UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Towards all-polymer surface acoustic wave chemical sensors for air quality monitoring Man, Gabriel 2009

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

Item Metadata

Download

Media
24-ubc_2010_spring_man_gabriel.pdf [ 85.92MB ]
Metadata
JSON: 24-1.0068869.json
JSON-LD: 24-1.0068869-ld.json
RDF/XML (Pretty): 24-1.0068869-rdf.xml
RDF/JSON: 24-1.0068869-rdf.json
Turtle: 24-1.0068869-turtle.txt
N-Triples: 24-1.0068869-rdf-ntriples.txt
Original Record: 24-1.0068869-source.json
Full Text
24-1.0068869-fulltext.txt
Citation
24-1.0068869.ris

Full Text

 Towards all-polymer surface acoustic wave chemical sensors for air quality monitoring    by   Gabriel Man   B.A.Sc., The University of British Columbia, 2007     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) December 2009    © Gabriel Man, 2009     ii Abstract Volatile organic compounds (VOCs) are a precursor to the formation of ground-level ozone and airborne particulate matter, both of which are hazardous to human health. Currently in Canada, other air pollutants such as ozone and nitrous oxides are measured by an air quality monitoring network in real-time, while VOCs are collected in canisters and sent to a central laboratory for analysis.  This is a time-consuming and non real-time method, and due to the spatial variability of air pollution, many points of measurement are needed.  A distributed point sensor network could address the resolution and real-time challenges, but would impose an added operating expenditure burden on air quality monitoring agencies.  Low-cost, yet sensitive chemical sensors could contribute to lowering operating expenditures of a network’s sensing units over the installed lifetime of the units.  The objective of this work was to lay the groundwork for a sensing platform from which low-cost yet sensitive chemical sensors can be developed.  The sensing platform is an all-polymer surface acoustic wave (SAW) device, and the materials selected for its fabrication are Polyvinylidene Fluoride (PVDF) for the sensor substrate and Poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate) (PEDOT:PSS) for the interdigital transducer electrodes.  In this work, an apparatus and a process for preparing piezoelectric PVDF film was developed.  PVDF-based resonators were successfully demonstrated.  In addition, repeatable processes for inkjet micropatterning highly electrically conductive PEDOT:PSS electrode tracks on PVDF were developed for three inkjet nozzle orifice sizes (20, 30, 40 µm).  For tracks micropatterned using the same process, the electrical resistances have a standard deviation of 8.5% of the average.  The electrical conductivity of micropatterned tracks is approximately 150 S/cm, or one-sixth  iii of the manufacturer’s claimed bulk film conductivity.  Using the 30 µm nozzle, the smallest electrode track width that can be micropatterned repeatably is 75 µm.  A track width of 55 µm was achieved using the 20 µm nozzle.  iv Table of Contents  Abstract ............................................................................................................................... ii Table of Contents............................................................................................................... iv List of Tables .................................................................................................................... vii List of Figures .................................................................................................................... ix Acknowledgements.......................................................................................................... xiii Dedication ........................................................................................................................ xiv 1. Introduction..................................................................................................................... 1 1.1 Motivation................................................................................................................. 2 1.2 Objectives ................................................................................................................. 3 1.3 Sensing technology for Volatile Organic Compounds ............................................. 4 1.3.1 Overview............................................................................................................ 4 1.3.2 Selection of sensor technology .......................................................................... 7 1.4 Principles of operation of Surface Acoustic Wave sensors ...................................... 9 1.4.1 Acoustic wave propagation in an isotropic material.......................................... 9 1.4.2 Surface Acoustic Wave excitation and detection............................................. 10 1.4.3 Perturbations in Surface Acoustic Wave propagation for gas sensing ............ 12 1.5 Prior art ................................................................................................................... 14 2. Sensor materials ............................................................................................................ 17 2.1 Selection of sensor device materials ....................................................................... 17 2.2 Influence of material properties on SAW sensor performance............................... 19 2.2.1 Device sensitivity............................................................................................. 20 2.2.2 Stability............................................................................................................ 24  v 2.3 Selection methodology for sensing layers .............................................................. 27 2.3.1 Overview of solubility-prediction systems ...................................................... 27 2.3.2 Application of the HSP to select polymeric sensing layers ............................. 30 3. Sensor design and fabrication ....................................................................................... 32 3.1 Substrate.................................................................................................................. 32 3.1.1 The piezoelectric phase of PVDF .................................................................... 32 3.1.2 Film thickness requirements ............................................................................ 33 3.1.3 Formation of the piezoelectric substrate .......................................................... 33 3.1.4 Processing of commercially-available thick film ............................................ 36 3.1.5 Methods used to characterize the piezoelectric constants................................ 48 3.1.6 Summary of results .......................................................................................... 59 3.2 Transducers ............................................................................................................. 61 3.2.1 Basic requirements........................................................................................... 61 3.2.2 Selection of inkjet microprinting as the fabrication technique ........................ 61 3.2.3 Materials and equipment.................................................................................. 62 3.2.4 Challenges and considerations for inkjet micropatterning of conductive traces ................................................................................................................................... 65 3.2.5 Development of an inkjet microprinting-based electrode deposition process . 74 3.2.6 RC charging considerations ............................................................................. 92 3.3 Towards the fabrication of a SAW device.............................................................. 93 3.3.1 Process integration considerations................................................................... 93 3.3.2 Loss of electrode conductivity phenomenon ................................................... 95 4. Conclusions and future work ........................................................................................ 97  vi 4.1 Summary of results ................................................................................................. 97 4.2 Future work............................................................................................................. 98 4.2.1 Proposed piezoelectric substrate characterization procedures ......................... 98 4.2.2 Reduction of electrode track width .................................................................. 99 4.2.3 Investigation into how PVDF material properties could limit SAW device performance .............................................................................................................. 99 4.2.4 Investigation into the loss of electrode electrical conductivity phenomenon 100 4.2.5 Fabrication and characterization of a SAW device prototype ....................... 101 4.2.6 Development of a technique for addressing the temperature sensitivity of PVDF ...................................................................................................................... 101 4.2.7 Verification of sensing layer selection methodology..................................... 102 References....................................................................................................................... 103 Appendices...................................................................................................................... 112 Appendix A – Experimental procedures for PVDF spin coating ............................... 112 Appendix B – Matlab script for calculating evaporation time.................................... 113 Appendix C – Matlab script for calculating the RC charging time constant .............. 114 Appendix D – Autodrop macros for multi-layer patterning ....................................... 116     vii List of Tables  Table 1 - Ambient concentrations of 10 VOCs whose concentrations were abnormally high on days when high ozone concentrations were present (maximum one hour concentration of at least 50 ppb) at the Chilliwack, BC monitoring station [3] ................. 2 Table 2 - Typical material properties of PVDF [33]......................................................... 21 Table 3 - Typical Rayleigh velocities for substrate materials traditionally used for SAW devices [9]......................................................................................................................... 21 Table 4 – Relative dielectric permittivity of PVDF measured at different frequencies [33] ........................................................................................................................................... 23 Table 5 - Starting values for PVDF heating and poling process parameters .................... 42 Table 6 - Typical piezoelectric stain constant values for PVDF [33]............................... 49 Table 7 – Comparison of deposited drop parameters for untreated and surface treated PVDF (30 µm nozzle, same process)................................................................................ 68 Table 8 - Average contact diameter, and contact diameter variability for drops printed using different solutions.................................................................................................... 72 Table 9 - Comparison of dots printed using two different grades of PEDOT:PSS, using the 30 µm nozzle and the same process (a drop spacing of 140 µm yields a fully discontinuous line) ............................................................................................................ 73 Table 10 – Track parameters for five microprinted tracks such as the one in Figure 28.. 82 Table 11 - Track statistics for two tracks printed using a 20 µm nozzle .......................... 88 Table 12 – Ranking of plasma treatments by contact diameter, contact diameter standard deviation, and interdrop spacing standard deviation......................................................... 89  viii Table 13 - Summary of optimal microprinting process parameters, and deposited drop characteristics with optimal plasma treatment .................................................................. 91 Table 14 – Calculated IDT parameter values ................................................................... 93   ix List of Figures Figure 1 - Three-dimensional view of a segment of β phase PVDF................................. 33 Figure 2 - Convection toaster oven with drill press vice on a cooking rack, supported by an aluminum block underneath......................................................................................... 40 Figure 3 – (a) First electrode setup (b) second electrode setup ........................................ 43 Figure 4 - Third electrode setup........................................................................................ 44 Figure 5 - Equipment for heating and poling PVDF films, with the convection oven and HV supply inside fumehood ............................................................................................. 48 Figure 6  - PVDF-based resonator fabricated by depositing 50 nm of gold on both sides of the film.............................................................................................................................. 51 Figure 7 - Holder for PVDF samples................................................................................ 54 Figure 8 - PVDF film clamped onto steel plate using magnets ........................................ 54 Figure 9 - Screenshot of surface profiler software showing surface roughness measurement ..................................................................................................................... 56 Figure 10 – Single-point LDV data showing the effects of higher poling field strength . 58 Figure 11 – Single-point LDV data showing the effects of longer poling time (data for the 18 kV/mm, 30 mins poling is not visible since no signal was obtained from the sample)58 Figure 12 – Single-point LDV data showing the effects of poling time after heating has stopped .............................................................................................................................. 59 Figure 13 – Comparison of measurements taken by two LDV’s on the same sample (45 kV/mm, 30 mins, 10 min cool down) ............................................................................... 59 Figure 14 - Location inside plasma cleaner chamber where PVDF wafers are placed .... 63  x Figure 15 – Surface profile of 3 lines demonstrating repetition accuracy of Microdrop. After printing each line shown in the figure, the stage is moved up to several centimeters away to print several other lines, before printing the next line in the figure.   Process: 30 µm nozzle, Clevios 1000 + 5 wt% DMSO ink, and plasma-treated PVDF...................... 65 Figure 16 – Line of drops printed with a discontinuous drop spacing, showing shifting of drop position after being deposited (30 µm nozzle, 90 µm drop spacing) ....................... 68 Figure 17 - PEDOT:PSS line printed on PVDF using a single layer patterning process . 70 Figure 18 – Comparison of surface profiles of two tracks printed using the same process (a) 2 minutes per layer drying time and no ventilation  (b) 2 minutes per layer drying time with ventilation ................................................................................................................. 71 Figure 19 – Comparison of surface profiles of tracks showing how adding a small amount of surfactant significantly increases surface wetting (a) Clevios 1000 + 5wt% DMSO + 0.5wt% Triton X-100 (b) Clevios 1000 + 5wt% DMSO.................................................. 74 Figure 20 - Contact diameter, and contact diameter standard deviation (shown as error bars), as a function of plasma treatment time and power level......................................... 77 Figure 21 - Interdrop spacing standard deviation as a function of plasma treatment time and power level ................................................................................................................. 77 Figure 22 – Surface profiles of lines microprinted with (a) 70 µm drop spacing and (b) 80 µm drop spacing, showing drop coalescence.................................................................... 78 Figure 23 – Surface profiles of lines microprinted with (a) 90 µm drop spacing and (b) 100 µm drop spacing, showing drop coalescence............................................................. 79 Figure 24 – Surface profiles of lines microprinted with (a) 110 µm drop spacing and (b) 120 µm drop spacing, showing no drop coalescence........................................................ 79  xi Figure 25 – Surface profiles for alternate overlaying with (a) 110 µm and (b) 120 µm drop spacing ...................................................................................................................... 80 Figure 26 – Surface profiles for alternate overlaying with (a) 140 µm and (b) 160 µm drop spacing ...................................................................................................................... 81 Figure 27 – Surface profile of track produced with 3 layers and 120 µm drop spacing using the 30 µm nozzle ..................................................................................................... 81 Figure 28 – Surface profile of a typical track microprinted using multi-layer process and 30 µm nozzle..................................................................................................................... 83 Figure 29 – Surface profiles of single layer patterned pads (a) 100 µm drop spacing (b) 120 µm drop spacing......................................................................................................... 84 Figure 30 – Surface profile of single layer patterned pad (140 µm drop spacing) ........... 84 Figure 31 – Surface profiles of square pads patterned with two layers (a) 100 µm drop spacing (b) 110 µm drop spacing...................................................................................... 85 Figure 32 – Surface profiles of square pads patterned with two layers (a) 120 µm drop spacing (b) 130 µm drop spacing...................................................................................... 85 Figure 33 - Contact diameter vs. plasma treatment time, for deposited drops printed using the 20 µm nozzle.  The error bars represent the contact diameter standard deviation...... 86 Figure 34 - Interdrop spacing variability vs. plasma treatment time, for deposited drops printed using the 20 µm nozzle......................................................................................... 87 Figure 35  - Surface profiles of two tracks printed with a 20 µm nozzle, using an early stage multi-layering process (a) 0-50-0-50-0-50-0-50 (b) 0-33-67-50-33-67-50-50........ 88 Figure 36 – Surface profile of 5 electrode finger pair printed using a 20 µm nozzle and an 8 layer process................................................................................................................... 89  xii Figure 37 - Contact diameter average and standard deviation (error bars) as a function of the plasma treatment time and power level for the 40 µm nozzle .................................... 90 Figure 38 - Interdrop spacing variability as a function of the plasma treatment time for the 40 µm nozzle............................................................................................................... 90           xiii Acknowledgements I offer my enduring gratitude to my two advisors, Dr. Konrad Walus and Dr. Boris Stoeber, for their guidance and support, for giving me the freedom to explore my own project, and for stimulating my interest in sensors.    By constantly asking challenging questions, they have motivated me to become a better researcher who can anticipate some of those questions and is a little better prepared to answer some of them.  I would particularly like to thank Dr. Konrad Walus for sparking my interest in research via my senior undergraduate capstone project, and for encouraging me to pursue graduate studies.  In retrospect I believe it was the right decision for me.  I thank Dr. Boris Stoeber for his attention to detail; he has motivated me to develop a rigorous approach to scientific inquiry.  I would like to acknowledge Kaan Williams for his contributions to the development of the spin coating process and preliminary stretching and heating apparatus.  I offer my thanks to Dr. Frank Ko and Dr. Heejae Yang for letting me use their high voltage supply.  In addition, I would like to acknowledge the support of four agencies: the Canadian Police Research Centre (CPRC), the Natural Sciences and Engineering Research Council of Canada (NSERC), the Fraser Basin Council with their British Columbia Clean Air Research (BC CLEAR) program, and the Canada Foundation for Innovation (CFI).  xiv Dedication     To my parents, Alfred and Cecilia Man   1 1. Introduction Volatile Organic Compounds (VOCs) are a known precursor to the formation of ground- level ozone and secondary particulate matter, both of which have negative impacts on human health [1].  In addition, many VOCs emitted by anthropogenic sources such as fossil fuel-based combustion engines, solvent evaporation, and chemical manufacturing are known carcinogens.  As a result, monitoring of VOCs is necessary, both to detect elevated levels of the gases and to serve as a tool for formulating effective air management strategies and policies.  What complicates VOC detection is the presence of many classes of VOC’s in the atmosphere: alkanes, alkenes, alkynes, aromatics, aldehydes, ketones, alcohols, esters, and some chlorinated compounds [2].  To formulate effective air quality management strategies, it is necessary to distinguish between different VOCs for two reasons.  The first reason is because the VOCs emitted by anthropogenic sources (controllable) are often different than the VOCs emitted by biogenic sources (less controllable), and the second is due to the substantial variability in the contribution of individual VOCs to ozone formation, which is due to their varying photochemical reactivities [2].  