{"Affiliation":[{"label":"Affiliation","value":"Science, Faculty of","attrs":{"lang":"en","ns":"http:\/\/vivoweb.org\/ontology\/core#departmentOrSchool","classmap":"vivo:EducationalProcess","property":"vivo:departmentOrSchool"},"iri":"http:\/\/vivoweb.org\/ontology\/core#departmentOrSchool","explain":"VIVO-ISF Ontology V1.6 Property; The department or school name within institution; Not intended to be an institution name."},{"label":"Affiliation","value":"Chemistry, Department of","attrs":{"lang":"en","ns":"http:\/\/vivoweb.org\/ontology\/core#departmentOrSchool","classmap":"vivo:EducationalProcess","property":"vivo:departmentOrSchool"},"iri":"http:\/\/vivoweb.org\/ontology\/core#departmentOrSchool","explain":"VIVO-ISF Ontology V1.6 Property; The department or school name within institution; Not intended to be an institution name."}],"AggregatedSourceRepository":[{"label":"Aggregated Source Repository","value":"DSpace","attrs":{"lang":"en","ns":"http:\/\/www.europeana.eu\/schemas\/edm\/dataProvider","classmap":"ore:Aggregation","property":"edm:dataProvider"},"iri":"http:\/\/www.europeana.eu\/schemas\/edm\/dataProvider","explain":"A Europeana Data Model Property; The name or identifier of the organization who contributes data indirectly to an aggregation service (e.g. Europeana)"}],"Campus":[{"label":"Campus","value":"UBCV","attrs":{"lang":"en","ns":"https:\/\/open.library.ubc.ca\/terms#degreeCampus","classmap":"oc:ThesisDescription","property":"oc:degreeCampus"},"iri":"https:\/\/open.library.ubc.ca\/terms#degreeCampus","explain":"UBC Open Collections Metadata Components; Local Field; Identifies the name of the campus from which the graduate completed their degree."}],"Creator":[{"label":"Creator","value":"Petel, Yael","attrs":{"lang":"en","ns":"http:\/\/purl.org\/dc\/terms\/creator","classmap":"dpla:SourceResource","property":"dcterms:creator"},"iri":"http:\/\/purl.org\/dc\/terms\/creator","explain":"A Dublin Core Terms Property; An entity primarily responsible for making the resource.; Examples of a Contributor include a person, an organization, or a service."}],"DateAvailable":[{"label":"Date Available","value":"2020-08-17T22:22:10Z","attrs":{"lang":"en","ns":"http:\/\/purl.org\/dc\/terms\/issued","classmap":"edm:WebResource","property":"dcterms:issued"},"iri":"http:\/\/purl.org\/dc\/terms\/issued","explain":"A Dublin Core Terms Property; Date of formal issuance (e.g., publication) of the resource."}],"DateIssued":[{"label":"Date Issued","value":"2020","attrs":{"lang":"en","ns":"http:\/\/purl.org\/dc\/terms\/issued","classmap":"oc:SourceResource","property":"dcterms:issued"},"iri":"http:\/\/purl.org\/dc\/terms\/issued","explain":"A Dublin Core Terms Property; Date of formal issuance (e.g., publication) of the resource."}],"Degree":[{"label":"Degree (Theses)","value":"Doctor of Philosophy - PhD","attrs":{"lang":"en","ns":"http:\/\/vivoweb.org\/ontology\/core#relatedDegree","classmap":"vivo:ThesisDegree","property":"vivo:relatedDegree"},"iri":"http:\/\/vivoweb.org\/ontology\/core#relatedDegree","explain":"VIVO-ISF Ontology V1.6 Property; The thesis degree; Extended Property specified by UBC, as per https:\/\/wiki.duraspace.org\/display\/VIVO\/Ontology+Editor%27s+Guide"}],"DegreeGrantor":[{"label":"Degree Grantor","value":"University of British Columbia","attrs":{"lang":"en","ns":"https:\/\/open.library.ubc.ca\/terms#degreeGrantor","classmap":"oc:ThesisDescription","property":"oc:degreeGrantor"},"iri":"https:\/\/open.library.ubc.ca\/terms#degreeGrantor","explain":"UBC Open Collections Metadata Components; Local Field; Indicates the institution where thesis was granted."}],"Description":[{"label":"Description","value":"In order to guide the development of innovative materials and their applications, a better understanding of the mechanisms that drive their unique properties is necessary. It has been widely observed that the well-established theory of driven and self-diffusion in highly diluted solutions does not directly apply to higher concentrations. The deviation from the existing theory of transport in high concentration materials is responsible, in part, for some interesting phenomena that present opportunities for innovative applications. Due to a limited number of direct measurement techniques, the mechanisms behind these phenomena remain unknown.\r\n\r\nIn this dissertation, nuclear magnetic resonance (NMR) is deployed as a tool to track the migration of magnetically visible species in complex systems. We track the transport of chromophores in electro-optical devices that can change their light-transmission properties with the application of voltage. We investigate room temperature ionic liquids and electrolyte salts in piezoionic materials, which are the basis for artificial nerves and muscles. Finally, we explore the initiating factors of crosslinker diffusion in vitrimers, a class of polymers that presents an opportunity for truly recyclable plastics. \r\n\r\nWe use the well-established technique of pulsed field gradient NMR (PFG-NMR) to measure self-diffusion and extend our measurements to electrophoretic mobility by using a new, low-cost, home-built electrophoretic NMR (eNMR) probe. eNMR development still faces a variety of application challenges. We overcome some of them by setting the driven diffusion in a direction perpendicular to the majority of undesired flows such as convection currents or bubbles. Using this new probe, we successfully measure electrophoretic mobilities of individual ions which accurately predict conductivities in concentrated solutions. \r\n\r\nBy measuring both driven and self-diffusion in a variety of materials, we explain some of the transport mechanisms that are behind unique material behaviours. In all the systems investigated, we find that some interaction between the ions, solvent, polymer, or a combination of the three, create interesting phenomena that alter the description of diffusion and mobility from known theory.","attrs":{"lang":"en","ns":"http:\/\/purl.org\/dc\/terms\/description","classmap":"dpla:SourceResource","property":"dcterms:description"},"iri":"http:\/\/purl.org\/dc\/terms\/description","explain":"A Dublin Core Terms Property; An account of the resource.; Description may include but is not limited to: an abstract, a table of contents, a graphical representation, or a free-text account of the resource."}],"DigitalResourceOriginalRecord":[{"label":"Digital Resource Original Record","value":"https:\/\/circle.library.ubc.ca\/rest\/handle\/2429\/75578?expand=metadata","attrs":{"lang":"en","ns":"http:\/\/www.europeana.eu\/schemas\/edm\/aggregatedCHO","classmap":"ore:Aggregation","property":"edm:aggregatedCHO"},"iri":"http:\/\/www.europeana.eu\/schemas\/edm\/aggregatedCHO","explain":"A Europeana Data Model Property; The identifier of the source object, e.g. the Mona Lisa itself. This could be a full linked open date URI or an internal identifier"}],"FullText":[{"label":"Full Text","value":"Transport in Ion Dense Mediaand Recyclable PolymersbyYael PetelB.Sc., Ben Gurion University of the Negev, 2010M.Sc., Weizmann Institute of Science, 2014A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinThe Faculty of Graduate and Postdoctoral Studies(Chemistry)THE UNIVERSITY OF BRITISH COLUMBIA(Vancouver)August 2020c\u00a9 Yael Petel, 2020The following individuals certify that they have read, and recommend to the Faculty ofGraduate and Postdoctoral Studies for acceptance, the dissertation entitled:Transport in Ion Dense Media and Recyclable Polymerssubmitted by Yael Petel in partial fulfilment of the requirements forthe degree of Doctor of Philosophyin ChemistryExamining Committee:Carl A. Michal, Associate Professor, Physics and Astronomy, UBCSupervisorSuzana K. Straus, Professor, Chemistry, UBCSupervisory Committee MemberGren N. Patey, Professor, Chemistry, UBCUniversity ExaminerEdmond Cretu, Associate Professor, Electrical & Computer Engineering, UBCUniversity ExaminerLouis A. Madsen, Professor, Chemistry, Virginia TechExternal ExaminerAdditional Supervisory Committee Members:Michael O. Wolf, Professor, Chemistry, UBCSupervisory Committee MemberAndrew W. MacFarlane, Associate Professor, Chemistry, UBCSupervisory Committee MemberiiAbstractIn order to guide the development of innovative materials and their applications, a betterunderstanding of the mechanisms that drive their unique properties is necessary. It hasbeen widely observed that the well-established theory of driven and self-diffusion in highlydiluted solutions does not directly apply to higher concentrations. The deviation fromthe existing theory of transport in high concentration materials is responsible, in part,for some interesting phenomena that present opportunities for innovative applications.Due to a limited number of direct measurement techniques, the mechanisms behind thesephenomena remain unknown.In this dissertation, nuclear magnetic resonance (NMR) is deployed as a tool to trackthe migration of magnetically visible species in complex systems. We track the transportof chromophores in electro-optical devices that can change their light-transmission prop-erties with the application of voltage. We investigate room temperature ionic liquids andelectrolyte salts in piezoionic materials, which are the basis for artificial nerves and mus-cles. Finally, we explore the initiating factors of crosslinker diffusion in vitrimers, a classof polymers that presents an opportunity for truly recyclable plastics.We use the well-established technique of pulsed field gradient NMR (PFG-NMR) tomeasure self-diffusion and extend our measurements to electrophoretic mobility by using anew, low-cost, home-built electrophoretic NMR (eNMR) probe. eNMR development stillfaces a variety of application challenges. We overcome some of them by setting the driveniiiAbstractdiffusion in a direction perpendicular to the majority of undesired flows such as convectioncurrents or bubbles. Using this new probe, we successfully measure electrophoretic mobil-ities of individual ions which accurately predict conductivities in concentrated solutions.By measuring both driven and self-diffusion in a variety of materials, we explain some ofthe transport mechanisms that are behind unique material behaviours. In all the systemsinvestigated, we find that some interaction between the ions, solvent, polymer, or a com-bination of the three, create interesting phenomena that alter the description of diffusionand mobility from known theory.ivLay SummaryIn order to help guide the development of new materials, we must obtain a better under-standing of the mechanisms behind unique material properties. One of the drivers of theunique properties in artificial muscles, electro-optical devices and recyclable polymers, ismolecular transport. One difficulty of investigating molecular transport in these uniquematerials is that we cannot look inside them while they are being used in devices. Nuclearmagnetic resonance (NMR) is a tool that can change that; similarly to how an MRI canobtain images of living people to see what is inside, NMR can \u201clook\u201d inside materials, andtrack how the molecules move inside while a device is operating. Using NMR, we can learnhow molecular transport influences the properties of these materials.My research focuses on the development of this method, extending its ability, andenabling measurements of transport in new materials.vPrefaceEvery single part of this dissertation was done with help, supervision and support fromDr. Carl A. Michal.The home-built PFG self-diffusion probe and resonance driver circuit used in chapter4-7 were originally built by Tso and Michal [1] and improved by Michan [2]. I have repairedand improved the gradient coil of the probe, increasing the maximal pulse field gradientstrength by 17.8%. The self-diffusion pulse sequences were written by Carl Michal andAlison Michan.The eNMR probe described in chapter 3 and used in chapter 5 was designed and builtby me. The eNMR sample holder was designed and built by me. The applied voltagedriving circuit was designed by Carl Michal and built by me. The eNMR pulse sequencesand the connected micro-controller program were written by me.Chapter 4 of this dissertation