glass allowed a convenient conductive surface for the nano- wires to grow on. The smoothness of the substrate resulted in very well aligned nanowires after growth. However, the use of ITO is not ideal most simply because it is expensive. Transparent conductors are highly sought after for applications such as solar cells and displays, whereas a strain sensor does not benefit from be- ing transparent. Nearly any other conductive material could be used provided the surface is smooth enough to facilitate well-aligned nanowire growth. Since the for- mation of a Schottky contact at at least one end of the nanowire is a requirement 57 for separation of charge, the connection at the base of the nanowire does not nec- essarily have to be ohmic, as with ITO. Another concern may be the use of glass as a substrate because it is fragile and may be damaged by excessive force. This is less of a concern if the device can be miniaturized and packaged in a smaller enclo- sure where it would not be subjected to large shock forces. Many silicon devices (silicon wafers are similarly fragile to glass) such as accelerometers and pressure sensors can be enclosed in such protective packages. The conductive polymer PEDOT:PSS was chosen based on its high work func- tion, which forms a Schottky contact with ZnO. Unfortunately, since it is a polymer and is soft compared to zinc oxide (Young’s modulus of 2.5 GPa for PEDOT:PSS vs. 140 GPa for ZnO), it is expected that the electromechanical conversion ef- ficiency of the device will be adversely affected due to damping in the polymer layer. A stiffer electrode material would provide greater mechanical coupling to the nanowires, making it easier to stress the wires and increase the output signal. Some possible alternatives to PEDOT:PSS would be gold, which has a work func- tion from 5.1-5.5 eV and Young’s modulus of 78 GPa, or platinum, which has a work function from 5.1-5.9 eV and Young’s modulus of 168 GPa [50]. These materials however are considerably more expensive than PEDOT:PSS since they are precious metals, and deposition of these materials is considerably more com- plex, requiring sputter deposition equipment compared with simple drop-casting of PEDOT:PSS solution. 5.3.3 Characterization In this work the characterization of a few key parameters were limited by the test setup and equipment available, including the measurement for resonant frequency and high-frequency vibration sensitivity. The piezoelectric stack actuator used in the vibration tests could not allow for frequency measurements higher than 90 kHz due to its low resonant frequency. Time did not allow for measurements of the free vibration of the sensor for frequencies below 90 kHz using the stack actuator, but this would be a fairly simple measurement to set up with the available equipment. Measurement of the response of higher frequency vibrations will require an actu- ator with a higher resonant frequency and lower capacitance so that the resonant 58 frequency does not affect the charge output in the measurement frequency range, and so that the voltage delivered to the actuator is not attenuated at higher actuation frequencies. 59 Chapter 6 Conclusions A dynamic strain sensor based on zinc oxide nanowires was fabricated and demon- strated. The sensor was shown to have a piezoelectric coefficient larger than the bulk value for zinc oxide, and a much higher pressure sensitivity compared to sim- ilar ZnO nanowire devices. The sensor displayed a linear response as expected to different amplitude vibrations when a high preloading force was applied, but dis- played a nonlinear response for low frequency, low amplitude loadings. For low force, it is possible that the stress from both uniaxial compression of vertically aligned nanowires and bending strain from angled nanowires are significant, re- sulting in a nonlinear