For an air quality monitoring solution to be effective, the sensing units, and ultimately the chemical sensors, must be able to differentiate between the many different VOCs that exist in the atmosphere.  In addition to being selective, the sensors also need to be very sensitive and possess low detection limits.  According to data provided by Environment Canada, ambient VOC concentrations are typically in the sub-ppb concentration range, as shown in Table 1.  2 Table 1 - Ambient concentrations of 10 VOCs whose concentrations were abnormally high on days when high ozone concentrations were present (maximum one hour concentration of at least 50 ppb) at the Chilliwack, BC monitoring station [3] VOC Ambient concentration 1-Butene 0.1 ppb Ethylene 0.6 ppb m/p-Xylene 0.2 ppb Isoprene 0.8 ppb Toluene 0.4 ppb Propylene 0.2 ppb Isopentane 0.6 ppb Styrene 0.1 ppb α-Pinene 0.3 ppb  The ten VOCs listed in Table 1 are those for which the concentrations were noticeably higher than their average daily concentration on days when ozone levels were particularly high (one hour concentration of at least 50 ppb) at the Chilliwack, BC air quality monitoring station.  Data from the Chilliwack station was used because it is usually downwind of the Vancouver downtown urban core.  The chemical sensors would need to be able to detect sub-ppb concentrations, either directly or with the use of a preconcentrator [4].  1.1 Motivation British Columbia has approximately 100 continuous and 50 non-continuous ambient air monitoring stations [5].  Presently, VOCs are sampled at the non-continuous monitoring stations, and the canisters are collected by technicians in the field after a specific duration of time (i.e. 24 hours) and sent to a central lab for gas chromatography-mass spectrometry (GC-MS) analysis.  This method is costly, time-consuming, and cannot provide real-time measurements, which is a major disadvantage due to the temporal  3 variability of air pollution.  Real-time in air quality monitoring is typically defined as sample intervals ranging from 15 minutes to an hour, which current VOC sampling techniques cannot achieve.  In addition, due to the spatial variability of air pollution, measurements must be made at many points.  A distributed, real-time air quality monitoring network can address these challenges, but a distributed sensor network imposes increased operating costs due to the need for regular sensor replacement and sensing unit maintenance.  An inexpensive point sensor could facilitate a large monitoring network at a lower overhead.  1.2 Objectives The three high-level requirements for the chemical sensor technology, as mentioned in Sections 1 and 1.1, are: • selectivity – able to distinguish VOCs listed in Table 1, or classes of VOCs at the minimum, • low detection limits – sub-ppb concentrations possibly in conjunction with a preconcentration stage, • low cost.  As discussed in the following section, the Surface Acoustic Wave (SAW) platform was selected for this work.  The sensing platform would form the basis of low-cost, yet sensitive and selective chemical sensors.  To achieve the high-level requirement of low- cost, polymeric materials were used in the fabrication of the SAW device.  The objective  4 of this work was to develop the processes which would be used in the fabrication of the SAW sensing platform.    The processes developed and described in this thesis are the: (1) development of a low-cost apparatus and a repeatable process for preparing polymer- based sensor substrate material, (2) development of repeatable, high-resolution electrode micropatterning processes.  1.3 Sensing technology for Volatile Organic Compounds 1.3.1 Overview An assessment of sensor technologies for detecting both inorganic and organic compounds was performed previously [6].  VOC sensor technologies can be grouped into five major classes based on detection mechanism: electrical capacitive, electrical conductance-based, ionization-based, gravimetric, and optical sensors.  The first class, capacitive sensors (or chemicapacitors), undergo a change in capacitance due to a change in analyte concentration, consume little power and are low-cost due to their simple construction.  Chemicapacitors typically incorporate polymer-based sensing layers as the chemically selective material, and have been demonstrated to detect analytes down to sub-ppm concentrations with response times ranging from less than a second to more than 30 minutes.  The second class of VOC sensor technologies are electrical conductance-based sensing devices, which have been implemented with a variety of sensor materials and device geometries.  The detection mechanism varies across this sensor class.  Examples of  5 conductance-based sensors include chemical resistors (chemiresistors), chemically sensitive field-effect transistors (chemFETs), amperometric electrochemical sensors, semiconducting metal oxide (SMO) sensors, thermal sensors, and nanomaterial-based sensors.  Polymer-based chemiresistors are simple in architecture, low-cost, capable of sub-50 ppm-level detection, and have reasonably quick response times ranging from a few seconds to a minute.  ChemFET’s can be fabricated using standard CMOS fabrication processes, and are simple to integrate with associated transducer or signal processing circuitry.  Demonstrated sensors can detect down to one ppm of analyte. Amperometric electrochemical sensors are generally recognized for their low power consumption, fast response times (10 seconds to a couple of minutes), low cost, small size, selectivity, stability, and sensitivity.  The sensing technology is mature, a linear relationship between output electrical current and analyte concentration typically exists over a wide range, and the sensors possess low detection limits that are typically at the sub-ppm level.  SMO sensors are also a mature technology, have high sensitivity to combustible gases, are compact and durable, and are relatively inexpensive to produce. Detection limits range in the low ten’s of ppm concentrations, and response times are typically less than one minute.  Thermal gas sensors have been in use for more than 50 years, primarily in the detection of combustible gases; the sensors are simple to fabricate, and their detection limits range between ten’s to hundred’s of ppm.  Response times for thermal gas sensors generally range between 10-15 seconds.  Nanomaterial conductance-based sensors have been investigated in recent years, and reported devices in the literature possess sub-200 ppm detection limits and 5 to 50 second response times.  6 The performance of these types of sensors is expected to improve as highly controllable fabrication processes are developed.  Ionization sensors are the third class of VOC sensing technology, and the two relevant technologies are ion mobility spectrometry (IMS) and photoionization detection (PID). IMS is a highly sensitive technique which offers low detection limits in the ten’s to hundred’s of ppb’s for most analytes.  IMS technology is able to detect a wide range of compounds with good selectivity, and offers response times on the order of ten’s of seconds.  PID is a mature and commercially successful technology, and at present, is the preferred choice for monitoring total VOC levels.  The advantages of PID include extremely low detection limits (10’s of ppb’s) and fast response, typically less than 20 seconds.  The fourth class of VOC sensing technology are gravimetric sensors.  The two most relevant types of gravimetric-based sensors are acoustic wave sensors and microcantilever sensors.  Both types of sensors detect analytes via a sensing layer, which is typically polymeric.  There are four types of acoustic-based sensors, but thickness shear mode (TSM, also called quartz crystal microbalance or QCM) and SAW sensors have been most heavily exploited in chemical sensing.  Characteristics of TSM and SAW sensors include demonstrated low detection limits (ppb’s to low ppm’s), excellent sensitivity, small size, a simple architecture, relatively low cost, and reasonably fast response times (10-60 seconds).  Microcantilever-based sensors have demonstrated  7 detection limits at ppb or ppt (parts per trillion) levels, and response times on the order of a couple minutes.  Optical VOC sensors are the fifth class of sensing technology, and include chemiluminescent sensors, colorimetric sensors, fluorescence sensors, and infrared spectrometers.  Optical sensors are generally considered to have the best selectivity and detection limits and in several cases, are able to identity components of a chemical mixture in one measurement.  The main advantage of optical detection is the remote sensing capability of some of the sensing devices.  With remote sensing, it is possible to directly localize the source of the target gas, and arguably the most popular optical technique is Fourier Transform Infrared (FTIR) spectrometry.  Nearly all compounds can be detected using FTIR, and the technology offers sub-ppm detection limits and sub- 20 second response times.  1.3.2 Selection of sensor technology Each VOC sensing technology possesses unique performance and cost advantages and disadvantages, and no sensing technology is superior in all areas.  From the five classes of sensor technologies mentioned in the previous section, the technologies which offer a good compromise between selectivity, low detection limits, and low cost are: • chemicapacitors, • chemiresistors, • chemFETs, • electrochemical sensors,  8 • acoustic wave sensors, • microcantilever sensors.  ChemFETs are susceptible to minor variations in the environment.  A 0.1 °C rise in temperature has an equivalent effect of 30-1000 ppm of analyte, and 1 ppm of water vapour in the air is equivalent to 0.5-13 ppm of analyte.  As a result, chemFETs are not suitable for ambient air monitoring applications since humidity levels are expected to fluctuate constantly.  Amperometric electrochemical sensors are unable to detect organic compounds with non-electroactive functional groups.  If electrochemical sensors are used for air quality monitoring, they will need to be complemented by another sensing technology.  The remaining viable technologies are chemicapacitors, chemiresistors, acoustic wave sensors and microcantilever sensors.  Typical detection limits for chemicapacitors and chemiresistors are in the low ppm range while the gravimetric sensor technologies have been demonstrated to detect ppb-level concentrations.  Consequently, the two sensor technologies of interest are acoustic wave and microcantilever sensors.  Between the two gravimetric sensor technologies, the fabrication of SAW sensors is more straightforward since the SAW devices can be fabricated with additive processes whereas microcantilevers are typically fabricated with subtractive processes.  Due to their demonstrated low detection limits to VOCs, simple architecture, and straightforward fabrication, SAW devices are an attractive platform for building sensitive VOC sensors.   9 1.4 Principles of operation of Surface Acoustic Wave sensors Surface acoustic waves (SAW), where the acoustic energy is confined primarily at the surface of a solid, were first discovered by Lord Rayleigh in 1887 [7].  As a result, this mode of acoustic propagation is also known as the Rayleigh wave, and is very similar to the Rayleigh earthquake which travels along the ground surface of the Earth [8]. Rayleigh wave energy penetrates approximately one wavelength into the propagation medium, and the waves are independent of longitudinal and shear bulk acoustic wave (BAW) modes, which propagate at different velocities [9].  The following sections will give a brief overview of acoustic wave propagation in isotropic materials, describe how Rayleigh waves are induced in piezoelectric materials, and how they can be perturbed.  The terms Rayleigh wave and SAW will be used interchangeably.  The sensing capabilities of SAW-based sensor devices depend on the perturbation resulting from the interaction of the SAWs with the sensing layer.  For a more in-depth discussion and mathematical treatment of acoustic and ultrasonic wave fundamentals and applications, the reader is referred to the following references [7,9].  1.4.1 Acoustic wave propagation in an isotropic material  Two types of fundamental waves can propagate in an isotropic material: longitudinal (also known as dilational, compressional, primary, or P-) waves and transverse (also called shear, secondary, or S-) waves [10].  For longitudinal waves, the direction of polarization, or material deformation, is the same as the direction of propagation.  The deformation can either be compression or expansion, or a combination of the two.  Shear  10 waves deform the propagation medium with a direction that is perpendicular to the direction of propagation, and the motions include both shear and rotation.  Shear waves with shear planes that are perpendicular to each other are treated as independent types.  In seismology for example, the shear planes are referenced to the Earth’s surface and the shear wave that deforms the surface horizontally is called a shear horizontal (SH) wave, and the shear vertical (SV) wave deforms the surface vertically [10].  When bulk propagating waves reach the boundary of the propagation medium, two different types of waves arise.  Surface acoustic waves travel on the surface of a semi-infinite solid, and plate waves can propagate in thin solid plates [7].  1.4.2 Surface Acoustic Wave excitation and detection  Excitation and the synchronous frequency Most SAW devices exploit the piezoelectric effect, a linear electromechanical effect, to create propagating SAWs on the surface of a piezoelectric material.  A small number of devices use other types of coupling, such as magnetic transduction using a magnetostrictive thin film, to create SAWs [11].  As for the piezoelectric effect, the mechanical strain j E ijjiji TsEdS +=  (1.1) is related to the applied electric field Ej and the applied mechanical stress Tj.  The surface charge density j T ijjiji ETdD ε+=  (1.2) is related to the applied stress Tj and applied electric field Ej.  The coupling coefficients dij are known as the charge piezoelectric coefficients, sij are the elastic compliance  11 constants, and εij are the permittivity constants.  Together Equations 1.1 and 1.2 are known as the piezoelectric constitutive relations, and completely describe the interplay between mechanical variables such as stress and strain, and electrical variables such as surface charge density and electric field [7].  The piezoelectric coefficients are described using reduced notation, where i denotes the component of the charge density or electric field in the Cartesian reference frame (x1, x2, x3), j=1, 2, and 3 corresponds to the x1, x2, and x3 axes, and j=4, 5, and 6 corresponds to the shear stress and strain.  The application of stress to produce charge density (Equation 1.2) is called the direct piezoelectric effect, and the application of an electric field to produce strain (Equation 1.1) is called the converse piezoelectric effect.  Aside from the d strain constants, the piezoelectricity of materials can also be described in terms of the g stress constants.  Utilizing the converse piezoelectric effect, a periodic electric field can be applied to a piezoelectric substrate via a suitable transducer to generate a periodic strain field.  The standing surface acoustic wave gives rise to propagating SAWs.  Excitation of SAWs in a piezoelectric substrate using interdigital transducers was first demonstrated by White and Voltmer in 1965 [12].  The interdigital transducer is most efficient when the transducer periodicity matches the SAW wavelength at an excitation frequency called the synchronous frequency d vf 00 = , (1.3) where v0 is the SAW propagation velocity and d is the transducer periodicity [7].    12 Detection An interdigital transducer, similar to one used to induce propagating SAWs, can also be used as a SAW detector.  A propagating SAW in a piezoelectric material generates a propagating layer of bound charge at the surface and hence a propagating evanescent electric field [7].  The SAW is detected via direct piezoelectric coupling.  1.4.3 Perturbations in Surface Acoustic Wave propagation for gas sensing  The perturbations in wave propagation characteristics, specifically wave velocity and attenuation, are exploited in SAW sensors [7].  Since a SAW propagating in a piezoelectric material mechanically deforms the piezoelectric medium and generates an electrical potential, perturbations can arise from both mechanical and electrical coupling between the device and the sensing layer.  Mechanical interactions include mass loading and elastic/viscoelastic effects, and electrical interactions include acoustoelectric interactions [7].  Gas sensing can be performed if the sensing layer increases its mass or changes its electrical conductivity upon exposure to the target analyte.  Mechanical coupling Mass loading is a common perturbation mechanism where propagating SAWs are perturbed by changes in the mass per unit area on the device surface, or surface mass density (ρs).  A key performance metric for mass loading-based sensors is mass sensitivity   13         ++= P v P v P vv c z yx m ωωω pi 20 2 0 2 00 2  (1.4)  where v0 is the SAW propagation velocity, vx0, vy0, vz0 are the SAW surface particle velocities, ω is the operating frequency, and P is the power density (power/area) carried by the wave [7].  As the operating frequency increases, wave energy is confined closer to the surface and the surface particle velocities increase in proportion to (Pω)1/2.  The quantities in the parentheses (i.e. vx02/ωP) are normalized surface particle velocities and are dependant only on the substrate material.  The mass sensitivity determines the relative change in SAW propagation velocity as the wave interacts with the sensing layer, given by sm fc v v ρ∆−=∆ 0 0 . (1.5) If the SAW sensor is incorporated into a phase-locked oscillator loop so that the SAW propagation velocity change ∆v / v0  results in a phase shift between the transmitted and received signals and hence a frequency change ∆f / f0 = ∆v / v0 , then the sensitivity to changes in surface mass density 2 0fcd fdS m s −= ∆ = ρ  (1.6) has a quadratic dependence on operating frequency.  Although the mass sensitivity is intrinsic to the propagation medium, for increased perturbation to the propagating SAW and hence increased sensitivity, the SAW sensor can be operated at a higher synchronous frequency.  The operating frequency of the device is dictated by the SAW propagation  14 velocity - a property of the propagation medium – and the geometry of the interdigital electrodes used to drive the device.  Electrical coupling A propagating SAW in a piezoelectric medium is accompanied by an evanescent electric field [7].  This electric field can be perturbed by the presence of a conductive film on the surface of the SAW, although this perturbation mechanism is not as common as mass- loading.  An example of a device that exploits acoustoelectric coupling is a carbon nanotube-coated SAW sensor for detecting carbon dioxide [13].  The concentration of carbon dioxide modulates the electrical conductivity of the self-assembled nanotube film, which changes the amplitude of the SAWs.  1.5 Prior art  In 1965, White and Voltmer demonstrated direct piezoelectric surface wave transduction using interdigital transducers [12].  Following their seminal work, SAW devices have proliferated in many fields ranging from telecommunications to sensors, and SAW devices have become an established platform for chemical and biological sensing. Examples of conventional SAW gas sensors can be found in a book chapter by Ippolito et al. [14].  In this work, SAW sensors which are fabricated using traditional materials (substrate material is quartz, lithium niobate, etc., and electrode material is aluminum, gold, etc.) and traditional fabrication processes (photolithography, electron beam deposition, etc.) are referred to as conventional SAW sensors.  As mentioned by Grate in a review published in 2000, many review articles, and at least two books, exist on  15 acoustic wave chemical sensors for gas and liquid-phase sensing [15].  However, the conventional SAW device continues to be a popular sensing platform.  Atashbar et al. created hydrogen sensors by first depositing an intermediate layer of aluminum nitride onto 64° YX-LiNbO3-based two-port resonators, and then coating the device with a sensing layer consisting of polyaniline nanofibers [16].  Hsu et al. designed a two-port resonator using coupling-of-modes theory, and then fabricated the resonator from ST-cut quartz, aluminum gratings, and a polyimide protective layer [17].  The resonator was converted into an ammonia sensor by depositing a polyaniline/WO3 composite onto the area between the intput/output transducers.  SAW-based sensors have also been successfully converted into UV detectors.  Wang et al. deposited UV-sensitive zinc oxide (ZnO) nanorods onto a 128° YX-LiNbO3-based two-port SAW oscillator, making use of the acoustoelectric effect to obtain a maximum frequency shift of 40 kHz when exposed to a UV source of 365 nm [18].  The three aforementioned works were published recently (2009), and there are other examples in the literature.  Polymer-based SAW devices have been investigated by Preethichandra and Kaneto, who applied the devices to bending curvature detection [19].  The researchers fabricated the sensors by depositing gold electrodes onto a Polyvinylidene Fluoride (PVDF) substrate (Sigma Aldrich P-0807) using thermal vapour deposition.  The dimensions of the SAW sensor are not given in the article.  A phase angle difference of 1.1 to 2.6 degrees was obtained when the bent sensor’s radius of curvature was varied from 2 to 6 cm.  The output amplitude decreased by approximately 50% as the radius of curvature increased from 2 to 6 cm.  The input interdigital transducer of the sensor was intentionally not  16 excited at its resonance frequency, and was excited at 4 kHz instead.  The reasons for doing so were not explained.  