Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Evaporation cast thin film carbon nanotube strain gauges Au, Daniel Tak Yin 2013

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

Item Metadata


24-ubc_2013_fall_au_daniel.pdf [ 3.08MB ]
JSON: 24-1.0074118.json
JSON-LD: 24-1.0074118-ld.json
RDF/XML (Pretty): 24-1.0074118-rdf.xml
RDF/JSON: 24-1.0074118-rdf.json
Turtle: 24-1.0074118-turtle.txt
N-Triples: 24-1.0074118-rdf-ntriples.txt
Original Record: 24-1.0074118-source.json
Full Text

Full Text

i  Evaporation Cast Thin Film Carbon Nanotube Strain Gauges by Daniel Tak Yin Au B.ASc., The University of Waterloo, 2011  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE in The Faculty of Graduate and Postdoctoral Studies (Electrical Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) August 2013  ? Daniel Tak Yin Au, 2013 ii  Abstract This work describes the research performed on synthesising and measuring the gauge factor of evaporation cast thin film carbon nanotube strain gauges. The main characteristics pursued of the strain gauges are inexpensive, easily manufactured and reasonably sensitive. Carbon nanotubes have exhibited a high gauge factor due to their intrinsic piezoresistivity and were incorporated into evaporation cast films to try to take advantage of the high sensitivity. Another direction taken to improve the sensitivity is alignment of carbon nanotubes in the thin film. Previous work produced an evaporation cast carbon nanotube strain gauge with a relatively high gauge factor. However, it was not reproducible and the research encompassed extends from the previous work. A number of ink compositions with different carbon nanotube and surfactant loadings were used to synthesise thin films of carbon nanotubes on a polyimide substrate. Variations of evaporation casting were used to decrease the evaporation rate in attempts of carbon nanotube alignment through a self-organising liquid crystal phase during evaporation. Other methods of inkjet printing and air flow evaporation casting were also attempted to achieve alignment. Electrical connections using a conductive polymer and metal wires were fabricated onto the samples for electrical measurements. A four-point probe resistance measurement under the application of strain was used to elicit the gauge factors. The strain gauge design was modified from previous work for more reliable electrical connections and for higher applied strains. A procedure for electrical measurements coupled with the application of strain was devised and the gauge factors achieved varied between 0.1 and 4.0 with a median of 1.1 ?0.1. The median gauge factor was reproducible and exhibited by several samples fabricated with different types of evaporation casting. The decrease in evaporation rate  iii  did not result in either alignment or relatively high gauge factors. In general, alignment was not achieved with the other methods of air flow evaporation and inkjet printing.      iv  Preface This work is based on and is a continuation of unpublished work by Dan Sik Yoo and includes many theoretical contributions from Simon Beyer and Dr. Konrad Walus.    v  Table of Contents  Abstract ........................................................................................................................................... ii Preface............................................................................................................................................ iv Table of Contents ............................................................................................................................ v List of Tables ................................................................................................................................ vii List of Figures .............................................................................................................................. viii Acknowledgements ......................................................................................................................... x Dedication ...................................................................................................................................... xi 1 Introduction ............................................................................................................................. 1 1.1 Strain and strain gauges.................................................................................................... 1 1.2 Geometric resistance effects, piezoresistivity and gauge factor ...................................... 2 1.3 Gauge factor of carbon nanotubes .................................................................................... 4 1.4 Mechanisms of resistance change in CNT composites .................................................... 9 1.5 Existing CNT strain gauges.............................................................................................. 9 1.6 Goals and motivation ..................................................................................................... 14 1.7 Previous accomplishments ............................................................................................. 15 1.8 Contributions .................................................................................................................. 16 1.9 Chapter overview ........................................................................................................... 18 2 Fabrication Procedures .......................................................................................................... 19 2.1 Ink preparation and substrates ........................................................................................ 19 2.2 CNT film synthesis......................................................................................................... 20 2.3 Electrical connections .................................................................................................... 23 2.4 PEDOT deposition ......................................................................................................... 24 3 Analytical Procedures ............................................................................................................ 26  vi  3.1 Polarized light and scanning electron microscopy ......................................................... 26 3.2 Gauge factor measurement ............................................................................................. 28 4 Results and Discussion .......................................................................................................... 32 4.1 CNT ink and substrate .................................................................................................... 32 4.2 CNT film fabrication ...................................................................................................... 36 4.3 Reproducing results of previous work ........................................................................... 43 4.4 Imaging (PLM and SEM) for alignment ........................................................................ 43 4.5 Gauge factor measurements ........................................................................................... 47 4.6 Initial resistances ............................................................................................................ 58 4.7 Influence of individual CNT piezoresistivity ................................................................. 61 5 Conclusion ............................................................................................................................. 66 5.1 Overview ........................................................................................................................ 66 5.2 Recommendations for future work ................................................................................. 68 Bibliography ................................................................................................................................. 70 Appendix ....................................................................................................................................... 75     vii  List of Tables Table 4.1.1: Gauge factors achieved with each ink composition ................................................. 34 Table 4.2.1: Gauge factors achieved with different evaporation casting (EC) methods .............. 40 Table 4.5.1: Summary of gauge factors of fabricated CNT film strain gauges ............................ 57 Table 4.7.1: Estimated values for hypothetical resistances of strain gauges ................................ 63     viii  List of Figures Figure 1.3.1: Indirect contact method for straining a carbon nanotube .......................................... 6 Figure 1.5.1: Diagram of the sensing elements............................................................................. 10 Figure 1.5.2: Image of the CNT forest strain gauge ..................................................................... 11 Figure 1.5.3: SEM image of an aerosol sprayed SWCNT thin film strain gauge......................... 12 Figure 1.7.1: Basic strain gauge design ........................................................................................ 16 Figure 2.2.1: Image of the sealed sample submerged in a waterbath ........................................... 22 Figure 2.4.1: PEDOT ink deposition images ................................................................................ 25 Figure 3.1.1: Reference reflective PLM image of CNTs .............................................................. 27 Figure 3.2.1: Sample CNT film with electrical connections ........................................................ 29 Figure 4.1.1: Plot of gauge factor measurements with their corresponding strain ....................... 33 Figure 4.1.2: Graph of displacement vs time of the two different samples .................................. 35 Figure 4.1.3: Sample showing unpinned contact line and non-uniform evaporation ................... 36 Figure 4.2.1: Inkjet printing CNT films exhibiting challenges of matching evaporation............. 38 Figure 4.2.2: Air flow cast samples .............................................................................................. 39 Figure 4.2.3: Plot of a strain measurement performed on a blank substrate ................................. 42 Figure 4.2.4: Plot of a strain measurement performed on a complete device ............................... 42 Figure 4.4.1: Reference set of reflective PLM images of an inkjet printed CNT film ................. 44 Figure 4.4.2: Reflective PLM image comparison 1 ...................................................................... 45 Figure 4.4.3: Reflective PLM image comparison 2. ..................................................................... 45 Figure 4.4.4: SEM images of an air flow cast sample .................................................................. 46 Figure 4.4.5: Reference set of SEM images of an inkjet printed CNT film ................................. 46 Figure 4.5.1: Graph of displacement and load versus time........................................................... 48  ix  Figure 4.5.2: Graph of resistance versus time for determining the base resistance ...................... 50 Figure 4.5.3: Graph of relative change in resistance versus strain ............................................... 52 Figure 4.5.4: Plot of gauge factor measurements with their corresponding strain 1 .................... 53 Figure 4.5.5: Plot of gauge factor measurements with their corresponding strain 2 .................... 55 Figure 4.6.1: Plots of resistance values of various samples 1 ....................................................... 59 Figure 4.6.2: Plot of resistance values of various samples 2 ........................................................ 60      x  Acknowledgements I am grateful to the faculty, staff and fellow students who have provided countless advice, motivation and inspiration. I owe particular thanks to my two supervising professors, Dr. Edmond Cretu and Dr. John Madden, for their contributions in knowledge, direction and most importantly, time. I would also like to thank Dr. Konrad Walus for the overall project opportunity and the NSERC Strategic grant which helped to fund my research and graduate education. Particular colleagues I would like to thank include John Berring, Simon Beyer, Ron Linklater, Seyed Mohammad Mirvakili and Mirza Saquib for their various advice and contributions to my research as well as their friendship throughout my time at the University of British Columbia. Special thanks to my parents for their financial support and to my friends for their emotional support throughout my years of education.    xi  Dedication   For my parents, family, friends and supporters  1  1 Introduction An important factor often overlooked or taken granted for in day to day life is the structural properties of an object such as bridges, skyscrapers or even automobiles. One such property is the strain of a structural element like a beam or a cross section of a frame. There are various methods to measure strain and a low impact strain gauge using novel materials is encompassed in this thesis. 1.1 Strain and strain gauges Strain is the term used to measure the deformation of an object in dimensions due to an external or internal force. A positive change in dimensions is labelled as tensile strain and a negative change in dimensions is referred to as compressive strain. The forces inducing strain can vary from applied external forces, bodily forces or internal forces. In the case of piezoelectric materials, mechanical deformation is induced by an electric force. An example of strain is the deformation of skin when a person smiles. The skin is stretched or compressed due to deformation of the facial structure leading to tensile and compressive strain of their skin. Strain is present in everyday objects but is mostly neglected. However, in situations where stress in an object is important, it is useful determine the strain. For example, knowing the strains of aircraft parts can provide information about structural integrity which can be used as preventative measures and decrease the chances of accidents [1]. However, comprehensive strain data would require an abundance of strain gauges and this fuels the pursuit of designing an easily manufactured strain gauge of low cost. Strain gauges are devices used to measure strain. A basic strain gauge consists of a zigzag pattern of metal attached to an insulating substrate; a metal foil strain gauge. As the substrate is  2  deformed, the metal lines in the pattern will stretch or compress following the deformation. This stretching and compression of the metal will alter its atomic structure thereby affecting its electrical resistance. The change of resistance is correlated to the amount of strain the substrate experienced. Through this correlation, the strain can be determined by measuring the change in resistance with external electronics. Metal foil strain gauges are relatively low cost compared to semiconductor strain gauges. They are also accurate over wide static and dynamic ranges but are limited in performance including a large temperature dependence, low resistance and a low gauge factor [2]. A key importance in strain gauges is using an electronically conductive material whose resistance will change greatly due to mechanical strain. 1.2 Geometric resistance effects, piezoresistivity and gauge factor The effect of electrical resistance changing due to a mechanical deformation is called the piezoresistive effect where piezo is based on the Greek word piezein meaning to squeeze or press. The resistance of a rectangular material is represented by [3]:              (   )  where    is electrical resistance,    is resistivity of the material and     and   are length, width and thickness respectively. If the rectangular material is stretched, the relative change in resistance can be represented by the change in dimensions as well as resistivity [3]:                          (   )   3  The change in thickness and width can be related to the change in length by Poisson?s ratio,   assuming the geometric changes are isotropic [3]:                   (   )  If we consider strain of the material along the length of the material:           (   )  we can rewrite Equation 1.2 as follows:     ?           ?      (   ) Considering Equation 1.5, the first two terms (    ) represent the resistance change from geometrical changes. This is the dominant effect for metal foil strain gauges where the intrinsic resistance is low. The third term (     ) represents the resistance change from the change in resistivity of the material and this is the dominant effect for semiconducting materials. The resistivity of the material changes from strain because the strain on the atomic structure affects the band gaps and band energy levels thereby altering the intrinsic resistivity [3], [4]. Note that the above explanation is a generalized picture and does not consider composite materials where other possible mechanisms of resistance change due to deformation such as electrical tunnelling are possible. For strain gauges, it is desirable to maximize the relative change in resistance due to strain and therefore the sensitivity. The gauge factor (GF) is used to represent the sensitivity of piezoresistance and is typically represented as [5?7]:  4         ?      (   )  A higher GF represents a greater magnitude of resistance change due to strain and vice versa. It follows that a good strain gauge would have a high gauge factor and the piezoresistive material is the key determinant. The conventional metal foil strain gauges exhibit a low GF ranging from 2 to 5 [2], [5], [7?11]. This low GF among other limitations of metal foil strain gauges lead to the pursuit of developing strain gauges with other piezoresistive materials such as silicon and carbon nanotubes. Development with doped silicon strain gauges has exhibited GFs of approximately 200 [12], [13]. On the other hand, experiments with carbon nanotubes (CNTs) have shown intrinsic GFs of up to 2900 and relative resistance changes of up to two orders in magnitude [14], [15]. Even graphene has been used in the development of a strain gauge with a GF of approximately 300 [16]. As such, CNTs and graphene are attractive materials for development of new types of strain gauges. 1.3 Gauge factor of carbon nanotubes Carbon nanotubes were first discovered in 1991 and have been determined to possess highly desirable material properties for different applications in general. CNTs can be imagined by ?rolling up? a piece of graphene and depending on the angle that it is ?rolled in?, the CNT can exhibit different properties. The chirality and diameter of CNTs largely affects its material properties and of particular importance, its electrical properties. CNTs can exhibit metallic, quasi-metallic (small band gap) or semi-conducting electronic behaviour depending on its chirality. They have also exhibited superconductivity when they have tiny diameters. Another important classification of CNTs is whether they are multi-walled or single-walled. This refers to  5  the number of layers the CNT consists of. For example, if there is more than one layer, it is considered a multi-walled CNT.  Aside from useful electronic properties, CNTs also exhibit high durability through mechanical properties such as high capacity for reversible tensile or compressive strain and high yield strength among others. These desirable properties lead to the attempt in adopting CNTs into transistors for their electrical efficiency or to material composites for their mechanical attributes [17]. Other than sensing strain, there are developments aimed at incorporating CNTs to sense gases, temperature, pressure, light and pH [18]. In one experiment, a metallic single-walled CNT (SWCNT) was isolated and used to bridge a microfabricated trench on a silicon substrate. The ends of the SWCNT were attached to electrodes for electrical measurements and experimentally deemed to be immobile. An atomic force microscope (AFM) tip was used to push the SWCNT thereby bending the SWCNT and straining it. The literature reported a global strain value of 3% with a relative change in resistance of approximately two orders in magnitude. The initial resistance of the metallic SWCNT was reported to be approximately 200 k? [15]. With a simple calculation shown in Equation 1.7, this represents a gauge factor of approximately 3300 for an isolated SWCNT:                 ?            (   )  Due to the use of an AFM tip to probe the SWCNT, it has been suggested that the high gauge factor could be the result of a local deformation of the nanotube due to tip contact as opposed to the result of uniform strain throughout the SWCNT [13], [15], [19].  6  This experiment led to more scientific investigations involving a more uniform strain inducement as well as theoretical modelling to determine the basis of the piezoresistive nature of CNTs. Their scheme for inducing strain in the CNT is shown in Figure 1.3.1 below. One such experiment focused on the differences between three types of SWCNTs; metallic, quasi-metallic and semi-conducting (SC-CNT). Quasi-metallic is used to describe a small band gap semi-conducting CNT (SGS-CNT). This experiment used a SWCNT to connect the free end of a cantilever to the other side of the trench which isolated the cantilever. By probing the cantilever with an AFM tip, the SWCNT was uniformly strained thereby avoiding local deformations or interactions. The GFs of the metallic CNT, SGS-CNT and SC-CNT are in the ranges of 40 to 60, 600 to 1000 and 150 respectively [13].  Figure 1.3.1: Indirect contact method for straining a carbon nanotube resulting in a more uniform strain inducement [13]. *Reprinted with permission from J. Cao, Q. Wang, H. Dai, ?Electromechanical Properties of Metallic, Quasimetallic, and Semiconducting Carbon Nanotubes under Stretching?, Physical Review Letters, Volume 90, Page 157601, 2003. Copyright 2013 by the American Physical Society. The idea of using a microfabricated cantilever to investigate CNT GFs was also used by another group. They fabricated a cantilever positioned perpendicular to and over a suspended CNT. The nanotube was strained by using an AFM tip to push down on the cantilever. The authors mentioned this set-up could also be affected by the local deformations of the CNT located at the  7  edges of the cantilever but they stressed the agreement of their experimental results with theoretical simulations and also previous data from strained CNTs that did not involve local deformations. This group used a metallic SWCNT with an initial resistance of approximately 315 k? which exhibited a GF as high as 2900 [14]. The mechanism behind the piezoresistive nature of CNTs has been discussed to be a result of the changes in band gaps and band energy levels due to the strain induced on the CNT. The strain changes the atomic lattice of the CNT, for example changing the inter-atomic distances. This affects the overlapping of electron orbitals which determine the band energy levels and band gaps. This change in band gap and band energy levels determines the intrinsic resistivity of the CNT. It follows that a change in atomic lattice from strain induces a change in resistivity of the CNT [4], [14], [19?23]. One theoretical investigation concluded that a compressive strain leads to a decrease in band gap and therefore an increase in conductivity and vice versa [4]. Another theoretical investigation showed the intrinsic resistivity of CNTs is based on their band gap which is based on their chirality. Therefore, the intrinsic resistivity of CNTs of different chiralities can be determined theoretically. Furthermore, the theoretical gauge factor of CNTs of different chiralities can be determined with the incorporation of deformation in the atomic lattice of CNTs. The general trends reported are that zigzag CNTs, chiral indices of (n, 0), exhibit the maximum magnitude gauge factors and decrease with increasing chiral angle until the gauge factor reaches zero with armchair CNTs, chiral indices of (n, n). It was also shown that negative gauge factors can exist depending on the chirality of the CNTs [12], [23]. Overall, the piezoresistivity of CNTs have been shown to exhibit a very high sensitivity in some experiments which could be due to local deformations. When the local deformations are accounted for or removed from the experimental set-up, the sensitivity is still relatively high for  8  quasi-metallic CNTs decreasing with semiconducting CNTs and lowest with metallic CNTs. The fact that the gauge factor of CNTs can be theoretically determined by their chiral indices makes filtering of CNTs for use in strain gauges a promising direction. However, the work in this thesis uses SWCNTs of varying chiralities and considering the possibility of CNTs with a negative gauge factor and the fact that electrical conduction tends to take the path of least resistance, a mixture of chiralities might restrict the achievements of CNT thin films with high positive gauge factors.  There are three main methods to synthesize CNTs and they include arc discharge, laser ablation and chemical vapour deposition (CVD). The last method, CVD, is the most favourable due to its ability to grow CNTs on a desired substrate as well as its scale up potential for industrial synthesis [17]. As of the time of writing, the cost of CVD grown SWCNT from a company called Cheap Tubes Inc. (, one of the lowest cost providers, ranges from 110 USD to 300 USD per gram depending on purity and functionalization. Although this cost is high compared to bulk metals and polymers (one thousand to ten thousand times lower), the CNTs can be considered inexpensive due to the low weight fraction of CNTs used per strain gauge. For example, approximately 40 ?g of SWCNTs are used in sensors described in this work, corresponding to a range of costs between 0.44 to 1.2 US cents per device calculated using the previously given SWCNT costs from Cheap Tubes Inc.. A typical metal foil strain gauge sells for approximately 6.26 to 11.76 USD per device depending on bulk price breaks from Digikey ( Alternatively, semiconductor based strain gauges cost in the range of 8 to 136 USD per device depending on design from Micron Instruments ( The low cost and availability of CNTs along with their high GFs position them as a promising piezoresistive material for strain gauges.  9  1.4 Mechanisms of resistance change in CNT composites The resistance change mechanisms in CNT films may be different from CNT composites but the mechanisms of resistance change in CNT composites discussed in literature can provide a perspective of mechanisms at work in CNT films. For CNT composites composed of a network of randomly oriented CNTs, literature refers to three major mechanisms which affect the change in resistance. These three major mechanisms are a change in tunneling between neighbouring CNTs, a change in the internal conductive network formed by CNTs and the CNT?s inherent piezoresistivity [24]. The mechanisms were considered in numerical simulations and it was determined that the CNT?s piezoresistivity has the least influence (4.85% influence when the intrinsic CNT gauge factor is assumed to be approximately 2.1) on the change in resistance due to strain [24]. It was also suggested based on the simulation results that the influence of CNT piezoresistivity on the change in resistance is limited even if the intrinsic CNT piezoresistivity is increased. This was investigated by artificially amplifying the piezoresistivity of CNTs in the simulation. The results of the amplification (intrinsic CNT gauge factor assumed to be approximately 21) showed the influence of CNT piezoresistivity reaching a limit of 8.88% with varying CNT weight fractions and at 6,000 ?? [24]. The other 91.2% of change in resistance is a result of the other two mechanisms. Note that the assumed intrinsic CNT gauge factors of approximately 2.1 or 21 are low compared to the reported intrinsic CNT gauge factors in literature of up to 1000 for quasi-metallic CNTs [13]. Also, the literature refers to CNT composites which are different to the CNT films referred to in this work. 1.5 Existing CNT strain gauges One of the highest GFs exhibited by a strain gauge incorporating CNTs has a form factor that is most impractical relative to the thesis goals in terms of simplicity of design and ease of  10  fabrication. Instead of a thin film, the architecture relies on connecting electrodes and bridging cantilevers with individual SWCNTs and can only make very short devices. The sensor fabrication requires the use of microfabrication techniques such as photolithography and wet etching to make several cantilever beams of varying lengths. A CVD technique is used to grow a CNT to bridge the pre-fabricated cantilevers allowing for a single CNT to be used as the piezoresistive material. The architecture is shown below in Figure 1.5.1. This sensor utilizing the piezoresistivity of individual SWCNTs reported a GF of 744. This value is within the range of reported quasi-metallic SWCNTs gauge factors (600 to 1000) mentioned previously in section 1.3. This device outperforms other strain gauges using CNTs, metal foils and even silicon thereby exemplifying the potential of CNT strain gauges [10].   Figure 1.5.1: Diagram of the sensing elements which take advantage of custom grown carbon nanotubes (1-electrical connections, 2-short cantilever beam, 3-long cantilever beam, 4-carbon nanotubes, 5-Silicon base, 6-CNT growth catalyst, and 7-Silicon dioxide thin film) [10]. *Reprinted from Sensors and Actuators A: Physical, Volume 176, C. Su, T. Liu, N. Chang et al., ?Two dimensional carbon nanotube based strain sensor?, Pages 124-129, 2012 with permission from Elsevier. Two other form factors of CNT strain gauges include a CNT forest embedded on the surface of a parylene film and a high pressure compressed pure CNT pellet. The CNT forest strain gauge also requires microfabrication techniques and the CNT forest is also grown using a CVD technique. This device is shown in Figure 1.5.2. It includes a suspended CNT forest supported by a parylene  11  membrane which is strained. The electric conduction occurs through the CNT forest and strain on the membrane also strains the CNT forest leading to a change in the conductance. The GF achieved by this device was reported to be 4.52 [25]. The pure CNT pellet strain gauge uses high pressures of 200 MPa and 300 MPa to compress pure CNT powder into a pellet of size 10?5?1 mm (Length?Width?Height). The tablet is then glued onto an elastic beam for strain transference. It follows that strain on the pellet will affect the conductive paths formed by the compressed CNTs in the pellet. This device exhibited initial resistances ranging from 6 to 9 ? and GFs ranging from 50-80 at strains of up to 3000 ? [26]. The CNT pellet strain gauge is promising as an easily fabricated strain gauge but it requires too much CNT to synthesize thereby increasing the cost per device.  Figure 1.5.2: Image of the CNT forest strain gauge accompanied with SEM images of the CNT forest [25]. *Reprinted with permission from A. Bsoul, M. Sultan Mohamed Ali, A. Nojeh et al., ?Piezoresistive strain sensing using carbon nanotube forests suspended by Parylene-C membranes?, Applied Physics Letters, Volume 100, Page 213510, 2012. Copyright 2012 American Institute of Physics. In spite of some good GFs and novel techniques of developing CNT strain gauges, none of the above fulfils the criteria this thesis seeks which are ease of fabrication, high sensitivity and inexpensive costs. Compressing a large amount of pure CNTs into a pellet requires too much CNT and would be expensive and not an effective use of CNTs. Using microfabrication  12  techniques is reasonable due to the existing infrastructure in other industries but it is not as simple as evaporative casting or printing. There are CNT strain gauges which utilize such fabrication methods and will be discussed below.  One group fabricated devices using aerosol sprayed SWCNTs resulting in a thin film of SWCNTs in a random orientation. However, this group?s fabrication is relatively complex and requires photolithography and wet etching to etch their SWCNT film to achieve their sensing lines as shown in Figure 1.5.3 below.   Figure 1.5.3: SEM image of an aerosol sprayed SWCNT thin film strain gauge device utilizing photolithography and wet etching [6]. *Reprinted from Sensors and Actuators A: Physical, Volume 180, D. Lee, H. Hong, M. Lee et al., ?A prototype high sensitivity load cell using single walled carbon nanotube strain gauges?, Pages 120-126, 2012 with permission from Elsevier. This device was reported to attain GFs of approximately 60 for both tensile and compressive strain [6]. Another group using layer by layer aerosol sprayed SWCNTs achieved GFs ranging from 1.44 to 5.05 using strains of up to 40,000 ??. Some of the differences which could contribute to the huge variation in GF for devices synthesized using aerosol sprayed SWCNTs include an absence of sensing patterns, alternate method of SWCNT suspension in solution, different substrate, different thin film thickness and different printing procedures [27]. Another device that also used aerosol spray directly printed their sensing lines without requiring etching  13  but this device used MWCNTs. The reported GF was a relatively low 0.71 which could be attributed to the use of MWCNTs or the concentration of CNTs used amongst other factors mentioned previously [28]. Aerosol printing is a fast and easy synthesis technique but evaporative casting is a method that does not require printing instruments or air flow.  