charge output. Table 6.1 shows the important characteristics of the demonstrated device. Table 6.1: Properties of demonstrated zinc oxide nanowire strain sensor (Sen- sor A) Quantity Measured Value d33 5-11.5 pC/N Cp 79 pF (at 1 kHz) Rp 500 kΩ-2 MΩ Max. charge sensitivity 1.32 pC/kPa Est. voltage sensitivity 16.7 mV/kPa The hydrothermal growth mechanism of zinc oxide nanowires was investigated to determine the ideal seed deposition method and substrate in order to obtain 60 high quality vertically aligned nanowires. It was observed that increased substrate smoothness increased the alignment of the nanowires, as rough surfaces caused the nanowires to grow at randomly oriented angles. ITO-coated glass slides and silicon wafers were the smoothest substrates tested, which resulted in the highest quality nanowires. Seed layer uniformity and thickness also affected the align- ment of nanowires, with more uniform seed layers yielding the highest quality nanowires. Inkjet printing was found to provide the best seed layer for aligned nanowire growth since the deposited layer was uniform and thicker than for spin- coating. PEDOT:PSS conductive polymer was determined to be the best available material for the device as it provided the largest Schottky contact with ZnO, which is required for charge separation to occur. The fabrication of nanowires using hy- drothermal growth was found to be simple and repeatable, but the assembly of the rest of the demonstrated device including the seed layer deposition, electrode deposition and packaging led to inconsistencies in device performance. Although the ability to measure charge from strain was demonstrated, the consistency and performance of the sensor can likely be optimized by better controlling the fabri- cation processes and by selecting more suitable substrate and electrode materials. Gold or platinum electrodes may be better alternatives to PEDOT:PSS conductive polymer because they are stiffer and should form a Schottky contact with ZnO due to their high work functions, however, they will have a higher cost and a more complex fabrication process. The optimization of the sensor design will depend on the ability to characterize fundamental device properties such as resonant fre- quency and charge sensitivity over a larger frequency range. The development of an electromechanical model of the sensor will also aid in optimizing sensor perfor- mance by parameterizing the sensor geometry, resonant frequency, etc., in order to maximize sensitivity. As a piezoelectric sensing material, zinc oxide nanowires show promise with potentially large piezoelectric coefficients and simple fabrication methods. Com- pared with PZT—the most widely used piezoelectric material for sensing and actu- ating, zinc oxide nanowires have potentially higher piezoelectric coefficients due to nanoscale enhancements of piezoelectric polarization, and are intrinsically piezo- electric so they do not have to be poled. ZnO nanowires can also tolerate much larger strains than PZT and can therefore operate over a larger range for both sens- 61 ing and actuation. The use of piezoelectric zinc oxide nanowires shows promise as a dynamic strain sensor but further characterization and testing in an SHM appli- cation will be required in order to optimize the sensor for commercial use. 62 Chapter 7 Future Work In order to advance the design of the zinc oxide nanowire strain sensor, more thor- ough characterization of the fundamental device properties is required. An elec- tromechanical model of the sensor should be developed to aid in the tuning of the resonant frequency in order to match the response of the sensor to the type of vi- bration to be sensed. The ability to estimate the charge output and sensitivity to a variety of strain inputs for different sensor geometries and configurations would allow