Related all-polymer devices include a PVDF-based thin film speaker, and PVDF-based bimorph transducers.  Schmidt et al. fabricated bimorph transducers from two 30 to 34 µm thick piezoelectric PVDF sheets, and inkjet printed PEDOT:PSS electrodes onto the PVDF sheets [20].  The inkjet printing was performed using a commercial HP 5850 desktop printer and the minimum line width achieved was 150 µm.  A minimum of 2 to 3 layers was needed in order to obtain conductive electrodes, and the researchers typically printed 10 layers.  The bimorph transducer fabricated by Lee et al. consists of two 45 µm piezoelectric PVDF films held together with a 40 µm thick epoxy adhesive layer, and screen-printed PEDOT:PSS electrodes [21].  The tip displacement of the transducer at resonance (27 Hz) was 7.0 mm, and the transducer was operated successfully at 1 MHz. Lee et al. also fabricated a PVDF-based thin film speaker with screen-printed PEDOT:PSS electrodes [22].   17 2. Sensor materials 2.1 Selection of sensor device materials To meet the objective of cost-efficiency, polymeric materials were selected for use in fabricating the SAW sensor.  Polymer-based substrate materials have an added advantage in that sheets of polymer substrate can be easily tailored to suit arbitrary sizes of sensor arrays.  An electroactive piezoelectric polymer, PVDF, was selected for the sensor substrate.  The air-stable, highly electrically-conductive doped polymer PEDOT:PSS was selected for the interdigital electrodes.  PEDOT:PSS is readily available commercially as an aqueous dispersion which can be solution processed in an inexpensive manner. Various types of electrically insulating, elastomeric polymers such as polyethylene-co- vinyl-acetate (PEVA) and polyisobutylene (PIB) were selected for the sensing layers. Due to the cross-selectivity of the polymeric sensing layer materials, an array-based approach to sensing is needed.  The SAW sensors in the array will each feature a different sensing layer, a methodology for selecting the sensing layer will be discussed in Section 2.3.  The thermoplastic fluoropolymer, PVDF or PVF2, possesses large piezoelectric coefficients, approximately ten times larger than that of any other synthetic polymer, and today accounts for nearly all of the commercially significant piezoelectric polymer applications [23].  PVDF was first discovered to be piezoelectric in 1969, after films of the polymer were stretched several times their original length at temperatures ranging between 100-150 °C, and static electric fields of approximately 300 kV/cm were applied along their thicknesses [24].  In addition to its piezoelectric properties, PVDF is also  18 pyroelectric, and is known as a ferroelectric material, although only its piezoelectric properties are of interest in this work.  The Curie temperature of PVDF is 80 °C, above which its piezoelectric properties will disappear [23].  PVDF has many useful properties, such as good chemical resistance and mechanical strength, and thus has been explored in applications such as pyroelectric sensors, pressure sensors, audio-frequency transducers, sonar hydrophones, and ultrasonic transducers [23,25].  Sonar applications are common since the acoustic impedance of PVDF is close to that of water [26].  Another advantage of PVDF is that the polymer is easy to process.  PVDF can be melt-processed without the need for processing aids, stabilizers, or additives, or it can be solution-processed in various common polar solvents such as N-methyl-2-pyrrolidone (NMP), dimethylformamide (DMF), dimethyl sulfoxide (DMSO), etc [27].  Aside from the homopolymer PVDF, copolymers of PVDF such as polyvinylidene fluoride-trifluoroethylene (PVDF-TrFE) and polyvinylidene fluoride-tetrafluoroethylene (PVDF-TFE), have been shown to possess strong and stable piezoelectric, pyroelectric, and ferroelectric properties [23].    PVDF-TrFE possesses several advantages over PVDF and PVDF-TFE, including higher Curie temperature (90-100 °C for PVDF-TrFE vs. 80 °C for PVDF), and increased mechanical strength and chemical resistance.  Attempts to source PVDF-TrFE revealed that the material is less common than PVDF and is substantially more expensive [28].  The choice of whether to use the homopolymer PVDF or the copolymers PVDF-TrFE/PVDF-TFE was heavily influenced by cost and availability, and to a lesser degree by the piezoelectric dij coefficients reported in the  19 literature, which are a measure of the electromechanical energy conversion potential of the material as described in Section 1.4.2.  Polymer-based organic conductors were first discovered in 1977 [29], and significant commercialization of intrinsically/inherently conductive polymers (ICPs) began in the 1990’s.  PEDOT:PSS is one of the most well-known and widely used ICPs, due to its high electrical conductivity and stability in air.  Other ICPs include the polyanilines, polypyrroles, polythiophenes, polyphenylenes, and poly(p-phenylene vinylene)s, and of these the polythiophenes and poly(p-phenylene vinylene)s have been demonstrated to be the most environmentally stable [30].  PEDOT:PSS is readily available from H.C. Starck (Clevios) and Afga (Orgacon).  According to H.C. Starck, the electrical conductivity of films obtained from Clevios formulations ranges between 10-2 to 103 S/cm, which compares favourably to in-situ polymerized PEDOT with conductivities ranging between 1 to 104 S/cm, and germanium with 10-1 to 102 S/cm [31].  2.2 Influence of material properties on SAW sensor performance The materials selected for fabricating the sensor will heavily influence its performance. Typical performance criteria for a sensor include sensitivity, selectivity, reversibility, response time, dynamic range, and stability to varying environmental parameters such as temperature [7].  The focus of this work was on the SAW device itself, without the chemical sensing layer, and therefore several of the factors described above will not be further discussed in great detail.  Sensitivity can be defined at two different levels, the intrinsic mass sensitivity of the SAW device, and the overall sensitivity of the SAW  20 sensor system which is a function of the device, coating, and in some cases the supporting hardware [7].  Of the two, the intrinsic mass sensitivity will be important for this work.  Selectivity and reversibility are chiefly determined by the properties of the sensing material.   Response time depends on sensing layer properties (thickness, sorption/reaction kinetics, etc.) and the properties of the overall system such as sample volume and characteristics of the gas delivery system [7].  The response time of the SAW device itself is neglibly short.  A sensor’s dynamic range is the concentration interval over which a clear relationship exists between concentration and response, and is constrained by limit of detection at the low end, and saturation effects at the high end [7]. Limit of detection is determined partially by the inherent sensitivity of the SAW device, while the saturation limit is dictated by the sorptive/reactive capacity of the sensing layer. Finally, environmental effects such as the chemical resistance of the SAW device materials, and the effects of temperature on the stability of the SAW device will be discussed.  2.2.1 Device sensitivity Influence of SAW propagation velocity on operating frequency As mentioned in Section 1.4.3, it is desirable to have the highest possible synchronous frequency for the best SAW device sensitivity.  The synchronous frequency depends on the transducer periodicity, which is designed and is limited by the fabrication technology, and the SAW propagation velocity, which is intrinsic to the substrate material and places an upper limit on the operating frequency.  The SAW propagation velocity  21 )(0 vfVV T=  (2.1) can be approximated using the shear wave velocity )1(2 1 v EVT + = ρ  (2.2) where E, v and ρ are the Young’s modulus, Poisson’s ratio and bulk density [32].  The equation v v vf + + = 1 12.187.0)(  (2.3) is an approximate solution to the Rayleigh equation for Poisson ratio values between 0 to 0.5.  Typical material properties of PVDF are listed in Table 2.  Table 2 - Typical material properties of PVDF [33] Quantity Value Young’s modulus (E) 1200 – 1600 MPa Poisson’s ratio (v) 0.383 Density (ρ) 1750 kg/m3  Using an average Young’s modulus, the Rayleigh velocity was approximated to be V0 = 505 m/s.  This velocity is approximately 15% of the Rayleigh velocities in ceramic piezoelectric materials, as shown in Table 3.  Table 3 - Typical Rayleigh velocities for substrate materials traditionally used for SAW devices [9]  Orientation Rayleigh velocity Lithium niobate (LiNbO3) Y, Z 3488 m/s Lithium tantalate (LiTaO3) Z, Y 3329 m/s Y, X 3159 m/s Quartz (SiO2) ST, X 3158 m/s  The slow SAW propagation velocity in PVDF is a challenge for achieving high device sensitivity and resolution of small mass loading changes.  If the transducer periodicity,  22 and hence the SAW wavelength, is assumed to be 250 µm, then Equation 1.3 yields an operating frequency of approximately 2 MHz.  Conventional SAW devices typically operate in the 10s to 100s of MHz.  Taking a mass resolution calculation example from Ballantine et al., a 100 MHz SAW oscillator on ST-cut quartz with an intrinsic mass sensitivity cm = 1.29 X 10-6 cm2-s/g will have a sensitivity 20fcS m−=  = -13 Hz-cm2/ng [7].  With a typical SAW oscillator stability of 1 Hz and a standard signal-to-noise ratio of 3, the limit of mass resolution S fRm ∆= 3  = 3 Hz / (13 Hz-cm2/ng) = 0.23 ng/cm2 [7].  Since PVDF-based SAW devices are expected to operate at substantially lower operating frequencies than conventional SAW devices (i.e. 2 MHz for PVDF-based SAW vs. 100 MHz for quartz-based SAW), their sensitivities appear to be vastly inferior to that of conventional SAW devices (i.e. 502 or 2500 times less sensitive).  However, the intrinsic mass sensitivity cm is also proportional to the normalized surface particle velocities (Equation 1.4).  For a given wave energy, the surface particle velocities of PVDF are expected to be higher than that of other substrate materials such as quartz and lithium niobate since the polymer is considerably less stiff.  Therefore the sensitivity of a PVDF- based SAW device is not as poor as it seems.  A reasonably detailed search for PVDF surface particle velocities did not yield any results.  If obtained, the intrinsic mass sensitivity and limit of mass resolution can be calculated.      23 Influence of electrode RC charging on operating frequency RC charging time is one of the rate limiting mechanisms that could potentially limit the operating frequency of the SAW device.  The electrical conductivity of the interdigital transducers will influence the RC constant.  To obtain an approximate RC time constant, the interdigital transducer fingers were treated as parallel stripline transmission lines [34]. RC charging calculations, which are based on measurements of the electrode finger electrical conductivity, are presented in Section 3.2.7.  Frequency dependence of substrate material properties and the effects on operating frequency The material properties of PVDF, such as its elastic, dielectric, and piezoelectric properties are frequency-dependant [35].  Dielectric permittivity is the ability of the material to polarise in response to an electric field, and the relative dielectric permittivity is the ratio of the material’s dielectric permittivity to the dielectric permittivity of the vacuum.  Table 4 shows how one of the relative dielectric permittivity constants of PVDF decreases with frequency.  Although the frequency dependence of PVDF material properties was not investigated, it is expected to play a role in determining the maximum operating frequency of the SAW device.  Table 4 – Relative dielectric permittivity of PVDF measured at different frequencies [33] ε33/ε0 Frequency Measurement method 8.4 – 13.5 60 Hz ASTM D150 7.4 – 13.2 1 kHz ASTM D150 6.0 – 7.6 1 MHz ASTM D150   24 2.2.2 Stability Temperature Effects The usable temperature range of the SAW sensor will be governed by the temperature sensitivity and temperature-dependant aging/degradation of the PVDF substrate, PEDOT:PSS electrodes, and polymeric sensing layers.  The effects of temperature on the two main materials used to fabricate the SAW sensor – PVDF and PEDOT:PSS – will be discussed in this section.  PEDOT:PSS films are highly stable, and can be heated in air at 100°C for over 1000 hours with minimal changes to electrical conductivity [36].  After 48 hours at 150 °C and in the presence of air and moisture, irreversible structural changes in PEDOT films can occur [37].  Continuous degradation starts at 150 °C and occurs until major decomposition takes place at 390-450 °C [38].  In the absence of moisture and in an inert nitrogen atmosphere, researchers have noted that the electrical conductivity of PEDOT:PSS films increased after being heated, with increasing conductivity as the treatment temperature was increased from 100 to 250 °C [37].  However, in the same study, an optimum heating duration was observed, after which the beneficial effects of increased electrical conductivity saturated and then decreased.  PVDF is sensitive to temperature changes.  Researchers investigated the response of a PVDF film-based pressure sensor from -25 to 65 °C, and found the output signal to be a strong function of temperature [39].  Temperature changes cause changes in the SAW propagation velocity, due to changes in the density of the substrate [7].  Like SAW  25 sensors fabricated from traditional piezoceramic materials, a PVDF-based SAW sensor will require some form of temperature compensation.  As for temperature-dependant degradation, a report on the isothermal aging of PVDF has shown that below 60 °C, no decay of the piezoelectric properties of PVDF takes place [40].  Between 60 to 160 °C, piezoelectric decay is linearly dependant on temperature, and is due to the decay of poling-induced polarization.  Between the PVDF substrate and the PEDOT:PSS electrodes, the temperature constraint on the PVDF will likely limit the usable temperature range of the SAW sensor.  Humidity PEDOT:PSS is known as a hygroscopic material, and as the PSS content increases, so does the expected water uptake [37].  The polymer has been used to fabricate humidity sensors which are effective up to 80% relative humidity (rH) [41].  The resistivity of the PEDOT:PSS film increases linearly from approximately 13 to 42 Ω-cm as the humidity increases from approximately 44 to 76% rH.  As for mechanical properties, at 23% rH the material can be regarded as brittle while at 40% and 50% rH it becomes more ductile [42].  The Young’s modulus at 23% rH is 2.8 ± 0.5 GPa, at 40% rH is 1.9 ± 0.02 GPa, and at 55% rH is 0.9 ± 0.2 GPa.  From Table 2, the average Young’s modulus of PVDF is 1.4 GPa, so depending on the relative humidity the PEDOT:PSS electrodes may become stiffer than the substrate and constrain it slightly.  Humidity control may be necessary. Higher relative humidities will lead to more pliable electrodes, but lower electrode conductivity and longer RC charging times which may limit the SAW device operating  26 frequency.  Lower relative humidity will result in stiffer electrodes which may constrain the substrate and limit the operating frequency, but possess higher electrical conductivity.  PVDF has low cross-sensitivity to humidity.  Researchers measured the output signal of a PVDF film-based pressure sensor at 20, 48, and 75% relative humidity, at a constant temperature, and found the signal deviated by less than 10% [39].  The water absorption of PVDF ranges from 0.03 to 0.06% [33].  Chemical resistance PEDOT:PSS is one of the most chemically resistant and environmentally stable intrinsic conductive polymers, and is stable in an oxygen-containing atmosphere even at 100 °C [36].  H.C. Starck, the manufacturer of a commercially-available aqueous dispersion of PEDOT:PSS, states that films formed from their product is resistant against common organic solvents such as toluene, acetone, MEK, ethanol, ethylacetate, DMSO, and DMF, as well as basic solutions such as 10% sodium hydroxide [43].  Researchers have tested the chemical resistance of PVDF by immersing the film into various chemicals such as 98% ethylacetate, 99% tetrahydrofuran, 99% chloroform, 99.9% acetonitrile, and 95-97% sulphuric acid for one week, then measuring the film’s output voltage [39].  While the output signal of the PVDF film was reduced slightly - to varying degrees for different chemicals - it was not significantly affected.  PVDF is not compatible with various polar organic solvents such as DMSO and DMF, which were used in this work to create PVDF solutions for spin coating.  To verify the resistance of  27 PVDF to a given compound, chemical resistance charts offered by commercial manufacturers of PVDF such as Arkema can be used [44].  2.3 Selection methodology for sensing layers 2.3.1 Overview of solubility-prediction systems As the sensing layer materials are polymers and are cross-selective, a sensing array scheme will be employed to differentiate between individual VOC analytes, or classes of VOCs at the minimum.  Provided at least one of the sensing layers has some affinity for the analyte, if the sensing materials are sufficiently dissimilar in terms of their response to individual analytes, then a unique signature can be generated for each analyte.  The response of a polymeric sensing layer can include mass increase due to analyte absorption, density changes due to swelling, and/or changes in dielectric permittivity, but the main response which will be exploited is absorption-induced mass change since the SAW device is being used as a mass-loading platform.  To select an optimal set of sensing layers which will enable the sensor array to distinguish the VOC analytes of interest, a selection methodology for the sensing layers is needed.  A polymer absorbs an analyte when its intermolecular self-interaction is overcome by the intermolecular interaction between the polymer and the analyte.  When a sufficiently large amount of analyte is present, the polymer is essentially dissolved.  As a result, the affinity of a polymeric sensing material for a VOC can be predicted using solubility parameters, which offer a quantitative measure of solvent behaviour and are related to the attractive strength amongst the molecules in a material [45].  The application of solubility  28 parameters to the process of selecting appropriate polymeric sensing layers has previously been discussed by Eastman et al. [45], Patel et al. [46], Loui et al. [47], Grate and Abraham [48], and others.  The term solubility parameter was first used by Hildebrand and Scott, and builds on earlier work by Scatchard and others [49].  The Hildebrand solubility parameter V E =δ (2.4) is defined as the square root of the cohesive energy density, where E is the measurable energy of vaporization and V is the molar volume of the pure solvent [49].  Cohesive energy density is the energy needed to completely remove a unit volume of molecules from their neighbours to infinite separation (ideal gas).  The Hildebrand parameter was originally devised to describe the free energy of mixing of non-polar and non-associating fluids, but the concept has been extended to solvents and polymers [45].  Two fluids with similar δ values are expected to mix well, and a polymer with a similar δ value to an analyte is expected to have a strong affinity for and show a strong response to the analyte (i.e. volume increase due to swelling).  Aside from the Hildebrand parameter and various other single parameter solubility- prediction systems such as the partition coefficient, several multi-parameter solubility parameter systems such as the Hansen solubility parameters (HSP) and the Linear solvation energy relationship (LSER) exist.  Multi-parameter solubility-prediction systems are used to overcome the shortcomings to using a single parameter system such as the Hildebrand solubility parameter.  Vapours with substantial dissimilarities, such as toluene (polarisable, non-hydrogen-bonding) and ethyl acetate (dipolar, hydrogen-bond,  29 basic), have similar Hildebrand parameters, so it is customary to break down the Hildebrand parameter into components which quantify the different interactions [48]. The interactions include dispersion (induced-dipole/induced-dipole), dipole induction (dipole/induced-dipole), dipole orientation (dipole/dipole), and hydrogen-bonding interactions.  The term “Van der Waals” interactions are not used because its use is ambiguous; some authors use it to describe dispersion interactions only while others refer to dispersion, induction, and orientation interactions [48].  The most extensive multi- parameter solubility prediction system was described by Karger et al., and is based on five parameters: dispersion, induction, orientation, proton donor (Lewis acid), and proton acceptor (Lewis base) [50].  However, the Karger solubility parameters have not received much attention in polymer-liquid systems [51].  The HSP, while not the most comprehensive multi-parameter system, is one of the most popular because it can accurately describe most compounds with three parameters, and HSP values are generally available for most solvents and polymers.  Consequently the HSP system was chosen as the basis for selecting the polymeric sensing layers.  The HSP system consists of three parameters: dispersion δD, dipole orientation δP, and hydrogen bonding δH.  The square of the sum of the HSP is equal to the square of the Hildebrand parameter 2222 HPD δδδδ ++=  (2.5) Equation 2.5 is based on the fact that the sum of the energies due to the three interactions is equal to the total cohesion energy HPD EEEE ++=  (2.6)  30 Dividing Equation 2.6 by the molar volume V gives Equation 2.5, based on Equation 2.4. A convenient list of HSP values for solvents and polymers can be found in a book by Hansen [49].  2.3.2 Application of the HSP to select polymeric sensing layers To determine the affinity of compound 1 (polymer or analyte) to compound 2 using the HSP, the solubility parameter distance  2 12 2 12 2 12 2 )()()(4)( HHPPDDRa δδδδδδ −+−+−=  (2.7)  can be calculated [49].  