is based on the published paper: [3] V. Woehling, G.T.M.Nguyen, C. Plesse, Y. Petel, Y. Dobashi, J.D.W. Madden, C.A. Michal and F. Vidal,Multifunctional Materials, 2, 045002 (2019) https:\/\/doi.org\/10.1088\/2399-7532\/ab56a2Figures 4.1, 4.2 and 4.8 are used with permission from Woehling et al. (2019) of which I aman author. In this work, the experiments were suggested by Carl Michal and Giao T. M.Nguyen of Universite\u00b4 de Cergy-Pontoise. Samples were all prepared by Vincent Woehlingof Universite\u00b4 de Cergy-Pontoise. All NMR experiments were performed and analyzed byme. Carl Michal and I developed the data interpretation together.viPrefaceChapter 5 of this dissertation is a collaboration between the Michal lab and Maddengroup of Electrical Engineering at UBC. The ideas for these experiments are a result of anongoing discussion between all collaborators. All samples were prepared by Yuta Dobashiand me. The vast majority of voltage measurements presented in this dissertation weredone by Yuta. All NMR and eNMR experiments were performed and analyzed by me.Great discussions between all collaborators have led to the mechanism descriptions in thischapter. Viscosity measurements were done by me with the help of Anindya Lal Royfrom Dr. Konrad Walus\u2019s lab at Quantum Matter Institute, UBC. The experimental workpresented in this chapter was done during 2018 and presented in a conference on April2019. A similar lithium conduction mechanism was suggested in a research paper fromanother group, published in April 2019 [4], including computational work and conductivitydata. We were not aware of this paper until the final stages of this dissertation preparationand reached a similar suggested mechanism independently. We believe our experimentalwork, which directly measures lithium mobility, strengthens the conclusions reached byboth groups.Chapter 6 was done in collaboration with SWITCH Materials Inc. All sample materialswere prepared by Dr. Glen Bremner of SWITCH Materials Inc. All NMR experiments andanalysis were done by me. The applied voltage set-up was designed and constructed by mewith the help of Carl Michal. Measurements of transitions time for the optical filters thatare presented in Table 6.2 were done by Dr. Glen Bremner. Interpretation of the data wasdone by me and Carl Michal, with consultation from SWITCH Materials Inc.Chapter 7 of this dissertation is a collaboration between the Michal lab and the Wolfgroup of Chemistry at UBC. The idea for the initial experiments came from Taylor D.Wright and his advisor, Dr. Michael Wolf. All samples have been synthesized by Taylorand the procedure has been published in [5]. Schemes 7.1, 7.2, A.1 and Figure 7.7 are usedviiPrefacewith permission from Taylor D. Wright. All NMR experiments were performed by me.Sample preparation and sealing have been done by me with assistance from Taylor. Allinterpretation of the data has been done by me and Carl Michal, with discussions withTaylor and Dr. Wolf leading to further suggested experiments.To help produce the final version of this dissertation, comments to the draft have beenmade by committee members Dr. Suzana Straus and Dr. Andrew MacFarlane.viiiTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiiList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xivList of Schemes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviiiList of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xixList of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxiAcknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxivDedication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xxv1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 High Concentration Solutions and Room Temperature Ionic Liquids . . . . 2ixTable of Contents1.1.1 Ion Interactions in Media . . . . . . . . . . . . . . . . . . . . . . . . 41.1.2 Diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91.1.3 Electrophoresis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131.1.4 Relation Between Electrophoresis and Self-Diffusion . . . . . . . . . 161.2 Research Aims and Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . 192 Magnetic Resonance Transport Measurements: Theory and Background 222.1 Pulsed Field Gradient-NMR . . . . . . . . . . . . . . . . . . . . . . . . . . 232.2 eNMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262.2.1 Signal Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272.2.2 eNMR Limitations and Probe Design Solutions . . . . . . . . . . . 292.2.3 Pulse Sequence Based Solutions . . . . . . . . . . . . . . . . . . . . 353 Magnetic Resonance Transport Measurements: Methods . . . . . . . . 393.1 NMR Spectrometers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393.2 Home built PFG Probe and Driving Circuit . . . . . . . . . . . . . . . . . 403.3 Data Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 413.3.1 Temperature Control . . . . . . . . . . . . . . . . . . . . . . . . . . 423.4 Bruker Imaging 3-axes Gradient Set . . . . . . . . . . . . . . . . . . . . . . 443.5 eNMR Probe Construction and Validation . . . . . . . . . . . . . . . . . . 453.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453.5.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 453.6 Data Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 493.6.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 504 Diffusion of EMI-TFSI Dilutions in Propylene Carbonate . . . . . . . . 524.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52xTable of Contents4.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 564.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 574.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 685 Ion Transport in Touch Sensor Electrolytes . . . . . . . . . . . . . . . . . 705.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 705.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 725.2.1 Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . 725.2.2 Viscosity Measurements . . . . . . . . . . . . . . . . . . . . . . . . . 725.2.3 Self-Diffusion Measurements . . . . . . . . . . . . . . . . . . . . . . 735.2.4 eNMR Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . 735.2.5 Conductivity and Piezoionic Measurements . . . . . . . . . . . . . . 755.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 765.3.1 Self-Diffusion in Electrolyte Samples . . . . . . . . . . . . . . . . . . 775.3.2 Self-Diffusion in Polymer Electrolyte Samples . . . . . . . . . . . . 845.3.3 Driven Diffusion in Solution Electrolyte Samples . . . . . . . . . . . 885.3.4 Driven Diffusion in Polymer Electrolyte Samples . . . . . . . . . . . 965.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 986 Chromophore Diffusion in Optical Filter Devices . . . . . . . . . . . . . . 1006.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1006.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1026.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1066.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1187 Vitrimers - Transport of Thermally Exchangeable Molecules . . . . . . 1207.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120xiTable of Contents7.2 Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1227.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1277.4 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1388 Conclusions and Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . 1398.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1398.2 Recommended Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . 1428.2.1 eNMR Probe Construction and Validation . . . . . . . . . . . . . . 1428.2.2 Diffusion of EMI-TFSI Dilutions in Propylene Carbonate . . . . . . 1428.2.3 Ion Transport in Touch Sensor Electrolytes . . . . . . . . . . . . . . 1438.2.4 Chromophore Diffusion in Optical Filter Devices . . . . . . . . . . . 1448.2.5 Vitrimers - Transport of Thermally Exchangeable Molecules . . . . 1458.3 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147AppendicesA Vitrimer Sample Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . 170xiiList of Tables4.1 Ratio of D coefficients (D+D\u2013) for EMI-TFSI in solution and IPN . . . . . . . 626.1 D comparison of selected chromophores in crosslinked samples . . . . . . . . 1126.2 Transition time and D comparison . . . . . . . . . . . . . . . . . . . . . . . 1136.3 Chromophore D comparison with and without applied V . . . . . . . . . . . 1187.1 D of vitrimer sample PD1a vs. temperature and time. . . . . . . . . . . . . 1277.2 Initial Ds in PDMS backbone vitrimers . . . . . . . . . . . . . . . . . . . . 1317.3 D of heterogeneously loaded vitrimer vs. time . . . . . . . . . . . . . . . . . 1337.4 Ds of vitrimer sample PD1c under different mechanical stress . . . . . . . . 136xiiiList of Figures1.1 Structure of common RTILs . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2 Types of ion pairs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.3 Supra-molecular ionic structures . . . . . . . . . . . . . . . . . . . . . . . . 72.1 Pulsed Field Gradient(PFG)-echo pulse sequence diagram . . . . . . . . . . 232.2 Pulsed Field Gradient Stimulated Echo (PFG-STE) pulse sequence diagram 252.3 Spin echo eNMR pulse sequence diagram . . . . . . . . . . . . . . . . . . . . 282.4 Conventional arrangements for eNMR sample cells . . . . . . . . . . . . . . 302.5 Conventional arrangements for eNMR sample cells with capillary tubes . . 332.6 Artifact flow suppression CPMG-eNMR pulse sequence diagram . . . . . . 362.7 Double Stimulated Echo-eNMR pulse sequence diagram . . . . . . . . . . . 373.1 Signal intensity vs. magnetic field gradient for calibration measurements . . 423.2 Temperature control set-up illustration . . . . . . . . . . . . . . . . . . . . . 433.3 Diagram of constructed eNMR probe . . . . . . . . . . . . . . . . . . . . . . 473.4 Charged particle distribution diagram . . . . . . . . . . . . . . . . . . . . . 483.5 eNMR pulse sequence diagram . . . . . . . . . . . . . . . . . . . . . . . . . 493.6 Measured phase shift under applied voltage . . . . . . . . . . . . . . . . . . 513.7 Zeroth order phase vs. constant current . . . . . . . . . . . . . . . . . . . . 514.1 Experimental setup for sensor characterization . . . . . . . . . . . . . . . . 54xivList of Figures4.2 EMI-TFSI sensor and actuator mode measurements . . . . . . . . . . . . . 544.3 EMI-TFSI ionic conductivity vs concentration . . . . . . . . . . . . . . . . . 554.4 Structural formula of RTIL EMI-TFSI and co-solvent PC . . . . . . . . . . 564.5 19F spectra of 3 M EMI-TFSI\/PC in PEO-NBR IPN . . . . . . . . . . . . . 574.6 Solution and IPN 1H spectra comparison for 3 M EMI-TFSI in PC . . . . . 584.7 2.50 M EMI-TFSI\/PC self-diffusion sample of TFSI\u2013 anion in PEO-NBR IPN 594.8 Self-diffusion summary for EMI-TFSI\/PC solution and IPN . . . . . . . . . 604.9 Estimated conductivity derived from self-diffusion measurements. . . . . . . 654.10 Ratio of directly measured conductivity \u03c3mes to estimated \u03c3NMR . . . . . . 664.11 Co\u2013solvent PC self-diffusion vs. RTIL concentration . . . . . . . . . . . . . 675.1 Structures of Li+ and bis(trifluoromethane)-sulfonimide (TFSI\u2013) ions . . . 725.2 Structure of Propylene Carbonate (PC) . . . . . . . . . . . . . . . . . . . . 735.3 Poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) co-polymer . 735.4 Voltage response to deformation of PVDF-HFP with varying LiTFSI conc. 765.5 2 M LiTFSI\/PC self-diffusion sample of TFSI\u2013 anion in solution . . . . . . . 795.6 2 M LiTFSI\/PC self-diffusion sample of Li+ cation in solution . . . . . . . . 795.7 D of LiTFSI vs. electrolyte concentration . . . . . . . . . . . . . . . . . . . 805.8 Dynamic viscosity and derived effective r of LiTFSI solutions vs. concentration 815.9 Total conductivity derived from PFG-NMR self-diffusion measurements ofLiTFSI vs. salt concentration in PC . . . . . . . . . . . . . . . . . . . . . . 835.10 Transport number derived from self-diffusion measurements of LiTFSI vs.salt concentration in PC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 845.11 2 M LiTFSI\/PC self-diffusion sample of TFSI\u2013 anion in polymer . . . . . . 855.12 2 M LiTFSI\/PC self-diffusion sample of Li+ cation in polymer . . . . . . . . 