the device to be optimized to different applications. A robust testing apparatus for measuring the charge output of the sensor for free vibrations should be designed in order to determine the resonant frequency of the device and to confirm the accuracy of the electromechanical model. The laser Doppler vibrometer coupled with a piezoelectric stack actuator with a high resonance frequency would be a good starting point for such an apparatus. If the resonant frequency of the sensor is too high for the stack actuator, mass could be added to the top of the device in order to lower its resonant frequency into a mea- surable range. The existing stack actuator can be used to measure the response to free vibrations below 90 kHz, as in the clamped vibration tests. An actuator with a small capacitance will reduce the attenuation of applied voltage at higher frequencies, allowing greater vibrational amplitudes to be applied during the test. The output charge from the device could be measured using the charge amplifier, and the resonance frequency also determined from the measured charge. The stack actuator must be able to provide constant amplitude vibrations over the measure- 63 ment frequency range so that its resonance does not affect the charge output of the sensor. Once the model is verified, a test simulating the use of the sensor in a struc- tural health monitoring application should be developed. The same stack actuator used in the free vibration testing apparatus could be used to send acoustic vibra- tions through a test structure, with the vibrations measured by the nanowire sensor some distance away. Alternatively, the nanowire sensor could be modified to act as an actuator, in order to develop a completely nanowire-based monitoring system. More suitable electrode and substrate materials will have to be investigated so that the electromechanical coupling is maximized. The current PEDOT:PSS electrode has not been proven to be a good electrode material for actuation of the nanowire device. An example SHM test structure could be a concrete beam with a known struc- tural defect such as a crack or a void, placed in between the stack actuator and nanowire sensor. An actuator can be used to send an acoustic signal through the structure, and the sensor will measure the acoustic signal as it interacts with the structural defect. Such tests will allow the determination of different types of struc- tural damage from the measured signal of the nanowire sensor. In order to have a complete SHM monitoring solution, software should be developed to analyze the measured signal from the sensor, and determine the type and severity of the struc- tural defect, possibly by cross-referencing the measured signal with a library of known signatures. 64 Bibliography [1] V. Giurgiutiu, Structural Health Monitoring: with Piezoelectric Wafer Active Sensors. Academic Press, Dec. 2007. → pages 1 [2] D. Inaudi, “Overview of 40 bridge structural health monitoring projects,” SMARTEC SA, Roctest Ltd, pp. 1–8, 2007. → pages 1 [3] K. Worden, C. R. Farrar, G. Manson, and G. Park, “The fundamental axioms of structural health monitoring,” Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, vol. 463, pp. 1639–1664, June 2007. → pages 1 [4] S. K. Ramamoorthy, Y. Kane, and J. A. Turner, “Ultrasound diffusion for crack depth determination in concrete,” The Journal of the Acoustical Society of America, vol. 115, no. 2, p. 523, 2004. → pages 3 [5] J. N. Kudva, C. Marantidis, J. D. Gentry, and E. Blazic, “Smart structures concepts for aircraft structural health monitoring,” pp. 964–971, Sept. 1993. → pages 4 [6] C. B. Van Way, J. N. Kudva, J. N. Schoess, M. L. Zeigler, and J. M. Alper, “Aircraft structural health monitoring system development: overview of the Air Force/Navy smart metallic structures program,” pp. 277–285, May 1995. → pages 4 [7] Z. L. Wang, “Ten years’ venturing in ZnO nanostructures: from discovery to scientific understanding and to technology applications,” Chinese Science Bulletin, vol. 54, pp. 4021–4034, July 2009. → pages 5, 14 [8] S. Xu, Y. Qin, C. Xu, Y. Wei, R. Yang, and Z. L. Wang, “Self-powered nanowire devices,” Nature Nanotechnology, vol. 5, pp. 366–373, Mar. 2010. → pages 5, 55, 56 65 [9] Y. Qin, X. Wang, and Z. L. Wang, “Microfibre-nanowire hybrid structure for energy scavenging,” Nature, vol. 451, pp. 809–813, Feb. 2008. → pages 6 [10] G. Zhu, A. C. Wang, Y. Liu, Y. Zhou, and Z. L. Wang, “Functional electrical stimulation by nanogenerator with 58 V output voltage,” Nano Letters, vol. 12, pp. 3086–3090, June 2012. → pages 6 [11] Y. Zhang, K. Yu, D. Jiang, Z. Zhu, H. Geng, and L. Luo, “Zinc oxide nanorod and nanowire for humidity sensor,” Applied Surface Science, vol. 242, pp. 212–217, Mar. 2005. → pages 7 [12] J. Zang, C. M. Li, X. Cui, J. Wang, X. Sun, H. Dong, and C. Q. Sun, “Tailoring zinc oxide nanowires for high performance amperometric glucose sensor,” Electroanalysis, vol. 19, no. 9, pp. 1008–1014, 2007. → pages 7 [13] L.-J. Bie, X.-N. Yan, J. Yin, Y.-Q. Duan, and Z.-H. Yuan, “Nanopillar ZnO gas sensor for hydrogen and ethanol,” Sensors and Actuators B: Chemical, vol. 126, pp. 604–608, Oct. 2007. → pages 7 [14] S.-J. Chang, T.-J. Hsueh, I.-C. Chen, and B.-R. Huang, “Highly sensitive ZnO nanowire CO sensors with the adsorption of Au nanoparticles,” Nanotechnology, vol. 19, p. 175502, Apr. 2008. → pages 7 [15] R. J. Brook, R. W. Cahn, and M. B. Bever, Concise encyclopedia of advanced ceramic materials. Oxford; New York; Cambridge, Mass., USA: MIT Press, 1991. → pages 9 [16] H. Lee, J. Park, S. A. Han, D. Lee, K. Kim, N. Lee, J. Park, Y. Seo, S. Lee, and Y. J. Choi, “The stress-dependent piezoelectric coefficient of ZnO wire measured by piezoresponse force microscopy,” Scripta Materialia, vol. 66, pp. 101–104, Jan. 2012. → pages 9, 17, 54 [17] J. Karki, “Signal conditioning piezoelectric sensors,” Texas Instruments Application Report SLOA033A, pp. 1–6, 2000. → pages 11, 12, 13 [18] S. Baruah and J. Dutta, “Hydrothermal growth of ZnO nanostructures,” Science and Technology of Advanced Materials, vol. 10, p. 013001, Jan. 2009. → pages 14, 25 [19] S. State, “Wurtzite polyhedra.” Online, http://en.wikipedia.org/wiki/File:Wurtzite polyhedra.png, accessed Feb 2013, Apr. 2008. → pages 14, 15 66 [20] R. Agrawal, B. Peng, and H. D. Espinosa, “Experimental-computational investigation of ZnO nanowires strength and fracture,” Nano Letters, vol. 9, pp. 4177–4183, Dec. 2009. → pages 15 [21] M. Riaz, O. Nur, M. Willander, and P. Klason, “Buckling of ZnO nanowires under uniaxial compression,” Applied Physics Letters, vol. 92, no. 10, p. 103118, 2008. → pages 15 [22] R. Yang, Y. Qin, L. Dai, and Z. L. Wang, “Power generation with laterally packaged piezoelectric fine wires,” Nature Nanotechnology, vol. 4, pp. 34–39, Nov. 2008. → pages 16, 29, 30 [23] Z. L. Wang, “Piezoelectric nanogenerators based on zinc oxide nanowire arrays,” Science, vol. 312, pp. 242–246, Apr. 2006. → pages 16 [24] X. Wang, J. Song, J. Liu, and Z. L. Wang, “Direct-current nanogenerator driven by ultrasonic waves,” Science, vol. 316, pp. 102–105, Apr. 2007. → pages 16 [25] Z. Gao, J. Zhou, Y. Gu, P. Fei, Y. Hao, G. Bao, and Z. L. Wang, “Effects of piezoelectric potential on the transport characteristics of metal-ZnO nanowire-metal field effect transistor,” Journal of Applied Physics, vol. 105, no. 11, p. 113707, 2009. → pages 17, 70 [26] R. Zhu, D. Wang, S. Xiang, Z. Zhou, and X. Ye, “Piezoelectric characterization of a single zinc oxide nanowire using a nanoelectromechanical oscillator,” Nanotechnology, vol. 19, p. 285712, July 2008. → pages 17 [27] Z. L. Wang, “Zinc oxide nanostructures: growth, properties and applications,” Journal of Physics: Condensed Matter, vol. 16, pp. R829–R858, June 2004. → pages 19 [28] R. Agrawal and H. D. Espinosa, “Giant piezoelectric size effects in zinc oxide and gallium nitride nanowires. A first principles investigation,” Nano Letters, vol. 11, pp. 786–790, Feb. 2011. → pages 19 [29] C. Lin and Y. Li, “Synthesis of ZnO nanowires by thermal decomposition of zinc acetate dihydrate,” Materials Chemistry and Physics, vol. 113, pp. 334–337, Jan. 2009. → pages 20 [30] M. Kamalasanan and S. Chandra, “Sol-gel synthesis of ZnO thin films,” Thin Solid Films, vol. 288, pp. 112–115, Nov. 1996. → pages 24 67 [31] S. Rani, P. Suri, P. Shishodia, and R. Mehra, “Synthesis of nanocrystalline ZnO powder via sol-gel route for dye-sensitized solar cells,” Solar Energy Materials and Solar Cells, vol. 92, pp. 1639–1645, Dec. 2008. → pages 24 [32] G. Kenanakis, D. Vernardou, E. Koudoumas, and N. Katsarakis, “Growth of c-axis oriented ZnO nanowires from aqueous solution: The decisive role of a seed layer for controlling the wires’ diameter,” Journal of Crystal Growth, vol. 311, pp. 4799–4804, Dec. 2009. → pages 24 [33] M. Ashfold, R. Doherty, N. Ndiforangwafor, D. Riley, and Y. Sun, “The kinetics of the hydrothermal growth of ZnO nanostructures,” Thin Solid Films, vol. 515, pp. 8679–8683, Oct. 2007. → pages 25 [34] L. E. Greene, B. D. Yuhas, M. Law, D. Zitoun, and P. Yang, “Solution-grown zinc oxide nanowires,” Inorganic Chemistry, vol. 45, pp. 7535–7543, Sept. 2006. → pages 27 [35] C. Xu, P. Shin, L. Cao, and D. Gao, “Preferential growth of long ZnO nanowire array and its application in dye-sensitized solar cells,” The Journal of Physical Chemistry C, vol. 114, pp. 125–129, Jan. 2010. → pages 27 [36] Y. Park, V. Choong, Y. Gao, B. R. Hsieh, and C. W. Tang, “Work function of indium tin oxide transparent conductor measured by photoelectron spectroscopy,” Applied Physics Letters, vol. 68, no. 19, p. 2699, 1996. → pages 30 [37] K. Sugiyama, H. Ishii, Y. Ouchi, and K. Seki, “Dependence of indium-tin-oxide work function on surface cleaning method as studied by ultraviolet and x-ray photoemission spectroscopies,” Journal of Applied Physics, vol. 87, no. 1, p. 295, 2000. → pages 30 [38] A. Moujoud, S. H. Oh, H. S. Shin, and H. J. Kim, “On the mechanism of conductivity enhancement and work function control in PEDOT:PSS film through UV-light treatment,” physica status solidi (a), vol. 207, pp. 1704–1707, May 2010. → pages 31 [39] O. Dimitriev, D. Grinko, Y. Noskov, N. Ogurtsov, and A. Pud, “PEDOT:PSS films—Effect of organic solvent additives and annealing on the film conductivity,” Synthetic Metals, vol. 159, pp. 2237–2239, Nov. 2009. → pages 31 [40] J. Song, Y. Zhang, C. Xu, W. Wu, and Z. L. Wang, “Polar charges induced electric hysteresis of ZnO nano/microwire for fast data storage,” Nano Letters, vol. 11, pp. 2829–2834, July 2011. → pages 32 68 [41] S. H. Ko, D. Lee, N. Hotz, J. Yeo, S. Hong, K. H. Nam, and C. P. Grigoropoulos, “Digital selective growth of ZnO nanowire arrays from inkjet-printed nanoparticle seeds on a flexible substrate,” Langmuir, vol. 28, no. 10, pp. 4787–4792, 2012. → pages 39 [42] D. Soltman and V. Subramanian, “Inkjet-printed line morphologies and temperature control of the coffee ring effect,” Langmuir, vol. 24, pp. 2224–2231, Mar. 2008. → pages 40 [43] S. Park, C.-B. Yun, Y. Roh, and J.-J. Lee, “PZT-based active damage detection techniques for steel bridge components,” Smart Materials and Structures, vol. 15, pp. 957–966, Aug. 2006. → pages 56 [44] G. Park, H. H. Cudney, and D. J. Inman, “Impedance-based health monitoring of civil structural components,” Journal of Infrastructure Systems, vol. 6, pp. 153–160, Dec. 2000. → pages 56 [45] F. Wu and F.