From experimental data, the constant of 4 in Equation 2.7 was found convenient and enabled the representation of the solubility data as a sphere encompassing the matching solvents [49].  Intuitively, the dispersion (induced dipole/induced dipole) interactions for most compounds are quite weak, which may explain why the constant of 4 is needed to make the solubility region spherical.  Once the Ra value has been calculated, the affinity of a polymer to an analyte can be calculated using the relative energy difference Ro RaRED = (2.8) where Ro is interaction radius for the polymer.  The interaction radius, when plotted, defines the solubility region/sphere where all of the analytes which the polymer has affinity for are within the sphere.  RED values of zero to less than 1.0 indicate high affinity.   31 With the RED values it is possible to calculate the relative expected responses of a sensor array consisting of sensors coated with different polymeric sensing layers, to a given analyte.  Although this work focused primarily on the design and fabrication of the SAW sensing platform, the selection of optimal sensing layers is essential to the creation of sensors and will require further investigation.     32 3. Sensor design and fabrication 3.1 Substrate 3.1.1 The piezoelectric phase of PVDF PVDF is semi-crystalline, with typically 50-60% of the material in crystalline form depending on what processing techniques were used [23].  The unit cell of PVDF is [CH2CF2]n.  It is a polymorphic material, and four phases have been identified: α, β, γ, and δ phases (also referred to as phases II, I, III, and IV in the order they were discovered) [27].  The phase of interest is the polar β phase (Figure 1), with a polar unit cell due to the aligned fluorine atoms and the polymer chains arranged in a distorted, planar zigzag, all-trans conformation.  There are multiple techniques to obtain a higher percentage of the β phase in the material: drawing/stretching, high voltage poling, or a combination of the two in conjunction with heating.  Maximum piezoelectric constants were obtained by simultaneously stretching PVDF films by 4.5 times their original length, heating the films at 80 °C, and high voltage poling them at 0.55 MV/cm [52]. After poling, the room-temperature stability of the polarization is excellent, however the polarization (and hence the piezoelectricity) degrades with increasing temperature [23]. At or above the Curie temperature of 80 °C, the poling-induced piezoelectric activity of PVDF will decrease [26].    33  Figure 1 - Three-dimensional view of a segment of β phase PVDF  3.1.2 Film thickness requirements  The primary requirement is that the PVDF film should possess a thickness of at least once the SAW wavelength.  The reason for this is the strain energy density generated by a propagating Rayleigh wave is contained within one SAW wavelength of the surface, which acts like a waveguide [7].  A film of insufficient thickness would act more as a medium for flexural plate wave propagation.  Assuming that the droplets ejected from a 30 µm orifice nozzle would spread to approximately twice the diameter of the orifice upon contact with the substrate, the minimum electrode track width would be 60 µm. With a track width and inter-electrode spacing of 60 µm, the SAW wavelength would be 240 µm, and the film thickness should be a minimum of 240 µm.  3.1.3 Formation of the piezoelectric substrate Several methods to form the PVDF substrate were considered and attempted, such as inkjet microprinting, spin coating, solvent-casting into a mould, and melt moulding.     34 3.1.3.1 Inkjet microprinting PVDF is soluble in polar aprotic solvents such as DMF and DMSO [53].  As a result, inkjet microprinting was considered for fabricating the PVDF thin film substrate. Investigators such as Zhang et al. have demonstrated the printing of PVDF-TrFE microdots on silicon substrates, and controlled the diameter and uniformity of the dots by modifying the substrate surface chemistry and ink formulation [54].  Typical concentrations of PVDF-based inks are around 1.0 wt%.  There are two primary challenges with fabricating the PVDF substrate using inkjet microprinting.  The first is the necessity to optimize the deposition process for a sufficiently thick film of uniform thickness.  The second is the amount of time needed to fabricate the substrate via multi-layer printing, due to the low polymer weight-loading in the ink.  Consequently, other approaches to substrate fabrication were considered.  3.1.3.2 Spin coating Spin coating is a common microfabrication technique used to apply thin films to flat substrates, and is suitable for this application because a uniform film is desired.  Work on this fabrication method in our laboratory resulted in successful fabrication of thin (approximately 20-30 µm thick) PVDF films, and stretching, heating, and high-voltage poling of the films [28].  The experimental procedures described in Appendix A build on our previous work, and features additional optimizations.   35 Well-formed translucent white PVDF films were obtained via spin coating on glass slides.  However, it was challenging to obtain uniform films thicker than 30 µm, despite using high wt% (i.e. 30 wt%, 40 wt% solutions did not dissolve completely) PVDF solutions.  The 30 wt% solutions were a solid gel at room temperature and needed to be pre-heated prior to spin coating.  Decreasing the spin speed led to the formation of non- uniform films that were at most 40 µm thick.  A multi-layer spin coating process was attempted to build up a thick film from sequential spin coating of thin films.  The first layer was spun successfully, but the next layer caused the first layer to de-adhere from the glass slide.  3.1.3.3 Solvent-casting Sufficiently thick films were difficult to obtain using spin coating.  A glass mould, consisting of a 4” glass wafer, and walls formed from 3” x 1” microscope glass slides 1.5 mm high, was constructed for solvent-casting.  The walls were affixed to the glass wafer using epoxy, and the volume of the mould was equivalent to the volume of a 3” x 2” glass slide of the same height.  For solvent-casting, solutions of 30 wt% PVDF in DMF/DMSO were prepared using the same procedures as the solutions prepared for spin coating.  The fabrication procedure consisted of pouring the solution into the mould, then sweeping the top of the mould with a plastic squeegee to scrape off excess solution.  The next step was to form the PVDF film by allowing the volatile solvent to evaporate, with or without heating on a hotplate.  Observations of the dried films indicated that the rate of  36 solvent evaporation had a strong influence on the quality of the film.  Solvent casting could be used to produce thicker films than spin coating, but it was challenging to obtain well-formed and uniformly thick films.  3.1.3.4 Melt moulding With a melting point of slightly over 155 °C [33], PVDF is easily thermally moulded. Glass moulds were constructed, and both PVDF pellets (Sigma-Aldrich 427152) and powder (Sigma-Aldrich 182702) were melted in a convection oven.  The melt viscosity of PVDF is quite high, and despite compressing the melt, the melted PVDF did not flow sufficiently to fill in the mould.  3.1.4 Processing of commercially-available thick film  Acquisition of PVDF Commercially available film was chosen as a more promising alternative to custom- formed film.  Although not essential, the purchased film would ideally be pre-stretched and poled, to facilitate faster prototyping of the SAW sensor.  Researchers experimenting with PVDF have sourced their films from a variety of commercial sources.  Kumar and Periman used 100 µm thick PVDF-TrFE films from Solvay S.A. (Brussels, Belgium) [55].  Lee et al. obtained 45 µm thick β-phase PVDF from Kureha (Iwaki City, Japan) [56].  Oreski and Wallner used 37 µm thick PVDF (with 10-20% PMMA copolymer fraction) films from Isovolta AG (Werndorf, Austria) [57].  Dargaville et al. obtained 76 µm thick PVDF films from Westlake Plastics (Lenni, PA), and 55 µm thick PVDF films from Terphane Inc. (Bloomfield, NY) [58].  In another report, Dargaville et al. list  37 various commercial and non-commercial sources for PVDF-based polymers [27]. Piezoelectric PVDF films can be obtained from MSI USA (Littleton, MA) and Ktech Corp (Albuquerque, NM), but the thickest films are 110 µm.  Shirinov and Schomburg used 25 µm thick bi-axially stretched PVDF film from Piezotech SA (Hesingue, France) and 28 µm thick mono-axially stretched PVDF film from Measurement Specialties (Hampton, VA) [39].  Preethichandra and Kaneto fabricated PVDF-based SAW sensors using 135 µm thick PVDF films (P-0807) purchased from Sigma-Aldrich [19].  The Sigma-Aldrich films were used without further preparation and seemed to be piezoelectric already.  Thick films 500 µm or thicker with or without pre-poling were challenging to procure. Goodfellow, a specialist supplier of small to medium quantities of research materials, did sell PVDF films in thicknesses of up to 2.0 mm.  However, the thickest uniaxially/biaxially oriented, piezoelectric PVDF film sold was 110 µm.  According to Goodfellow, films of 0.5 – 2.0 mm were very challenging to orient and pole [59]. McMaster-Carr, a frequently-used supplier for sourcing research and industrial grade parts, also stocked PVDF film ranging in thicknesses from 0.003 – 0.02” (75 – 500 µm). The 500 µm film (part 8675K25) is intended for industrial applications.  It was relatively inexpensive at $13/ft2, and was therefore chosen for this work.  The film is extruded from Solvay Solef or Arkema Kynar PVDF resins and is not piezoelectric as shipped since the PVDF is primarily in the α-phase, so an apparatus used to orient, heat, and pole the film was needed.   38 Apparatus to orient, heat, and pole the PVDF film Many of the reported poling tools were designed and built in-house.  Kaura et al. used a stage driven by a geared motor to stretch the 50 and 100 µm thick films, a hotplate located underneath the mid-section of the film for heating, and a corona needle connected to a high voltage supply for poling [52].  Huan et al. built an apparatus consisting of rollers rotating at different speeds to stretch the 110 µm thick film, and metallic plates to heat and contact pole the film [60].  Another investigator in our lab constructed a heating, stretching, and poling apparatus, and was successful in heating and stretching 20 µm thick PVDF films which were obtained from spin coating [28].  The apparatus consisted of a modified drill press vice (Mastercraft #55-5906-6) with 4” travel, for clamping and manually stretching the film, and two metallic plates for contact heating and high voltage poling of the film.  The ground plate featured a resistive strip heater and a thermistor, and the other plate was used simply for applying the high voltage.  Using the apparatus, thin films could be heated and then uniaxially stretched, but as the film thickness increased (i.e. 40 µm), the films slipped out from the clamps on the vice before they could be stretched.  An additional challenge was the difficulty in maintaining good thermal contact between the film and the heated ground electrode as the film was being stretched, and the mid-sections of the films often became brittle and fractured during stretching despite low draw rates.  Convection heating has advantages over contact heating since it provides more uniform heating.  A generic 1200 W convection toaster oven (Oster model #6058-033) was acquired and modified.  The modifications included:  39 • removing a rectangular section in the back of the oven so an electrically insulating plate could be mounted to feed the high voltage into the oven via a banana jack, • drilling a hole out of the left side of the oven so the handle of the drill press vice could protrude out and be manipulated during heating, • connecting the chassis of the oven to the ground wire of a standard 3-prong mains plug, so the chassis can be safely earth grounded.  The electrically-insulating plate mounted in the back of the oven is made of G-7 Garolite, a woven-glass-fabric laminate with silicone resin binder which offers excellent heat and arc resistance.  The temperature regulation of the oven was qualitatively checked by placing a thermometer inside the oven.  In convection heating mode, the oven takes approximately 2 minutes to reach 175 °F/80 °C from room temperature and regulates within ± 3 °C.  To heat and stretch a PVDF film, the vice was placed on a rack inside the oven as shown in Figure 2, and once the film (and vice) had reached the target temperature, the handle of the vice would be turned slowly, thus uniaxially stretching the film.  Initial tests with 20 µm PVDF films were successful, and the films were stretched to at least 300% of their original length.  The draw rate during stretching needs to be quite low or the film may break.   40  Figure 2 - Convection toaster oven with drill press vice on a cooking rack, supported by an aluminum block underneath  The next step was to heat and stretch the 500 µm PVDF film purchased from McMaster- Carr.  PVDF is normally a very stiff polymer.  Its tensile strength at yield, at room temperature, ranges from 20-40 MPa, and its Young’s modulus ranges from 1.2-1.6 GPa [33].  At temperatures of 80-100 °C, it softens somewhat, but is still quite stiff.  Despite heating the film for at least 15 minutes after the target temperature had been reached, sufficient stress could not be achieved using the vice and clamps to stretch the stiff film. The smooth PVDF film would slip out from the clamps, so a couple of modifications were made to the clamps.  Firstly, two rubber layers were introduced between the bottom of the clamp and the aluminum top plate underneath the wing-nuts, to offer increased grip on the PVDF.  The film still slipped.  Secondly, the top plates on the clamps were Aluminum supporting block Drill press vice with 4” travel Wing-nuts to hold down the clamp plate  41 swapped out for plates which had been bent into a V-shape.  Once the wing-nuts were tightened, the V-shaped plates would pinch the PVDF.  Surface indentations on the film away from the middle were not of concern since only the middle section was poled.  The V-shaped plates were also not effective at preventing slippage of the film.  Various other options were considered, including introducing staggered, punched-out slots on the two sides of the film for gripping the film.  However, this avenue was not pursued since there were concerns over how uniform the stretching would be across the width of the film.  Due to the challenges imposed by stretching thick PVDF film, the PVDF film was prepared using just heating and poling as an intermediate step, followed by characterization of the piezoelectric coefficients to determine if the film is sufficiently piezoelectric for the end application.  Development of a process for heating and poling PVDF films The initial process development for heating and poling revolved around the optimization of the following parameters: • heating and poling duration, • poling field strength, • heating temperature.  The starting values for the parameters, shown in Table 5, were selected to be close to the optimal reported values [52].  For the poling field strength, it is assumed that the film will be in good contact with the ground and high voltage electrodes and there are negligible  42 air gaps.  Once the heating and poling period had elapsed, the oven door would be opened and the film would be allowed to cool for 10 minutes, with the poling field still applied. Preliminary data indicated that the cooling down process (temperature ramp-down profile, duration of applied field while cooling) does have a noticeable effect on the piezoelectricity of the film.  The optimization of the cool-down process will be an important and future step to improve sensor performance.  Table 5 - Starting values for PVDF heating and poling process parameters Parameter Value(s) Heating and poling duration 30, 60, 90 minutes Poling field strength 30, 40, 50, 60 kV/mm Heating temperature 80, 100°C   The three main challenges during process development were: 1. film breakdown due to electrode geometry, 2. dipping of the high voltage output of the in-house built flyback converter supply, 3. varying electrical properties of the PVDF film from McMaster-Carr.  One of the process development challenges was creating an electrode setup that could be used to reliably heat and pole PVDF film.  Even though the edges of the high voltage electrode were buffed to remove sharp edges, breakdown of the film still occurred.  The first electrode set up is shown in Figure 3a.  A polished, square steel plate served as the ground plate, and was in electrical contact with earth ground via the cylindrical aluminum stand and the chassis of the oven.  The high voltage electrode was a square copper plate, with a flap for the high voltage wire to connect to.  A neodymium-iron-boron disk magnet was used to hold the copper high voltage electrode, PVDF film, and steel ground  43 electrode together.  The challenges with this setup were dielectric breakdown due to electric field enhancement at the edges of the electrode, and warping of the film during heating since the outer region of the PVDF was unsupported.  The warping tended to lift the high voltage electrode and increase the air gap, therefore reducing the effective poling field strength across the film.  (a) (b) Figure 3 – (a) First electrode setup (b) second electrode setup   The second electrode setup used is shown in Figure 3b.  The square ground electrode plate was substituted with a circular steel plate with the same diameter as the PVDF wafer, and a larger high voltage electrode plate was used to maximize the active poled region.  Four cylindrical magnets were used to hold the electrode down, and ensured good electrical and thermal contact between the plates and the film.  The challenges with this setup were nonuniform heating of the film, resulting in nonuniform thermal Copper high voltage electrode Neodymium- iron-boron disk magnet Cylindrical aluminum stand Warped PVDF film Steel ground electrode plate  44 expansion and warping as shown in Figure 3b, and breakdown due to field enhancement despite extensive buffing to obtain smooth rounded edges.  The third electrode setup is shown in Figure 4, and consisted of the same ground electrode plate as the previous setup, with a tall cylindrical disk with rounded edges serving as the high voltage electrode.  Several circular and cylindrical neodymium-iron- boron magnets were used to hold the high voltage electrode down firmly, but not so firmly as to compress the film during heating and increase the likelihood of breakdown. The magnets were also used to make the electrical connection.  This electrode setup was successfully used with poling voltages up to 30 kV (60 kV/mm, with 500 µm thick film), although at the higher voltages frequent arcing will occur between the top electrode and the outer edges of the ground plate.   Figure 4 - Third electrode setup Ground electrode plate High voltage electrode PVDF film Magnets  45 The second process development challenge was the inability of the in-house built high voltage flyback converter to source significant output current.  As the PVDF film is heated and the leakage current through the film increases, the high voltage would dip. This issue was resolved when a commercial 0-30 kV high voltage supply (Model ES30R/PRG/M754, Gamma Research, Ormond Beach, FL) - capable of providing up to 200 µA at 30 kV – was used.  Current leakage was reduced by the use of an electrically insulated high voltage connector wire, with insulation rated up to 40 kVDC and heat resistant up to 80°C.  The third, and most challenging to control, complication to developing a repeatable process for heating and poling PVDF films has been variations in the electrical properties of the PVDF films.  All of the films obtained from McMaster-Carr meet the specifications for thickness (stated: 508 µm, actual: 500-520 µm), but there are no dielectric breakdown specifications.  A simple test using two wafers cut from two different rolls, and performed with the same setup and parameters, showed one film could withstand 50 kV/mm, while the other film experienced dielectric breakdown at 35 kV/mm.  Using the naked eye it is difficult to distinguish between the two samples, but by holding them up to the light it was observed that the film which withstood a higher poling field strength is slightly more opaque than the other film.  Email correspondence with McMaster-Carr revealed that the PVDF films are extruded from either Arkema Kynar or Solvay Solef PVDF resins, which may explain the differences in electrical properties.  McMaster does not distinguish between films extruded from one resin or the other, so an alternate supplier of PVDF is necessary.  46 Experimental procedures Experimental procedures for heating and poling the PVDF film have been developed to achieve repeatable results.  The PVDF film is shipped from McMaster-Carr in a roll.  The preparation of the PVDF film for heating and poling involves: (1) tracing out wafer(s) on an unrolled sheet of PVDF, and cutting out the wafers using scissors, (2) cleaning the wafer with isopropanol followed by a DI water rinse; acetone should not be used because it is incompatible with PVDF [61], (3) blow drying the film with nitrogen to remove excess water from the surfaces; water beads off of the hydrophobic surface quite easily, (4) placing the wafer in between two flat surfaces and annealing at 60-80°C for at least 30 minutes to flatten the wafer; the surfaces would ideally be metallic for more uniform heating, and be covered with clean aluminum foil to minimize contamination of the PVDF wafer, (5) cleaning the wafer with an isopropanol and DI water rinse if necessary.  Once a relatively flat PVDF wafer has been obtained, the next stage is to set up the heating and poling apparatus.  The film is aligned with the circular ground plate, and the high voltage electrode with magnets is placed on top.  The entire assembly is placed on the cylindrical stand inside the oven, and the high voltage wire is connected from the high voltage supply to the banana jack feedthrough on the back of the oven.  