Another group used a biopolymer to disperse and suspend their CNTs to make evaporative cast films and also bucky papers which are typically pure CNT films prepared using vacuum filtration techniques. They investigated conductivity of the films but did not perform any gauge factor measurements. They suggest that the electrical properties of these films are reliant on the CNT to polymer volume fraction and that the mechanical properties of the film are not heavily affected by the CNT to polymer volume fraction and instead reliant on the total mass of the film [29]. One group which fabricated a strain gauge using Bucky paper made of pure 100% CNT was able to achieve a GF of 7 but this device showed non-linear behaviour above 500 ?? [30].  In terms of strain gauges synthesized without the use of microfabrication techniques, there are aerosol printed CNT strain gauges, filtered Bucky paper strain gauges and evaporation cast films using biopolymers. These are all relatively simple manufacturing processes and have exhibited gauge factors ranging as low as 0.71 up through to 7. The literature most similar to the work in this thesis is the evaporation cast CNT using bio-polymers as the surfactant. These devices also seem to fit the criteria desired in this project but the paper did not report gauge factors. A couple differences between these works in literature and the work in this thesis are the method of evaporation casting and attempts on aligned CNT films. This past work suggests that further development of SWCNT strain gauges with the goals of simple fabrication methods and high sensitivity over a wide strain range is an appropriate objective.  14  1.6 Goals and motivation A couple of the often discussed applications of CNT strain gauges are for biological monitoring and structural health monitoring (SHM). Some strain gauges are being developed to attach to the top of the skin to monitor joint motion, tissue swelling or emotional expression [5], [31]. Another is being developed for monitoring bone strains as a part of post-surgical therapy [8]. Structural health monitoring is envisioned as the formation of a network of sensors to monitor structural integrity in real time. This can be applied to aircraft parts as mentioned before or for bridges, pipelines and any other structures where structural integrity is important [1], [32?34]. To fulfill this vision, the goal is to develop a CNT strain gauge that is easily manufactured, inexpensive, and able to function over a wide range of strains depending on the application.  The conventional metal foil strain gauges fulfill two of these criteria but have their own relative shortcomings including low GF, large device size, and unfavourable sensitivity to temperature change. MWCNTs have been shown to increase their GF as temperature increases while other piezoresistive materials such as polysilicon have a decreasing GF with increasing temperature [35], [36]. For biological implantation, metal foil strain gauges are difficult to implant inside the body and are not biocompatible [8]. When implementing strain gauges in large numbers for structural health monitoring, installation becomes cumbersome, costly and labor intensive. Difficulty in installation leads to the motivation of strain sensing via thin film sensors [28]. Sensitivity of CNTs is another motivational factor for developing thin film CNT strain gauges. As mentioned previously, conventional metal foil strain gauges have a GF from 2 to 5 while it has been demonstrated that a single SWCNT can exhibit a gauge factor up to 2900. It is a goal to incorporate this level of sensitivity into a thin film CNT strain gauge. Other CNT strain gauges  15  in literature have reported GFs as low as 0.71 and as high as 744 depending on the strain gauge form factor and type of CNT used [10], [28]. It is possible to suspend CNTs in solution and use them as an ink to be evaporation cast or printed. A common method of making CNT suspensions is to use a surfactant such as sodium dodecyl sulphate to suspend the CNT in water [27]. Since the solvent is water, the ink is not volatile or dangerous to use. The liquid CNT ink can be easily used for evaporative casting or printing onto a substrate. The safety and ease of CNT ink handling along with evaporative casting or printing enables high throughput and simple manufacturing methods. This contributes towards the motivation for developing CNT strain gauges.  1.7 Previous accomplishments This thesis is a continuation of existing work that was started but discontinued due to unique and difficult circumstances of the previous student. A short summary of the previous accomplishments this thesis launches from includes the development of a basic strain gauge design used for determining the GF, the investigation of two different types of surfactants and the achievement of GFs ranging from 2 to as high as approximately 30. Shown below in Figure 1.7.1 is an example of the strain gauge design used for measuring GFs. In the below setup, CNT ink is evaporation cast onto a polyimide substrate followed by deposition and attachment of the electrical connections. The electrical connections involve the evaporation casting of a conductive polymer, PEDOT:PSS (PEDOT), followed by the adhesion of metal wires onto the PEDOT using a conductive metal epoxy. The device is supported with polyimide tape at the connections and also with glued sandpaper at the ends to improve gripping of the substrate for measurements.  16   Figure 1.7.1: Basic strain gauge design showing a strip of evaporation cast CNT down the middle of a polyimide substrate. PEDOT, silver epoxy and tape is used for the electrical connections to the CNT film for electrical measurements. Sandpaper at glued to the ends of the substrate is also shown. The surfactants previously investigated to suspend CNTs in water were sodium cholate (SC) and sodium dodecyl sulphate (SDS). Sodium dodecyl sulphate was suggested as a superior surfactant in terms of suspension stability, and hydrophilic interactions with the polyimide substrate. The reported GFs of devices fabricated with SDS were superior to those with SC. The SC devices exhibited GFs ranging between 2 to 3 while the SDS devices exhibited GFs ranging between 1 and 29.8. However, the GF of 29.8 could not be reproduced and the same sample exhibited a GF of 19.3 in a second measurement which is approximately a 33% decrease in sensitivity. The achievements of the previous student set the foundation for the work in this thesis and a general overview of the continued work is given in the following section. 1.8 Contributions As previously mentioned, the goals are to develop an easily manufactured, sensitive and low cost strain gauge using CNT thin films. To achieve this, one of the main directions was to incorporate the high sensitivity from the intrinsic piezoresistivity of individual CNTs into a macro CNT film effect. At first, the strain gauge design was improved with various modifications including an  17  overhaul of the electrical connections to the CNT film. This can be seen in Figure 3.2.1. The previous design had some faults that were addressed and details are discussed in a later chapter. At the same time, reliable and detailed methods for strain gauge fabrication and electrical measurement for determining the gauge factors were established. The strain gauges synthesised to replicate the relatively high gauge factor exhibited in the previous work was unsuccessful but the gauge factors determined were within the range of gauge factors achieved in previous work. The attempt to replicate the relatively high gauge factor was followed by many experiments consisting of adjusting the CNT ink composition and trying different methods of fabrication to take advantage of the individual CNT sensitivity. These results are shown in Table 4.5.1. The fabrication methods of evaporation casting were modified in order to decrease the evaporation rate as advised by the previous researcher. Overall, gauge factors between 1.0 and 1.2 at a strain of 5000 ?? were reproduced throughout several devices with a couple devices achieving a higher gauge factor of 4.0 at higher strains. Also, during the time of the previous work, another student was developing a technique for CNT alignment with inkjet printing [37]. Thus, the development of an aligned CNT film strain gauge was also attempted. For this, aerosol evaporation casting method was used in place of the inkjet method due to complications using the custom inkjet printer developed by the other student. Most of these samples exhibited lower gauge factors within the range of 0.1 to 1.0 at strains of up to 30,000 ??. The obtained results have been discussed with theoretical work about the mechanisms of resistance change. Connections with CNT thin film compositions and the resistance change mechanisms to explain the disparity between the achieved gauge factors in these devices and the gauge factor of individual CNTs is discussed in later chapters.  18  1.9 Chapter overview The following chapters will detail or discuss the fabrication procedures, analytical procedures and the results along with a discussion. The next chapter focuses on the strain gauge synthesis and includes details such as the CNT ink preparation and the different methods and modifications used to create the CNT thin film followed up by the procedure for attaching the electrical connections. The analytical procedure chapter discusses the different techniques used in evaluating the strain gauges including imaging techniques and electrical measurement. The imaging techniques were used to determine alignment of the CNTs and the electrical measurement was used to determine the gauge factor as well as other basic electrical characteristics of the devices. The obtained results are presented and discussed in their respective chapter and covers the breadth of reproducing previous work, strain gauge design and synthesis, modifications made to the electrical connections, data obtained from the analytical procedures and a thorough discussion on the obtained gauge factors.  19  2 Fabrication Procedures The basic approach to the synthesis of CNT thin film strain gauges is outlined in this section and the reproduction of such strain gauges should be possible with the detailed information below. The process flows logically starting from CNT ink preparation to thin film casting followed by the electrical connection fabrication to prepare each device for testing. The different evaporation casting methods is also outlined as well as the air flow casting method. 2.1 Ink preparation and substrates In the method used to prepare carbon nanotube containing inks, the SWCNTs are suspended in distilled water using surfactants to form an aqueous CNT ink. The preparation consists of dispersing the SWCNTs (Product # 704148, Sigma Aldrich, and surfactant in distilled water using sonication with an ultrasonic probe. The sonication is performed at a power level of 2.5 for 20 hours while switching between 0.5 s ON and 0.5 s OFF resulting in a total ON time of 10 hours. The sonication is meant to uniformly disperse SWCNT bundles and allow the surfactant to fully encapsulate individual CNTs. The suspension is then centrifuged for 2 hours at 17,000RPM which is roughly equivalent to 20,000 g. After centrifugation, the top 80% of the supernatant is removed from the centrifuge tubes and collected in a separate vial for subsequent deposition. The rest is disposed of according to safety measures. The purpose of centrifugation and removing only the top supernatant is to remove any remaining SWCNT bundles, residual catalysts from CNT synthesis, and ultrasonic probe tip particle contaminants from the ink. Two surfactants, sodium dodecyl sulphate (SDS) and sodium cholate (SC), with concentrations varying between 0.5 wt.% to 2 wt.% were used for ink preparation.  20  The ratio of surfactant to CNT was also varied but a typical ratio of 5:1 was mostly used. Both of these surfactants have been shown to reliably suspend SWCNTs in water [38], [39]. Two substrates, polyimide and polyethylene terephthalate (PET) were considered for ink deposition. The selection criteria included the substrate?s mechanical properties and hydrophobic interaction with the ink. A high yield strength and low Young?s modulus was desirable to allow for a high strain in the GF measurements. The requirements regarding the hydrophobic interactions are more flexible because the surfactant concentration also determines the hydrophobicity. However, the important detail is the capability of maintaining a pinned contact line for the ink when deposited onto the substrate. This requires the interaction between the ink and substrate to be neither too hydrophobic nor hydrophilic. The surface of the substrate could also be plasma-treated to decrease the hydrophobicity. When required, plasma treatment is performed using a small handheld plasma generator and passing the device ten times over the substrate with a total duration of roughly 20 s. The results of the substrate selection are discussed in a later chapter. 2.2 CNT film synthesis A few different ink deposition methods are utilized for thin film synthesis, evaporation casting, inkjet printing, and air flow casting. Evaporation casting was also performed under a few different environmental conditions such as inside a fume hood with and without a cover, on top of a Peltier substrate and also sealed within a water bath.  Evaporation casting is the simplest method involving deposition of the ink onto the substrate with a micropipette by hand. All depositions used 10 ?L of ink forming a line approximately 2-3 mm in width and 20 mm in length. This was then left to dry under various conditions. A pinned  21  contact line is crucial in retaining the cast shape while drying. At first, the drying conditions were with and without a cover in the fume hood. The next condition was in an unsealed plastic container left to dry at room temperature. The iteration after was using a Peltier substrate to cool the plastic substrate in order to slow the evaporation rate for a more uniform film in an attempt to achieve alignment through self-organization in a liquid crystal phase. This type of self-organization has been exhibited for evaporation casting resulting in ?coffee rings? of aligned SWCNTs [40], [41]. The final evolution involved sealing the sample in a plastic container that had an air vent and submerging the container under chilled or room temperature water. A prototype of this setup is shown in Figure 2.2.1. This method was able to keep the environmental temperature stable and cool enough to slow the evaporation rate significantly from approximately 0.5 hours in ambient conditions to approximately 72 hours. Another technique attempted to induce alignment was tilted-evaporation. Alignment is possible by allowing the sample to evaporate while tilted to induce liquid crystal ordering through shear flow [41?43]. Some of the challenges encountered for evaporation casting included an unpinned contact line of the ink with the substrate, uneven evaporation of the deposited ink, unstable temperatures with the Peltier substrate, lack of alignment and inconsistent drying within each respective drying condition.   