-K. Chang, “Built-in active sensing diagnostic system for civil infrastructure systems,” pp. 27–35, July 2001. → pages 56 [46] M. Sun, W. Staszewski, R. Swamy, and Z. Li, “Application of low-profile piezoceramic transducers for health monitoring of concrete structures,” NDT & E International, vol. 41, pp. 589–595, Dec. 2008. → pages 56 [47] C. K. Leung, N. Elvin, N. Olson, T. F. Morse, and Y.-F. He, “A novel distributed optical crack sensor for concrete structures,” Engineering Fracture Mechanics, vol. 65, pp. 133–148, Jan. 2000. → pages 56, 57 [48] X. Chen and F. Ansari, “Fiber optic stress wave sensor for detection of internal flaws in concrete structures,” Journal of Intelligent Material Systems and Structures, vol. 10, pp. 274–279, Apr. 1999. → pages 56 [49] D. C. Betz, G. Thursby, B. Culshaw, and W. J. Staszewski, “Acousto-ultrasonic sensing using fiber bragg gratings,” Smart Materials and Structures, vol. 12, pp. 122–128, Feb. 2003. → pages 56 [50] D. R. Lide, CRC handbook of chemistry and physics : a ready-reference book of chemical and physical data : 2008-2009. Boca Raton; London; New York: CRC Press, 2008. → pages 58 69 Appendix A COMSOL Models of Nanowire Strain Models of bending and tensile strain on a zinc oxide nanowire were simulated in COMSOL Multiphysics 3.4 software in order to determine the direction and rela- tive magnitude of the electric field for both cases. The material constants used were as published from [25] and from the COMSOL material library where unspecified. The constants are provided in Tables A.1-A.3. Table A.1: Elastic constants, ci j (GPa), for ZnO nanowire model c11 = c22 = 207.0 c21 = 117.7 c13 = c23 = 106.1 c33 = 209.5 c44 = 44.8 c55 = c66 = 44.6 The length of the nanowire for both cases was 1200 nm and the inner circle diameter was 86.6 nm, given a hexagonal edge length of 100 nm. For the stretch- ing case, 85 nN of tensile force was applied to both ends of the nanowire. For the bending case, the bottom of the nanowire was fixed and 85 nN of force was applied to the free end of the nanowire. For the stretching case, the piezoelectric 70 Table A.2: Piezoelectric coupling matrix, ei j (C/m2), for nanowire model e31 = e32 =−0.51 e33 = 1.22 e24 = e15 =−0.45 Table A.3: Relative permittivity matrix, εi j (×ε0 F/m), for nanowire model ε11 = 7.77 ε22 = 7.77 ε33 = 8.91 potential develops across the length of the nanowire and is larger in magnitude than for the bending case. For bending, the piezoelectric potential develops across the thickness of the nanowire, while the base remains grounded. See Figure 3.2 for the visualization of piezoelectric potential in both cases. 71 Appendix B Matlab Code for Clamped Vibration Test clear all; close all; clc; instrreset; sensor_ID = input('Enter sensor identifier: ', 's'); sensitivity = input('Enter charge amplifier sensitivity (pC/V): ','s'); preload = input('Enter preloading force (N): ', 's'); datestr=date; f_start = 90000; int = -100; f_end = 1000; frequency = f_start:int:f_end; amplitude = frequency; [M,N]=size(frequency); N data=zeros(N,3); time=4./frequency; %record n cycles in oscilloscope num_avg = 1; filename = strcat(sensor_ID,'_',sensitivity,'pCV_',num2str(f_end),'-',... num2str(f_start),'Hz_',preload,'N_',datestr); 72 voltagedata = strcat(filename,'_voltage.txt'); wavename = strcat(filename,'_wave.txt'); wavetimesname = strcat(filename,'_wavetimes.txt'); figname = strcat(filename,'_actuation_voltage.fig'); figname2 = strcat(filename,'_Cvsf.fig'); figname3 = strcat(filename,'_Pvsf.fig'); %GPIB address for signal generator (input signal to actuator) gen = gpib('ni',1,8); %GPIB address for multimeter (to measure voltage applied to actuator) volt = gpib('ni',1,22); %Address of oscilloscope (to measure output from sensor) osc=icdevice('Agilent546XX.mdd','USBInstrument1'); fopen(gen); fopen(volt); connect(osc); %Set up oscilloscope invoke(osc.Measurements,'Abort'); osc.Channel(1).Enabled = 'on'; %sensor output osc.Channel(1).Coupling = 'Agilent546XXVerticalCouplingAC'; osc.System.TimeoutMilliseconds=1000000; %Set