86xvList of Figures5.13 Self-diffusion measurements of LiTFSI vs. electrolyte concentration embed-ded in PVDF-HFP co-polymer . . . . . . . . . . . . . . . . . . . . . . . . . 875.14 Total conductivity and transport numbers derived from PFG-NMR self-diffusion measurements of LiTFSI vs. salt concentration in Gel samples . . 885.15 NMR spectra of 2.0 M LiTFSI in PC as a function of applied electric field . 905.16 Electrophoretic mobility measurements for LiTFSI vs. salt concentration inPC as measured by eNMR . . . . . . . . . . . . . . . . . . . . . . . . . . . . 915.17 Transport numbers vs. salt concentration in PC derived from eNMR . . . . 955.18 zeff derived from eNMR measurements vs. salt concentration in PC . . . . . 965.19 Total conductivity derived from eNMR mobility measurements of LiTFSIvs. salt concentration in PC . . . . . . . . . . . . . . . . . . . . . . . . . . . 976.1 Chromophore inter-conversion general structure . . . . . . . . . . . . . . . . 1016.2 Chromophores structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1036.3 Schematic and wiring diagram of gelled sample between electrodes . . . . . 1056.4 General molecular structure of blue chromophore and 1D 19F NMR spectraof S158 for both light and partially dark state . . . . . . . . . . . . . . . . . 1066.5 19F NMR measurements of S164 chromophore in the open state dissolvedin RhodiasolvR\u00a9 IRIS\/BC . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1096.6 General structure of the open state chromophore in parallel and anti-parallelarm positions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1106.7 S164 19F NMR measurements vs. temperature . . . . . . . . . . . . . . . . 1116.8 Self-diffusion measurements vs. temperature for 4 chromophores . . . . . . . 1136.9 Uncrosslinked full formulation sample S164 left at 250 K overnight . . . . . 1146.10 Chromophore diffusion example for S109 . . . . . . . . . . . . . . . . . . . . 1156.11 Chromophore diffusion example with applied voltage of S109 . . . . . . . . 116xviList of Figures7.1 PDMS-NH2 polymer structure . . . . . . . . . . . . . . . . . . . . . . . . . 1247.2 Exchangeable crosslinker 1,12-dodecane-bis-\u03b2-ketoenamine n-butyl . . . . . 1247.3 Non-exchangeable crosslinker triethylene glycol dimethacrylate. . . . . . . . 1247.4 Exchangeable tag 2,2,2-Trifluoroethylacetoacetate . . . . . . . . . . . . . . . 1247.5 Nonexchangeable tag 2,2,2-trifluoroethyl methacrylate . . . . . . . . . . . . 1247.6 4-trifluoromethyl-4\u2019-methylacetoacetate azobenzene tag . . . . . . . . . . . 1257.7 Steps of heterogeneously loading a sample with coloured tag molecule . . . 1267.8 Sample PD1a D as heated from room temperature to 100 \u25e6C . . . . . . . . 1287.9 Sample PD1b D at 60 \u25e6C before and after 10 hours . . . . . . . . . . . . . . 1307.10 Relaxation times of sample PD1 vs temperature . . . . . . . . . . . . . . . 1327.11 Explanation of the mechanical stress in the sample at different steps . . . . 1367.12 Pictures of sample PG1 before and after physical stress application . . . . . 137A.1 2,2,2-Trifluoroethylacetoacetate. . . . . . . . . . . . . . . . . . . . . . . . . . 170A.2 1,12-dodecane-bis-\u03b2-ketoenamine n-butyl . . . . . . . . . . . . . . . . . . . 170A.3 4-trifluoromethyl-4\u2019-methylacetoacetate azobenzene . . . . . . . . . . . . . . 171A.4 PDMS-NH2 polymer structure . . . . . . . . . . . . . . . . . . . . . . . . . 172xviiList of Schemes7.1 General mechanism of thermally exchangeable covalent bonds in vitrimers. 1227.2 Vinylogous urethane exchange reaction of a pendant molecule . . . . . . . . 123A.1 Synthesis steps for F labelled tag . . . . . . . . . . . . . . . . . . . . . . . . 172xviiiList of AbbreviationsBC butylene carbonateDLS dynamic light scatteringeNMR electrophoretic NMRFID free induction decayIPN interpenatrating polymer networkLN2 liquid nitrogenMRI magnetic resonance imagingNBR nitrile butadiene rubberNMR nuclear magnetic resonanceNOE nuclear Overhauser effectPC propylene carbonatePEO poly(ethylene oxide)PFG-NMR pulsed field gradient NMRRF radio frequencyxixList of AbbreviationsRMS root mean squareRPM restrictive primitive modelRTIL room temperature ionic liquidSNR signal to noiseSTE stimulated echoTEAP tetramethylammonium hexafluorophosphatexxList of Symbols\u03b1 Correction parameter to electrophoretic migration used for ionic conductivity calculation(Eq. 1.10 )b Parameter from Stoke-Einstein equation (Eq. 1.5). Reflects the boundary conditions forthe velocity of the surrounding at the surface of the solute. Depends on the frictionbetween the solvent and soluteB0 Static magnetic field\u03c7 Ionic strengthC Concentration\u2206 Diffusion time: the time delay between the magnetic field gradient pulses\u03b4 Magnetic field gradient pulse durationD Self-diffusion coefficientE Electric fielde Electron charge\u000f Dielectric constant\u000f0 Premittivity of vacuumxxiList of Symbols\u03b7 Kinematic viscosityF Faraday constantF Forcef Friction drag factorg Magnetic field gradient strength\u03b3 Gyromagnetic frequencyI CurrentJ Particle flux\u03ba\u20131 Debye lengthkB Boltzmann constant\u03bbB Bjerrum length\u00b5 Electrophoretic mobility coefficientNA Avogadro\u2019s number\u03bd Drift velocity\u03c9 Larmor frequencyq Electric chargeQ Heatr Effective radiusxxiiList of SymbolsR Resistence\u03c3 Ionic conductivityT1 Spin-lattice relaxation timeT2 Spin-spin relaxation time\u03c4 Echo timeT Temperaturet Timeti Transport number for a particular ionic speciesU Electric potentialv VelocityV Voltagez Charge valencezeff Effective chargexxiiiAcknowledgementsFirst and foremost, I would like to thank my advisor Carl. Your knowledge knows nolimits and it was a pleasure learning from someone who is a specialist in so many fields.You combine professionalism with kindness and patience in a way that very few can. I feellucky to have had the chance to learn from you. You are truly an amazing mentor.To all the past and present students of Room 100, thank you for always being thereto lend a hand, be a shoulder to cry on, and to be quick to hand over a beer. Theconversations we had were sometimes ridiculous and sometimes enriching, but they werealways a pleasure.To my Vancouver friends, thank you for welcoming me so quickly into your lives andfor pronouncing hummus the proper way (at least when I\u2019m around).To my friends back home, thank you for supporting me through this adventure. I stillwish for a direct flight between us, but sadly, that\u2019s not part of my research.Dave, your love is the thing that got me here in one piece. I am so happy we foundeach other.To my dearest family, I owe every single accomplishment to your love, support, andfaith in me. You are the ones who made me believe that nothing is impossible.And to Pippin, thanks for all the licks.xxivDedicationxxvChapter 1IntroductionMany interesting and useful properties seen in new materials are based on the transportof molecules and ions. An understanding of this transport is essential to guide the de-velopment of innovative materials. To achieve better control over mechanical properties,reaction rates, and conductivities one must obtain better understanding of the molecularmechanisms which result in such properties. Many experimental and theoretical studieshave been devoted over the last century to the diffusivity of dilute solutions [6, 7, 8, 9, 10]yet, much less to the investigation of concentrated solutions. Although it has been widelyobserved that classical theory does not directly apply to the diffusion of particles outsidethe realm of infinitely diluted solutions [11, 12], a lot remains unknown due to the relativelysmall number of experimental techniques by which diffusion may be measured [13, 14].In this dissertation, nuclear magnetic resonance (NMR) is deployed as a tool to trackthe migration of magnetically visible molecules and ions, in a variety of media, to try andisolate the mechanisms of both self and driven diffusion. We will discuss the thermallyand electrically driven diffusion of ions in varying concentration solutions and electrolyteswollen polymer gels, the diffusion of chromophore molecules in novel optical devices, andthe stress induced diffusion of crosslinker molecules in a new class of recyclable polymers.The experiments presented in the following chapters were performed in collaborationwith several research and industry groups that produce unique polymer-based materialsfor a variety of applications. The samples contain solutions that are soaked or embedded11.1. High Concentration Solutions and Room Temperature Ionic Liquidsin semi-solids such as polymers, gels, and gel electrolytes. These semi-solids combine be-haviours of solids and liquids, having relatively fast diffusion and a firm physical structure.This combination of properties enables the production of flexible, transparent conductivedevices such as solar cells, batteries, screens and sensors. Still, due to strong interactionsbetween the ions and their surrounding environment, many questions remain unansweredregarding the diffusion and conduction mechanisms in this type of medium. With the ever-increasing importance of technology based on these materials, a detailed understanding ofthe microscopic mechanisms of mass transport is essential.In all of these projects, some interaction between the ions, solvent, polymer or a com-bination of the three create interesting phenomena that alter the diffusion and mobilityof the dissolved species. By combining strong magnetic field gradients and short gradientpulses, pulsed field gradient NMR can provide measurements of diffusion coefficients fora high range of concentrated solutions and semi-solid samples. Using this technique toobtain specific diffusion coefficients for each species in a variety of polymer based materialswill help to better explain some of the underlying phenomena. Applying an electric fieldto the samples allowed us to measure mobility constants and observe ionic conductivityin-situ and help shed light on the mechanisms and interactions driving them.1.1 High Concentration Solutions and Room TemperatureIonic LiquidsMany applications of electroactive gels, including some of the materials to be investigatedherein, are based on one of the most interesting types of ion-dense materials: ionic liquids.Room temperature ionic liquid (RTIL), such as EMI-PF6, EMIM-BF4 (see structures inFig. 