22   Figure 2.2.1: Image of the sealed sample submerged in a waterbath without cooling prototype set up (no air vent). The sealed sample container is at the bottom of the large glass container and is being held down by a glass jar filled with water. Inkjet printing was shown to be able to create surface aligned CNT films by a colleague, Simon Beyer, and this method was attempted as well [37]. A custom inkjet printer was used to print lines of approximately 2-3 mm wide and 20 mm long. However, the printer was very challenging to use and was plagued with problems ranging from random freezes, malfunctions with stage positioning and nozzle clogging. In addition, for alignment, the evaporation and deposition rates had to be matched and this proved to be a difficult and time-consuming endeavor without the prerequisite skills and knowledge acquired by the previous researcher. Further, the ink ideally had to consist of medium length SWCNTs requiring special care during the ink preparation. It is also not possible to determine the amount of ink used in printing. This technique only exhibited long-range alignment on the surface with local areas of alignment within the film [37]. Together with the difficulty of mastering this technique, different attempts to achieve alignment including the air flow casting, tilted-evaporation and the water bath set-up were pursued. Air flow casting is a method for depositing the ink in a fashion seemingly like aerosol printing. However, the ink droplets are only partially dispersed into tinier droplets in the air and a majority of the dispersion by the air flow occurs after the ink has hit the substrate. The ink was dropped  23  into a laminar airflow of between 2.5-15 psi which passed over the plastic substrate. Approximately 150 ?L was used for each sample but the actual deposited amount is undetermined. The droplets undergo shear forces on the substrate due to the air flow resulting in deposited CNTs corresponding to the flow vector of the ink [43], [44]. This method was used in attempts to make CNT films that were aligned parallel and perpendicular to the applied strain. The challenges encountered with this technique were the lack of alignment, difficulty reproducing films of similar quality and inability to determine how much of the ink was deposited. The technique was also wasteful of the ink due to the nature of being sprayed across the substrate from the applied air flow. However, it has the advantage of being a fast deposition method requiring almost no time for evaporation. An example of a samples synthesised in this the above methods are shown in Figure 4.1.3 and Figure 4.2.2 in a later chapter. 2.3 Electrical connections Four electrical connections were made for each sample to connect to the electrical instruments for GF measurements. The electrical connections to the CNT film consisted of a conductive polymer line, conductive nickel epoxy, tiny electrical wires and crafts glue. The conductive polymer is used to connect the CNT film to a vacant space on the substrate where the metal wires were glued with conductive nickel epoxy. The crafts glue was applied with a hot glue gun to glue the metal wire firmly to the substrate.  The conductive polymer used was poly(3,4-ethylenedioxythiophene) poly(styrenesulfonate) (PEDOT) from Heraeus (product Clevios PH 1000, The PEDOT was mixed with 0.2 wt.% Triton X100 and 5 wt.% dimethyl sulfoxide to improve wettability and electrical conductivity respectively [45]. The PEDOT lines were formed by evaporation casting at first but were improved by using the custom inkjet printer due to  24  challenges with dried CNT displacement and PEDOT ink wetting on the CNT film surface. These issues are specifically discussed in the next section. Fewer complications were encountered while printing PEDOT with the custom inkjet printer and this is attributed to a shorter printing time and an ink with no particles that would jam the nozzle. However, freezing and stage malfunctions still occurred on occasion but with no detrimental effects on printing PEDOT lines as the process could easily be restarted or continued. Before printing, an ethanol wash was performed to prevent the wetting of PEDOT ink onto the CNT film surface. It consisted of submerging the CNT film in ethanol for 10 min with some agitation of the film while submerged at the beginning and at the end of the wash. After ethanol washing, the custom inkjet printer was used to print fine PEDOT lines to connect from the CNT film to a vacant area. In the vacant area, a contact pad was formed for adhesion to the metal wires. This was performed by evaporation casting of a 2.5 ?L droplet of PEDOT ink placed at the end of each line. The droplets were allowed to dry for at least 3 hours resulting in a round contact pad. The metal wires were then glued to the contact pads using a conductive nickel epoxy (Epoxies, The epoxy was allowed a full 24 hours to dry before any electrical testing occurred. Finally, the metal wires were further glued to the substrate with a hot glue gun typically used for arts and crafts for mechanical support to remove any stress to the PEDOT connection and conductive epoxy from handling the wires. 2.4 PEDOT deposition As part of the electrical connections, PEDOT was at first evaporation cast but this caused damage to the CNT films. The challenges involved were the surface absorption of the PEDOT ink onto the majority of the CNT film and the displacement of the deposited CNT ink. The surface absorption of the PEDOT ink occurred as a result of a strong hydrophilic interaction  25  between the dried CNT film and the PEDOT ink. The surfactant from the CNT ink left on the surface of the CNT film contributed greatly to this hydrophilic interaction. To remove the surfactants on the surface of the film, the CNT films were washed in ethanol. The cleansing of the surface surfactant helped to diminish PEDOT ink absorption onto the CNT films. Note that this process only removed the surfactant on the surface of the thin film.  The displacement of the dried CNT ink occurred because of the aqueous nature of the PEDOT ink. A relatively large volume of the ink could re-suspend the deposited and dried CNTs along with the surfactants and displace the CNTs from its original dried position. As a result of this displacement, a gap or crack can form between the CNT film and the PEDOT as can be seen in the left image in Figure 2.4.1. Cracks like this could be how a relatively high GF was attained in previous work. However, this effect is unreliable, difficult to control and not constantly reproducible. To prevent this, following PEDOT depositions were performed with the custom ink jet printer. This allowed for deposition by tiny droplets which would evaporate and dry on top of the CNT film avoiding any CNT ink absorption and displacement. The difference between the two methods and their effects are shown in Figure 2.4.1.   Figure 2.4.1: PEDOT ink deposition images; evaporation cast PEDOT ink resulting in CNT film displacement on the left and inkjet printed PEDOT resulting in fine lines and absence of CNT film displacement on the right.    26  3 Analytical Procedures This chapter introduces and explains the observation and electrical characterisation techniques used to investigate the synthesised carbon nanotube thin films. The imaging techniques of reflective polarised light microscopy and scanning electron microscopy is described followed by an explanation of the electrical measurement system for resistance measurements and the straining mechanism. 3.1 Polarized light and scanning electron microscopy Polarized light microscopy (PLM) and scanning electron microscopy (SEM) were used to search for any indications of alignment. Note that both of these methods were only used to investigate the surface of the sample for alignment. The subsurface cannot be analyzed with the type of PLM used and was not analyzed with SEM. Reflective polarized light microscopy was employed, where the polarized light source is above the sample only. After the polarized light is reflected off the surface of the sample, the resulting image will have a birefringent pattern in the presence of CNT alignment. For example, when the polarizer and microscope analyzer are oriented perpendicular to each other, a CNT film with a circular alignment will exhibit a refraction pattern in the shape of a dark cross. The dark areas are from aligned regions perpendicular and parallel to the analyzer while the bright areas are from aligned regions 45? to the analyzer [37]. This is shown in Figure 3.1.1 below.   27   Figure 3.1.1: Reference reflective PLM image of CNTs in circular alignment clearly exhibiting the birefringent pattern of a dark cross [37]. *Reprinted with permission from S. Beyer, K. Walus, ?Controlled orientation and alignment in films of single-walled carbon nanotubes using inkjet printing?, Langmuir, Volume 28, Pages 8753-8759, 2012. Copyright 2013 American Chemical Society. Consider a rectangular CNT film with CNT alignment along the axis, the whole sample will become dark or bright as the CNT film is rotated relative to the analyzer. If the CNT film only exhibited partial alignment, there will be a birefringent pattern of dark and bright spots. These spots will change from dark to bright as the CNT film is rotated relative to the analyzer. If there is an absence of alignment, there will be no birefringent pattern and no change of brightness from the reflected light with rotation. A Hitachi S4700 FESEM was used to take scanning electron microscopy images. In SEM imaging, the sample should be conductive and exhibit chemical stability amongst other factors to obtain clear and well resolved images. However, the CNT films contain surfactant which is a polymer and may be affected by bombardment of electrons resulting in effects such as heating which can lead to melting and off-gassing. Since the substrate, polyimide, is also plastic, the majority of the sample is non-conductive preventing any charges from escaping and resulting in charge accumulation. Charge accumulation can distort the image and prevent features from being well resolved during the imaging process. This problem was anticipated and encountered while  28  imaging. To address these problems, the solution chosen was thin film chromium deposition. Chromium is deposited onto the sample in a very thin layer of approximately 2 nm via sputtering, which is a type of physical vapour deposition. This provides conductive pathways for charge to escape thereby preventing charge accumulation. However, this extra layer of metal may conceal some features of the sample. It is also irreversible making the sample unusable for testing. After sputtering a thin layer of chromium, the samples were imaged via SEM to investigate the surface for alignment. 3.2 Gauge factor measurement In order to determine the gauge factor (GF) of the synthesized samples, a Bose ElectroForce Load Frame System 3100 ( was used to induce strain on the sample. The Bose uniaxial system consists of a load cell at one end and a motor at the other end with clamps at each end. The load cell is for smaller forces and operates within the range of up to ?30 N. With the load cell and the motor, the system can operate using force feedback or displacement feedback meaning the mechanical instructions can consist of setting precise forces or displacements. One of the challenges encountered was sample slipping between the polyimide substrate and the clamps. This was prevented by utilizing clamps from a different Bose system which provided a tighter grip on the sample and was able to prevent slippage. The set-up is shown in Figure 3.2.1.  29   Figure 3.2.1: Sample CNT film with electrical connections undergoing strain in the Bose uniaxial system. To acquire electrical data to relate with the strain and determine the GF, a four-point probe resistance measurement was performed simultaneously as the sample was strained by the Bose system. This was performed using previously developed Labview software which controlled a Keithley 2400 SourceMeter ( The software allowed user defined current amplitudes and the SourceMeter attempts to set this current between the two outer connections within a voltage compliance. This voltage range or compliance could also be user defined with a maximum of 20V. The voltage was then measured by the SourceMeter through the two inner connections to determine the resistance. The resistance data was consolidated with force, displacement, axial command data from the Bose and recorded by the Labview software at user defined intervals. The intervals were defined at 200 ?s. However, the measurements were not taken exactly at every 200 ?s due to the delay between communicating with the SourceMeter and the Bose system resulting in an actual interval averaging 201.3 ?s. The user defined current amplitude varied between 0.01 mA and 1 mA in order to achieve a measurable voltage, which was dependant on the samples. The maximum compliance of 20 V was used. Some challenges encountered with this set-up were freezing of the Labview software and computer due to unknown reasons. It commonly occurred during measurements lasting over 6 hours and such  30  measurements were subsequently avoided. A possibility was that there was too much data to store but no common time frames could be determined amongst the freezing occurrences. The measurements were performed using displacement feedback which allowed for exact displacements to be controlled. The Bose system software is able to carry out a recipe of instructions such as ramping up or down, holding a certain displacement or cycling between set displacements at a controllable rate. Each sample was tested with a pre-stress at 1 MPa to ensure the sample was taut in the sample holder and the induced displacements were transferred with no slack. The displacement at which the sample experienced 1 MPa of stress was determined by measuring the cross-section of the sample to calculate the required load. The sample was then slowly strained until the appropriate load was achieved in the real-time data shown in the Bose system software. The displacement at which the load was achieved was recorded and incorporated into the measurement instructions as the pre-stress displacement. The typical recipe involved ramping up to the pre-stress displacement, holding for 5 min followed by 5 cycles of strain with 5 min between each cycle, followed by another hold for 15 min and ending with ramping down to zero displacement. The 5 min between each strain cycle allowed the sample to relax and allow any creep effects to settle. This procedure is shown in Figure 4.5.1 in the following chapter. The acquired consolidated data from the Bose system and the SourceMeter was analyzed in Microsoft Excel. The resistance, load and displacements were plotted versus time to check for any discontinuities or incongruence. The resistance was typically not stable during the tests and would increase or decrease throughout the measurement. The origin of the drift is uncertain but instrumentation error is likely not the cause because a simple test measurement with hard resistors did not elucidate any drift in the readings. It is possible that the drift is a result of local heating effects within the CNT thin film because a test to check the  31  dependency of the CNT thin film?s resistance from heat showed a decreasing resistance with an increase in temperature. A linear or quadratic fit was performed to determine the base resistance. Using the formula obtained from the line fit, a base resistance was calculated for each data point and the change in resistance was determined by subtracting the base resistance from the measured resistance. The change in resistance was then divided by the base resistance to obtain the relative change in resistance. The relative change in resistance was plotted versus strain only for the range of data points corresponding to the strain cycles from which GFs were elicited. To review, the imaging techniques of reflective polarised light microscopy and scanning electron microscopy were used to investigate the CNT thin films for presence of alignment to determine if the fabrication procedures were able to produce alignment. The electrical measurement coupled with the Bose system was used to induce strain on a sample and record the resistance at the same time in order to determine the gauge factors of the synthesised samples.   32  4 Results and Discussion This chapter discusses the results from the various procedures of synthesising the strain gauges to their analysis. The selection of materials, ink composition and fabrication method are discussed in the first few sections followed by the modifications done to improve the electrical connections of the strain gauges from the original design. The contents flow into the data obtained from the imaging techniques and electrical measurements and the discussion of the data. The chapter finishes off with a prominent discussion of gauge factors and likeliness of piezoresistivity as a resistance change mechanism at work in carbon nanotube thin films. 4.1 CNT ink and substrate Recall that two options existed each for the CNT ink and substrate. These two are related in the sense that their hydrophobic interactions largely determine the ability of retaining a pinned contact angle during evaporation and attaining uniform evaporation. For the CNT ink, the surfactant sodium cholate (SC) was favoured over sodium dodecyl sulfate (SDS) due to its more favourable hydrophobic interactions and success rate with fabricating uniformly evaporated and contact pinned CNT films. However, samples using SDS were also investigated in attempts to replicate the relatively high GFs achieved in previous work. Ink compositions with surfactant concentrations varying between 0.05 wt.