up averaging of 256 cycles osc.Acquisition.AcquisitionType='Agilent546XXAcquisitionTypeAverage'; osc.Acquisition.NumberOfPointsMin=1000; numpoints = osc.Acquisition.NumberOfPointsMin; osc.Acquisition.NumberOfAverages=256; waveform_zno = zeros(N,numpoints); times = waveform_zno; Yrange = 8; %initial Y-axis range osc.Channel(1).Range = Yrange; %Set up voltmeter fprintf(volt,'CONF:VOLT:AC 10, 0.00001') %Determine attenuated voltage at max frequency ampl = 11.5; amplitude(1) = ampl; fprintf(gen,'FUNC 0'); 73 fprintf(gen,'FREQ %i',frequency(1)); fprintf(gen,'AMPL %d VR',ampl); fprintf(gen,'OUTE 1'); pause(0.5); %allow output to settle osc.Acquisition.TimePerRecord=(time(1)); invoke(osc.Measurements,'Initiate'); pause(256*time(1)); %record voltage applied to actuator fprintf(volt, 'READ?') % Retrieve Voltage data Vt(1) = scanstr(volt,',','%f'); % Save data to array target = Vt(1); %save voltage waveform from oscilloscope waveform=invoke(osc.Measurement1,'FetchWaveform'); data(1,1) = frequency(1); data(1,2) = amplitude(1); data(1,3) = Vt(1); waveform_zno(1,:) = waveform; for i = 2:N i %display current iteration fprintf(gen,'FREQ %i',frequency(i)); pause(0.1); fprintf(volt, 'READ?') % Retrieve Voltage data current = scanstr(volt,',','%f') % Save data to array while current < 0.99*target || current > 1.01*target if current < 0.75*target ampl = 1.15*ampl while ampl>14.14 ampl = 0.95*ampl; end fprintf(gen,'AMPL %d VR',ampl); pause(0.1); fprintf(volt, 'READ?') % Retrieve Voltage data current = scanstr(volt,',','%f'); elseif current < 0.90*target ampl = 1.05*ampl while ampl>14.14 ampl = 0.95*ampl; end fprintf(gen,'AMPL %d VR',ampl); 74 pause(0.1); fprintf(volt, 'READ?') % Retrieve Voltage data current = scanstr(volt,',','%f'); elseif current < 0.99*target ampl = 1.005*ampl while ampl>14.14 ampl = 0.95*ampl; end fprintf(gen,'AMPL %d VR',ampl); pause(0.1); fprintf(volt, 'READ?') % Retrieve Voltage data current = scanstr(volt,',','%f'); elseif current > 1.25*target ampl = 0.75*ampl fprintf(gen,'AMPL %d VR',ampl); pause(0.1); fprintf(volt, 'READ?') % Retrieve Voltage data current = scanstr(volt,',','%f'); elseif current > 1.10*target ampl = 0.90*ampl fprintf(gen,'AMPL %d VR',ampl); pause(0.1); fprintf(volt, 'READ?') % Retrieve Voltage data current = scanstr(volt,',','%f'); elseif current > 1.01*target ampl = 0.99*ampl fprintf(gen,'AMPL %d VR',ampl); pause(0.1); fprintf(volt, 'READ?') % Retrieve Voltage data current = scanstr(volt,',','%f'); end end amplitude(i) = ampl; osc.Acquisition.TimePerRecord=(time(i)); invoke(osc.Measurements,'Initiate'); pause(256*time(i)); %record voltage applied to actuator Vt(i) = current; % Save data to array target = Vt(1); waveform=invoke(osc.Measurement1,'FetchWaveform'); 75 data(i,1) = frequency(i); data(i,2) = amplitude(i); data(i,3) = Vt(i); waveform_zno(i,:) = waveform; i = i-1; end fprintf(gen,'OUTE 0'); figure; h = plot(data(:,1),data(:,3)); xlabel('Target Voltage (V)'); ylabel('Actual Voltage (V)'); for n=1:N times(n,:)=(1:numpoints).*time(n)./numpoints; end save(voltagedata, 'data', '-ascii'); save(wavename,'waveform_zno','-ascii'); save(wavetimesname,'times','-ascii'); saveas(h,figname); disconnect(osc); delete(osc); fclose(gen); fclose(volt); for n = 1:N Y(n,:) = fft(waveform_zno(n,:),1000); %complex volt-seconds Pyy(n,:) = Y(n,:).*conj(Y(n,:))/1000; %Vˆ2*s freq(n,:) = (1./(times(n,1000)/1000))/1000*(0:500); samplefreq(n)=(1./(times(n,1000)/1000)); [ampl(n) ind(n)] = max(Pyy(n,2:500)); time(n) = freq(n,ind(n)+1); RMS_time(n) = sqrt((sum(waveform_zno(n,:).ˆ2)/... length(waveform_zno(n,:)))); RMS_fft(n) = sqrt(sum(abs(Y(n,:)/length(Y(n,:))).ˆ2)); end figure; 76 h2 = plot(frequency./1000,RMS_fft.*str2num(sensitivity)); xlabel('Frequency (kHz)'); ylabel('Charge (pC)'); saveas(h2,figname2); 77


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items