1.1, EMI-TFSI (Fig. 4.4), and many others, are salts that have melting points lower21.1. High Concentration Solutions and Room Temperature Ionic Liquidsthan room temperature. Most of the interest in RTILs has been due to their negligiblevapour pressure and low flammability. This, among other unique properties, makes themgood candidates as non-aqueous polar solvents that are also non-coordinating [15]. Theseproperties have made RTILs candidates to replace organic solvent based electrolytes inelectrochemical technologies such as super capacitors, batteries, organic dye sensitized solarcells, and actuators. However, although RTILs are comprised of ions entirely, the roomtemperature conductivity of pure RTILs is currently too low for them to become profitablereplacements for existing organic electrolyte solutions. By introducing a co-solvent, theionic conductivity of RTILs can be tuned and improved. Common existing electrolytesare made from salts dissolved in organic solvents. These mixtures usually suffer fromlimited salt solubility, which also limits the working temperature range [16, 17]. The factthat RTILs are completely miscible with organic solvents presents an advantage that alsoallows for a wider range of stable working temperatures.Figure 1.1: The structure of commonly used room temperature ionic liquids (RTILs).One of the advantages of RTILs as a group is the large variety of physio-chemicalproperties they have. By choosing the correct combination of ions, it is possible to change31.1. High Concentration Solutions and Room Temperature Ionic Liquidschemical and physical properties, optimizing them specifically for a desired use [18]. Yet,predicting the physio-chemical properties from the chemical structure is complicated. Oneof the current challenges results from the complexity of interactions between the ionsof RTILs. These unique interactions distinguish the physical behaviour of RTILs fromtraditional solvents and consequently, makes their design difficult.In general, solvents are classified according to their polarity and dielectric constant.These scales are helpful in predicting the solvent\u2019s properties such as its ability to dissolveionic solute. RTILs have shown contradictory results, depending on the method used tomeasure polarity and their solvent behaviour does not follow the polarity scale [19]. Itis agreed upon that a quantitative description of the ion association can be used as auseful indicator to characterize RTILs. A good correlation was shown between \u201cionicity\u201d(level of ionic association) and physio-chemical properties such as polarity, surface tension,density, and conductivity [20]. RTILs\u2019 relatively low conductivity and its dependence ondilution leads to the assumption that they form ion pairs or higher-order supramolecularstructures [21, 22].In order to enable optimization of ionic conductivity, an accurate technique is neededfor measuring microscopic diffusive transport. This method must be usable with differentcombinations of co-solvents and RTILs, under a variety of physical and chemical conditions.1.1.1 Ion Interactions in MediaIn order to better understand ion transport in electrolyte solutions, a variety of interactionsmust be considered. Interactions between ions, solvents and other components in solutionwill affect the motion of ions and will play a major part in determining the transport. Thissection will present some of the existing theories of interactions of ions in solution andporous media.41.1. High Concentration Solutions and Room Temperature Ionic LiquidsIon-Solution InteractionsWhen describing an electrolyte solution, one must think of the local electric field distribu-tion around each ion and how it affects the surrounding molecules. As known from basicelectrostatics, any charged particle creates an electric potential that decays with distance.The extent of this electric potential depends on both the charge distribution of the particleitself as well as the surrounding dielectric constant. The Debye length, or electrostaticscreening length, is a measure of the electrostatic effect a charged particle has in solu-tion (see Eq. 1.1). This measure of distance arises from the Poisson-Boltzmann equationsolved for a spherical charge in a continuous medium under the restrictive primitive model(RPM) [23]. In this classical model, equal numbers of positive and negative ions in solutionare modelled as hard spheres [24]. The Debye length describes the scale for the exponentialdrop of potential with distance from the ion due to the dielectric screening effect of themedia [25]. For electrolyte solutions, the Debye length is usually noted as \u03ba\u20131 and for amonovalent symmetrical electrolyte is given as\u03ba\u20131 =\u221a\u000f0\u000fkBT2NAe2\u03c7(1.1)where \u000f0 is the permittivity of vacuum, \u000f the solution dielectric constant, kB the Boltzmannconstant, T the absolute temperature, e the electron charge, \u03c7 the ionic strength, and NAis Avogadro\u2019s number.Keeping the RPM in mind, the potential from a central ion attracts counterions in asphere with radius equal to \u03ba\u20131. This sphere surrounds the ion with a \u201ccloud\u201d of moleculeswith a total effective charge opposite to that of the central ion. This \u201ccloud\u201d is composedof both solvent and counterions that screen the electric potential from the central ion. Thecounterions in this cloud are in constant movement and only have an average, effective51.1. High Concentration Solutions and Room Temperature Ionic Liquidscharge; they are not fixed to the molecule [26]. In a highly diluted solution, this cloudconsists entirely of solvent molecules and is termed a solvation layer. Together with theionic radius, the Debye length determines the size of the solvation layer around an ion,which in turn will determine the ion effective size, weight, and its tendency to aggregatewith other ions in solution.Figure 1.2: Types of ion pairs. Dark circles represent solvent molecules and light circlesrepresents ions. The Debye lengths of the ions must be at least as long as the distancebetween the ions in order for the ion-pairs to exist. Therefore, the Debye length for anion in a contact ion pair can be short, while it must be much longer for ions in a solventseparated pair. An illustration of the Debye length for the anion is marked as a greydashed line.In addition to the interactions with the solvation layer, ions also experience long-rangeelectrostatic forces. The counterions attract each other according to Coulomb\u2019s law, makingit possible for ion-pairs to form. This attraction is masked by the permittivity of thesolution. One of the theories to define ion pairs was suggested by Bjerrum [27]. TheBjerrum length, \u03bbB, given by\u03bbB =z\u2013z+e24pi\u000f0\u000fkBT(1.2)determines the cut-off distance at which associated ions with charges z+ and z\u2013 are consid-61.1. High Concentration Solutions and Room Temperature Ionic Liquidsered paired. \u03bbB is the distance at which the electrostatic interaction energy is comparableto the thermal energy. If the distance between the ions is smaller than \u03bbB the ions areconsidered paired and will form a long-lasting aggregate.This means the ions are arranged in a non-rigid structure, such as those in Fig. 1.3,in which the ions\u2019 motion is correlated for a time longer than would be taken for non-associated species to diffuse over the distance between them [28, 29]. In this range ofdistance, the solvation layers might be present to varying extent and several types ofion pairs, as described in Fig. 1.2, can be distinguished. In case of a highly concentratedsolution, such as an RTIL, ion pairs might form bigger, supramolecular structures consistingof more than two ions as can be seen in Fig. 1.3. When ions are bound in these structures,their effective radius, r, changes to some combination of the ions and solvent. The netcharge of this entity is the sum of its ionic components and, as for an ion-pair, might beneutral. For water at room temperature with 1 M ionic strength \u03bbB \u2248 2\u03ba\u20131.Figure 1.3: Supra-molecular ionic structures. Dark circles represent anions and lightcircles represents cations which are held together by electrostatic forces represented bydotted lines. Figure adapted from Dupont et al. [22] with permission.71.1. High Concentration Solutions and Room Temperature Ionic LiquidsIon-Polymer InteractionsSince the discovery of the first ion conducting polymer four decades ago [30], and the demon-stration of an all-solid-state battery cell [31], significant effort has been spent developingefficient polymer electrolytes. Polymer electrolytes are a safer, more durable replacementfor liquid electrolytes in electrochemical devices. Polymer electrolytes can be described asmembranes with transport properties comparable to liquid ionic solutions, which are usedin the vast majority of current Li-ion batteries. The interactions between the ions and thepolymer alter the ion motion, affecting conductivity and device performance.Polymer gel electrolytes are cross-linked polymers swollen with solvent and thereforepossess both the physical firmness of a solid and the diffusive properties of liquid. Themajor interactions between polymers and small molecules in these materials arise frompolymer solvation. Depending on the polymer and the solvent, these interactions varyfrom weaker van der Waals bonds to stronger electrostatic-based bonds or even coordina-tion bonds in some cases [32]. In some cases, polymers are directly charged and thereforeproduce a surrounding electric field. For example, polyelectrolytes are polymers that losea charged group when dissolved, leaving the backbone of the polymer charged [33]. An-other example is conjugated polymers. This family of organic polymers are consideredconductors when doped and semiconductors when undoped. The process of doping usuallyinvolves an oxidation reaction that increases the mobility of electrons in the conjugatedpi-orbital. When maximally-doped, they may hold up to one charge per three monomerunits. Applying an electric potential may lead to a total positive charge on the polymerbackbone that corresponds to a capacitance of up to 100 F\/g [34]. In order to balance thischarge, negative ions from the solution will bond with the polymer and, if the conditionsallow it, an influx of negative ions will penetrate the polymer, causing it to swell. Theinteraction between polymer chains and the ions is electrostatic and is similar to the ion-81.1. High Concentration Solutions and Room Temperature Ionic Liquidssolution forces that were previously described. However, even neutral polymers have somecharge distribution that causes molecular binding of ions from the solution.Depending on the solvent, the ion charge and size, and the polymer charge distribution,polymer-ion interactions will have different effects on the distribution of ions in solution.Some interactions might be strong enough to leave the ions immobile while others may notbe strong enough to create a long-lasting association.1.1.2 DiffusionNow that the interactions ions experience in solution have been considered, it is possible todiscuss ion motion in different media. Diffusion is generally divided into two major types\u2013driven by a potential gradient or by thermal energy. In the following sections we willdiscuss thermally driven self-diffusion and diffusion that is driven by an electric potentialgradient: electrophoresis.Self-DiffusionTranslational diffusion, or self-diffusion as it will be referred to in this dissertation, isdiffusion under zero chemical potential gradient, that is, under conditions of uniform con-centration [35]. It arises from random, thermal motions and collisions, and is characterizedthrough the self-diffusion coefficient, D. Self-diffusion, or Brownian motion, results in theincreasing displacement of particles as a function of time. While the mean displacement iszero in an isotropic medium, the mean squared displacement after time t is given by [36]:< d2 >= 2DNt. (1.3)91.1. High Concentration Solutions and Room Temperature Ionic Liquidswhere N is the dimensionality of the motion. Microscopically, D is given by the Einstein-Smoluchowski-Sutherland equation [36, 37, 6],D =kBTf. (1.4)where f, the friction drag factor, describes the drag force on a particle moving with avelocity v in a viscous fluid, f = \u2013Fdrag\/v. f depends on all the interactions mentioned inSection 1.1.1, and is unique to any solution and physical condition.In its more common form, Eq. 1.4 is known as the Stokes-Einstein equation for aspherical molecule [25]:D =kBTbpi\u03b7r(1.5)where \u03b7 is the kinematic viscosity of the medium, r is the effective radius of the diffus-ing particle, and b is a dimensionless number that reflects the interactions between themolecule of interest and its surroundings. b ranges from 4 (no interaction) to 6 (stronginteraction) [37]. When considering the strong interaction limit, this equation is known asStokes\u2019 Law.Even though Eq. 1.5 is widely used to describe self-diffusion, experimental data hasshown b to be outside the predicted range of 4-6. This is a result of several approximationsthat are not necessarily obeyed in certain solutions. Eq. 1.5 assumes the actual shape of theion is spherical, the solvation layer rigid, and a solute\/solvent size ratio of at least 4 [8]. Inaddition, it assumes all interactions between the solute molecules are negligible. For mostdiluted salt solutions, these approximations are accurate enough. Yet, as discussed, thereare a number of interactions in an ionic solution that affect both the ion and the mediumin question. For those solutions which diverge from these assumptions, the Stokes-Einsteinequation does not hold and diffusion coefficients cannot be predicted accurately.101.1. High Concentration Solutions and Room Temperature Ionic LiquidsEffect of Media on DiffusionAs expected, polymer chains