% and 2 wt.% were prepared. The lower concentrations were prepared to lower the overall CNT content of thin films thereby decreasing the surface concentration of CNTs. The surfactant to CNT ratio was 5:1 for most inks but a 1:1 ratio was also prepared in attempts to investigate the effect on the GF. Table 4.1.1 below shows the highest GF attained with each ink composition along with a typical gauge factor which is the median of gauge factors observed from repeated measurements at various strains and from samples with the same ink composition. The average initial resistance taken between samples of the same ink  33  composition is also shown. Focusing on the surfactant used (SDS or SC), it seems there is a slight tendency that a higher strain is required to elicit similar GFs in the samples using SDS compared with samples using SC as shown in Figure 4.1.1.  However, this observation is purely speculative and is not rigorously defined.  Figure 4.1.1: Plot of gauge factor measurements with their corresponding strain for various samples synthesised with different surfactants and ink compositions. Samples with SC are shown to require less strain to achieve a relatively high gauge factor. Otherwise, there is no evident correlation between the ink composition and the GFs, typical GFs and initial resistance. However, it is important to consider the ink compositions are not the same as the CNT film concentrations. The CNT film concentrations will depend on the ratio between the surfactant and the CNT since the water will have completely dried. The two ratios made were 5:1 and 1:1 for the ink compositions. As such, the CNT films are approximately 20 wt.% CNT or 50 wt.% CNT with the rest of the film ideally containing the surfactant. These concentrations are  34  slightly higher since the surfactant on the surface of the film was washed off with ethanol as previously described. Table 4.1.1: Gauge factors achieved with each ink composition irrespective of deposition method Ink Composition Highest GF (strain ??) Typical GF Initial Resistance (?) 0.4 wt.% CNT, 2.0 wt.% SDS 1.3 (15,000) 0.95 ?0.35 1,616 0.2 wt.% CNT, 1.0 wt.% SDS 4.0 (20,000) 2.55 ?1.45 9,950 0.4 wt.% CNT, 2.0 wt.% SC 4.0 (14,000) 2.5 ?1.5 502.5 0.2 wt.% CNT, 1.0 wt.% SC 1.5 (20,000) 0.8 ?0.7 189,614 0.1 wt.% CNT, 0.1 wt.% SC 2.6 (7,000) 2.3 ?0.3 1,870 0.05 wt.% CNT, 0.05 wt.% SC 1.4 (5,000) 1.4 1,570  For the substrate, polyimide was chosen over polyethylene terephthalate (PET). The polyimide provided a more favourable hydrophobic interaction and was thinner than PET allowing for higher strains to be tested with the Bose system due to a limited force range. The graph in Figure 4.1.2 shows the displacements achieved when applying a 3 N load on the polyimide and PET film samples. The polyimide sample exhibited a displacement of approximately 0.11 mm which is more than five times the displacement the PET sample exhibited which is approximately 0.02 mm.  35   Figure 4.1.2: Graph of displacement vs time of the two different samples undergoing five cycles of strain at a 3 N load. The use of SC as a surfactant along with ionization of the polyimide substrate allowed for the challenges of the non-uniform drying and non-pinning contact of the CNT ink after deposition to be overcome most of the time. Even though SC showed better characteristics, other factors such as temperature or perhaps surface charge would affect the process and result in a film shown on the left in Figure 4.1.3. Both samples in this figure were evaporation cast on a level surface and the accumulation of ink shown on the left sample is possibly due to the non-uniform drying effects. Figure 4.1.3: Sample showing unpinned contact line and non-uniform evaporation on the left, sample showing a well pinned contact line and uniform evaporation while retaining desired shape on the right.  36    Figure 4.1.3: Sample showing unpinned contact line and non-uniform evaporation on the left, sample showing a well pinned contact line and uniform evaporation while retaining desired shape on the right. It is worth noting that the cost of using the polyimide film which is in the form of a polyimide tape costs approximately 0.30 US cents per device. This is based on the cost of a 3300 m roll of polyimide tape which costs 20 USD. This leads to a cost of approximately 0.30 US cents when each device uses 5 cm of polyimide tape in length. In terms of device costs, the price of the nickel epoxy used in the electrical connections is the dominating factor. It is estimated that approximately 1.5 US cents worth of epoxy is used per device. Considering the cost of each strain gauge from a materials perspective, each strain gauge fabricated in this work is worth only 5 US cents per device assuming that the electrical wires and the PEDOT costs 2 US cents per device. 4.2 CNT film fabrication A successful evaporation casting method would result in CNT films similar to the one shown on the right in Figure 4.1.3. Throughout the different evaporative methods utilized, the evaporation rate for each method was monitored by observing the amount of time taken for the film to dry. The evaporation rates decreased in the following order, casting inside the fume hood, inside a plastic container at room temperature, submerged in a water bath, on a Peltier substrate, and  37  submerged in a chilled water bath. The drying times were respectively approximately 0.5 hour, 1 hour, 12 hours, between 24 to 48 hours, and up to 72 hours. The decrease in evaporation rate was an attempt to achieve an aligned CNT film through self-organisation in a liquid crystal phase as described by Simon Beyer and other literature [37], [40], [41]. It was hypothesized that a lower evaporation rate would allow the CNT ink to spend more time in a liquid crystal phase allowing more time for the self-organising effect to occur thereby leading to alignment. The investigation of effects of alignment in a CNT film on the GF was desired but unfortunately, the evaporation casting methods did not produce any well aligned CNT films as determined by PLM and SEM imaging and the gauge factors were also low. The images showing lack of alignment are presented in the following section. Other methods of achieving alignment were also attempted including air flow casting, tilted-evaporation casting, and inkjet printing. The tilted-evaporation casting is evaporation casting with the sample left at a tilt to induce shear flow of the CNT ink as introduced in the fabrication chapter. This method resulted in non-uniform CNT film evaporation with a bulk of the CNT ink evaporating at the bottom of the slope resulting in an exponential gradient of CNT film concentration. The prototype sample film could not be used for a strain device due to the intense gradient of CNT film concentration and as a result, this method was not used to make any CNT film strain gauges.  Inkjet printing was shown by my colleague, Simon Beyer, to be able to create surface aligned and sub-surface semi-aligned films [37]. However, as mentioned previously, it was not possible to recreate such CNT films with the technical problems persisting in the custom inkjet printer and the absence of adapted skills for printing CNT ink. Ignoring the technical problems of the inkjet printer, the major challenge was achieving the appropriate printing conditions and  38  avoiding nozzle jamming by the CNTs. In order to achieve the desired evaporation effects that control the CNT alignment, the rate of CNT ink deposition must be equal to the rate of evaporation. However, it was incredibly difficult to match the deposition and evaporation rates since the evaporation rate constantly fluctuated because the local temperature increased due to the running instruments. If the current speed of printing was too slow, the droplets would evaporate as soon as it hit the substrate. If the speed of printing was too high, it resulted in other problems such as droplet aggregation due to unpinned contact lines as can be seen in Figure 4.2.1. The ideal printing rate is constantly fluctuating due to the environmental temperature and the speed of printing was not easily controlled. A couple CNT films using inkjet printing had electrical connections fabricated but the initial resistance of these samples were in the gigaohm magnitude and there was no electrical response to strain.  Figure 4.2.1: Inkjet printing CNT films exhibiting challenges of matching evaporation and deposition rates, droplet aggregation and unpinned contact lines. The air flow casting method is the third method attempted to induce alignment in CNT films. It was performed with air flow parallel to the direction of strain and also perpendicular as shown in Figure 4.2.2. The method resulted in very thin films that are nearly transparent due to the nature of the deposition method. The first couple samples were performed at a relatively low 2.5 PSI  39  and the air flow direction was parallel to the direction of strain; higher pressures were used for later samples. The use of air flow casting made it difficult to achieve a uniform layer and some larger drops of CNT ink immobilized on the substrate at a low 2.5 PSI. Also, scotch tape was originally used as a mask to prevent deposition on areas where CNT ink was not desired. Following samples were performed with the air flow perpendicular to the direction of strain and utilized a higher 15 PSI. This allowed the preparation of more uniform films without larger drops of CNT ink being immobilized on the substrate. Note that the samples with the air flow perpendicular to the direction of strain did not use scotch tape as masks because it significantly affected the air flow. Instead, the areas where CNT ink deposition was not desired were removed with a Kimwipe ? and ethanol after deposition.   Figure 4.2.2: Air flow cast samples with direction of air flow parallel to the direction of strain on the left and perpendicular to the direction of strain on the right. An air flow cast sample fabricated with the air flow parallel to the strain direction at 15 PSI was investigated under SEM and the surface did not exhibit any solid evidence of alignment. Some areas imaged exhibited a random orientation of CNTs and other areas exhibited a general direction of very weak alignment but it is not substantial proof of alignment. This is shown and discussed further below.  40  Table 4.2.1 below shows the different evaporation methods used and the respective highest GF attained using each method. The evaporation rates of the methods where the films are dried in the fume hood or in a drawer or in an un-chilled water bath are similar. The greatest disparity in evaporation rate is between these methods and the method using the chilled water bath. From the decrease in evaporation rate, we can see that the GF has dropped by more than a half. For example, two samples which were dried in a water bath exhibited gauge factors in the range of 1.4 to 2.6 while two different samples dried in a chilled water bath exhibited gauge factors in the range of 0.6 to 1.0. This lends to the hypothesis that a slower evaporation rate may not be beneficial as originally hypothesised regardless of alignment.  Alternatively with the air flow evaporation casting, the first sample with a low air flow exhibited a GF similar to the other simple evaporation cast films. However, the rest of the air flow evaporation cast samples exhibited a lower gauge factor than the simple evaporation cast films. In addition, the sample with perpendicular air flow resulted in one of the lowest gauge factor devices fabricated. As mentioned previously, the air flow evaporation cast samples did not reliably exhibited any general alignment on the surface. Table 4.2.1: Gauge factors achieved with different evaporation casting (EC) methods irrespective of ink composition Deposition Method Highest GF (strain ??) Typical GF Initial Resistance EC dried in box in fume hood  1.3 (5,000) 1.05 ?0.25 873 ? EC dried in box in fume hood (5 ?L) 4.0 (14,000) 2.5 ?1.5 1,467 ? EC dried in box in drawer 4.0 (20,000) 2.55 ?1.5 9,950 ? EC dried in water bath 2.6 (7,000) 2.0 ?0.6 1,720 ? EC dried on Peltier substrate 1.2 (5,000) 1.1 ?0.1 530 ?  41  Deposition Method Highest GF (strain ??) Typical GF Initial Resistance EC dried in chilled water bath 1.0 (7,500) 0.8 ?0.2 1,219 ? Air flow EC at 2.5 PSI (parallel) 1.5 (20,000) 1.25 ?0.25 890 k? Air flow EC at 5.0 PSI (parallel) 0.6 (20,000) 0.35 ?0.25 141 k? Air flow EC at 10 PSI (parallel) 0.6 (20,000) 0.5 ?0.1 2,555 ? Air flow EC at 15 PSI (parallel) 1.0 (20,000) 0.85 ?0.15 37,300 ? Air flow EC at 15 PSI (perpendicular) 0.5 (20,000) 0.4 ?0.1 65,700 ?  To determine the effect on the mechanical properties of the substrate from the deposition of CNT ink and electrical connections, strain measurements were performed to elicit stress-strain curves. Figure 4.2.3 below is the stress-strain data for a blank substrate before any deposition of CNT ink or electrical connections. The substrate was determined to have a tensile modulus of approximately 0.5 GPa by taking the slope. The tensile modulus of a full device complete with CNT ink, PEDOT, conductive epoxy and craft glue was determined to be approximately 0.7 GPa as shown in Figure 4.2.4. This difference of 0.2 GPa shows that the complete device experiences more stress from strain due to the increase in tensile modulus. Note that the variance of the substrate dimensions can lead to a ?0.04 GPa difference which is less than the observed difference between a blank and complete device.  42   Figure 4.2.3: Plot of a strain measurement performed on a blank substrate to determine the tensile modulus which was determined to be approximately 0.5 GPa.  Figure 4.2.4: Plot of a strain measurement performed on a complete device to determine the tensile modulus which was determined to be approximately 0.7 GPa.  43  4.3 Reproducing results of previous work It was stated previously that previous work exhibited gauge factors within the range of approximately 2 to 30. The highest gauge factor from the previous work was reported to be 29.8 which were irreproducible. This sample used an ink composition of 0.4 wt.% CNT, 2 wt.% SDS and was fabricated using evaporation casting. The sample also used evaporation casting for the PEDOT electrical contacts and is the same design as shown in Figure 1.7.1. An attempt to reproduce this work using the same ink composition and fabrication method was unsuccessful and exhibited GFs within the range of 0.6 to 1.3. In fact, the GFs achieved from all fabricated strain gauges ranged from 0.4 to 4. It is suspected that the high GF attained previously is not a piezoresistive effect or a geometric effect. A possibility is that there were discontinuities or cracks in the CNT film due to the evaporation casting of PEDOT. These cracks would contribute to a large relative resistance change upon strain resulting in a relatively high GF and might also explain the large variation in observed gauge factors. The detrimental effect of evaporation casting PEDOT on the CNT films was shown and discussed in the fabrication chapter. 4.4 Imaging (PLM and SEM) for alignment The evaporation cast samples were investigated using reflective polarized light microscopy (PLM) and the air flow evaporation cast samples with scanning electron microscopy (SEM). Neither technique produced samples with long range alignment on the surface. The evaporation cast samples produced evidence of short range alignment but it was non-uniform and did not persist throughout the samples while the air flow evaporation cast samples did not show any conclusive alignment.   For the normal evaporation casting, images of the same sample were compiled into two PLM image comparisons shown in Figure 4.4.2 and Figure 4.4.3 below. The two images were taken 45?  44  rotated from each other but was rotated using an image editing program for ease of comparison. As previously mentioned, reflective PLM reveals surface alignment via birefringent patterns of bright and dark areas. If alignment is present, these bright and dark areas should be complimentary to each other when rotated 45? from each other as shown in the reference images obtained from colleague, Simon Beyer, in Figure 4.4.1.   