Heating and poling is performed with the convection oven inside the fumehood, as shown in Figure 5. The minor corona discharge which occurs inside the oven as well as around the high  47 voltage supply produces ozone, which is unpleasant even in small amounts.  In addition, if the PVDF were to thermally degrade for any reason, dehydrofluorination occurs, releasing extremely dangerous hydrogen fluoride gas [33].  After the apparatus has been set up, the high voltage is set.  The commercial supply features a voltmeter panel which is quite accurate.  After the poling voltage is set, the desired heating temperature is set and the poling process is initiated.  After the process has completed, the centre region of the PVDF wafer covered by the top electrode will be piezoelectric, and the regions extending outwards from the centre are expected to be piezoelectric to a lesser degree (not verified).  The experimental procedures for integrating the heating and poling of the PVDF substrate with the micropatterning of the electrodes are similar to, but slightly different than the aforementioned procedures, and will be described in more detail in Section 3.3.1.          48  Figure 5 - Equipment for heating and poling PVDF films, with the convection oven and HV supply inside fumehood   3.1.5 Methods used to characterize the piezoelectric constants  PVDF has five independent piezoelectric constants d31, d32, d33, d24, and d15 [35].  The piezoelectric constants most relevant to the fabrication of a SAW sensor are d33, d31, and d32 because d33 describes the longitudinal effect and d31 and/or d32 describe the transverse effect.  Typical values for the strain constants of uniaxially stretched PVDF are given in Table 6.  The negative sign of d33 is due to convention and indicates that compression occurs instead of extension when a field is applied in the standard direction x3, and the direction of x3 is taken to be along the thickness of the PVDF film and along the same axis as the poling electric field.  If uniaxially stretched, PVDF is an anisotropic material Convection oven DC power supply to power high voltage flyback converter High voltage flyback converter  49 as demonstrated by the large differences in its d31 and d32 constants, so for efficient SAW propagation the waves should travel in the x1 direction.  Since the preparation process currently does not involve stretching, the material is expected to be fairly isotropic with a smaller magnitude of d31, and a smaller difference between d31 and d32.  The isotropy of the material was not investigated, and anisotropy in the material is possible due to the extrusion process used to manufacture the film.  Table 6 - Typical piezoelectric stain constant values for PVDF [33] Constant Value (pC/N or pm/V) d31 28 d32 4 d33 -35   The material properties of PVDF have been measured using different techniques, which has led to discrepancies in the reported piezoelectric properties [62].  One technique, exploiting the direct piezoelectric effect, involves vacuum depositing thin metallic electrodes onto the film, attaching a weight to the film and measuring the induced charge using an electrometer [52].  The converse piezoelectric effect has also been exploited, by applying an electric field between two sputtered or evaporated electrodes and measuring the strain generated via a strain gauge [27].  The piezoelectric coefficients d31 and d32 can be obtained by measuring the deformation of a bimorph [27].  The challenge with measurements based on the direct piezoelectric effect is that some of the polarization generated by the application/removal of stress can be masked by charge leakage before it is measured; as a result, the converse effect is generally more accurate for measuring the piezoelectric constants [63].   50 In spite of the variation in electrical properties of the PVDF film sourced from McMaster-Carr, poled PVDF films were successfully obtained.  For verifying whether the poled film was piezoelectric, PVDF-based resonators were fabricated and actuated in the audible frequency range.  For characterization, it was decided that characterizing d33 alone would be sufficient initially.  The magnitude of the d33 calculated from measurement in relation to the reported value would indicate the effectiveness of the poling process.  The three methods used to characterize the d33 of the prepared PVDF films were optical profiling, surface profiling, and measuring with a Laser Doppler vibrometer.  Actuation of a PVDF-based resonator at audible frequencies The resonator method was used to verify whether the film was piezoelectric or not.  To fabricate a resonator, electrodes were deposited on both sides of the film to excite the film in the thickness direction.  Electron beam evaporation was used to deposit 50 nm of gold on both sides of poled PVDF films, as shown in Figure 6.  The active piezoelectric region is overlapped by the top and bottom electrodes.  The top finger electrode was formed by masking with Scotch tape prior to evaporation.  The bottom electrode required minimal masking, and the only region that was masked was the portion directly underneath the top electrode and close to the edge of the film, to avoid shorting the two electrodes.          51  Figure 6  - PVDF-based resonator fabricated by depositing 50 nm of gold on both sides of the film   The resonator was connected to the output of a high voltage signal amplifier (Tabor Electronics Model 9400), which was connected to a function generator (HP 33120A). The output amplitude of the signal generator was set to 400 V peak-to-peak.  The PVDF resonator was successfully actuated in the audible frequency range, and the output frequency was verified to be the same as the excitation frequency via a simple condenser microphone circuit, measured using an oscilloscope.  Actuation of a resonator fabricated from unpoled PVDF film was not audible, indicating the poling was successful.  A PVDF resonator was also fabricated with PEDOT:PSS thin film electrodes and also successfully actuated in the audible frequency range.  The PEDOT:PSS films were drop- casted onto both sides of the PVDF film.  The active poled region was overlapped by both the top and bottom electrodes, but the electrodes extend to opposing sides of the film to avoid short-circuiting when connecting. Top and bottom electrode overlapped piezoelectric region Top electrode Bottom electrode  52 Optical Profiler Static displacement resulting from the converse piezoelectric effect, if measured accurately, yields good measurements for determining the d33 constant.  The challenge with measurements based on the direct piezoelectric effect is potential charge leakage, yielding inaccurate measurements [63].  An optical profiler (Veeco Wyko NT1100) was used to measure the static displacement.  Assuming optimal poling, with a reported d33 magnitude of 35 pC/N or pm/V and film thickness of 500 µm, an electrical potential of 100 V applied in the thickness direction would result in strain of 6107 −× , or static displacement of 3.5 nm.  The optical profiler features two imaging modes, Vertical Scanning Interferometry (VSI) and Phase Shift Interferometry (PSI).  VSI mode is suitable for larger vertical ranges but its main limitation is a vertical resolution of 3 nm [64].  PSI mode features a vertical resolution of 0.1 nm, but cannot be used with samples with height discontinuities greater than 150 nm.  With expected static displacements of a few nanometres, the PSI mode is more suitable.  To measure static displacement, a reference surface is needed.  Gold-metallized PVDF films, similar to the ones fabricated as resonators, were modified for the Wyko measurements.  To actuate the film, a standard DC supply, connected to a high voltage signal amplifier (Tabor Electronics Model 9400) was used.  Once the images were obtained, the static displacement would be calculated by taking the difference between the surface profiles of the film in its non-actuated and actuated states.  There were two major challenges associated with measuring the static displacement using an optical profiler: substrate drift and the profiler’s vertical range/resolution trade-off.  53 Substrate drift was noticed when focusing the profiler on the sample.  After the optical profiler had been focused, the PVDF film would drift, causing the interference fringes to drift on-screen.  Within a few seconds the profiler would be out of focus.  Substrate drift was not observed with hard substrates such as a glass slide.  Measures were taken to eliminate the drift, including clamping the metallized film in a holder (Figure 7) or onto a plate (Figure 8) prior to imaging, and leaving the sample for up to half a day to reach thermal equilibrium with its surroundings before it was imaged.    Care was taken not to constrain the actuated region being imaged while the film is clamped.  The measures helped to reduce the substrate drift rate significantly, so that two consecutive images of the film in its non-actuated state, taken one after one another and subtracted, yielded zero height change.  The substrate would drift out of focus on the order of a minute instead of tens of seconds.   The second major challenge to characterizing the static displacement using the optical profiler was the profiler’s vertical range/resolution trade-off.  Assuming optimal poling, applying a reasonably high voltage across the film such as 200 V would result in static displacement of less than 10 nanometres.  Initial attempts to image regions consisting of the metallized piezoelectric PVDF and electrically insulated, metallized PVDF, were not successful.  However, there were noticeable shifts in the interference fringes on-screen immediately after the active region was actuated, indicating that the film was piezoelectric.  Stylus-based profiler measurements revealed that the height discontinuities between the electrically active and insulated regions were several hundred nanometres high, making the PSI mode ineffective since it is incompatible with samples containing  54 height discontinuities of 150 nm or greater.  The VSI mode was tried, but was unable to detect any height changes due to its limited vertical resolution.   Figure 7 - Holder for PVDF samples    Figure 8 - PVDF film clamped onto steel plate using magnets     Measurement window Plate for electrically grounding and clamping the PVDF film Magnet for clamping PVDF film  55 Surface Profiler The second static displacement method used to characterize the d33 of the poled film was surface profiling using a stylus-based profiler (Veeco Dektak 150).  The step height between the metallized and unmetallized PVDF would be measured for two cases, the film in its actuated and non-actuated state, and the difference between the two step heights would be the static displacement.  To accurately measure the step height, the surfaces must be very smooth.  Measurements were taken to characterize the surface roughness of the film, and indicated that the surface roughness is quite substantial, as shown in Figure 9.  The measurement shown in Figure 9 is a scan crossing the interface between the metallized (left, red cursor) and unmetallized (right, green cursor) PVDF. The step is present, but with surface roughness heights ranging from 20-100 nm, it would be challenging to detect slightly less than 10 nm of static displacement.  The curvature in the sample is exaggerated due to the length scales of measurement, but is expected since the film originally came in a roll.                    56  Figure 9 - Screenshot of surface profiler software showing surface roughness measurement   Laser Doppler vibrometer Due to the challenges associated with the measurement of static displacement, a dynamic characterization method was employed.  A single-point Laser Doppler vibrometer (Polytec OFV-5000) was used to measure surface displacement as a function of frequency f.  The output of the LDV is an analog voltage which is proportional to the surface velocity v that allows determining the displacement f v x pi2 = . (3.1)  Gold- metallized region Un-metallized PVDF Average step height between cursors is 98.7 nm  57 Since the expected surface displacement will be on the order of a few nanometres, a velocity decoder (VD-02, range 5 mm/sV) instead of a displacement decoder was utilized since it has a better vertical resolution.  A function generator (HP 33120A) was connected to a high voltage signal amplifier (Tabor Electronics Model 9400), which drove the metallized PVDF film directly.  For clamping the sample, the holder shown in Figure 7 was used, which features a window for the measurement beam to pass through, and space for the alligator clips to connect to the film on the side.  The repeatability of the clamping was not investigated, but should be for future work.  Gold-metallized PVDF films were successfully characterized using the LDV, and the results are shown in Figures 10, 11, 12, and 13.  Figure 10 shows how a higher poling field strength (45 kV/mm vs. 30 kV/mm) yields a resonator which has increased surface displacement.  The optimal poling field strength was reported to be 55 kV/mm [52], so the observed trend seems reasonable.  Figure 11 shows the effects of a longer poling time (30 mins vs. 60 mins).  The sample poled for 30 minutes at 18 kV/mm was not sufficiently piezoelectric to yield a clean measurement signal.  The cool-down rate and/or duration of applied field during cooling does have an effect on the piezoelectricity of the PVDF, as shown in Figure 12.  This is an area which requires further investigation.  A multi-beam LDV (Polytec MSA-500) operated in single-point mode was used to verify the results of the single-point LDV, and their measurements of the same sample did not agree, as shown in Figure 13.  As a result of the discrepancy, and also the dynamic effects that need to be accounted for, the d33 for the measured films was not calculated for the measured data.  However, the single-point LDV data for the samples is still useful  58 because the effects of varying poling parameters can be deduced from the relative magnitudes of the surface displacements.  0 50 100 150 200 250 300 0 5 10 15 20 25 30 35 Frequency (kHz) Pe ak - to - pe ak  di sp la ce m en t (n m ) 30 kV/mm, 30 mins poling 45 kV/mm, 30 mins poling  Figure 10 – Single-point LDV data showing the effects of higher poling field strength   0 5 10 15 20 25 30 35 40 45 50 0 5 10 15 20 25 30 Frequency (kHz) Pe ak - to - pe ak  di sp la ce m en t (n m ) 18 kV/mm, 30 mins poling, no signal 18 kV/mm, 60 mins poling  Figure 11 – Single-point LDV data showing the effects of longer poling time (data for the 18 kV/mm, 30 mins poling is not visible since no signal was obtained from the sample)   59 0 50 100 150 200 0 10 20 30 40 50 60 Frequency (kHz)Pe ak - to - pe ak  di sp la ce m en t (n m ) 35 kV/mm, 30 mins poling, 30 mins slow cooling (oven door closed) with field applied 40 kV/mm, 30 mins poling, standard 10 min cooling  Figure 12 – Single-point LDV data showing the effects of poling time after heating has stopped  0 50 100 150 200 250 300 0 10 20 30 40 50 60 Frequency (kHz) Pe ak - to - pe ak  di sp la ce m en t (n m ) Multi-beam LDV in single-point mode Single-point LDV  Figure 13 – Comparison of measurements taken by two LDV’s on the same sample (45 kV/mm, 30 mins, 10 min cool down)  3.1.6 Summary of results PVDF resonator PVDF films were poled and metallized to form resonators.  Actuation of a resonator fabricated from unpoled film was not audible, indicating the poling was successful.  The  60 PVDF film of the loudest resonator produced to date was poled with the following parameters: • poling voltage: 22.5 kV (45 kV/mm for a 500 µm thick film) • poling temperature: 80 °C • duration of poling: 30 minutes • cool-down process: 10 minute cool-down with the electric field applied; there was no controlled temperature ramp-down, the door of the oven was simply opened  The optimal heating temperature during poling was reported to be 80 °C [52], but experiments were conducted both at 80 °C and 100 °C in this work to determine if the elevated temperature has any effect on the magnitude of the d33 constant.  Films poled at 100 °C were very susceptible to thermal meltdown due to electrical conduction – the PVDF burns for 5-10 seconds as opposed to dielectric breakdown which is extremely quick - hence for subsequent poling only 80 °C was used.  Laser Doppler vibrometer All of the poling was conducted at 80 °C, and unless stated otherwise, the cool-down procedure consisted of opening the oven door and applying the electrical poling field for 10 minutes while the film cools.  The films were actuated with 400V peak-to-peak signals.  Although the values of the measured data have not been verified as accurate, the following trends can be derived from the relative magnitudes: • increasing poling field strength up to the reported optimal poling field strength (55 kV/mm) is expected to result in films with stronger properties,  61 • poling for a longer duration is expected to result in more piezoelectric films; a saturation time may exist and requires further investigation, the report which presented the optimal poling parameters did not include the poling duration [52].  3.2 Transducers 3.2.1 Basic requirements One of the requirements for the interdigital transducers is to achieve the smallest electrode track width and spacing possible.  The track width and spacing determines the transducer periodicity, which influences the operating frequency and ultimately the sensitivity of the SAW device as mentioned in Section 2.2.1.  The second requirement is that the electrical conductivity of the electrode material should be sufficiently high so RC charging does not limit the operating frequency of the SAW device.  Design optimization of the interdigital transducers and the use of elements such as reflectors are separate topics which were not investigated in this work.  3.2.2 Selection of inkjet microprinting as the fabrication technique In addition to its environmental stability and high conductivity, PEDOT:PSS was selected as the electrode material because it is readily available from commercial sources. PEDOT:PSS is available from Afga (Orgacon) and H.C. Starck (Clevios) as a water- based dispersion, so a solution compatible patterning technique was desired.  Inkjet microprinting is such a technique.  In addition, its advantages include the ability to precisely deposit picolitre volumes of solutions, its direct-write ability eliminates the  62 need for masks, and since it is a non-contact deposition method, opportunities for contamination are reduced [65].  3.2.3 Materials and equipment The PVDF film used in the microprinting process development is the same film used in the heating and poling process, and is sourced from McMaster-Carr (part #8675K25). The film comes with a different finish on each side, and the microprinting was performed on the smoother side.  Surface modification was performed with a plasma cleaner (Harrick Plasma PDC-001).  The plasma cleaner features three power settings; for low power, 7.16 W is applied to the RF coil, for medium power 10.15 W, and for high power 29.6 W [66].  The intensity of treatment was found to have a substantial dependence on sample location within the plasma cleaner chamber, so for repeatability the sample was always placed in the same location (Figure 14).  Aqueous dispersions of PEDOT:PSS were sourced from H.C. Starck.  Clevios PH 750, with a manufacturer claimed bulk film conductivity of over 650 S/cm, was used initially [67].  When supplies needed to be replenished, Clevios PH 1000, due to its claimed bulk film conductivity of over 900 S/cm, was acquired.  Higher electrical conductivity is desirable because RC charging may be a limiting factor for high frequency operation of the SAW sensor.  The surface wetting properties of the two grades of aqueous PEDOT:PSS dispersions are nearly identical, as will be demonstrated below.   63  Figure 14 - Location inside plasma cleaner chamber where PVDF wafers are placed   An optical profiler (Veeco Wyko NT1100) was used to image the samples.  For all of the images, the 5X objective and either the 0.5X or 1.0X field of vision lens was used.  Once an image was acquired, measurements were taken using the Vision software.  The inkjet microprinting equipment utilized in this work is a partially modified research and production grade Microdrop Autodrop station from Microdrop Technologies (Norderstedt, Germany).  The Autodrop platform consists of an XYZ stage with a workspace of 200 x 200 x 80 mm, control electronics, a camera system, and a Windows PC running the Autodrop software.  The microprinting station was designed to be used with a variety of Microdrop dispenser heads.  However, dispenser heads from MicroFab Technologies (Plano, TX) were found to be less prone to clogging when printing with polymer-based solutions, and for the same nozzle orifice, produce slightly smaller Location for plasma treatment of PVDF wafer  64 droplets.  As a result, MicroFab dispensers (20, 30, 40 µm) were adapted for use with the Microdrop platform, using in-house built enclosures and electrical connectors.  Investigators such as Polasik and Schmidt [20], and others have utilized consumer inkjet printers for depositing polymers.  As Polasik and Schmidt noted, if the PEDOT:PSS ink is not sufficiently dilute and the viscosity is too high, the inkjet dispensers would be unable to expel any ink.  Consumer printers are designed for standard inks with a narrow range of low viscosities, whereas the piezoelectric dispenser heads from both Microdrop and MicroFab are capable of dispensing inks of a wider viscosity range, which offers more flexibility in the choice of ink.  For example, H.C. Starck offers a Clevios P Jet formulation specifically for inkjet printing with a dried film electrical conductivity of 200 S/cm, while its Clevios PH 1000 formulation for antistatic coating applications is more conductive at 900 S/cm.  The viscosities of Clevios P Jet and PH 1000 are 15 and 50 mPa·s respectively.  In addition to offering greater flexibility in the choice of printing ink, the Microdrop platform is more suitable for multi-layer microprinting.  