retard any diffusive motion. How this happens has beeninvestigated but is still an area of active research. Many of the proposed mechanisms haveeither an unexplained empirical factor or are only applicable to a narrow range of solventsand polymers. The models that have been proposed for calculating the diffusion factorof solutes in a polymer network can be combined into three general families of models.These are known as the \u2018Average Free Volume\u2019, \u2018Hydrodynamic\u2019 and \u2018Obstruction\u2019 modelfamilies [38].Diffusive motion primarily occurs within and between the solvent-filled regions boundedby the polymer chains. The free volume models are unique in that they do not considerdiffusion as a molecular, thermally-activated process. Instead, they assume diffusion is aresult of random redistribution of free volume spaces within the polymer mesh, a propertyof the polymer dynamics. Molecules can move from one free space to another if theireffective cross-section is smaller than the opening between spaces. The existence of voidsarises from a statistical redistribution of the free volume. The more general descriptiondefines the free volume as the volume of a system at a given temperature minus the volumeof the same system at 0 K, where there is no polymer motion [39]. The empirical ratio thatdetermines the free volume effect on the diffusion coefficient is dependent on many factors,depending on the specific theory, but is mostly the size of the diffusive molecule in relationto the size of the passage between pores\u2013 a factor that is dependent on the density of crosslinks between the polymer chains.In the \u2018Obstruction\u2019 family of models, the diffusion of the polymer molecules is assumedto be negligible. The path of solute molecules is highly tortuous, due to the mesh of polymerchains obstructing their path. Consequently, the relevant parameters for these models arethe volume fraction of the solute and polymer and the solvent\u2019s volume and shape [40, 41].111.1. High Concentration Solutions and Room Temperature Ionic LiquidsIn addition, an empirical factor is added to some of those models to include solvent polymerinteractions that may bind the solvent molecules to the polymer[38]. The tortuous natureof the path means that the diffusion length may no longer scale with the square root oftime so that the diffusion measured depends on the duration of the measurement.The hydrodynamic models consider the hydrodynamic drag force experienced by thesolute and are based on the Stokes-Einstein equation, taking into account the interactionspresent in the whole system to determine the friction drag factor. According to thesemodels, the polymer chains increase the drag on the molecules by slowing the solvent flowin the proximity of the polymer. Some of the models do take into account the binding ofions to the charged groups on the polymer chains [42]. By summing all the friction factorsbetween any molecule and the total polymer, a ratio between the free solution diffusion tothe diffusion in a polymer network is determined.To summarize, the effect a polymer network has on solute diffusion is complicated andhard to predict, especially when taking into account the numerous factors that affect boththe solute molecule and the polymer mesh. The models that exist today are helpful, butnot sufficient to present the full picture.Diffusion MeasurementsBecause self-diffusion is a process that takes place even at equilibrium, it does not neces-sarily result in an observable change in the system. In order to follow the self-diffusionmotion, some kind of labelling is required to allow tracking of a molecule. In NMR, thespatial labelling of molecules is achieved by using a magnetic field gradient in a non-invasivemanner. By combining NMR\u2019s ability to distinguish chemical environments with spatialresolution, the tracing of different components in solution is possible. As NMR can observemagnetically-visible nuclei in any phase, it has become a popular technique to measure dif-121.1. High Concentration Solutions and Room Temperature Ionic Liquidsfusion for gases, liquids and solids. This well-established method is known as pulsed fieldgradient NMR (PFG-NMR) and it will be discussed in Section 2.1.Another method that can measure self-diffusion is dynamic light scattering (DLS).This technique can measure the velocity at which particles move by measuring the timedependent fluctuations in the intensity of scattered light. This method is widely used andcan measure self-diffusion of particles and micelles easily [13]. For the case of RTIL andother solvated molecules,DLS cannot separate between similar sized species and thereforemeasures an average diffusion constant termed \u201cMutual Diffusivity\u201d [43, 44].1.1.3 ElectrophoresisIn contrast to self-diffusion, electrophoresis describes particle motion that is driven ina directional, non-random manner. Electrophoresis describes the migration of chargedparticles in a spatially uniform electric field [36] and can be described as a \u201cdirected diffusionprocess\u201d [45]. This mobility of charged ions is of high interest as it depends on both thecharged species properties and the chemical environment, and so can give information onboth.A particle of charge q experiences force, F, due to an electric field, E, according to thefollowing relation:F = qE. (1.6)The particle is accelerated to velocity v by this force and will experience a drag force fromthe medium, denoted by the friction drag factor f which was introduced in Eq. 1.4. Thesecombined forces result in a constant drift velocity \u03bd,\u03bd =FFdrag=qEf=zeEf, (1.7)131.1. High Concentration Solutions and Room Temperature Ionic Liquidswhere z is the charge number of the particle in solution. To characterize this motion acoefficient is defined: \u00b5, the electrophoretic mobility. This transport quantity describesthe motion in question with units of m2s\u00b7V and is defined as the drift velocity, \u03bd of an ion insolution in response to an electric field of unit strength:\u00b5 =\u03bdE=zef. (1.8)The electrophoresis friction drag factor f is affected by the same interactions as theself-diffusion, only, in addition, we need to consider the external electric field and the effectit has on the charge \u201ccloud\u201d that is surrounding the ion (see Section 1.1.1). The externalpotential has a tendency to alter the charge distribution around the ion and to move theionic cloud itself in addition to the ion. Being oppositely charged, the cloud will tend tomove in the opposite direction from the ion, and, because of the attractive forces betweenthem, the migration of the ion will be attenuated by this effect, called the AsymmetryEffect [25].Ionic ConductivityIonic conductivity describes the ability of a solution to carry electric current. It arises fromthe motion of charge carriers in solution, and therefore depends on \u00b5, the electrophoreticmobility. The total ionic conductivity, \u03c3, is the sum of each charge carriers\u2019 specific con-ductivity in solution:\u03c3 =\u2211iFCi|zi|\u00b5i, (1.9)where F is the Faraday constant, C the concentration, and \u00b5 is the specific mobility foreach species and is denoted by the subscript i.Yet, it is known that some interactions in the solution affect the charge each ion carries.141.1. High Concentration Solutions and Room Temperature Ionic LiquidsSome even create charge neutral aggregates that have no contribution to the total conduc-tivity. Weaker ion associations that are not strong enough to create persistent aggregatescan lead to correlated ion motion that similarly reduces the effective charge an ion carries.Eq. 1.9 needs to be adjusted to include a correcting factor that takes into account the ratioof the ions that are charged and contribute to the conductivity,\u03c3 =\u2211i\u03b1iCiF|zi|\u00b5i. (1.10)Here the correction parameter \u03b1 is unique to electrophoretic migration measurements. Itarises from the fact that only charged particles will migrate in response to an electricpotential. \u03b1 can be combined with z to give the effective charge, zeff . This correctionfactor can shed some light on ionic solution behaviour as it depends on the ratio of freeions to associated ones. Another common description of this correction parameter, mostoften found in the solid-state ionic conductor literature, is called the Haven ratio, whichis the ratio between the conductivity derived from self-diffusion coefficients and the actualionic conductivity [46, 47, 48, 49].Due to the difference in charge contribution and the potentially different mobility ofeach ion in solution, the current is not necessarily carried equally by the different ionicspecies. The transport number, ti is defined as the fraction of the current that is carriedby a particular species [50]. It describes the contribution each species has to the totalcurrent:ti =\u03b1iCiF|zi|\u00b5i\u2211i\u03b1iCiF|zi|\u00b5i. (1.11)The sum of transport numbers from all the charge carriers in solution is unity(\u2211iti = 1).Transport numbers are important design parameters for an electrolyte solution and can151.1. High Concentration Solutions and Room Temperature Ionic Liquidshelp determine its potential applications. For example, a strong single-ion conductivepolymer requires that one of the ions remain immobilized while the current is carried bythe counterions alone\u2013in other words, a transport number of unity for one of the ions [51].In contrast, for an efficient capacitor, similar transport numbers for each of the ions wouldbe an advantage [52, 53].Electrophoresis MeasurementsThe majority of today\u2019s existing methods to measure mobility and transport number areonly useful for liquid solutions and\/or are not selective in their detection. They are notuseful for multi-component solutions or polymer based electrolytes. The most famousmethods to measure transport numbers are the moving boundary and Hittof methods.Both of these fail when it comes to porous media (extensively discussed in [54, 55]). Anadditional problem is the inability to distinguish between ions in a polyelectrolyte solution.DC and AC methods rely on the relation between conductivity and mobility. They areuseful for porous media, but do not distinguish between the different ionic species. Amethod which is both selective and allows for measurement in polymer networks is theradio tracer method, but it adds the complication of nuclear labelling [14]. DLS can beused to measure the velocity at which particles move by measuring the time dependentfluctuations in the intensity of scattered light. By applying a voltage to the sample, DLScan measure the electrophoretic mobility of particles and solvated molecules, but cannotseparate between similar sized species [13, 43, 44].1.1.4 Relation Between Electrophoresis and Self-DiffusionIt is straightforward to relate migration due to electrophoresis and diffusion. Accordingto the Boltzmann distribution, in equilibrium the number of particles with charge q at161.1. High Concentration Solutions and Room Temperature Ionic Liquidstemperature T under an electric potential U isN = N0 exp(qUkBT). (1.12)The particle spatial gradient is\u2207N = qUkBT\u2207U \u00b7N0 exp(qUkBT)=qUkBT\u2207U \u00b7N, (1.13)and according to Fick\u2019s first law, the flux of particles J due to thermal energy is \u2013D\u2207N.The flux of particles under an electric field is the number of particles times their velocity.The total flux in the system is equal toJtotal = Jdrift + J(D) = \u00b5\u2207U \u00b7N \u2013 D\u2207N. (1.14)This is zero at equilibrium. By plugging in Eq. 1.13 we derive the Nernst-Einsteinrelation for charged particles:D =kBTzeffe\u00b7 \u00b5. (1.15)This relation between self-diffusion and electric mobility is completely general and is validfor any isothermic system and chemical condition.Currently, accurate D values obtained by PFG-NMR are used to estimate mobility usingthe Nernst-Einstein relation. Plugging the relation in Eq. 1.9, results in an estimation forionic conductivity and transport numbers:\u03c3 =\u2211iz2i CieFDikBT. (1.16)These equations are often used under the assumption that zeff is equal to the ion valence.171.1. High Concentration Solutions and Room Temperature Ionic LiquidsThis assumption includes an inherent error that increases with the ionic strength of thesolution. It is therefore expected that this assumption does not hold for the case of RTILs.And, as expected, ionic conductivity measurements of RTILs deviate by up to \u2248 70%from conductivities derived from measured self-diffusion coefficients and the Nernst-Einsteinrelation, indicating the existence of ion-ion associated species [56, 57, 58]. By rearrangingEq. 