Figure 4.4.1: Reference set of reflective PLM images of an inkjet printed CNT film clearly exhibiting the effect of CNT alignment on the reflectance, the images are rotated 45? from each other [37]. *Reprinted with permission from S. Beyer, K. Walus, ?Controlled orientation and alignment in films of single-walled carbon nanotubes using inkjet printing?, Langmuir, Volume 28, Pages 8753-8759, 2012. Copyright 2013 American Chemical Society. In the first comparison shown in Figure 4.4.2, a birefringent pattern is evident and although the patterns are not completely complimentary to each other, they do not overlap. This suggests partial random local alignment on the surface but alignment did not persist throughout the sample. In the second set of images found in Figure 4.4.3, there is an absence of a birefringent pattern suggesting a complete absence of alignment. This shows that in the same sample, there is partial alignment at one end and no alignment at the other end of the CNT film. Images taken of the middle section of the CNT film showed an absence of alignment. This suggests that evaporation casting was not able to produce aligned CNT films but exhibited random localized alignment in some areas.  45   Figure 4.4.2: Reflective PLM image comparison of two images 45? from each other exhibiting partial alignment due to presence of birefringent pattern (visible zebra stripes of dark and light gray).  Figure 4.4.3: Reflective PLM image comparison of two images 45? from each other exhibiting absence of alignment due to absence of any birefringence. For an air flow cast sample where the direction of air flow was parallel to the direction of strain, SEM images were taken to investigate any presence of surface alignment. The results exhibited areas with a possible weak alignment while most other areas exhibited no alignment. In Figure 4.4.4, on the left is an SEM image exhibiting random orientation deposition while the image on the right exhibits a slight general direction of alignment possibly from the air flow casting deposition. However, both images are quite similar and in the end, there is no clear presence of alignment. Therefore, none of the air flow evaporation cast samples produced have reliably  46  exhibited alignment. More ideal SEM images obtained from colleague, Simon Beyer, exhibiting alignment are shown in Figure 4.4.5 as a reference.   Figure 4.4.4: SEM images of an air flow cast sample; on the left is an imaged area exhibiting a mostly random orientation and on the right is an imaged area with a general direction of weak alignment as indicated by the arrow.   Figure 4.4.5: Reference set of SEM images of an inkjet printed CNT film clearly exhibiting surface alignment [37]. *Reprinted with permission from S. Beyer, K. Walus, ?Controlled orientation and alignment in films of single-walled carbon nanotubes using inkjet printing?, Langmuir, Volume 28, Pages 8753-8759, 2012. Copyright 2013 American Chemical Society. The presence of larger particles on the CNT film is evident from the images. These are likely to be the surfactant used in the CNT ink. In addition, judging from the scale, the ?sticks? shown could be CNT bundles due to their size. The diameters are nearing 50 nm which is too large for individual SWCNTs whose diameters should be approximately 1 nm. The chromium coating was  47  deposited to be approximately 2 nm and does not account for the large discrepancy. It is likely that the depositions imaged are CNT bundles as opposed to single CNTs. This could have an impact on the electrical behaviour of the CNT films in response to strain. The presence of bundles and surfactant lumps would affect the electrical conductive pathways and any mechanisms in resistance change. 4.5 Gauge factor measurements Many CNT films were made using the different methods and even within the same method, there were variations. The films which retained their contact lines and dried uniformly were made into simple strain gauges by fabricating the four electrical connections as shown previously. These samples were tested by straining the sample and concurrently measuring the resistance with a four point probe method. A large range of strains were applied ranging from 3,000 ?? to 30,000 ??. One of the reasons for the large variation of applied strains was because the strain gauges were constantly being improved by a different polyimide film and reducing the dimensions enabling larger displacements with the same load.  Overall, the gauge factors (GF) ranged between 0.1 and 4.0 which is relatively low compared to other novel CNT/polymer composite strain gauges or silicon strain gauges. However, it is comparable to strain gauges fabricated with existing metal foil gauges, with as-grown CNT forests and with CNT/polymer composite coatings on yarns. The latter two types of devices have reported GFs of approximately 5.0 in literature [8], [11], [18], [25]. In addition, other thin film CNT strain gauges report similar GFs ranging from approximately 0.7 to 7.0 but one thin film CNT strain gauge requiring microfabrication techniques report a gauge factor of approximately 60 [6], [7], [27], [28], [46?48].  48  For determining the gauge factor, as mentioned previously, the data obtained was analyzed and the resistance, load and displacements were plotted versus time to check for any discontinuities or incongruences as shown in Figure 4.5.1. Since the measured resistance was not stable, a line fit was performed using Microsoft Excel to determine the base resistance as shown in Figure 4.5.2. A couple other sample graphs are included in the Appendix. Finally, the calculated relative change in resistance was plotted versus strain to determine the gauge factor from the slopes as shown in Figure 4.5.3. Note that the graphs with relative change in resistance versus strain only use data isolated from each individual cycle of strain.  Figure 4.5.1: Graph of displacement and load versus time to check for discontinuity and incongruence between the applied strain and measured load. Note the pre-stress of 1 MPa shown from the first ramp and the subsequent cycles of approximate max stress 11.7 MPa and strain 10,000 ??. From Figure 4.5.1, there are no discontinuities or incongruence between the induced displacement and the measured load. However, a little viscoelastic deformation is apparent from  49  the gradual decrease in the measured load peaks with each cycle of displacement. It is also seen in the difference in the measured load before and after the first strain cycle. The measured load is lower at the pre-stress displacement after the first cycle compared to before. Both of these behaviours mostly diminished after the first two or three cycles. Since a plastic thin film is used as the substrate, some viscoelastic behaviour is likely to be present. Throughout all of the GF measurements performed, these behaviours were present. To minimize the effect of viscoelasticity on the GF measurements, displacement feedback was used instead of force feedback. If force feedback were used, the Bose system would have continued stretching the sample to achieve the directed load value. This could have resulted in a greater displacement with each strain cycle resulting in increasing strains. Figure 4.5.2 shows the resistance measurements throughout the course of the ramping and strain cycles. It can be seen that the ramping to and from the pre-stress displacement value does not result in any changes in resistance for this sample. Recall the pre-stress displacement value is the displacement at which the sample experiences 1 MPa of stress. This suggests that a stress of greater than 1 MPa is required to induce a change in resistance. However, not all tested samples required more than 1 MPa of stress to exhibit any resistance change. Notice in Figure 4.5.2 that the resistance shows a gradual decrease after each cycle of strain for the first three cycles and the effect is diminished thereafter. This matches the viscoelastic behaviour of the substrate mentioned previously and is also likely a result of the viscoelastic behaviour.  50   Figure 4.5.2: Graph of resistance versus time for determining the base resistance through a linear line fit. It is also important to note that the resistance is not completely stable and shows a gradual decrease. The reason for this effect was not discovered completely but could be due to many factors such as environmental effects of temperature or humidity changes in the room throughout the 45 minute measurement. A quick test with a heat gun showed that the resistance would decrease from the application of heat on the sample and increase after the removal of heat with the data shown in the Appendix. Another possibility is an effect of a charge or potential build-up in the CNT film or in the electrical connections since these measurements was performed with a DC bias. However, a 13 hour long test measuring only the resistance without strain did not show a continuous decrease of resistance but also an increase in resistance over time and this data is shown in the Appendix.  51  With the first few measurements, some samples exhibited a curve similar to a logarithmic decrease in resistance determined to last over the span of two hours. These samples had a relatively stable resistance after the two hour span thus some GF measurements included a two hour application of the electrical signal before strain was applied. However, later samples did not exhibit a relatively stable resistance even after two hours and a couple simple tests of measuring the resistance over a thirteen hour period did not elucidate any stabilizing or periodic effects (same tests for investigating charge or potential build-up). With respect to the GF, the effect of an unstable resistance lowered the accuracy of the line fits and affected the calculated GF by a small amount. As such, the remaining measurements continued without any application of the electrical signal for two hours, to reach a stabilised resistance, before applying strain.  In terms of the effects on sensing strain, a drifting unstable resistance can be taken into account for with data analysis techniques such as the ones used to calculate the gauge factors in this work. A baseline for the resistance can be determined and as long as the change in resistance due to strain is greater than the gradual change in resistance due to the drift, a strain can be determined. Another possible method would be to have two strain gauges in the same circuit but allow only one device to be affected by strain and use the other to determine the base resistance. However, this would only work if both devices experience the same drift effects which may not be the case.  52   Figure 4.5.3: Graph of relative change in resistance versus strain from each of the five strain cycles; the average of the slope for each cycle is used to determine the gauge factor which is approximately 1.5 for this sample. The data shown above in Figure 4.5.3 shows five separate lines for each of the applied strain cycles. As can be seen, there is a little bit of hysteresis from the tensile straining and de-straining but it is minor. The lines also overlap each other which represent a small change in the base resistance over time. The response of the change in resistance from strain is also mostly linear. The gauge factor is determined by averaging the slopes of these lines. For the sample in Figure 4.5.3, the GF was averaged to be approximately a relatively low 1.5. An overview of the measured GFs is shown in Table 4.5.1 below and samples referred to below are found in this table. The highest GF exhibited was 4.0 and was measured from two samples, both fabricated by normal evaporation casting (EC), one with ink composition 0.4 wt.% CNT, 2.0 wt.% SC and the other with 0.2 wt.% CNT, 1.0 wt.% SDS. Focusing on the samples using a CNT ink composition 0.4 wt.% CNT, 2.0 wt.% SC (samples 5 to 11), the measured gauge factors  53  are all in the range of 1.0 to 1.2 with an applied strain of 5000 ?? regardless of the fabrication method. Note that EC with a Peltier substrate was one of the methods encompassed in this sample set and was meant to slow the evaporation rate possibly leading to alignment by self-organisation in a liquid crystal phase. However, from this small range of measured GFs, it suggests that slowing the evaporation rate ultimately does not lead to an increase in GF. Furthermore, the two samples fabricated with the slowest evaporation rate of chilled water bath evaporation casting, which uses the CNT ink composition 0.4 wt.% CNT, 2.0 wt.% SDS (samples 1 and 2 from Table 4.5.1), exhibited GFs lower than the EC samples dried at room temperature. The observation that a slower evaporation rate does not lead to a higher GF is highlighted in Figure 4.5.4. However, note that the difference in GFs is not substantial and could be due to experimental error or variation.  Figure 4.5.4: Plot of gauge factor measurements with their corresponding strain for various samples synthesised with different fabrication methods. Fabrication methods with active cooling to slow the evaporation rate and the air flow  54  evaporation cast samples have a lower gauge factor (at similar strains) in general compared to those fabricated at room temperature. Within each fabrication method, a higher strain generally resulted in a higher gauge factor. The EC methods utilizing slower evaporation did not produce any long range alignment and the alternative air flow EC method attempted (samples 12 to 16) to produce alignment also did not reliably produce any alignment. The measured GFs were all 1.5 or lower with all samples except sample 12 exhibiting GFs 1.0 or below. These samples have poorer performance in general compared to the samples made with other techniques as shown in Figure 4.5.4. Note that sample 12 exhibited results as an outlier to the other air flow EC samples. It was fabricated with a relatively low air flow rate which may have resulted in a film more characteristic of non-air flow EC samples allowing it to behave more similar to those samples. Another observation from these measurements is that the GFs increased as the strain increased, this has been reported in numerical models and their corresponding experimental results [24]. This effect is persistent in most of the samples where multiple measurements were performed and is lightly exhibited in the general overview of gauge factors presented in Figure 4.5.4 within each fabrication method. The effect is also generally linear except in samples 4, 9 and 15. In sample 15 fabricated using air flow EC, the GF only increased between strains 10,000 ?? and 20,000 ?? but not between 20,000 ?? and 30,000 ??. This was the only sample tested at a strain of 30,000 ?? for air flow EC and the other similarly fabricated samples were tested with a max strain of 20,000 ??. For samples 4 and 9, they have a common property that the total mass of CNT and surfactant are halved compared to other samples excluding samples made with the air flow EC method. They are the two samples that achieved the highest gauge factor of 4.0. Sample 4 was made with an ink which had half the concentration of CNT and surfactant than normal and sample 9 was made by depositing only 5 ?L instead of the normal 10 ?L. Some possibilities responsible for this  55  effect include the application of a higher strain and a lower total mass of CNTs and surfactants (resulting in a thinner film or a lower density film). Samples 17 and 18 also have a lower CNT and surfactant mass but they also have a different ratio of surfactant to CNT of 1:1 as opposed to 5:1 for all the other samples. This means the dried film is composed of approximately 50 wt.% CNT and 50 wt.% surfactant. The GFs measured for these samples are slightly higher than the median gauge factor as shown in Figure 4.5.5. However, there is not enough conclusive data to support this claim. The higher GFs could be a result of the lower CNT mass, density or the differing ratio of surfactant to CNTs.  