Consumer desktop printers are typically equipped with paper sheet feeders which are not designed to precisely align the sheets for repeated same-sheet printing.  With the XYZ positioning system on the Microdrop platform, the stage moves and not the sheet/substrate/sample, and the manufacturer claims an x/y positioning accuracy of ± 5 µm and repetition accuracy of ± 1 µm [68]. These claims were not verified, but for this application which involves  65 typical drop spacings of approximately 100 µm, the repetition accuracy of the Microdrop positioning system was demonstrated to be quite good, as shown in Figure 15.   Figure 15 – Surface profile of 3 lines demonstrating repetition accuracy of Microdrop.  After printing each line shown in the figure, the stage is moved up to several centimeters away to print several other lines, before printing the next line in the figure.   Process: 30 µm nozzle, Clevios 1000 + 5 wt% DMSO ink, and plasma-treated PVDF.   3.2.4 Challenges and considerations for inkjet micropatterning of conductive traces  In the following sections, considerations relevant to the microprinting process development will be discussed, such as the: • hydrophobicity of the PVDF and the need for surface modification, • strategy for multi layer micropatterning, Middle line offsetted by half of the drop spacing (55 µm) from top and bottom lines Drop spacing of all lines is 110 µm  66 • effects of evaporation time on process repeatability, • selection of the conductivity enhancer for the PEDOT:PSS ink, • effects of substituting a different grade of PEDOT:PSS ink, • effects of adding a surfactant to the PEDOT:PSS ink.  3.2.4.1 Substrate surface modification Like other fluoropolymers, PVDF possesses a very hydrophobic surface.  Spin coating was used to deposit PEDOT:PSS electrodes onto PVDF films, but the aqueous dispersion of PEDOT:PSS was spun off of the PVDF even at a low spin speed of 300 RPM.  In addition, preliminary inkjet microprinting of PEDOT:PSS (H.C. Starck Clevios 750) onto PVDF indicated that the drops shifted slightly in position once the droplets came into contact with the surface, due to the poor adhesion (Figure 16).  Investigators such as Lee et al. have surface modified PVDF to make the surface more hydrophilic and improve the adhesion between PEDOT:PSS and PVDF [56].  The technique entails the introduction of hydrophilic carbon-oxygen functional groups onto the surface of the PVDF by bombarding the surface with ions and allowing the polymer to react with oxygen in the air.  Several different methods were used to surface modify the PVDF film to increase its hydrophilicity: corona treatment, UV-ozone, oxygen plasma treatment in a reactive ion etcher, and plasma treatment.  To gauge the effectiveness of the various surface modification techniques, qualitative contact angle measurements were obtained from photographs taken using a digital camera in close-up mode.  Corona treatment is a  67 standard surface treatment technique for polymers.  It uses electron avalanching to impact the surface of the polymer with sufficient energy to break molecular bonds.  Contact angle measurements of a fixed volume of water on two PVDF films – one modified with a handheld corona treatment unit – showed that the treatment increased the hydrophilicity of the PVDF.  The challenge with a handheld corona treatment unit is the non-uniformity of the surface modification.  A UV-ozone cleaner was used to treat PVDF film but the hydrophobicity did not change substantially.  Oxygen plasma generated by a reactive ion etcher (Trion PECVD/RIE) was very effective at increasing the hydrophilicity of PVDF.  Similar results were achieved using a plasma cleaner (Harrick Plasma PDC-001) which generates plasma from ambient air.  The plasma cleaner was used for subsequent surface modifications because it is simpler to operate and is readily accessible outside of a cleanroom.  Table 7 shows the effect of plasma surface treatment on the deposited drop parameters.  A smaller contact diameter (and potentially higher resolution) can be achieved when the PVDF is untreated, but surface treating the PVDF improves the contact diameter and interdrop spacing standard deviations, and the overall repeatability of the microprinting process.  The minimum drop spacing that avoids drop coalescence was also determined; the significance of this parameter will become evident in the following sections.   68  Figure 16 – Line of drops printed with a discontinuous drop spacing, showing shifting of drop position after being deposited (30 µm nozzle, 90 µm drop spacing)   Table 7 – Comparison of deposited drop parameters for untreated and surface treated PVDF (30 µm nozzle, same process) Deposited drop parameters Untreated PVDF Surface treated PVDF (30 seconds at high power) Contact diameter Avg: 68.9 µm Stdev: 9.7 µm Avg: 79.8 µm Stdev: 4.8 µm Minimum drop spacing for discontinuous line 90 µm 110 µm Interdrop spacing Avg: 91.4 µm Stdev: 13.1 µm Avg: 110.2 µm Stdev: 3.2 µm   Time dependence of surface modification The time dependence of the hydrophilic surface modification was examined using qualitative contact angle measurements.  Three PVDF films were surface modified simultaneously with the plasma cleaner, and the same volume of PEDOT:PSS was pipetted onto one film immediately, the second film 10 minutes after the treatment, and Shifting of drop position after being deposited PVDF not surface treated  69 the third film 30 minutes after the treatment.  The results showed that the surface- modified hydrophilic properties of the PVDF film tended to decrease slowly over time.  3.2.4.2 Strategy for multi layer micropatterning Single layer patterning was used initially to pattern the electrodes.  Uniform lines were obtained successfully, and the line shown in Figure 17 was printed using a 30 µm nozzle, with 70 µm drop spacing, plasma treatment time of 60 seconds at high power, and with a stage speed of 5 mm/s.  However, the success of the patterning depended heavily on the state of ambient environmental factors such as temperature and relative humidity, which determine the solvent evaporation time of the deposited drops.  Even with surface modification, the surface of the PVDF remains quite hydrophobic.  If freshly deposited drops are not allowed sufficient time to partially evaporate before the next drop is deposited, the drops merge and coalescence occurs.  For repeatability, a multi layer patterning process was developed which is insensitive to environmental conditions.  The multi-layer process development described in Section 3.2.5 is for the 30 µm nozzle. Multi layers are typically offsetted from each other in the direction of printing, and the notation used is a-b-c-d where a is zero for the first layer, b is the offset of the second layer from the first layer in the direction of printing, c is the offset of the third layer from the first layer in the direction of printing, etc.  The steps taken to develop repeatable microprinting processes for the 20 and 40 µm nozzles are similar, and the optimal process parameters for each will also be presented in Section 3.2.5.   70  Figure 17 - PEDOT:PSS line printed on PVDF using a single layer patterning process    3.2.4.3 Effects of evaporation time on process repeatability The aqueous dispersion of PEDOT:PSS ink is made up primarily of water, so the evaporation behavior of the ink was initially expected to be close to that of water.  The ink consists of: • 95 wt% Clevios PH 1000, with a solids content of 1.0 to 1.3 wt% [69] o the solids content consists of PEDOT:PSS and other additives • 5 wt% DMSO  Early process development of multi layer patterning indicated that the drying time needed for each layer was quite substantial.  The two tracks in Figure 18 were both printed using the same process (30 µm nozzle, 110 µm drop spacing, 60 seconds at low power plasma  71 treatment, four layers with 0-50-50-50 offsets, 2 minutes per layer drying time).  A more uniform track was obtained when cabinet ventilation was used (Figure 18b).  The expected evaporation time of a pure drop of water, with values for contact diameter, droplet volume, etc. obtained via measurement of PEDOT:PSS drops, was calculated using a non-linear approximation model by Schonfeld et al. [70].  The Matlab script implementing the model can be found in Appendix B. With a relative humidity of 50% and a temperature of 25 °C, the expected evaporation time for a typical drop dispensed from the 30 µm inkjet nozzle is 2.3 seconds.  For comparison, the expected evaporation time at 60 °C is 0.3 seconds.  (a) (b) Figure 18 – Comparison of surface profiles of two tracks printed using the same process (a) 2 minutes per layer drying time and no ventilation  (b) 2 minutes per layer drying time with ventilation   One possible explanation for the long observed evaporation time is the inhibited evaporation of water and DMSO due to the PEDOT:PSS.  To remove the evaporation time dependency from the micropatterning process, a long drying time (i.e. 5 minutes) was incorporated into the process.   72 3.2.4.4 Selection of PEDOT:PSS conductivity enhancer To enhance the conductivity of deposited PEDOT:PSS films, the manufacturer recommends adding a small amount of a secondary dopant – typically 5 wt% - of a high boiling point, polar solvent such as ethylene glycol or DMSO to the solution prior to coating or deposition.    The effect of the additive is independent of whether it remains in the film after drying or not; the mechanism of the conductivity enhancement is explained as the partial dissolution of the PEDOT stacks in the PEDOT:PSS complex by the polar solvent, creating opportunities for favorable morphological rearrangement and gel particle clustering [71].  Since PVDF is not chemically compatible with DMSO [61], ethylene glycol was used initially.  However, drops printed with Clevios 750 + 5wt% DMSO had smaller radii than drops printed with Clevios 750 + 5wt% ethylene glycol, as shown in Table 8.  A smaller contact diameter enables higher resolution printing, and the weight fraction of DMSO in the drop is quite low, so DMSO was selected as the conductivity enhancer.  Table 8 - Average contact diameter, and contact diameter variability for drops printed using different solutions Plasma treatment Clevios 750 + 5wt% DMSO Clevios 750 + 5wt% ethylene glycol 30 sec at high power Avg: 87.7 µm Stdev: 6.5 µm Avg: 94.2 µm Stdev: 4.3 µm 60 sec at low power Avg: 94.5 µm Stdev: 6.6 µm Avg: 99.6 µm Stdev: 5.9 µm none, pristine PVDF Avg: 47.8 µm Stdev: 5.8 µm Avg: 64.2 µm Stdev: 14.7 µm      73 3.2.4.5 Comparison of PEDOT:PSS types When Clevios PH 1000 – a more doped and electrically conductive grade of PEDOT:PSS - was substituted for Clevios PH 750, its surface wetting properties were investigated to ensure compatibility with existing patterning processes.  Measurements on the contact diameter and interdrop spacing standard deviation (Table 9) show the surface wetting properties of the two PEDOT:PSS dispersions are essentially the same.  Table 9 - Comparison of dots printed using two different grades of PEDOT:PSS, using the 30 µm nozzle and the same process (a drop spacing of 140 µm yields a fully discontinuous line)  Clevios 750 + 5 wt% DMSO Clevios 1000 + 5 wt% DMSO Contact diameter Avg: 70.9 µm Stdev: 2.7 µm Avg: 72.0 µm Stdev: 2.1 µm Interdrop spacing (140 µm) Avg: 141.2 µm Stdev: 3.1 µm Avg: 142.5 µm Stdev: 2.9 µm Height Not measured Avg: 572.3 nm Stdev: 56.6 nm   3.2.4.6 Effects of adding a surfactant to the PEDOT:PSS ink To make the aqueous PEDOT:PSS ink more compatible with the hydrophobic PVDF surface, a small amount (0.5 wt%) of Triton X-100, a non-ionic surfactant, was added to the ink.  Figure 19 shows how the addition of a small amount of surfactant increases the surface wetting abilities of the ink considerably.  With the exception of the printing ink, all process parameters are the same.  This effect is undesirable for micropatterning electrode tracks because thin line widths are crucial, but surfactant-added ink could be used to pattern connection pads to the electrodes.   74 (a) (b) Figure 19 – Comparison of surface profiles of tracks showing how adding a small amount of surfactant significantly increases surface wetting (a) Clevios 1000 + 5wt% DMSO + 0.5wt% Triton X-100 (b) Clevios 1000 + 5wt% DMSO   3.2.5 Development of an inkjet microprinting-based electrode deposition process  The development of a multi-layer inkjet microprinting process involves experimentally determining the: (1) optimal plasma treatment parameters (power level, treatment time) based on minimal contact diameter, contact diameter standard deviation, and interdrop spacing standard deviation of the deposited drops, (2) minimum drop spacing at which no coalescence occurs, (3) optimal drop spacing for alternate overlay in a multi-layer process, which will typically be slightly larger than the minimum no-coalescence drop spacing.  A microprinting process was first developed for the 30 µm nozzle, then for the 20 µm nozzle because it offers a smaller contact diameter and hence higher resolution.  The challenge with microprinting using the 20 µm nozzle is the rapid rate of nozzle clogging. If printing is stopped for more than 5 seconds, the evaporation of solvent from the solution at the tip will result in film formation and partial clogging.  Techniques for  75 addressing this challenge were devised but have not yet been tested.  As an interim solution, the 30 µm nozzle was used for microprinting, but due to a nozzle actuator failure and the non-availability of other 30 µm nozzles, a microprinting process for a 40 µm nozzle was developed.  3.2.5.1 General experimental procedures Prior to performing microprinting, the steps include: (1) cleaning the inkjet nozzle to ensure it is not clogged, (2) mounting the inkjet nozzle assembly onto the z-stage of the Microdrop station, and configuring the nozzle actuator parameters (driving voltage, pulse length) via the Autodrop software to achieve stable droplets, (3) cleaning a flat PVDF wafer with an isopropanol and distilled water rinse, then blow drying it with nitrogen, (4) plasma treating the PVDF wafer with appropriate parameters, (5) mounting the PVDF wafer onto the stage of the Microdrop station, (6) setting up the macro, or dispensing parameters, in the Autodrop software.  To avoid nozzle clogging, when the nozzle is not being used to pattern a sample, it should be left in the camera position to constantly eject droplets.  For the plasma treatment, the chamber is evacuated for 1 minute before the RF power is switched on, and a timer for keeping track of the treatment time should be started once the plasma is ignited.  As for the mounting of the PVDF wafer to the printing stage, to prevent the wafer from shifting during patterning (stage and sample move, inkjet nozzle is stationary), a steel plate was  76 attached to the stage using double-sided tape, and magnets were used to hold the wafer down at the edges.  With this setup, the droplet placement accuracy of the Microdrop micropatterning station was verified to be quite good.  Microdrop claims a stage positioning accuracy in the x and y directions of ± 5 µm.  3.2.5.2 Printing of lines with the 30 µm nozzle The process development described below is for the 30 µm nozzle.  The steps taken to develop repeatable microprinting processes for the 20 and 40 µm nozzles are similar, and the optimal process parameters for each will be presented in Section 3.2.5.6.  Step 1: Determining the optimal plasma treatment parameters for the 30 µm nozzle The first step in the development of the microprinting process is determining the optimal plasma treatment parameters.  Lines are printed onto PVDF film samples plasma treated with different processes, using a drop spacing which will assure a discontinuous line (i.e. 150 µm).  Figure 20 shows how the contact diameter, and contact diameter standard deviation, change as a function of the plasma treatment duration, and power level. Interdrop spacing standard deviation as a function of the plasma treatment time is shown in Figure 21.  The plasma treatment which yields the best compromise between contact diameter, contact diameter standard deviation, and interdrop spacing standard deviation was selected.  Sixty seconds at low power was selected to be the optimal plasma treatment process.  Although a smaller contact diameter can be achieved using high power plasma  77 treatment at 40 and 60 seconds, the 60 second low power treatment offers minimal contact diameter standard deviation and interdrop spacing standard deviation.  60 65 70 75 80 85 90 95 100 105 0 10 20 30 40 50 60 70 Plasma treatment time (sec) Co n ta c t d ia m e te r (u m ) Plasma LOW Plasma HIGH  Figure 20 - Contact diameter, and contact diameter standard deviation (shown as error bars), as a function of plasma treatment time and power level   2 4 6 8 10 12 14 16 0 10 20 30 40 50 60 70 Plasma treatment time (sec) In te rd ro p s ta n da rd  de v ia tio n  (u m ) Plasma LOW Plasma HIGH  Figure 21 - Interdrop spacing standard deviation as a function of plasma treatment time and power level    78 Step 2: Determining the minimum drop spacing at which no coalescence occurs for the 30 µm nozzle Using the optimal plasma treatment process, the minimum drop spacing at which no coalescence occurs is determined.  Drop coalescence can be seen in Figure 22 and Figure 23.  Many drops have merged, yielding fewer but larger drops and an apparent drop spacing that is substantially larger than the intended drop spacing.  Lines printed with a drop spacing equal to or exceeding the no-coalescence spacing are fully discontinuous as shown in Figure 24.  An Autodrop macro is useful for performing this task, and parallel lines with drop spacings ranging from two to four times the diameter of the inkjet nozzle orifice can be quickly microprinted.  The minimum suitable drop spacing for the 30 µm nozzle was determined to be 110 µm, with Clevios 1000 + 5 wt% DMSO ink and 60 seconds of low power plasma treatment.  (a) (b) Figure 22 – Surface profiles of lines microprinted with (a) 70 µm drop spacing and (b) 80 µm drop spacing, showing drop coalescence   79 (a) (b) Figure 23 – Surface profiles of lines microprinted with (a) 90 µm drop spacing and (b) 100 µm drop spacing, showing drop coalescence   (a) (b)  Figure 24 – Surface profiles of lines microprinted with (a) 110 µm drop spacing and (b) 120 µm drop spacing, showing no drop coalescence   Step 3: Determining the optimal drop spacing for alternate overlaying for the 30 µm nozzle The third step is determining the optimal drop spacing for alternate overlaying.  A second layer, which is offset by half the drop spacing of the first layer, and has double the drop spacing of the first layer, is printed on top of the first layer.  Drops deposited with the second layer will connect two neighboring drops in the first layer, as shown in Figures 25a and 25b.  The optimal drop spacing for alternate overlaying will typically be 10-20 µm larger than the minimum drop spacing at which the line is discontinuous.  A  80 possible explanation is that the slightly larger spacing enables more of the deposited drop to remain in contact with the substrate and not bead up as much around the already- deposited drops.  The minimum drop spacing with a 30 µm nozzle is 110 µm (Figure 25a), but a more uniform partial line is achieved with a drop spacing of 120 µm, as shown in Figure 25b.  With a larger drop spacing such as 140 or 160 µm, the droplets printed on the second layer merge with individual spots of the drop pattern from the first layer instead of connecting two adjacent spots (Figure 26a and 26b).  The optimal drop spacing for alternate overlaying using the 30 µm nozzle was found to be 120 µm (for Clevios 1000 + 5 wt% DMSO ink, 60 seconds of low power plasma treatment). Figure 27 shows a track with three layers formed from the first backbone pattern and two consecutive alternate-overlaid patterns.  (a) (b) Figure 25 – Surface profiles for alternate overlaying with (a) 110 µm and (b) 120 µm drop spacing    81 (a) (b) Figure 26 – Surface profiles for alternate overlaying with (a) 140 µm and (b) 160 µm drop spacing    Figure 27 – Surface profile of track produced with 3 layers and 120 µm drop spacing using the 30 µm nozzle   Step 4: Micropatterning tracks with a 30 µm nozzle Once the necessary process parameters have been determined, tracks can be micropatterned using a multi-layer microprinting process.  The process consists of: (1) a first layer microprinted using the optimal drop spacing (i.e. 120 µm for the 30 µm nozzle), (2) a second layer that is alternate overlaid onto the first layer, (3) a third layer that is alternate overlaid onto the second layer, fully connecting the track,  82 (4) a fourth layer to make the track width more uniform as well as to ensure the track is fully electrical connected, by halving the drop spacing (to 60 µm for the 30 µm nozzle) and doubling the number of droplets.  A substantial drying time (i.e. 5 minutes) should be allowed between layers, and the cabinet of the Microdrop micropatterning station should be ventilated.  Appendix D contains sample macros for a four layer micropatterning process.  Track microprinting process repeatability To demonstrate the repeatability of the process, seven identical and parallel tracks were fabricated.  One of the tracks is shown in Figure 28.  Of the seven tracks, five were electrically connected, the remaining two may have been interrupted by surface scratches on the industrial grade PVDF film.  The electrical resistances of the five connected tracks were all very similar, with a standard deviation of 8.5% of their average (Table 10).  Table 10 – Track parameters for five microprinted tracks such as the one in Figure 28 Track parameters Values Track width  Avg: 74.4 µm Stdev: 4.0 µm Track height Avg: 891.7 nm Stdev: 130.2 nm Electrical resistance Avg: 19.6 kΩ Stdev: 1.7 kΩ (8.5% of Avg) For tracks 20 mm in length on average   83   Figure 28 – Surface profile of a typical track microprinted using multi-layer process and 30 µm nozzle    3.2.5.3 Printing of pads with the 30 µm nozzle Preliminary single layer pad patterning was performed with surfactant-added PEDOT:PSS ink.  