1.15, it is possible to gain new knowledge of the level of ion association in solution.Using a combination of self-diffusion and mobility measurements enable determination ofzeff, the effective charge of the ion in question\u00b5D=zeffkBT. (1.17)This parameter gives us a direct indication of the level of association of a specific ion speciesin solution.Tokuda et al. [59] have measured a similar ratio and shown it to be a useful characteri-zation parameter for the physical properties for RTILs. It was also shown to explain someof the unusual behaviour of ion transport in RTILs in solution with aprotic co-solvents [21].In both of these papers, zeff was obtained by measuring the ratio between conductivity,measured by electrochemical impedance, and conductivity derived from D, using Eq. 1.16.D was obtained accurately, using PFG-NMR, but the impedance measurements are highlynon-selective and can only be used for a single ion with a non ionic solvent. Therefore theirmethod is limited by the conductance measurement to an electrolyte with only one pair ofcharge carriers.181.2. Research Aims and Outline1.2 Research Aims and OutlineThe work done in this dissertation aims to directly measure both self and driven diffusionin order to study the transport mechanisms of molecules in several unique materials. Thematerials investigated all show transport divergence from known theory which affects theirphysical properties. By selectively tracking both self and driven diffusion of molecules ina variety of materials we learn how transport contributes to their interesting mechanical,optical, and electrical behaviour. As discussed in this introduction, we expect the trans-port of molecules in concentrations above the highly diluted regime will diverge from theStokes-Einstein equation (Eq. 1.5). This divergence is predicted to be the source of the un-usual behaviour found in these materials. We expect the interaction between molecules inconcentrated materials to decrease their ability to diffuse as well as their ability to conductcurrent. By directly measuring these properties, we aim to increase our understandingof transport mechanisms in concentrated materials and guide the further development ofmaterials with interesting and useful properties.Chapter 2 provides an overview of existing magnetic resonance techniques that wereeither used in this dissertation or inspired new technique development. The purpose of thischapter is to provide the reader with an understanding of the methods used, necessary forthe discussion of results presented.In Chapter 3, we describe the development of a simple eNMR probe within a horizontalmagnetic field gradient set to allow for direct and selective measurements of electrophoreticmobility of magnetically visible species. This probe aims to measure mobility in a directionperpendicular to the majority of undesired flows such as convection currents or bubbles,resulting in accurate electrophoretic mobility measurements.Chapter 4 aims to investigate the \u201cpiezoionic effect\u201d observed in ionic electroactivepolymer based soft strain sensors. As the mechanism behind the piezoinic effect is ambigu-191.2. Research Aims and Outlineous, a more in-depth investigation of molecular transport in necessary. By utilizing thewell-established technique of PFG-NMR, we aim to explore how self-diffusion is affectedby the concentration of RTIL in both solution and polymer samples.The eNMR probe described in Chapter 3 is used in Chapter 5, along with PFG-NMR,to measure both self-diffusion and electrophoretic mobility of anions and cations in an elec-trolyte used for potential artificial muscles and nerves. Combining these measurements, wecalculate the effective chrage (zeff) for each of the ions in order to shed light on the uniqueconduction mechanism seen in lithium based salts. We observe how Li+ conductivitydepends dramatically on concentration, contributing only a small fraction at low concen-tration, but completely dominating at high concentration. To explain this dramatic changein conductivity, we suggest a potential mechanism that explains this unique behaviour.Chapter 6 describes transport measurements of electro-photochromic molecules embed-ded in polymer films which are used as optical filters which modulate solar energy absorp-tion in vehicle sunroofs. Using PFG-NMR, we measure self-diffusion of these chromophoresin a variety of conditions. These measurements were done in an effort to understand theperformance limitations of these films and guide the development of the automotive glazingindustry.In Chapter 7, we again utilize PFG-NMR to investigate the stress relaxation mecha-nism observed in a new class of recyclable polymers. These thermoset polymers, calledvitrimers, present an opportunity for efficient recycling via a thermally activated exchangeof molecular cross-links. The transport of crosslinker molecules in vitrimers is investigatedin this chapter, with emphasis on the driving force behind it. Through this investigation,we hope to learn of usage limitations for vitrimers and learn of the necessary conditionsrequired for easy recycling.Chapter 8 summarizes the lessons learned from these studies and how significant di-201.2. Research Aims and Outlinerect measurements of transport properties are for a full understanding of both conductionmechanisms and unique physical properties seen in concentrated materials. The chaptercontains suggestions for additional experiments that would help to further our understand-ing of transport in these concentrated systems and closes with an outlook to the future.21Chapter 2Magnetic Resonance TransportMeasurements: Theory andBackgroundThe purpose of this chapter is to provide an overview of the existing magnetic resonancemethods for measuring self-diffusion and electrophoretic mobility. The reader should havebasic knowledge of magnetic resonance theory in order to understand the majority of thecoming chapter. If this knowledge is missing, we advise to review the first two chaptersof Spin Dynamics by Malcolm H. Levitt [60]. For the case of electrophoretic mobility,this chapter describes several approaches and methods, each with its own advantages anddisadvantages. As electrophoretic NMR (eNMR) is still an emerging technique, there isno industry standard agreed upon and research labs are still experimenting with differentapproaches to voltage application methods, sample holder set-ups, and pulse sequences.This chapter begins with an overview of pulsed field gradient methods that will beused in Chapters 4, 5, 6, and 7. An understanding of this technique is crucial to theinterpretation of the results presented in the mentioned chapters. This overview is followedby a review of the state-of-the-art of eNMR techniques. Not all methods described in thischapter were used in the work of this dissertation, but they inspired ideas for experimentalset-ups and pulse sequences used in Chapters 3 and 5.222.1. Pulsed Field Gradient-NMR2.1 Pulsed Field Gradient-NMRThe well-established method of pulsed field gradient NMR (PFG-NMR) relies on similarprinciples to those underlying magnetic resonance imaging (MRI). An additional dimensionthat provides spatial information is encoded by introducing a magnetic field gradient,thereby making the Larmor frequency \u03c9 location-dependent:\u03c9z = \u03b3B(z) = \u03b3(B0 + gz) = \u03c90 + \u03b3gz, (2.1)where \u03b3 is the gyromagnetic ratio of the isotope, g is the magnetic field gradient strength(in Gcm) and B0 is the static magnetic field.Figure 2.1: Pulsed Field Gradient(PFG)-echo pulse sequence diagram. The row labelledas RF represents the radio frequency channel and the row labelled g represents themagnetic field gradient channel, both against time. \u03b4 is the sinusoidal gradient pulseduration and \u2206 is the time delay between the magnetic field gradient pulses and istermed \u201cdiffusion time\u201d.By applying a magnetic field gradient in the z direction, each nuclear spin is labelledaccording to its position along the z-axis. Over time as diffusion takes place and the spins232.1. Pulsed Field Gradient-NMRmove, they will develop a phase that is related to their displacement. This phase changewill enable the determination of the self-diffusion coefficient. A simplified pulse sequencefor diffusion measurements with PFG-NMR is illustrated in Fig. 2.1 based on the Stejskaland Tanner pulse sequence [61]. We will now elaborate on the phase progression of the spinsduring this pulse sequence. The initial pi2 pulse tips the entire magnetization of the nucleito the transverse plane, where it will begin precessing according to the Larmor frequency.By applying a short gradient pulse of duration \u03b4, each spin will acquire a phase accordingto its position \u03c6 = \u03c9z\u03b4. This initial position is now encoded to the spin\u2019s phase [62]. Thetotal magnetization is weighted by spin-spin relaxation, T2, and at the end of the first \u03c4is S(\u03c4\u2013) =\u2211S(0)e\u2013i\u03c9z\u03b4e\u2013i\u03c90\u03c4 e( \u2013\u03c4T2)where the z dependence represented in the phase is seenexplicitly. By applying a pi pulse, the phase will be reversed and the magnetization will beS(\u03c4+) =\u2211S(0)e+i\u03c9z\u03b4e+i\u03c90\u03c4 e( \u2013\u03c4T2). By reintroducing the gradient pulse of the same durationand strength, any spin that has not changed position will acquire the exact same phaseas it did in the first section resulting in S(2\u03c4) =\u2211S(0)e\u2013i\u03c9z\u03b4e+i\u03c9z\u03b4e( \u20132\u03c4T2)=\u2211S(0)e( \u20132\u03c4T2).This means the signal arising from spins that did not change their position is completelyrefocused. In contrast, any spin that changed location will have a total phase that dependson the magnitude of that position change. \u03c6 = +\u03c9z1\u03b4 \u2013 \u03c9z2\u03b4 = \u03c9\u03b4\u03b3g(z1 \u2013 z2). Integratingover all trajectories will result in a signal that is attenuated by a diffusion factor. For asinusoidal gradient pulse, this factor depends on \u03b4, g, \u2206 and the self-diffusion coefficient Daccording toS(2\u03c4) = S0 exp(\u20134pi2\u03b32g2\u03b42(\u2206 \u2013\u03b44)D)exp(\u20132\u03c4T2). (2.2)In order to allow the measurement of samples with short T2s, as is the case of mostviscous or solid samples [63], a stimulated echo (STE) RF pulse sequence may be used(see Fig. 2.2). This pulse sequence stores the magnetization along the z-axis (where no242.1. Pulsed Field Gradient-NMRFigure 2.2: Pulsed Field Gradient Stimulated Echo (PFG-STE) pulse sequence diagram.An additional acquisition delay of duration T was inserted to allow eddy currents todecay.spin-spin relaxation occurs) in between gradient pulses. As long as \u2206 is larger than T2,no magnetization will survive on the transverse plane, eliminating the dephasing effect ofT2 relaxation during the diffusion time. As a result, T2 relaxation will only be affectingthe spins during \u03c4 . This allows for longer diffusion times \u2206 to be used. Storing the mag-netization along the z-axis does create a factor of two loss in signal and the magnetizationundergoes spin-lattice relaxation with time constant T1 during the diffusion time. Theduration of \u03c4 is limited by the gradient pulse duration \u03b4. The total signal at the end ofPFG-STE NMR pulse sequence is given byS(2\u03c4) = S0 exp(\u20134pi2\u03b32g2\u03b42(\u2206 \u2013\u03b44)D)exp(\u20132\u03c4T2\u2013\u2206 \u2013 \u03c4T1). (2.3)To isolate the attenuation of the signal by the diffusion factor we normalize it. Normalizingthe signal with respect to that obtained with the gradient off will result in the Stejskal and252.2. eNMRTanner equation [61]:S(g)S(0)= exp(\u20134pi2\u03b32g2\u03b42(\u2206 \u2013\u03b44)D). (2.4)In order to validate the diffusion measurements obtained by PFG-NMR, many re-searchers use the Nernst-Einstein relation (Eq. 1.15) to calculate molar conductivity byusing the measured D coefficient. These derived conductivities are compared to measuredconductivity values from impedance measurements. These values are typically in agreementfor highly diluted solutions. In the case of higher concentration solutions (<\u223c 0.5 M) NMRdiffusion typically predicts higher conductivity values because, as stated before, withoutknowing zeff the Nernst-Einstein relation does not account for associated ionic species thatdo not contribute to the conductivity [56].2.2 eNMRThe concept of using NMR to measure field assisted transport was initially recognized byPacker in 1969 [64]. But the first successful experiments were made more than a decadelater in 1982 by Holtz et al. [65]. Experimental limitations, especially artifact currents,have left the field of electrophoretic NMR (eNMR) to slowly develop. eNMR has almostexclusively been developed for solution NMR, mostly because of the difficulty in measuringlong diffusion times for samples with short T2. In addition, diffusion in solids is in generalmuch slower and requires higher gradient strengths to be observed [66].The measurement of ionic mobility under an electric field is slowly developing and hasjust recently been commercialized (2013- solutions only [67], based on Hallberg et al. [68,69]). It is still considered an exotic technique that is performed in only a handful of labsaround the world.262.2. eNMR2.2.1 Signal EvolutionBy adding the electric field to the diffusion time of diffusion NMR pulse sequences, it ispossible to measure the drift velocity (\u03bd) as a coherent flow like motion with \u03bd determinedby the electric field strength and \u00b5. As in the PFG-NMR experiment, the spatially de-pendent Larmor frequency is \u03c6 = \u03c90 + \u03b3g\u03b4z(t) only now the position, z(t), is dependenton the velocity of the spins according to z(t) = \u03bdt, so the phase accumulated over time is\u03c6 = \u03c90 + \u03b3g\u03b4\u03bdt.The signal amplitude is then affected by both the flow process and thermal diffusion,and is given by [65]:S(g)S(0)= exp(\u20134pi2\u03b32g2\u03b42(\u2206 \u2013\u03b44)D)\u00d7 cos(\u03b3g\u2206\u03b4\u03bd), (2.5)or by plugging in the definition of \u00b5 from Eq. 1.8S(g)S(0)= exp(\u20134pi2\u03b32g2\u03b42(\u2206 \u2013\u03b44)D)\u00d7 cos(\u03b3g\u2206\u03b4\u00b5E). (2.6)In order to extract the electrophoretic mobility, \u00b5, the measured phase shift from thepresented eNMR pulse sequence (Fig. 2.3) is isolated to:\u03c6 = \u03b3g\u2206\u03b4\u00b5E (2.7)As we\u2019ve seen in the previous section, the stochastic process of self-diffusion causesa dephasing of the signal and therefore results in attenuation. In the case of coherentflow, driven by the electric field, the flow is constant, directional, and time dependent andtherefore results in phase modulation. This phenomenon can be compared to a constant,slow rotation of the NMR signal, similar to chemical shift, and therefore the result is a272.2. eNMRFigure 2.3: Spin echo eNMR pulse sequence diagram. An electric field is applied for theduration of \u2206 can be seen in row E. [65]phase modulation. As the spins are moving to a slightly different position in the magneticfield gradient, they have a slightly different Larmor frequency, similar to the explanationof a phase shift from the receiver being offset from the rotating frame. From this signalphase modulation, we can learn both the extent and directionality of the ion\u2019s motion.As both self-diffusion and driven diffusion occur at the same time, an optimization ofthe evolution time \u2206 is necessary. Other than the applied electric field, all the parametersthat affect the magnitude of the phase modulation are also responsible for the loss of signalresulting from self- diffusion. Increasing the applied field time will, therefore, decrease thesignal to noise ratio and might leave us with unmeasurable data. The same will happenwhen increasing the gradient strength or gradient pulse time. In order to achieve a usefuleNMR measurement, a minimal drift velocity \u03bd must be obtained and other time related282.2. eNMRparameters need to be carefully optimized.2.2.2 eNMR Limitations and Probe Design SolutionsOne of the major challenges in obtaining accurate eNMR data lies in the presence ofartifact flows. Any flow that does not arise from the electrophoretic process is consideredan artifact. Due to the integrative nature of NMR, it acquires an average of all the spinspresent in the detection volume. Therefore, all motions present in the coil in between thetwo gradient pulses are averaged and any non-electrophoretic currents are also averagedinto the NMR signal. Overcoming this challenge has led to the design of several eNMR-specific sample cells and pulse sequences to help reduce typical artifact flows or minimizetheir effects on the signal. Both the common artifact flows and typical methods to overcomethem are described in this section.There are two common approaches to a sample holder for the suppression of non-electrophoretic motion, each with its own advantages and disadvantages. One is basedon a U-tube cell that holds the electrodes above the RF coil. This set-up is useful forseveral reasons but suffers from a low fill factor and an inherent mixture of the mobilities\u2019direction. Because both arms of the U-tube are inside the RF coil, the charged ions movein opposite directions in each of the arms, making it impossible to distinguish the directionthey are moving (see Fig. 2.4a, [70, 71, 72]). The other successful approach is based on avertical capillary tube array (see Fig. 2.4b, [73, 58]). This set-up does not require uniqueglasswork and has a much higher fill factor. It can distinguish directionalities of \u00b5 butsuffers from other artifact flows as the electrodes pass through the detection volume. Inthis section, the challenges will be further described together with solutions that are basedon these sample cells.An inherent problem when adding any currents near the RF coil is the noise pick up292.2. eNMR(a) (b)Figure 2.4: Conventional arrangements for eNMR sample cells.(a) U-tube sample cell. (b) Vertical sample cell. The distance between the electrodes isroughly 3 cm. Taken from Furo\u00b4 et al. [74]because the probe acts as an antenna. The closer any part of the leads or the electrodeare to the RF coil, the more noise the antenna will pick up. Usually, this is a major issueas most eNMR probe designs require the electrodes to be inside the RF coil area. TheU-tube cell design overcomes this issue by setting the electrodes as far from the RF coil aspossible, and by not having any leads pass through it. In comparison, the vertical samplecell has an insulated lead passing through the coil. This set-up still works well by settingthe electric field axis as coaxial to the RF coil and thus perpendicular to the RF radiationdirection, minimizing the noise pick up by the coil [75]. Another solution to overcome someof the noise pick up in a vertical sample cell is the usage of an RF interference filter, asmade by Hallberg et al. [73, 68]. This filter system consists of a large ground box that issimilar to a one-point ground for the leads transferring the electrophoretic inducing currentand a low pass filter.Artifact flows arising from thermal convection are one of the major issues of using the302.2. eNMReNMR technique. Even in a standard solution NMR experiment, a temperature gradientwill result in uneven relaxation processes throughout the sample and results in distortedspectral lines. In eNMR experiments, the electrophoretic current passing through thesample will cause resistive heating according to Joule\u2019s first law: [45]Q = I2Rt = IVt (2.8)where Q is the produced heat, I, R and V are the current, resistance and voltage respectivelyand t is the current duration. This tends to be an important source of heat in comparisonto heat resulting from RF and gradient pulses. Although they can be important, thosetend to be small due to low duty cycles while the duty cycle of the applied electric field isoften much greater.There are two major sources of resistive heat in an eNMR sample cell. In the point ofcontact between the sample and the electrode there is an increase in resistance as a resultof the edge properties of both materials and the change between two types of currentconductors. The second source of heating comes from the conductivity of the sample.Higher conductivity samples can support an increased current, I, when a given voltage isapplied. They therefore can produce an increased amount of heat in the sample. Althoughresistive heating is relatively even throughout the sample, due to contact with the tubeglass walls, which serve as a heat sink, the temperature of the sample in contact with thetube may not be equal to the centre of the sample. The variety of areas with increasedheating or increased heat capacity create a temperature gradient along the sample. Thistemperature gradient causes a viscosity (\u03b7) and a density gradient in the sample whichlead to artifact flow, mostly in the direction parallel to gravity. In addition, the changein \u03b7 will cause a change of both self-diffusion and mobility coefficients which will not beconsistent for the entire sample, resulting in erroneous results. This is a significant limiting312.2. eNMRfactor for eNMR and until recently, the voltages used were very high (up to 1 kV [76, 72])and therefore samples were usually preferred to not exceed 50 mM in concentration to limittheir conductivity [77, 78]. One of the solutions to this uneven cooling is setting the samplein a polymer. By setting the sample in a polymer network, a mesh of heat sink that isin direct contact with the entire sample is created. This has two major benefits: 1) Thehigher thermal conductivity of the polymer helps to keep the temperature across the sampleuniform and 2) the polymer reduces bulk flow from any remaining sources. The methodof setting the sample in a stabilizing gel to minimize convection and heating of the samplehas proved to be effective but can only provide mobility coefficients of ions embedded in apolymer network [78].Another way to minimize the effect of convection currents is the use of capillary tubes.As can be seen in Fig. 2.5, a capillary tube array can be added to either type of conventionaleNMR cells. This solution was originally implemented by He et al. in an attempt toinvestigate proteins in highly conductive buffer solutions [70]. This capillary tube arraybreaks bulk convective current pathways and does not allow them to develop and affectthe eNMR signal. When implemented in a vertical cell as was done by Zhang et al. [58] andGouverneur et al. [79], the background convective flows are broken. This vertical set-upalso helps to increase the fill factor in the coil and by that increase the signal to noise ratio.In practice, many repetitions of the experiment are needed to achieve sufficient signalto noise ratio. Without care, heat may accumulate along with the accumulation of signal.Precautions such as constant cooling of the sample, increased relaxation delays betweenexperiments, and temperature control in the probe can help minimize the effects of resistiveheating by keeping the temperature of the sample constant.What seems to be the biggest issue in the implementation of eNMR is artifact flow thatarises not from convectivity, but from electro-osmosis; bulk solvent motion generated by an322.2. eNMR(a) (b)Figure 2.5: Conventional sample cells with capillary tubes added. (a) is taken with per-mission from [70] and (b) with permission from [58].externally applied electric field on an ionic solution. As has been discussed in Section 1.1.1,ions in solution interact with solvent molecules to form a Debye layer of counter chargearound a charged particle. A similar phenomenon occurs at the interface between chargedsurfaces and solutions, such as in the case of the sample holder glass wall. Due to Coulombattractions, a