Figure 4.5.5: Plot of gauge factor measurements with their corresponding strain for various samples differing by CNT content. The blue square represents the average of several samples fabricated with the respective CNT content in the legend while the green triangle and red diamond represent one sample each. To review, the samples using air flow evaporation casting elicited lower gauge factors compared to samples fabricated with simple evaporation casting. There were 7 samples using the same SC ink composition but different fabrication methods which all reproducibly exhibited a gauge factor of 1.1 ?0.1 at a strain of 5000 ??. This provides an indication that a lower evaporation rate does not result in a higher gauge factor. Also, samples with a lower overall CNT and surfactant content exhibited better gauge factors and although not conclusive, provides direction for further  56  studies. The relatively low gauge factors could be the result of the bundling of CNTs which is highly likely given the SEM image and the diameters of the sticks in the image. These diameters are approximately 50 nm while the CNTs used should be approximately only 1 nm in diameter. The bundling of CNTs is likely to affect the electrical characteristics thereby affecting the gauge factor. Another important factor is the variation in chiralities of the CNTs since CNTs can have a negative GF or exhibit no piezoresistivity depending on their chirality as mentioned earlier [20], [23]. Also, other factors such as the length of the CNTs can affect the number of contacts between individual CNTs affecting the number of conductive pathways available in the internal CNT network and how these conductive pathways change under strain. Discussed further below is the likeliness of individual CNT piezoresistivity as a dominant resistance change mechanism. The absence of piezoresistivity may explain the low GFs.   57  Table 4.5.1: Summary of gauge factors of fabricated CNT film strain gauges using different inks and evaporation casting (EC) methods at various levels of strain # Ink Deposition Gauge Factors (strain ??) Initial Resistance 1 0.4 wt.% CNT, 2.0 wt.% SDS EC in a chilled water bath 0.6 (5,000), 0.5 (8,000) 2075 ? 2 0.4 wt.% CNT, 2.0 wt.% SDS EC in a chilled water bath 0.8 (7,000) 363 ? 3 0.4 wt.% CNT, 2.0 wt.% SDS EC in a box in fume hood  0.8 (10,000), 1.3 (15,000) 2410 ? 4 0.2 wt.% CNT, 1.0 wt.% SDS EC in a box in drawer 1.1 (10,000), 4.0 (20,000) 9950 ? 5 0.4 wt.% CNT, 2.0 wt.% SC EC in a box in fume hood  1.2 (5,000) 450 ? 6 0.4 wt.% CNT, 2.0 wt.% SC EC in a box in fume hood  1.1 (5,000) 425 ? 7 0.4 wt.% CNT, 2.0 wt.% SC EC in a box in fume hood  1.2 (5,000) 500 ? 8 0.4 wt.% CNT, 2.0 wt.% SC EC in a box in fume hood  1.0 (5,000) 580 ? 9 0.4 wt.% CNT, 2.0 wt.% SC EC in a box in fume hood  (5 ?L) 1.0 (5,000), 2.8 (10,000), 4.0 (14,000) 1467 ? 10 0.4 wt.% CNT, 2.0 wt.% SC EC on Peltier substrate 1.0 (5,000) 630 ? 11 0.4 wt.% CNT, 2.0 wt.% SC EC on Peltier substrate 1.2 (5,000) 430 ? 12 0.2 wt.% CNT, 1.0 wt.% SC Air flow EC at 2.5 PSI (parallel) 1.2 (10,000), 1.3 (15,000), 1.5 (20,000) 890 k? 13 0.2 wt.% CNT, 1.0 wt.% SC Air flow EC at 5.0 PSI (parallel) 0.1 (10,000), 0.4 (15,000), 0.6 (20,000) 141 k? 14 0.2 wt.% CNT, 1.0 wt.% SC Air flow EC at 10 PSI (parallel) 0.4 (10,000), 0.6 (20,000) 2555 ? 15 0.2 wt.% CNT, 1.0 wt.% SC Air flow EC at 15 PSI (parallel) 0.7 (10,000), 1.0 (20,000), 1.0 (30,000) 37.3 k? 16 0.2 wt.% CNT, 1.0 wt.% SC Air flow EC at 15 PSI (perpendicular) 0.3 (10,000), 0.5 (20,000) 65.7 k? 17 0.1 wt.% CNT, 0.1 wt.% SC EC in a water bath 2.0 (5,000), 2.6 (7,000) 1870 ? 18 0.05 wt.% CNT, 0.05 wt.% SC EC in a water bath 1.4 (5,000) 1570 ?  58  4.6 Initial resistances Other factors such as film thickness, CNT dispersion, surfactant dispersion and their homogeneity throughout the film were not characterised, a connection between the performance of the device and the film characteristics through the initial resistance is attempted. The initial resistance can be used to gauge some characteristics of the film, for example the density of the CNT network of the film. A low initial resistance may indicate a strong network with many conductive paths and vice versa. These initial resistances of devices were also recorded in Table 4.5.1 above. They were recorded at the moment before the pre-stress occurred during the strain measurements and are indicative of the resistance of the sample throughout the strain measurements. Focusing separately on samples fabricated using the surfactant SDS and SC, there is unfortunately no reliable correlation between the resistance and measured gauge factors. However, the samples synthesised with SC surfactant tend to have a lower resistance in general than those with SDS surfactants. The data shows that a few samples with a higher resistance exhibited higher gauge factors but this could be a result of higher strain applied while measuring the GF, variances in measuring GF or between fabrications, and total CNT content. A connection between resistance and CNT content shows that samples with a lower CNT content in general exhibited a higher initial resistance. This seems independent of the CNT to surfactant ratio which varied between a couple of the samples synthesised using the SC surfactant (samples 17 and 18). From Table 4.5.1, samples 4, 9, 17 and 18 share the common characteristic of a lower total CNT content through a lower CNT ink weight percentage in the ink or a smaller amount of ink used in the synthesis. It can be seen in Figure 4.6.1 that the resistances of samples with a lower CNT content generally exhibited a resistance between 3-5 times higher when the CNT content is  59  halved. Recall in Figure 4.5.5, the weak speculation that samples with half the CNT content exhibited higher gauge factors than the average gauge factor at the same strain. The higher resistance with lower CNT content adds weight to this speculation. The correlation that a lower CNT content results in a higher initial resistance could be explained by a lower CNT surface area concentration since there are less CNTs between films of the same area. Assuming the CNTs are randomly oriented, this lower CNT surface area concentration may result in a weaker conductive network due to a lower number of conductive pathways throughout the film.   Figure 4.6.1: Plots of resistance values of various samples made with different overall CNT content. Sodium cholate surfactant samples are shown on the left and sodium dodecyl sulfate surfactant samples are shown on the right. For the SC samples, a difference of approximately 3 times the initial resistance is exhibited when the CNT content is halved. This difference is approximately 5 times the initial resistance for the SDS samples. For the samples fabricated using air flow evaporation casting, the initial resistances varied between 37.3 k? and 900 k? with one sample at 2555 ?. Excluding that one sample, the rest of  60  the initial resistances of airflow evaporation cast samples have resistances 1 to 3 orders of magnitude greater than the other evaporation cast samples. This can be seen in Figure 4.6.2.  Figure 4.6.2: Plot of resistance values of various samples made with different fabrication methods with a log scale for resistance. The air flow evaporation cast samples except for one sample exhibit a higher initial resistance than the other evaporation cast methods. The air flow cast samples generally have a higher initial resistance possibly because of the physics involved with the fabrication procedure. As mentioned previously, the droplets are under air flow and are pushed across the substrate. When this is done multiple times, especially when the air flow is perpendicular to the direction of strain, it enables a high variability in the uniformity of the film. It is also likely that the CNT content deposition has a high variability because of the fabrication procedure but tending towards a lower CNT content resulting in a high initial resistance. The air flow cast sample with the low resistance could be from the variability of the deposition process.  61  Due to an incomplete set of data with several factors of fabrication methods and ink compositions, it is difficult to draw a correlation between the initial resistance of the fabricated samples and their elicited gauge factors. However, the observations suggest lower CNT content results in a lower initial resistance. It is also observed that samples using SDS have a higher initial resistance in general compared to the SC samples. This could be due to different characteristics of how the two surfactants encapsulate the CNTs or also from the size of the molecules themselves. The two surfactants may also affect the bundling of CNTs which would affect the electrical characteristics. 4.7 Influence of individual CNT piezoresistivity As mentioned previously, there are three major mechanisms for resistance change. They are a change in tunneling between neighbouring CNTs, a change in the internal conductive network formed by CNTs and the CNT?s inherent piezoresistivity. It was also suggested in literature that individual CNT piezoresistivity may account for only a small percentage in overall resistance change for CNT composites. The small influence of individual CNT piezoresistivity in a composite could explain the huge disparity between the GF of individual CNTs (up to 1000) and the GF of the thin film CNT gauges presented in this paper (GF range of 0.1 to 4.0) and reported in literature (GF range of 0.7 to 60) [6], [13], [27], [28], [30]. Piezoresistivity is not a likely mechanism if the hypothetical resistance of a strain gauge where resistance is limited by individual CNTs is less than the experimental initial resistance of the strain gauges. A higher experimental strain gauge resistance would show that electrical conduction is dominated by other factors such as tunneling and CNT junctions as opposed to individual CNT resistances where piezoresistivity would occur. Equation 4.2 below is a crude calculation for a hypothetical strain  62  gauge where the ideal conduction path is through individual SWCNTs and ignores CNT junctions while taking precautions to err on a greater resistance:            (                 )(              )     (   )  Where   is the resistance of the CNT film,   is the length of the film between two of the four point probes (for comparison with experimental results),   is the cross section of the film,   is the conductivity of the SWCNT,    is a factor to increase the conductive path in the film by assuming the CNTs are positioned at an angle of 45 parallel to the length of the film,                   is a factor to assume only a certain percentage of the cross section of the CNT film is composed of CNTs and                is a factor representing the portion of CNTs which are of a certain type. The conductivity of metallic, quasi-metallic and semiconducting SWCNTs were obtained from literature where a group used four-point probe measurements to elicit conductivities on the order of 107 Sm-1, 107 Sm-1, and 106 Sm-1 respectively [49]. The                   is considered as 20% since most experimental samples have a CNT to surfactant ratio of 1:5 and it would also result in a greater hypothetical resistance. The                was determined from Sigma Aldrich product data and is considered to be 69% for semiconducting and quasi-metallic and 8% for metallic SWCNTs respectively. Since the thickness of the films was not measured, the thickness was estimated by determining the amount of surfactant and CNT in a typical sample film and by assuming the mass density and volumetric density to be equal to that of water. The calculation was followed up with the known width and length of each CNT film. These assumptions lead to calculations such as equation 4.2 and resistances shown in Table 4.7.1 below:  63    (        )  (        )(       )(       )(   )(    )        (   )  This crude approximation for strain gauge dominated by conduction through semiconducting SWCNTs results in a hypothetical resistance of 6 ?. Compared to some samples which exhibited similar characteristics whose average data is shown in Table 4.7.1, this is two orders of magnitude less, adding weight to the hypothesis that piezoresistivity is not a dominant mechanism in these CNT films. Using another calculation, the change in resistance of these hypothetical devices can be calculated by using GFs found in literature for each type of CNT exemplified in the equation below:               (   )  Using a GF of 150 for semiconducting CNTs from literature and assuming a 0.5% strain, the change in resistance can be calculated to be:                       (   )  Calculations for the other types of CNTs are included in Table 4.7.1 and it can be seen that the change in resistance is on the same order of magnitude as the experimental samples. Therefore, from a change in resistance perspective, the mechanism of piezoresistivity cannot be entirely ruled out. Table 4.7.1: Estimated values for hypothetical resistances of strain gauges and changes in resistance CNT type Conductivity CNT ratio Resistance  GF Change in Resistance Semiconducting 106 Sm-1 69% 6 ? 150 5 ?  64  CNT type Conductivity CNT ratio Resistance  GF Change in Resistance Quasi-metallic 107 Sm-1 69% 1 ? 1000 3 ? Metallic 107 Sm-1 8% 6 ? 60 2 ? Experimental data   500 ? 1.1 3 ?  In terms of conduction by tunneling, the sensitivity of a change in resistance from a tunneling mechanism is dependent on the CNT loading. The percolation threshold is indicative of an approximate CNT loading where conductive pathways are existent but not abundant. It is a boundary on the concentration of conductive particles such that the conductive particles are close enough to create a permanent conductive pathway (assuming a uniform dispersion of the particle in the matrix) but not an abundant number of pathways. For CNT loaded nanocomposite films, the percolation threshold is approximately 2 wt.% [50], [51]. Since the CNT thin films have a loading an order of magnitude greater, much higher than the percolation threshold, it is likely that electrical conductivity occurs through CNT junctions and contacts instead of being sensitive to tunneling. In light of these mechanisms, the combination of CNT types and a high CNT loading impair the mechanisms responsible for changes in resistance and could be the major reason for the low GFs. It is possible that the high CNT concentration and densities lower the resistance change sensitivity to tunneling and intrinsic CNT piezoresistivity mechanisms leaving only the conductive network mechanism. The strain induced changes to the conductive pathways is not expected to result in a large (non-linear) resistance change because of the many pathways available due to the high number of CNTs. The high CNT concentration and density has rendered the three main mechanisms of resistance change to be less effective thereby possibly  65  explaining the low GFs. Also, additional supporting literature reports that lower CNT density films result in higher GFs [46]. Considering the effect of alignment in CNT films with respect to these mechanisms, alignment may result in less overlap of CNTs and therefore a weaker conductive network. This effect would be favourable if, in the presence of alignment, strain still maintains its efficacy at breaking conductive pathways. The effect of alignment on tunneling is difficult to imagine as it depends on the number of in-plane tunneling contacts of the CNTs. For intrinsic piezoresistivity, alignment in the direction of strain might increase the strain experienced by the CNT. However, if the internal conductive CNT network still dominates conduction, the influence of intrinsic CNT piezoresistivity would still be low for high CNT loadings as mentioned previously.   66  5 Conclusion 5.1 Overview Encompassing this thesis, the goal was to investigate and develop a thin film CNT based strain gauge that is easy to fabricate, relatively inexpensive and exhibits a relatively high sensitivity. These novel CNT film strain gauges would be applied in applications where constant structural integrity is of importance for example in airplane fuselages and wings, in bridges or buildings, and in pipelines carrying oil. Biological applications are also viable for example, monitoring the structural integrity of bones post-surgery. Originally, it was hypothesized by Dr. Konrad Walus (NSERC Strategic Grant Application, 2010) that the very high piezoresistivity of carbon nanotubes could be taken advantage of in producing printable and very high performance strain gauges. It was suggested by him and seemingly verified through the work of Simon Beyer [37] and Dan Sik Yoo (unpublished) that good alignment of carbon nanotubes could produce high gauge factor devices. The method used to produce alignment involved evaporation casting and Dan Sik Yoo hypothesised that a better alignment could be achieved by decreasing the evaporation rate of evaporation cast films. To follow on from this early work, various methods of evaporation casting were investigated through decreasing the evaporation rates culminating in a chilled water bath method.  