Square pads were formed from a square matrix of points.  Figure 29 and Figure 30 show pads patterned using the same single layer process, but with a different drop spacing.  Minor drop coalescence is advantageous for microprinting pads – provided significant bead-up does not occur - since it reduces the number of layers needed to fabricate uniform pads.   84 (a) (b) Figure 29 – Surface profiles of single layer patterned pads (a) 100 µm drop spacing (b) 120 µm drop spacing    Figure 30 – Surface profile of single layer patterned pad (140 µm drop spacing)   Well-formed pads were challenging to obtain with single layer patterning, so multi-layer patterning was used.  The square pads shown in Figure 31 and Figure 32 were patterned using two layers, with the second pattern deposited directly on top of the first pattern with no offsets.  When a drop spacing below the minimum spacing (at which no coalescence occurs) is used, the drops merge successfully but the deposited liquid has a tendency to flow towards the edges.  At a drop spacing at or above the minimum spacing, the drops do not merge, as shown in Figure 32.   85 (a) (b) Figure 31 – Surface profiles of square pads patterned with two layers (a) 100 µm drop spacing (b) 110 µm drop spacing   (a) (b) Figure 32 – Surface profiles of square pads patterned with two layers (a) 120 µm drop spacing (b) 130 µm drop spacing   From the preliminary micropatterning of the pads, it is clear that the development of a repeatable process for microprinting pads will require more time.  For this work, pads were therefore formed by drop-casting a small amount of PEDOT:PSS dispersion onto the ends of the tracks.  3.2.5.4 Printing of lines with the 20 µm nozzle Figure 33 shows the contact diameter average and standard deviation as a function of the plasma treatment time, for two plasma treatment power levels.  Figure 34 shows how the interdrop spacing varies as a function of the plasma treatment time, for low and high  86 power plasma treatments.  The plasma treatment of 10 seconds at high power was determined to be most optimal, based on a small contact diameter, reasonably small variability in contact diameter, and small variability in interdrop spacing.  35 40 45 50 55 60 65 0 20 40 60 80 100 120 Plasma treatment time (sec) Co n ta ct  di am et er  (um ) Plasma LOW Plasma HIGH  Figure 33 - Contact diameter vs. plasma treatment time, for deposited drops printed using the 20 µm nozzle.  The error bars represent the contact diameter standard deviation.     87 0 2 4 6 8 10 12 14 16 18 0 20 40 60 80 100 120 Plasma treatment time (sec) In te rd ro p sp a c in g s ta n da rd  de v ia tio n  (um ) Plasma LOW Plasma HIGH  Figure 34 - Interdrop spacing variability vs. plasma treatment time, for deposited drops printed using the 20 µm nozzle   The highest resolution microprinting to date was performed using the 20 µm nozzle, and the thinnest track width and spacing achieved is 55 µm.  This track width is comparable to the smaller feature sizes fabricated using photolithography techniques.  The tracks shown in Figure 35 were printed using an early-stage multi-layering process (has drying time dependence), using an eight layer process.  Statistics for the tracks are shown in Table 11.  A multi-layer process with no drying time dependence is expected to produce comparable, if not better, tracks in terms of track uniformity and average width.  The tracks are identified by the offsets used to create them.  For example, 0-50-0-50 represents a four layer process with the second and fourth layer patterns offset from the first and third layer patterns – in the direction of printing - by 50 µm.     88 (a) (b) Figure 35  - Surface profiles of two tracks printed with a 20 µm nozzle, using an early stage multi- layering process (a) 0-50-0-50-0-50-0-50 (b) 0-33-67-50-33-67-50-50   Table 11 - Track statistics for two tracks printed using a 20 µm nozzle Drop offset sequence 0-50-0-50-0-50-0-50 0-33-67-50-33-67-50-50 Track width Avg: 55.2 µm Stdev: 6.2 µm Avg: 65.2 µm Stdev: 5.0 µm Track height Avg: 567 nm Stdev: 124 nm Avg: 585 nm Stdev: 95 nm   The most significant microprinting challenge with the 20 µm nozzle is the rapid clogging of the nozzle orifice.  Figures 36 shows an attempt at fabricating five pairs of electrode fingers; the direction of printing is from right to left.  After completion of a track, the x-y stage moves to the start of the next track, and during this time (~2 seconds) the inkjet nozzle partially clogs which leads to droplet ejection sideways.  As a result, the starting section of each track is distorted.      89  Figure 36 – Surface profile of 5 electrode finger pair printed using a 20 µm nozzle and an 8 layer process   3.2.5.5 Printing of lines with the 40 µm nozzle The results of the plasma treatment optimization are shown in Figure 37 and Figure 38. Since there is no plasma treatment process that is best in terms of smallest contact diameter average and variability, and interdrop spacing variability, the processes were ranked as shown in Table 12 and the 10 second treatment at low power was selected as the best compromise.  Table 12 – Ranking of plasma treatments by contact diameter, contact diameter standard deviation, and interdrop spacing standard deviation Rank (1 = best) Contact diameter Contact diameter standard deviation Interdrop spacing standard deviation 1 Low 30 (71.5 µm) High 10 (5.3 µm) High 30 (2.7 µm) 2 High 50 (73.2 µm) Low 10 (5.7 µm) High 10 (3.5 µm) 3 Low 10 (76.4 µm) Low 30 (6.1 µm) Low 10 (4.0 µm) 4 High 30 (85.8 µm) High 50 (8.0 µm) Low 50 (5.9 µm) 5 Low 50 (89.0 µm) High 30 (8.7 µm) Low 30 (7.6 µm) 6 High 10 (89.5 µm) Low 50 (10.1 µm) High 50 (8.7 µm)  90 60 70 80 90 100 0 10 20 30 40 50 60 Plasma treatment time (sec) Co n ta c t d ia m e te r (u m ) Plasma LOW Plasma HIGH  Figure 37 - Contact diameter average and standard deviation (error bars) as a function of the plasma treatment time and power level for the 40 µm nozzle   2 3 4 5 6 7 8 9 0 10 20 30 40 50 60 Plasma treatment time (sec) In te rd ro p s pa c in g s ta n da rd  de v ia tio n  (u m ) Plasma LOW Plasma HIGH  Figure 38 - Interdrop spacing variability as a function of the plasma treatment time for the 40 µm nozzle     91 3.2.5.6 Summary of results The optimal process parameters for microprinting with the 20, 30, and 40 µm nozzles are given in Table 13.  The optimal process parameters for the 20 µm nozzle is 10 seconds of plasma treatment at high power, and the optimal drop spacing is expected to be slightly larger than the minimum drop spacing of 100 µm.  Optimal process parameters for the 30 µm nozzle is 60 seconds of low power plasma treatment, and a drop spacing of 120 µm.  For the 40 µm nozzle, the optimal process parameters are 10 seconds of plasma at low power, and a drop spacing of 110 µm.  Drop characteristics associated with the optimal process parameters can also be found in Table 13.   Table 13 - Summary of optimal microprinting process parameters, and deposited drop characteristics with optimal plasma treatment  20 µm nozzle 30 µm nozzle 40 µm nozzle Optimal plasma treatment parameters 10 seconds at high power 60 seconds at low power 10 seconds at low power Minimum drop spacing with no drop coalescence 100 µm 110 µm 110 µm Optimal drop spacing for alternate-overlay Alternate-overlay not performed with nozzle yet 120 µm 120 µm Average contact diameter 41.9 µm 70.9 µm 76.4 µm Contact diameter standard deviation 4.1 µm 2.6 µm 5.7 µm Interdrop spacing standard deviation 5.5 µm 3.1 µm 4.0 µm      92 3.2.6 RC charging considerations  The electrical conductivity of a typical micropatterned PEDOT:PSS track was calculated to be approximately 150 S/cm, based on measurements of its dimensions and electrical resistance (Table 10).  The electrical resistivity  m m mmk l AR Ω×= × ××× ×Ω=×= − − −− 5 3 69 1037.6 1041.20 )1037.7410892()6.19(ρ  can be calculated from the electrical resistance of a track R, the approximate average rectangular cross-section of the track A, and the length l.  The actual electrical conductivity ρσ 1= = 150 S/cm is 1/6 of the manufacturer’s claimed bulk film conductivity of 900 S/cm for Clevios 1000 [69].  The strip line transmission model [34] was used to calculate the RC time constant of the input interdigital transducer.  The following simplifying assumptions were used: • The bus connectors which connect the interdigital fingers to the pad were not included in the calculation, they are assumed to be substantially more conductive than the fingers due to a larger width • Total capacitance was calculated entirely using the dielectric permittivity of PVDF (worst case scenario) since the relative dielectric permittivities of PVDF and air are 13 and 1 respectively • Electrical conductance of air, surface resistivity of PVDF, internal and external inductance of PEDOT:PSS electrodes were neglected.  93 The results for the following inputs: • 75 µm track width and a SAW wavelength of 300 µm • Length of IDT finger equal to 20 SAW wavelengths • 20 fingers per IDT • Operating frequency of 2 MHz can be found in Table 14.  Table 14 – Calculated IDT parameter values IDT parameters Values Resistance per finger 45.9 mΩ Capacitance per finger 55.3 pF Total IDT capacitance 2.21 nF Total IDT resistance 1.1 mΩ RC constant 2.54 ps Max expected operating frequency 62.8 GHz   As shown in Table 14, the maximum expected operating frequency was calculated to be 62.8 GHz, and is not a limiting factor for the PVDF-based SAW device, which will operate in the MHz range.  The Matlab script used to calculate the RC time constant can be found in Appendix C.  3.3 Towards the fabrication of a SAW device 3.3.1 Process integration considerations  To fabricate a SAW device, two approaches may be taken.  The PVDF substrate can be prepared first by heating and poling and the interdigital transducers can be micropatterned afterwards, or vice versa.   94 First step: preparation of the substrate If the PVDF is first poled, then the plasma cleaner cannot be used to surface modify the poled film for micropatterning.  The alternating RF field used to generate the plasma has been shown to decrease the film’s polarization.  As a check, a gold metallized PVDF resonator was placed inside the plasma cleaner, and after treatment, it was audibly quieter.  An alternative surface modification technique that could be used is corona treatment.  For process repeatability, a handheld corona treatment unit cannot be used and a fixed station is needed.  The high voltage poling process is also a form of corona treatment.  After heating and poling, the surface of the active piezoelectric region was observed to be significantly more hydrophilic than the non-poled regions.  If the degree of surface modification of the PVDF film via the poling process reaches saturation after a period of time – so the process is repeatable - it may be possible to forgo the surface modification step and micropattern the electrodes immediately after the poling process.  First step: micropatterning of the interdigital tranducers The advantage of performing the micropatterning process first is that controlling the surface modification will not be a concern.  As mentioned in the above section, the high voltage poling process is also a form of corona treatment, so the micropatterned side should not be in contact with the high voltage electrode to avoid PEDOT:PSS degradation.  The PVDF film can be mounted so that the micropatterned side is in contact with the ground electrode plate.  The heating temperature of 80 °C is not expected to  95 negatively impact the electrical conductivity of PEDOT:PSS films, as discussed in Section 2.2.2.  Since a repeatable electrode microprinting process was developed previously, this approach was taken.  3.3.2 Loss of electrode conductivity phenomenon  An unexpected phenomenon occurred when a PVDF film with microprinted PEDOT:PSS electrodes was heated and poled.  Previously electrically conductive tracks had become open-circuited.  This phenomenon was observed again when a second sample, prepared in the same way, had non-conductive tracks.  The PEDOT:PSS currently being used is highly doped, with a PEDOT:PSS ratio of 1:2.5 by weight [69].  The loss of electrical conductivity in highly doped conductive polymers due to the application of an external field has been investigated [72].  The effect is unexpected since external electrical fields are not expected to penetrate “metallic” conductive polymers due to the Debye screening length for metals.  Hsu et al. proposed that the loss of electrical conductivity is caused by interruptions in the low-dimensional carrier hopping network on the PEDOT backbone, which is caused by ionic charges compensating for some of the positively charged holes on the polymer backbone [72].  To restore electrical conductivity, it was hypothesized that exposure to a solution containing water and a conductivity enhancer (i.e. DMSO, ethylene glycol, etc.) would rearrange the morphology in a more favourable manner, just as the addition of DMSO to the PEDOT:PSS dispersion enhances the electrical conductivity of dried films.  The water would give the PSS mobility, while the DMSO would partially dissolve the  96 PEDOT stacks.  Preliminary results indicate that the solution (water and 5 wt% DMSO) may restore electrical conductivity, although the DMSO is incompatible with PVDF and partial dissolution of the surface was observed.  Ethylene glycol was substituted for DMSO, and preliminary results indicated that the solution may restore electrical conductivity but not to the same extent as DMSO and water (difference in electrical resistance of at least 3 orders of magnitude).  Further investigation is needed.  97 4. Conclusions and future work 4.1 Summary of results  An apparatus and a process for preparing piezoelectric PVDF film were developed. Although the degree of piezoelectricity was not characterized, the poled PVDF film was demonstrated to be piezoelectric via the fabrication of audible resonators, and was characterized using two Laser Doppler vibrometer units.  The measured values of the surface displacement did not agree between the two vibrometers, but the relative magnitudes of the surface displacements indicate that the piezoelectricity of the films increases as the poling field strength is increased, with other process parameters remaining the same.  Repeatable processes for micropatterning highly electrically conductive PEDOT:PSS electrode tracks were developed for three inkjet nozzle orifice sizes: 20, 30, and 40 µm. For tracks micropatterned using the same process, the electrical resistances have a standard deviation of 8.5% of the average.  The electrical conductivity of micropatterned tracks is approximately 150 S/cm, which is one-sixth of the manufacturer’s claimed bulk film conductivity.  Using the 30 µm nozzle, the smallest electrode track width that can be micropatterned repeatably is 75 µm.  A track width of 55 µm was achieved using the 20 µm nozzle, but the process is not yet fully repeatable as a result of rapid nozzle clogging caused by solvent evaporation of the ink at the nozzle tip.  The addition of a small amount (i.e. 0.5 wt%) of non-ionic surfactant increases the wetting properties of the PEDOT:PSS ink dramatically, which is not ideal for microprinting tracks since high resolution is desired, but may be useful for microprinting pads.  98 The optimal operating frequency of the SAW device is at the synchronous frequency, which is set by the SAW propagation velocity and transducer periodicity.  PVDF has a SAW velocity of approximately 500 m/s.  At the present, the thinnest electrode tracks that can be fabricated in a repeatable manner are 75 µm wide, so the transducer periodicity is 300 µm.  The achievable synchronous frequency using the parameters from this work is  d vf 00 = = ( 500 m/s ) / ( 300 µm ) ≈ 1.7 MHz   4.2 Future work 4.2.1 Proposed piezoelectric substrate characterization procedures  To determine the effects of process parameter changes, it is necessary to accurately characterize the piezoelectric properties of PVDF.  The measurement techniques used in this work (aside from Laser Doppler Vibrometry) are static in nature.  Investigators have also exploited dynamic measurements to obtain the piezoelectric strain constants.  Omote et al. prepared five different kinds of resonators (length extensional, thickness extensional, etc.), evaporated aluminum electrodes onto both surfaces of each resonator, obtained resonance curves with an impedance analyzer, and determined material constants by fitting curves simulated by Mason’s equivalent circuit to the observed resonance curves [73].  The proposed substrate characterization procedure should be based on both dynamic and static bimorph measurements.  To measure d33, a resonator of fixed dimensions would be  99 fabricated from the poled film, and be metallized on both sides with either gold or PEDOT:PSS.  The resonance curve would be obtained with an impedance analyzer.  To measure d31 and d32, a bimorph would be constructed from poled PVDF film.  4.2.2 Reduction of electrode track width  As mentioned in Section 4.1, the thinnest electrode track width that can be fabricated in a repeatable manner is 75 µm, using the 30 µm nozzle.  Electrode track widths of 55 µm were achieved using the 20 µm inkjet nozzle, but process repeatability was poor due to the rapid nozzle clogging.  Once that issue is addressed, track widths of 55 µm could be achieved, enabling the SAW device to operate at a higher synchronous frequency f0 = v0 / d = ( 500 m/s ) / ( 220 µm ) ≈ 2.3 MHz.  A potential solution is mounting a thin, highly water-absorbent sponge ring around the nozzle, increasing the relative humidity near the nozzle tip and slowing the evaporation rate of water from the PEDOT:PSS ink.  4.2.3 Investigation into how PVDF material properties could limit SAW device performance  As a ferroelectric polymer, PVDF has substantial hysteresis which results in significant energy dissipation [74].  This property is expected to limit the maximum operating frequency of a PVDF homopolymer-based SAW device, possibly to below 1.7 MHz. Other rate limiting mechanisms include RC charging, heating and dissipation within the PVDF, and ultimately the relaxation times of the polar backbone groups [74].  Further investigation is needed.   100 In this work, PVDF was not stretched while being heated and poled.  With the proposed substrate characterization procedures discussed in the previous section, it will be possible to determine the differences between the optimal reported piezoelectric constants and the constants obtained via optimal heating and poling.  If the difference is substantial, a stretching apparatus can be constructed for stretching the films, or pre-poled thick films can be sourced.  Homopolymer PVDF was selected over the co-polymer PVDF-TrFE due to its lower cost and higher piezoelectric constants.  However, PVDF-TrFE has several advantages over PVDF, including higher crystallinity, electromechanical coupling, and usable (Curie) temperature.  As a result, the ferroelectric polymer most commonly used for actuation is PVDF-TrFE [74].  In addition, PVDF-TrFE does not need to be poled as it naturally crystallizes into the desired β-phase.  If the material properties of homopolymer PVDF prove to be a limiting factor in the performance of the SAW device, PVDF-TrFE can be considered.  4.2.4 Investigation into the loss of electrode electrical conductivity phenomenon  As discussed in Section 3.3.2, the loss of electrode conductivity was observed after a micropatterned PVDF wafer was heated and poled.  If possible, preparation of the substrate first should be considered, as discussed in Section 3.3.1.  If that is not possible, then methods of restoring electrical conductivity in the electrodes need to be developed. One possible approach is to expose the electrodes to a solution intended to rearrange the  101 morphology of the PEDOT:PSS in a more favourable way, thus enhancing the electrical conductivity.  4.2.5 Fabrication and characterization of a SAW device prototype  If after further investigation the material properties of PVDF are deemed not to be a critical limiting factor for the performance of the SAW device, then fabrication of a SAW sensor can proceed.  Fabrication considerations are discussed in Section 3.3.1.  Once a SAW device prototype has been fabricated, electrical measurements can be performed on the output transducer to verify the presence of propagating SAWs.  In addition, a multi- beam Laser Doppler Vibrometer (Polytec MSA-500), which is capable of measuring in- plane and out-of-plane motion, can be used to characterize the device.  4.2.6 Development of a technique for addressing the temperature sensitivity of PVDF  As discussed in Section 2.2.2, PVDF is very sensitive to temperature, and temperature changes will cause density changes in the substrate thus changing the SAW propagation velocity.  Commercial SAW-based chemical detectors, such as the Electronic Sensor Technology zNose portable gas chromatographer, utilize a temperature control system to maintain the SAW sensor at a fixed temperature.  The temperature control system is also used to heat the sensor to desorb compounds trapped in the sensing layer.  If PVDF-based SAW sensors are to be used in air quality monitoring applications, a temperature control system is needed.  Care has to be taken to avoid exposing the PVDF to temperatures above 60 °C, since above that temperature piezoelectric decay occurs.  An alternative strategy is to actively monitor the temperature and continually adjust the excitation  102 frequency so the transducer is always excited at the synchronous frequency.  The disadvantage with this technique is that the intrinsic mass sensitivity, which depends on the SAW propagation velocity, of the SAW device will be temperature dependant.  4.2.7 Verification of sensing layer selection methodology  A sensing layer selection methodology based on the Hansen solubility parameters was proposed, but was not verified since the focus of this work was on the fabrication of a sensing platform.  Verification of the methodology entails applying polymer-based sensing layers to sensors and measuring their responses to a collection of VOC analytes. The sensors can either be chemiresistors, quartz crystal microbalance sensors, or even the PVDF-based SAW sensors, if working prototypes are available.  The methodology will be successfully verified if the relative magnitudes of the measured responses match the relative magnitudes of the predicted responses.   103 References  [1] “B.C. Air Action Plan - Environment - Province of British Columbia,” http://www.bcairsmart.ca/pollutants/, 2009. [2] D.V. Bates and R.B. Caton, A Citizen's Guide to Air Pollution, David Suzuki Foundation, 2002. [3] P. Schwarzhoff and K. Jones, “Email subject: Support for BC Clear Proposal,” Sep. 2008. [4] W.A. Groves, E.T. Zellers, and G.C. Frye, “Analyzing organic vapors in exhaled breath using a surface acoustic wave sensor array with preconcentration: Selection and characterization of the preconcentrator adsorbent,” Analytica Chimica Acta, vol. 371, Oct. 1998, pp. 131-143. [5] “B.C. Air Quality Network Description,” http://www.bcairquality.ca/assessment/network-description.html, 2009. [6] G. Man, B. Stoeber, and K. Walus, “An assessment of sensing technologies for the detection of clandestine methamphetamine drug laboratories,” Forensic Science International,  vol. 189, Aug. 2009, pp. 1-13. [7] D.S. Ballantine, R.M. White, and G.C. Frye, Acoustic Wave Sensors, Academic Press, 1997. [8] I. Towhata, Geotechnical Earthquake Engineering, Springer, 2008. [9] J.D.N. Cheeke, Fundamentals and applications of ultrasonic waves, CRC Press, 2002. [10] B.M. Lempriere, Ultrasound and Elastic Waves Frequently Asked Questions,  San Diego, CA, USA: Academic Press, 2002.  104 [11] K.A. Ellis, R.B. van Dover, T.J. Klemmer, and G.B. Alers, “Magnetically transduced surface acoustic wave devices,” J. Appl. Phys., AIP, 2000, pp. 6304- 6306. [12] R.M. White and F.W. Voltmer, “Direct piezoelectric coupling to surface elastic waves,” Applied Physics Letters,  vol. 7, 1965, pp. 314–316. [13] S. Sivaramakrishnan, R. Rajamani, C. Smith, K. McGee, K. Mann, and N. Yamashita, “Carbon nanotube-coated surface acoustic wave sensor for carbon dioxide sensing,” Sensors and Actuators B: Chemical,  vol. 132, May. 2008, pp. 296-304. [14] S.J. Ippolito, A. Trinchi, D.A. Powell, and W. Wlodarski, “Acoustic Wave Gas and Vapor Sensors,” Solid State Gas Sensing, 2009, pp. 1-44. [15] J. Grate, “Acoustic Wave Microsensor Arrays for Vapor Sensing,” Chemical Reviews,  vol. 100, Jul. 2000, pp. 2627-2648. [16] M. Atashbar, A. Sadek, W. Wlodarski, S. Sriram, M. Bhaskaran, C. Cheng, R. Kaner, and K. Kalantar-zadeh, “Layered SAW gas sensor based on CSA synthesized polyaniline nanofiber on AlN on 64° YX LiNbO3 for H2 sensing,” Sensors and Actuators B: Chemical,  vol. 138, Apr. 2009, pp. 85-89. [17] C.L. Hsu, C.Y. Shen, R.T. Tsai, and M.Y. Su, “Surface Acoustic Wave Ammonia Sensors Based on ST-cut Quartz under Periodic Al Structure,” Sensors,  vol. 9, pp. 980-994. [18] W. Wang, T. Wu, T. Chou, and Y. Chen, “A ZnO nanorod-based SAW oscillator system for ultraviolet detection,” Nanotechnology,  vol. 20, 2009, p. 135503. [19] D. Preethichandra and K. Kaneto, “SAW Sensor Network Fabricated on a  105 Polyvinylidine Difluoride (PVDF) Substrate for Dynamic Surface Profile Sensing,” Sensors Journal, IEEE,  vol. 7, 2007, pp. 646-649. [20] J.T. Polasik and V.H. Schmidt, “Conductive polymer PEDOT/PSS electrodes on the piezoelectric polymer PVDF,” Smart Structures and Materials 2005: Electroactive Polymer Actuators and Devices (EAPAD),  San Diego, CA, USA: SPIE, 2005, pp. 114-120. [21] C. Lee, J. Joo, S. Han, J. Lee, and S. Koh, “Poly (vinylidene fluoride) transducers with highly conducting poly (3, 4-ethylenedioxythiophene) electrodes,” Synthetic Metals,  vol. 152, 2005, pp. 49-52. [22] C.S. Lee, J.Y. Kim, D.E. Lee, J. Joo, B.G. Wagh, S. Han, Y.W. Beag, and S.K. Koh, “Flexible and transparent organic film speaker by using highly conducting PEDOT/PSS as electrode,” Synthetic Metals,  vol. 139, Sep. 2003, pp. 457-461. [23] M. Gad-el-Hak, MEMS: design and fabrication, CRC Press, 2006. [24] H. Kawai, “The piezoelectricity of poly (vinylidene fluoride),” Jpn. J. Appl. Phys, vol. 8, 1969, pp. 975–976. [25] G.M. Sessler, “Piezoelectricity in polyvinylidenefluoride,” The Journal of the Acoustical Society of America,  vol. 70, Dec. 1981, pp. 1596-1608. [26] P. Ueberschlag, “PVDF piezoelectric polymer,” Sensor review,  vol. 21, 2001, pp. 118-125. [27] G.D. Jones, R.A. Assink, T.R. Dargaville, P.M. Chaplya, R.L. Clough, J.M. Elliott, J.W. Martin, D.M. Mowery, and M.C. Celina, Characterization, performance and optimization of PVDF as a piezoelectric film for advanced space mirror concepts, Sandia report SAND2005-6846, Sandia National Laboratories, 2005.  106 [28] K. Williams, The design and construction of a poling station for piezoelectric thin films, EECE 400 report, Department of Electrical & Computer Engineering, The University of British Columbia, 2009. [29] C.K. Chiang, C.R. Fincher, Y.W. Park, A.J. Heeger, H. Shirakawa, E.J. Louis, S.C. Gau, and A.G. MacDiarmid, “Electrical Conductivity in Doped Polyacetylene,” Physical Review Letters,  vol. 39, Oct. 1977, p. 1098. [30] T.A. Skotheim and J.R. Reynolds, Handbook of Conducting Polymers, CRC Press, 2007. [31] H.C. Starck, “Conductivity range of ICP,” http://www.clevios.com/index.php?page_id=998. [32] M. Goueygou, Z. Lafhaj, and F. Soltani, “Assessment of porosity of mortar using ultrasonic Rayleigh waves,” NDT and E International,  vol. 42, 2009, pp. 353–360. [33] “Poly(vinylidene fluoride),” CHEMnetBASE Polymers: A Property Database, Chapman & Hall/CRC Press LLC, 2009. [34] Z. Popovic and B.D. Popovic, Introductory electromagnetics,  Upper Saddle River, NJ: Prentice-Hall, 2000. [35] Yongrae Roh and V. Varadan, “Characterization of all the elastic, dielectric, and piezoelectric constants of uniaxially oriented poled PVDF films,” Ultrasonics, Ferroelectrics and Frequency Control, IEEE Transactions on,  vol. 49, 2002, pp. 836-847. [36] L. Groenendaal, F. Jonas, D. Freitag, H. Pielartzik, and J.R. Reynolds, “Poly (3, 4- ethylenedioxythiophene) and its derivatives: past, present, and future,” Advanced Materials,  vol. 12, 2000, pp. 481–494.  107 [37] J. Huang, P.F. Miller, J.C. de Mello, A.J. de Mello, and D.D.C. Bradley, “Influence of thermal treatment on the conductivity and morphology of PEDOT/PSS films,” Synthetic Metals,  vol. 139, Oct. 2003, pp. 569-572. [38] R. Kiebooms, A. Aleshin, K. Hutchison, F. Wudl, and A. Heeger, “Doped poly(3,4- ethylenedioxythiophene) films: Thermal, electromagnetical and morphological analysis,” Synthetic Metals,  vol. 101, May. 1999, pp. 436-437. [39] A. Shirinov and W. Schomburg, “Pressure sensor from a PVDF film,” Sensors and Actuators A: Physical,  vol. 142, Mar. 2008, pp. 48-55. [40] A.G. Kolbeck, “Aging of piezoelectricity in poly(vinylidene fluoride),” Journal of Polymer Science: Polymer Physics Edition,  vol. 20, 1982, pp. 1987-2001. [41] M. Kus and S. Okur, “Electrical characterization of PEDOT:PSS beyond humidity saturation,” Sensors and Actuators B: Chemical,  vol. 143, Dec. 2009, pp. 177-181. [42] U. Lang, N. Naujoks, and J. Dual, “Mechanical characterization of PEDOT:PSS thin films,” Synthetic Metals,  vol. 159, Mar. 2009, pp. 473-479. [43] H.C. Starck, “Product information CLEVIOS F 141 M,” http://www.hcstarck.com/index.php?page_id=3364. [44] Arkema, “Kynar® PVDF Chemical Resistance Chart,” http://www.arkema- inc.com/print.cfm?pag=719. [45] M.P. Eastman, R.C. Hughes, G. Yelton, A.J. Ricco, S.V. Patel, and M.W. Jenkins, “Application of the Solubility Parameter Concept to the Design of Chemiresistor Arrays,” Journal of The Electrochemical Society,  vol. 146, Oct. 1999, pp. 3907- 3913. [46] S.V. Patel, T.E. Mlsna, B. Fruhberger, E. Klaassen, S. Cemalovic, and D.R. Baselt,  108 “Chemicapacitive microsensors for volatile organic compound detection,” Sensors and Actuators B: Chemical,  vol. 96, Dec. 2003, pp. 541-553. [47] A. Loui, T.V. Ratto, T.S. Wilson, S.K. McCall, E.V. Mukerjee, A.H. Love, and B.R. Hart, “Chemical vapor discrimination using a compact and low-power array of piezoresistive microcantilevers,” The Analyst,  vol. 133, 2008, pp. 608-615. [48] J.W. Grate and M.H. Abraham, “Solubility interactions and the design of chemically selective sorbent coatings for chemical sensors and arrays,” Sensors and Actuators B: Chemical,  vol. 3, Feb. 1991, pp. 85-111. [49] C.M. Hansen, Hansen solubility parameters: a user's handbook, CRC Press, 2000. [50] B.L. Karger, L.R. Snyder, and C. Eon, “Expanded solubility parameter treatment for classification and use of chromatographic solvents and adsorbents,” Analytical Chemistry,  vol. 50, Dec. 1978, pp. 2126-2136. [51] A.F.M. Barton, CRC handbook of polymer-liquid interaction parameters and solubility parameters, 1990. [52] T. Kaura, R. Nath, and M.M. Perlman, “Simultaneous stretching and corona poling of PVDF films,” Journal of Physics D. Applied Physics,  vol. 24, 1991, pp. 1848– 52. [53] J. Brandrup and E.H. Immergut, POLYMER handbook, John Wiley and Sons, cop., 1999. [54] S. Zhang, Z. Liang, Q. Wang, and Q. Zhang, “Electroresponsive Polymers and Their Applications,” Mater. Res. Soc. Symp. Proc. 889, 2006. [55] A. Kumar and M.M. Periman, “Simultaneous stretching and corona poling of PVDF and P(VDF-TriFE) films. II,” Journal of Physics D: Applied Physics,  vol. 26, 1993,  109 pp. 469-473. [56] C.S. Lee, J.Y. Kim, D.E. Lee, J. Joo, S. Han, Y.W. Beag, and S.K. Koh, “An approach to durable poly (vinylidene fluoride) thin film loudspeaker,” Journal of Materials Research,  vol. 18, 2003, pp. 2904-2911. [57] G. Oreski and G. Wallner, “Aging mechanisms of polymeric films for PV encapsulation,” Solar Energy,  vol. 79, Dec. 2005, pp. 612-617. [58] T. Dargaville, M. Celina, and R. Clough, “Evaluation of vinylidene fluoride polymers for use in space environments: Comparison of radiation sensitivities,” Radiation Physics and Chemistry,  vol. 75, Mar. 2006, pp. 432-442. [59] S. Kinning, “Email subject: PVDF,” Nov. 2009. [60] Y. Huan, Y. Liu, and Y. Yang, “Simultaneous stretching and static electric field poling of poly(vinylidene fluoride-hexafluoropropylene) copolymer films,” Polymer Engineering & Science,  vol. 47, 2007, pp. 1630-1633. [61] Porex Corporation, Porex PVDF chemical compatibility chart, 2007. [62] S. Sokhanvar, A. Zabihollah, and R. Sedaghati, “Investigating the effect of the orthotropic property of piezoelectric PVDF,” Transactions of the CSME,  vol. 31, 2007. [63] “IEEE standard on piezoelectricity,” ANSI/IEEE Std 176-1987, 1988. [64] J. Schmit and D. Chen, “AN541 - Greater Measurement Detail with High-Definition Vertical Scanning Interferometry,” 2007. [65] E. Tekin, P.J. Smith, and U.S. Schubert, “Inkjet printing as a deposition and patterning tool for polymers and inorganic particles,” Soft Matter,  vol. 4, 2008, pp. 703-713.  110 [66] Harrick Plasma, “Plasma Products : Plasma Cleaners,” http://www.harrickplasma.com/products_cleaners.php. [67] H.C. Starck, “Product information CLEVIOS PH 750,” http://www.hcstarck.com/index.php?page_id=3670. [68] microdrop Technologies, Autodrop Professional Positioning System MD-P-802, http://www.microdrop.de/wDeutsch/products/AD-Positioning_System_e_0508.pdf, 2008. [69] H.C. Starck, “Product information CLEVIOS PH 1000,” http://www.hcstarck.com/index.php?page_id=3671. [70] F. Schönfeld, K. Graf, S. Hardt, and H. Butt, “Evaporation dynamics of sessile liquid drops in still air with constant contact radius,” International Journal of Heat and Mass Transfer,  vol. 51, Jul. 2008, pp. 3696-3699. [71] T.A. Skotheim and J.R. Reynolds, Conjugated Polymers: Theory, Synthesis, and Characterization, CRC Press, 2006. [72] F. Hsu, V.N. Prigodin, and A.J. Epstein, “Electric-field-controlled conductance of ``metallic'' polymers in a transistor structure,” Physical Review B (Condensed Matter and Materials Physics),  vol. 74, Dec. 2006, pp. 235219-12. [73] K. Omote, H. Ohigashi, and K. Koga, “Temperature dependence of elastic, dielectric, and piezoelectric properties of 'single crystalline' ’films of vinylidene fluoride trifluoroethylene copolymer,” Journal of Applied Physics,  vol. 81, 1997, p. 2760. [74] J. Madden, N. Vandesteeg, P. Anquetil, P. Madden, A. Takshi, R. Pytel, S. Lafontaine, P. Wieringa, and I. Hunter, “Artificial muscle technology: physical  111 principles and naval prospects,” Oceanic Engineering, IEEE Journal of,  vol. 29, 2004, pp. 706-728.   112 Appendices  Appendix A – Experimental procedures for PVDF spin coating The experimental procedure for preparing PVDF solutions consists of the following steps: (1) Into a vial, dissolve 30 wt% PVDF beads (Sigma-Aldrich 427152-250g) in DMF or DMSO; at weight loadings above 30 wt% PVDF is difficult to dissolve (2) Using a hotplate, stir the mixture at 400 RPM and heat at 100°C for 4 hours until a clear, dissolved PVDF solution of moderate viscosity is obtained.  The experimental procedure for single-layer spin coating consists of the following steps: (1) Pre-heat the vial of PVDF solution to, or maintain the solution at 100°C.  The mixture will transition from a gel to a liquid of moderate viscosity. (2) Set up the spinner to spin at 300 RPM for 60 seconds, with an acceleration rate of 220 RPM/sec. (3) Mount a 3” x 2” glass slide onto the spinner, quickly take the vial and pour about 1 mL of the solution onto the slide, place the vial back onto the hotplate, and start the spinner. (4) Transfer the PVDF solution-coated slide to the hotplate and soft-bake at 100°C for 2 minutes, before cooling the sample in ambient air.   113 Appendix B – Matlab script for calculating evaporation time % Gabriel Man % Sep 2009 % Calculate evaporation time of Clevios PH 750 PEDOT + 5 wt% DMSO droplet  % From F. Schönfeld, K. Graf, S. Hardt, and H. Butt, “Evaporation dynamics of sessile liquid drops in still air with constant contact radius,” %   International Journal of Heat and Mass Transfer,  vol. 51, Jul. 2008, pp. 3696-3699. % Using 2.3 Non-linear approximation because initial contact angle is > 30 deg  clc;  % constants GasConstant=8.31447; % units [J/(mol K)] LiquidDensity=(0.95*1000)+(0.05*1101); % units [kg/m^3], Clevios PH 750 density ~ 1 g/cm^3, density of DMSO 1.1010 g/cm^3 Delta=1.994; % dimensionless constant in model MolarMass=((0.05*78.13)+(0.95*18))/1000; % units [kg/mol], molar mass of DMSO is 78.13 g/mol, water is 18 g/mol ScalingFactor=1; % when = 1, using 30 um nozzle.  If using 20 um nozzle, assume drop radius and contact radius scaled by 2/3.  % semi-constants DropRadius=35e-6*ScalingFactor; % units [m], typically around 35 um DropVolume=(4/3)*pi*DropRadius^3; % units [m^3] ContactRadius=50e-6*ScalingFactor; % units [m], ranges 50-60 um for PEDOT/EG on pristine PVDF  RelHumidity=input('Relative humidity? (0.0-1.0) '); TempC=input('Temperature? (deg C) ');  % variables DiffusionCoefficient=21.2e-6*(1+0.0071*TempC); % units [m^2/s] DeltaPressure=(1-RelHumidity)*(610.7*10^((7.5*TempC)/(237.3+TempC))); % units [Pa]  EvaporationTime=(DropVolume*ContactRadius^2*LiquidDensity*GasConstant*(273+T empC))/(((Delta*ContactRadius^3)+(DropVolume/4))*(2*DiffusionCoefficient*MolarM ass*DeltaPressure)); disp(' '); disp('Evaporation time (seconds): '); disp(EvaporationTime);  114 Appendix C – Matlab script for calculating the RC charging time constant  The Matlab script below can be used to calculate the RC time constant of an input interdigital transducer of a SAW device.  The fingers of the interdigital transducer are treated as strip line transmission lines.  The user needs to input: • the width of a typical finger (i.e. 70 µm, the SAW wavelength would then be 280 µm) • length of an IDT finger in number of SAW wavelengths (i.e. 20) • number of fingers in the input transducer (i.e. 20) • excitation frequency of the SAW device (i.e. 3 MHz)   % Gabriel Man % Oct 2009, updated Dec 2009 % Using strip line transmission line model, ignoring L and G % From Table 18.1, Introductory Electromagnetics by % Popovic and Popovic  clc; disp('Given the geometry of the IDTs on a SAW sensor and using a transmission line model'); disp('with simplifying assumptions, this m-script calculates the expected SAW max operating'); disp('frequency'); disp(' ');  % constants EpsilonNought=8.85419e-12; % units [F/m] EpsilonRelative=13; % unitless, worst case scenario E field goes completely through PVDF Epsilon=EpsilonNought*EpsilonRelative; % units [F/m] ElectricalConductivity=1.5e4; % units [S/m]  , Clevios 1000 + 5wt% DMSO conductivity measured to be 150 S/cm MuNought=4e-7*pi; % units [N/A^2] Mu=MuNought;  % input variables TrackWidth=input('Track width? (um) ');  115 FingerLength=input('Length of an IDT finger in wavelengths? (integer, should be > 10) '); NumberOfFingers=input('Number of fingers in the IDT? (integer) '); OperatingFrequency=input('Operating frequency of SAW device? (in MHz) '); disp(' ');  % calculate variables Omega=2e6*pi*OperatingFrequency; TrackWidth=TrackWidth*1e-6; SawWavelength=4*TrackWidth; FingerLength=FingerLength*SawWavelength; CapPerLength=Epsilon*(FingerLength/TrackWidth); Rs=sqrt((Omega*Mu)/(2*ElectricalConductivity)); RPerLength=2*Rs; CPerLength=CapPerLength*FingerLength;  % intermediate output disp('Resistance per finger (ohms)'); disp(RPerLength); disp('Capacitance per finger (Farads)'); disp(CPerLength);  % calculate impedance of one finger ImpedancePerFinger=(RPerLength)+(1/(i*Omega*CPerLength)); ZTotal=((NumberOfFingers*2)/ImpedancePerFinger)^-1; CTotal=abs(1/(i*Omega*imag(ZTotal))); disp('Total capacitance (Farads)'); disp(CTotal); disp('Total resistance (ohms)'); disp(real(ZTotal)); disp('RC constant'); disp(real(ZTotal)*CTotal); disp('Max expected operating frequency (Hz)'); disp(1/(2*pi*real(ZTotal)*CTotal));  116 Appendix D – Autodrop macros for multi-layer patterning  The macro below can be used to create a pattern consisting of 7 parallel lines of length 36.96 mm each (336 points spaced 0.11 mm apart), offsetted in the y-axis, using the Microdrop inkjet micropatterning station.  The fabrication of uniform, electrically connected tracks requires depositing 4 layers, with sufficient drying time (i.e. 3-5 minutes) between layers.  Use of the ventilation system mounted onto the cabinet of the Microdrop system is recommended.  The macro was designed to be used with the 40 µm inkjet nozzle and an aqueous PEDOT:PSS printing solution of Clevios 1000 + 5 wt% DMSO.  The optimal plasma treatment for the 40 µm nozzle is 10 seconds at low power, and the optimal drop spacing was experimentally determined to be 0.11 mm.  The following steps should be taken prior to microprinting. (1) Check that the PVDF film is clamped securely to the stage, so that it does not shift as the stage moves.  A reliable method for clamping the PVDF film is attaching a steel plate to the stage using double-sided tape, and using strong magnets at the edge of the film to clamp it to the plate. (2) Very important: ensure the z-axis setting of each dispense command does not cause the inkjet nozzle to come into physical contact with the sample and/or stage. (3) Set appropriate stage velocity and acceleration parameters via the XYZ stage dialog in Autodrop.  Typical values are vx= vy = 100 mm/s, vz = 80 mm/s, ax = ay = 90 mm/s2, az = 120 mm/s2.  117 (4) Obtain stable droplets by setting the droplet ejection parameters while in the camera position.  Unless the nozzle is actively patterning a sample, it should be constantly printing in the camera position, otherwise nozzle tip clogging will occur quickly (5 to 20 seconds). (5) Open the Dosing options dialog, go to the In Flight Dosing tab, the In Flight Dosing option should be checked and the X and Y stage speeds should be set.  Typical values are vx = vy = 16 mm/s.  These stage velocity settings are applicable when a pattern is being printed, the velocity settings in step 3 are for general stage movements. (6) Check that the trigger signal output of the stage controller unit is connected to the trigger input of the appropriate AD-E-110 dispenser driver.  After printing, the macros are designed to bring the nozzle back to the camera position stored for nozzle 2 (kam, 2) so that droplet ejection in the camera position can be quickly started to avoid clogging.  To execute the macros, do a copy-and-paste into the macro editor of the Autodrop software.   First layer – discontinuous track  set,dis,2,92.5,66.5,21.1,0.11,0.11,336,1,0,2,1,1 set,pup,0 dis set,dis,2,92.5,74.5,21.1,0.11,0.11,336,1,0,2,1,1 dis set,dis,2,92.5,85.5,21.1,0.11,0.11,336,1,0,2,1,1 dis set,dis,2,92.5,90.5,21.1,0.11,0.11,336,1,0,2,1,1 dis set,dis,2,92.5,95.5,21.1,0.11,0.11,336,1,0,2,1,1 dis set,dis,2,92.5,104.5,21.1,0.11,0.11,336,1,0,2,1,1 dis set,dis,2,92.5,112.5,21.1,0.11,0.11,336,1,0,2,1,1 dis  118  set,pup,1 kam,2   Second layer – alternate overlay 1  set,dis,2,92.445,66.5,21.1,0.22,0.22,168,1,0,2,1,1 set,pup,0 dis set,dis,2,92.445,74.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.445,85.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.445,90.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.445,95.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.445,104.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.445,112.5,21.1,0.22,0.22,168,1,0,2,1,1 dis  set,pup,1 kam,2  Third layer – alternate overlay 2  set,dis,2,92.555,66.5,21.1,0.22,0.22,168,1,0,2,1,1 set,pup,0 dis set,dis,2,92.555,74.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.555,85.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.555,90.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.555,95.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.555,104.5,21.1,0.22,0.22,168,1,0,2,1,1 dis set,dis,2,92.555,112.5,21.1,0.22,0.22,168,1,0,2,1,1 dis  set,pup,1 kam,2  Fourth layer – final layer, half of the drop spacing of the first layer, double the drops  set,dis,2,92.5,66.5,21.1,0.055,0.055,672,1,0,2,1,1 set,pup,0 dis set,dis,2,92.5,74.5,21.1,0.055,0.055,672,1,0,2,1,1 dis set,dis,2,92.5,85.5,21.1,0.055,0.055,672,1,0,2,1,1  119 dis set,dis,2,92.5,90.5,21.1,0.055,0.055,672,1,0,2,1,1 dis set,dis,2,92.5,95.5,21.1,0.055,0.055,672,1,0,2,1,1 dis set,dis,2,92.5,104.5,21.1,0.055,0.055,672,1,0,2,1,1 dis set,dis,2,92.5,112.5,21.1,0.055,0.055,672,1,0,2,1,1 dis  set,pup,1 kam,2 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0068869/manifest

Comment

Related Items