The range of gauge factors achieved in this study varied between 0.1 and 4.0, a disappointing result compared to the extremely high gauge factors claimed in individual carbon nanotubes of up to 1000 for quasi-metallic carbon nanotubes [13]. The highest gauge factors were exhibited by samples made from ink compositions using different surfactants and by the simplest method of evaporation casting which did not include any alterations to decrease evaporation rate. From  67  scanning electron microscopy and reflective polarized light microscopy, the attempts at achieving alignment resulted in only partial local areas of alignment on the surface at best. However, several samples exhibited a gauge factor of 1.1 ?0.1 up to a strain of 5000 ??. These gauge factors are reproducible within and across samples synthesised with variations of evaporation casting. The samples synthesised exhibited the general trend that a higher strain resulted in a higher gauge factor which is in accordance with literature [24]. Speculations from the data include that using sodium dodecyl sulfate over sodium cholate results in a higher initial resistance of the sample film and a higher strain requirement to elicit similar gauge factors. Also, samples fabricated with a lower overall carbon nanotube content tended to exhibit a greater initial resistance and a greater than average gauge factor. Unfortunately, no correlations between the initial resistance and gauge factors could be established. The samples fabricated with air flow evaporation casting mostly had a higher initial resistance than other samples likely due to the nature of the fabrication method. They also tended to exhibit lower than average gauge factors. The range of unexceptional gauge factors attained is within the same magnitude of most other gauge factors of carbon nanotube thin film strain gauges reported in literature (0.71 to 7.0) [27], [28], [30]. These results are not particularly surprising when the three mechanisms of resistance change are considered and also how the types of carbon nanotubes affect the resistance change. As discussed above, the intrinsic piezoresistivity of carbon nanotubes may have a small influence on the relative resistance change, tunneling is not a sensitive resistance change mechanism at high carbon nanotube loadings [24]. The initial resistances of the samples are also greater than the calculated resistance based on metallic single-walled carbon nanotubes suggesting carbon nanotube junctions and tunneling is the dominant factor in terms of resistance. These effects lend weight to the low gauge factors achieved in this work and reported in other literature.  68  In spite of the low gauge factors, the devices are inexpensive with a projected cost of only 5 US cents per device from a materials only perspective. This is much lower than common metal foil strain gauges which can be bought for as low as 6.26 USD. These devices are also fabricated with a very simple method of evaporation casting. Taking the very low cost and ease of fabrication into consideration, these carbon nanotube thin film strain gauges could be commercialised as an alternative to existing metal foil strain gauges. 5.2 Recommendations for future work Given the conclusions elucidated from the results, proposals for future direction to attain a higher sensitivity include using a lower carbon nanotube loading, controlling the carbon nanotube purity with respect to chirality and average length and attempting fabrication with increased evaporation rates or with different methods of deposition. More study pertaining to the effects of alignment on the resistance change mechanisms is also desired. It is also possible to move away from carbon nanotube thin films and investigate the use of conductive polymers such as PEDOT for use in thin film strain gauges. The decrease of carbon nanotube loading is to probe the possibility of the carbon nanotube film behaving like a carbon nanotube polymer composite thereby taking better advantage of resistance change mechanisms such as the intrinsic piezoresistivity of individual carbon nanotubes, tunnelling and maximising the effect of the change in the conductive network from applying strain by being closer to the percolation threshold. The increase in evaporation rate stems from the observation that decreased evaporation rates did not result in an improvement of gauge factors. For alignment, the samples investigated here exhibited only partial local alignment at best and other methods to induce alignment and their effects on the sensitivity are justified.  69  Further alternative directions of this research include the use of other different technologies such as microplotting and aerosol printing of the CNT ink. Microplotting relies heavily on surface interactions between the ink and the substrate surface overcoming capillary forces to plot a trail of ink. On the other hand, aerosol inkjet printing relies on a gas flow to disperse the ink into fine droplets. These methods take advantage of different deposition physics which may be useful for inducing alignment or may result in a different drying behaviour thereby affecting the properties of the carbon nanotube film with the possibility of changing the sensitivity.  70  Bibliography [1] P. Dharap, Z. Li, S. Nagarajaiah, and E. V Barrera, ?Nanotube film based on single-wall carbon nanotubes for strain sensing,? Nanotechnology, vol. 15, no. 3, pp. 379?382, Mar. 2004. [2] R. R. A. Chafy, M. H. Arafa, and A. M. K. Esawi, ?Fabrication of Carbon Nanotube/Low Density Polyethylene Composites for Strain Sensing,? Key Engineering Materials, vol. 495, pp. 33?36, Nov. 2011. [3] Y. Kanda, ?Piezoresistance effect of silicon,? Sensors and Actuators A: Physical, vol. 28, pp. 83?91, 1991. [4] O. S. Lyapkosova and N. G. Lebedev, ?Piezoresistive effect in single-walled carbon nanotubes,? Physics of the Solid State, vol. 54, no. 7, pp. 1501?1506, Jul. 2012. [5] N. Lu, C. Lu, S. Yang, and J. Rogers, ?Highly Sensitive Skin-Mountable Strain Gauges Based Entirely on Elastomers,? Advanced Functional Materials, vol. 22, no. 19, pp. 4044?4050, Oct. 2012. [6] D. Lee, H. P. Hong, M. J. Lee, C. W. Park, and N. K. Min, ?A prototype high sensitivity load cell using single walled carbon nanotube strain gauges,? Sensors and Actuators A: Physical, vol. 180, pp. 120?126, Jun. 2012. [7] M. Maiwald, C. Werner, V. Zoellmer, and M. Busse, ?INKtelligent printed strain gauges,? Sensors and Actuators A: Physical, vol. 162, no. 2, pp. 198?201, Aug. 2010. [8] Y.-T. Huang, S.-C. Huang, C.-C. Hsu, R.-M. Chao, and T. K. Vu, ?Design and fabrication of single-walled carbon nanonet flexible strain sensors.,? Sensors (Basel, Switzerland), vol. 12, no. 3, pp. 3269?80, Jan. 2012. [9] M. K. Njuguna, C. Yan, N. Hu, J. M. Bell, and P. K. D. V. Yarlagadda, ?Sandwiched carbon nanotube film as strain sensor,? Composites Part B: Engineering, vol. 43, no. 6, pp. 2711?2717, Sep. 2012. [10] C. C. Su, T. Liu, N. K. Chang, B. R. Wang, and S. H. Chang, ?Two dimensional carbon nanotube based strain sensor,? Sensors and Actuators A: Physical, vol. 176, pp. 124?129, Apr. 2012. [11] Q. Fan, Z. Qin, S. Gao, Y. Wu, J. Pionteck, E. M?der, and M. Zhu, ?The use of a carbon nanotube layer on a polyurethane multifilament substrate for monitoring strains as large as 400%,? Carbon, vol. 50, no. 11, pp. 4085?4092, Sep. 2012.  71  [12] R. J. Grow, Q. Wang, J. Cao, D. Wang, and H. Dai, ?Piezoresistance of carbon nanotubes on deformable thin-film membranes,? Applied Physics Letters, vol. 86, no. 9, p. 093104, 2005. [13] J. Cao, Q. Wang, and H. Dai, ?Electromechanical Properties of Metallic, Quasimetallic, and Semiconducting Carbon Nanotubes under Stretching,? Physical Review Letters, vol. 90, no. 15, p. 157601, Apr. 2003. [14] C. Stampfer, a Jungen, R. Linderman, D. Obergfell, S. Roth, and C. Hierold, ?Nano-electromechanical displacement sensing based on single-walled carbon nanotubes.,? Nano letters, vol. 6, no. 7, pp. 1449?53, Jul. 2006. [15] T. Tombler, C. Zhou, L. Alexseyev, J. Kong, H. Dai, L. Liu, C. Jayanthi, M. Tang, and S. Wu, ?Reversible electromechanical characteristics of carbon nanotubes under local-probe manipulation,? Nature, vol. 405, no. 6788, pp. 769?72, Jun. 2000. [16] J. Zhao, C. He, R. Yang, Z. Shi, M. Cheng, W. Yang, G. Xie, D. Wang, D. Shi, and G. Zhang, ?Ultra-sensitive strain sensors based on piezoresistive nanographene films,? Applied Physics Letters, vol. 101, no. 6, p. 063112, 2012. [17] S. Marinkovic, ?Carbon nanotubes,? Journal of the Serbian Chemical Society, vol. 73, no. 8?9, pp. 891?913, 2008. [18] R. Zhang, H. Deng, R. Valenca, J. Jin, Q. Fu, E. Bilotti, and T. Peijs, ?Carbon nanotube polymer coatings for textile yarns with good strain sensing capability,? Sensors and Actuators A: Physical, vol. 179, pp. 83?91, Jun. 2012. [19] M. a. Cullinan, R. M. Panas, C. M. DiBiasio, and M. L. Culpepper, ?Scaling electromechanical sensors down to the nanoscale,? Sensors and Actuators A: Physical, vol. 187, pp. 162?173, Nov. 2012. [20] M. a. Cullinan and M. L. Culpepper, ?Effects of chirality and impurities on the performance of carbon nanotube-based piezoresistive sensors,? Carbon, vol. 51, pp. 59?63, Jan. 2013. [21] E. Jeon, C. Park, J. A. Lee, M. Kim, K. Lee, H. So, C. Ahn, H. Chang, K. Kong, J. Kim, and J. Lee, ?Electromechanical properties of single-walled carbon nanotube devices on micromachined cantilevers,? Journal of Micromechanics and Microengineering, vol. 115010, no. 22, 2012. [22] E. Minot, Y. Yaish, V. Sazonova, J.-Y. Park, M. Brink, and P. McEuen, ?Tuning Carbon Nanotube Band Gaps with Strain,? Physical Review Letters, vol. 90, no. 15, p. 156401, Apr. 2003.  72  [23] M. a. Cullinan and M. L. Culpepper, ?Carbon nanotubes as piezoresistive microelectromechanical sensors: Theory and experiment,? Physical Review B, vol. 82, no. 11, p. 115428, Sep. 2010. [24] B. Hu, N. Hu, Y. Li, K. Akagi, W. Yuan, T. Watanabe, and Y. Cai, ?Multi-scale numerical simulations on piezoresistivity of CNT/polymer nanocomposites.,? Nanoscale research letters, vol. 7, no. 1, p. 402, Jan. 2012. [25] A. Bsoul, M. Sultan Mohamed Ali, A. Nojeh, and K. Takahata, ?Piezoresistive strain sensing using carbon nanotube forests suspended by Parylene-C membranes,? Applied Physics Letters, vol. 100, no. 21, p. 213510, 2012. [26] K. S. Karimov, F. A. Khalid, and M. T. S. Chani, ?Carbon nanotubes based strain sensors,? Measurement, vol. 45, no. 5, pp. 918?921, Jun. 2012. [27] Y. Miao, L. Chen, R. Sammynaiken, Y. Lin, and W. J. Zhang, ?Note: Optimization of piezoresistive response of pure carbon nanotubes networks as in-plane strain sensors.,? The Review of scientific instruments, vol. 82, no. 12, p. 126104, Dec. 2011. [28] B. Thompson and H.-S. Yoon, ?Aerosol Printed Carbon Nanotube Strain Sensor,? Proc. of SPIE, vol. 8346, p. 83461C?83461C?7, Apr. 2012. [29] A. Aldalbahi and M. in het Panhuis, ?Electrical and mechanical characteristics of buckypapers and evaporative cast films prepared using single and multi-walled carbon nanotubes and the biopolymer carrageenan,? Carbon, vol. 50, no. 3, pp. 1197?1208, Mar. 2012. [30] I. Kang, M. J. Schulz, J. H. Kim, V. Shanov, and D. Shi, ?A carbon nanotube strain sensor for structural health monitoring,? Smart Materials and Structures, vol. 15, no. 3, pp. 737?748, Jun. 2006. [31] T. Yamada, Y. Hayamizu, Y. Yamamoto, Y. Yomogida, A. Izadi-Najafabadi, D. N. Futaba, and K. Hata, ?A stretchable carbon nanotube strain sensor for human-motion detection.,? Nature nanotechnology, vol. 6, no. 5, pp. 296?301, May 2011. [32] P. a Withey, V. S. M. Vemuru, S. M. Bachilo, S. Nagarajaiah, and R. B. Weisman, ?Strain paint: noncontact strain measurement using single-walled carbon nanotube composite coatings.,? Nano letters, vol. 12, no. 7, pp. 3497?500, Jul. 2012. [33] A. a S. Mohammed, W. a Moussa, and E. Lou, ?High-performance piezoresistive MEMS strain sensor with low thermal sensitivity.,? Sensors (Basel, Switzerland), vol. 11, no. 2, pp. 1819?46, Jan. 2011. [34] F. Azhari and N. Banthia, ?Cement-based sensors with carbon fibers and carbon nanotubes for piezoresistive sensing,? Cement and Concrete Composites, vol. 34, no. 7, pp. 866?873, Aug. 2012.  73  [35] L. I. Yong, W. Wanlu, L. Kejun, and H. U. Chenguo, ?Piezoresistive effect in carbon nanotube films,? Chinese Science Bulletin, vol. 48, no. 2, pp. 11?13, 2003. [36] C. L. Cao, C. G. Hu, Y. F. Xiong, X. Y. Han, Y. Xi, and J. Miao, ?Temperature dependent piezoresistive effect of multi-walled carbon nanotube films,? Diamond and Related Materials, vol. 16, no. 2, pp. 388?392, Feb. 2007. [37] S. T. Beyer and K. Walus, ?Controlled orientation and alignment in films of single-walled carbon nanotubes using inkjet printing.,? Langmuir?: the ACS journal of surfaces and colloids, vol. 28, no. 23, pp. 8753?9, Jun. 2012. [38] M. S. Arnold, A. a Green, J. F. Hulvat, S. I. Stupp, and M. C. Hersam, ?Sorting carbon nanotubes by electronic structure using density differentiation.,? Nature nanotechnology, vol. 1, no. 1, pp. 60?5, Oct. 2006. [39] M. J. O. Connell, S. M. Bachilo, C. B. Huffman, V. C. Moore, M. S. Strano, E. H. Haroz, K. L. Rialon, P. J. Boul, W. H. Noon, J. Ma, R. H. Hauge, R. B. Weisman, and R. E. Smalley, ?Band Gap Fluorescence from Individual Single-Walled Carbon Nanotubes,? Science, vol. 297, no. 5581, pp. 593?596, 2002. [40] Q. Li, Y. T. Zhu, I. a Kinloch, and A. H. Windle, ?Self-organization of carbon nanotubes in evaporating droplets.,? The journal of physical chemistry. B, vol. 110, no. 28, pp. 13926?30, Jul. 2006. [41] S. Zhang and S. Kumar, ?Carbon nanotubes as liquid crystals.,? Small (Weinheim an der Bergstrasse, Germany), vol. 4, no. 9, pp. 1270?83, Sep. 2008. [42] E. Hobbie and D. Fry, ?Nonequilibrium Phase Diagram of Sticky Nanotube Suspensions,? Physical Review Letters, vol. 97, no. 3, p. 036101, Jul. 2006. [43] K. Iakoubovskii, ?Techniques of aligning carbon nanotubes,? Central European Journal of Physics, vol. 7, no. 4, pp. 645?653, May 2009. [44] J. Hedberg, L. Dong, and J. Jiao, ?Air flow technique for large scale dispersion and alignment of carbon nanotubes on various substrates Air flow technique for large scale dispersion and alignment of carbon nanotubes on various substrates,? vol. 143111, pp. 1?4, 2005. [45] G. Man, ?Towards all-polymer surface acoustic wave chemical sensors for air quality monitoring by,? UBC MASc Thesis, no. December, 2009. [46] Y. Miao, Q. Yang, L. Chen, R. Sammynaiken, and W. J. Zhang, ?Modelling of piezoresistive response of carbon nanotube network based films under in-plane straining by percolation theory Modelling of piezoresistive response of carbon nanotube network based films under in-plane straining by percolation theory,? Applied Physics Letters, vol. 063120, no. 101, 2012.  74  [47] K. J. Loh, J. P. Lynch, B. S. Shim, and N. a. Kotov, ?Tailoring Piezoresistive Sensitivity of Multilayer Carbon Nanotube Composite Strain Sensors,? Journal of Intelligent Material Systems and Structures, vol. 19, no. 7, pp. 747?764, Oct. 2007. [48] K. J. Loh, J. Kim, J. P. Lynch, N. W. S. Kam, and N. a Kotov, ?Multifunctional layer-by-layer carbon nanotube?polyelectrolyte thin films for strain and corrosion sensing,? Smart Materials and Structures, vol. 16, no. 2, pp. 429?438, Apr. 2007. [49] B. Gao, Y. Chen, M. Fuhrer, D. Glattli, and a. Bachtold, ?Four-Point Resistance of Individual Single-Wall Carbon Nanotubes,? Physical Review Letters, vol. 95, no. 19, p. 196802, Oct. 2005. [50]  a. Ferrreira, J. G. Rocha, a. Ans?n-Casaos, M. T. Mart?nez, F. Vaz, and S. Lanceros-Mendez, ?Electromechanical performance of poly(vinylidene fluoride)/carbon nanotube composites for strain sensor applications,? Sensors and Actuators A: Physical, vol. 178, pp. 10?16, May 2012. [51] Y. Kuronuma, T. Takeda, Y. Shindo, F. Narita, and Z. Wei, ?Electrical resistance-based strain sensing in carbon nanotube/polymer composites under tension: Analytical modeling and experiments,? Composites Science and Technology, vol. 72, no. 14, pp. 1678?1682, Sep. 2012.   75  Appendix Below are additional graphs pertaining to the discussion in the gauge factor measurements section: There are two sets of three graphs below as examples of other sets of data obtained from the measurements used to elicit the gauge factor.   76     77      78       79  The following two graphs are examples of data from measurements inappropriate for study.    80  The graph below shows the effect of heat using a heat gun before and after application. Notice the resistance decreases with the application of heat but increases again after the removal of heat.     81  The graph below shows the measurement of resistance over a long period of time to investigate for stabilizing or periodic effects. Both effects were not observed and the resistance seems to increase in noise.  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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


Related Items