fixed layer of counter-ions is built up against the charged surface of the glasswall. A bit further from the surface, a diffusive layer of ions is formed. This layer alsohas some net charge but is only loosely attracted to the surface and can diffuse easily.These two layers are called the electric double layer. When an electric field is applied, thediffusive layer will migrate, carrying the solvent molecules with it. Unlike an electrophoreticmigration, this will result in bulk flow of the neutral solvent.There are several techniques to overcome electro-osmosis in eNMR experiments. One isdecreasing the glass surface charge by either coating it with an anti-electroosmotic coating332.2. eNMR(e.g. poly(acrylic acid), polyimide or using a poly(methyl methacrylate) (Perspex) tubeinstead of glass [75, 79, 80]). The charged surface does not result in electro-osmotic flowbeyond 10 nm in high dielectric constant solutions [81]. Therefore, in tubes with largeenough volumes, the ratio of surface area to volume is low enough to only affect a smallnumber of molecules and the effect is minimal. For narrow tubes, such as the U-tubesample cell and capillary tubes, the effect is more prominent and a high quality coating isrequired to suppress such flow.Another potential issue in eNMR experiments is triggering electrochemistry for theexperiment\u2019s components. The application of voltage on a chemical sample can induceelectrochemistry, depending on the sample itself and the electrodes\u2019 material. If the elec-trochemical window of any of the materials in the sample is exceeded, redox reactions willtake place and result in chemical changes to the species themselves, their concentration,or the composition of the electrode. Electrochemistry might also cause the production ofgas at the electrodes, resulting in another artifact flow as gas bubbles try to escape thesample in the opposite direction to gravity causing turbulence that is greater than theeffects anticipated from the electrophoresis. This is especially an issue for samples con-taining solvents and does not usually cause an issue when using pure materials, such aspure RTILs. One solution as suggested by Stilbs et al., is using palladium as the electrode.It is known that Pd absorbs hydrogen at interstitial octahedral sites of its face-centeredcubic lattice [82, 83]. Using Pd electrodes can trap hydrogen as it is generated at the elec-trode and by that eliminates bubble formation. Another solution to bubble formation isto have the electrodes above the detection volume, such as in the U-tube shaped samplecell. This will ensure any flow and turbulence caused by the bubbles is above the detectionvolume and does not affect the NMR signal. In the case of severe bubble formation, anon-conductive gas gap might form in the sample, creating an electric blockage that will342.2. eNMRstop the electro-migration. For the specific case of the U-tube cell used with water as asolvent, electrochemistry causes a unique problem. For water, both electrodes will producegas, but at different volumes. This unbalanced gas production can cause a displacementof the whole water column in a U-tube shaped sample holder. To avoid these issues, boththe voltage and the type of electrodes must be selected carefully.More options for the suppression of bubble formation can be learned from the field ofcapillary electrophoresis in mass spectrometry as they deal with directional currents aswell [84].2.2.3 Pulse Sequence Based SolutionsOne of the ideas to suppress artifact flows that do not depend on the direction of the appliedcurrent, is to use a convection-compensated eNMR pulse sequence as suggested originallyby He and Wei [85] and used by many others [86, 75, 73, 58, 57]. The pulse sequence isshown in Fig. 2.6. This extension of the original eNMR pulse sequence can distinguishbetween migration that does or does not depend on electric field (\u03bdE vs. \u03bdeff). By splittingthe electrophoresis diffusion time into an even number of sections, while alternating thedirection of the applied electric field, one type of migration will cancel out. This pulsesequence is similar in concept to a Carr\u2013Purcell-Meiboom-Gill (CPMG) echo train [87, 88].The phase after the first diffusion block (see Fig. 2.6) is a result of a combination of thetwo migrations, \u03c6 = \u03b3g\u03b4\u2206(\u03bdE + \u03bdeff). Flipping the magnetization using a pi pulse resultsin a phase of \u03c61 = \u03b3g\u03b4\u2206(\u2013\u03bdE \u2013 \u03bdeff). At the even numbered diffusion block, the appliedelectric field is reversed in direction and so does \u03bdE. Any flow that arises from a processthat is not-electric field driven will not change direction. The phase for the even diffusionblocks is \u03c62 = \u03b3g\u03b4\u2206(\u2013\u03bdE + \u03bdeff). At the end of the pulse sequence, the total phase arisingfrom the flow processes will be a summation of both blocks \u03c6 = \u03c61 + \u03c62 = \u20132\u03b3g\u03b4\u2206\u03bdE or352.2. eNMRFigure 2.6: Artifact flow suppression CPMG-eNMR pulse sequence diagram.for n repetitions- \u03c6 = \u2013n\u03b3g\u03b4\u2206\u03bdE. Notice that the magnitude of the cancelled flow does notmatter and can be much larger than the electrophoretic flow. This allows for small motionsarising from electrophoresis to be distinguished in a cloud of noise from other migrations.By increasing the number of inversions in the pulse sequence, \u2206 can be decreased. Thisallows us to avoid acquiring after the flow in the sample has reached an equilibrium statein which the species are no longer flowing.Because of the extended time the magnetization spends on the transverse plane, CPMGbased pulse sequences will not be useful for samples with fast T2 relaxation. To overcomethis limitation, the echo pulse can be replaced with a stimulated echo. This pulse sequencewill enable artifact flow suppression for samples with short T2 in the same manner asCPMG-eNMR. The major downside in using an STE is the loss of signal due to the fact362.2. eNMRthat with each STE, the signal is reduced by half. For this reason, it is useful only whenusing a minimal number of repetitions. For the case of n=2, this pulse sequence is knownas the Double STE eNMR pulse sequence (DSTE) [75] and is shown in Fig. 2.7.Figure 2.7: Double Stimulated Echo-eNMR pulse sequence diagram.Although these pulse sequences can theoretically suppress any current artifact that isindependent of the electric field, one must remember two hardware limitations. 1) heating isan accumulative process independent of the sign of E. A temperature difference between thetwo sections might cause a migration difference 2) an RF coil is far from being completelyhomogeneous and so the applied pulses are not perfect, that is, not perfectly pi or pi2 forall the spins in the sample. These limitations will prevent a complete suppression of theartifact flows.In this chapter, an overview has been presented of the well-established method of PFGdiffusion NMR, which is used extensively in the measurements described in Chapters 4-7.372.2. eNMRThe state-of the art of the emerging technique of eNMR has also been reviewed, and laysthe foundation for work done in this dissertation to build (Chapter 3), validate (Chapter3) and apply (Chapter 5) a simple new design of eNMR probe that overcomes the mostprominent artifacts of previous probe designs by virtue of its geometry.38Chapter 3Magnetic Resonance TransportMeasurements: MethodsExperiments presented in this dissertation were performed using a variety of NMR probes,spectrometers and magnetic field gradient systems. This chapter describes the systemsthat were used, the purpose and abilities for each of these set-ups, and their validation.We will start by describing the pre-existing systems, including the PFG-NMR set-up forself-diffusion measurements and in Section 3.5, describe the construction of a new probeand voltage application system for electrophoretic diffusion measurements.3.1 NMR SpectrometersThree NMR spectrometers were used in the experiments presented in this dissertation.Two of the spectrometers are based upon Oxford Instruments (4.7 T and 8.4 T), 89 mmvertical-bore superconducting magnets with electronics consoles that were designed andbuilt in-house. The console for the 8.4 T system is described in detail in [89]. The widebore magnets simplified the creation of unique probes and use a variety of magnetic fieldgradient units that are described in the following sections. Typical pi2 pulse widths rangedbetween \u223c2-5\u00b5s for 19F and 1H to \u223c6-10\u00b5s for 7Li. The 8.4 T magnet, which was used forthe vast majority of experiments presented, was shut down and recharged during 2017. The393.2. Home built PFG Probe and Driving Circuit1H frequency of the magnet\u2019s field was reduced from 363 MHz to 360 MHz. The frequencychange was small enough that all probes and frequency generators handled with no issues.The third spectrometer used is a commercial 9.4 T Varian Unity Inova NMR spectrome-ter. This spectrometer is equipped with a commercial temperature control unit in additionto a high resolution probe. This combination of temperature control together with a highresolution probe allowed us to acquire a specific set of 19F-NMR spectra, which is presentedin Chapter 6. at varying temperature with relatively high resolution. The pi2 pulse widthsused in these acquisitions ranged between \u223c7-10\u00b5s.3.2 Home built PFG Probe and Driving CircuitThe measurement of self-diffusion presented in chapters 4, 5, 6, and 7 was done usingan existing PFG probe and resonance driver circuit. These were originally built by Tsoand Michal [1] and improved by Michan [2]. This probe and driving circuit combinationallows us to change the gradient coil current and, by that, the magnetic field gradientstrength in the z-direction. The magnetic field gradient pulses produced by this circuithave a sinusoidal time dependence and are half a period long, as shown in Fig. 2.1. whichallows for a complete shut-off at the end of the pulse. This ability minimizes interferenceand inaccuracies when measuring transport coefficients. Initially, the coil winding allowedfor gradient pulses between a complete shut off to 3090 Gcm . Rewinding of the coil in amore accurate and tight manner allowed us to increase the number of windings, increasingthe maximal magnetic field gradient by about 20% to 3642 Gcm . For further informationregarding the magnetic field gradient coil design see A. Michan\u2019s MASc thesis [2].The strength of this probe and driver circuit lies in the extremely short and stronggradient pulses. Strong magnetic field gradient is necessary to allow diffusion measurementsin semi-solids and high viscosity samples. By using a high g, smaller diffusion coefficients403.3. Data Interpretationcan be measured. The short gradient pulses allow \u03c4 in the STE based pulse sequences tobe shorter. The signal dependence of \u03c4 can be seen in Eq. 2.3. By allowing it to be asshort as 318\u00b5s, samples with short T2 can be measured. Combining these two benefitsallows for diffusion measurements in semi-solid samples of diffusion coefficients as low as\u2248 10\u201313 [m2s ] [2, 90]. The signal attenuation for experiments performed using this gradientset with sinusoidal gradient pulses followsS(2\u03c4)S(0)= exp(\u20134pi2D(\u2206 \u2013\u03b44)\u03b32\u03b42B2vV2), (3.1)where V is the programmed voltage to the driver (between 0\u201330 V), and Bv is the relationbetween voltage and the magnetic field gradient. A calibration of the Bv parameter wasdone by measuring a known sample of 0.05 M tetramethylammonium hexafluorophosphate(TEAP) in propylene carbonate (PC) and resulted in Bv = 121.4Gcm\u00b7V or 32.51Gcm\u00b7A . Anexample diffusion measurement of this calibration can be seen in Fig. 3.1.3.3 Data InterpretationIn order to extract self-diffusion coefficients from the data, the peak of interest from eachspectrum was integrated over its finite frequency width so the data points could be plottedas the peak integral vs. the magnetic field gradient strength. The error bars shown on thegraphs were calculated as the root mean square (RMS) of the noise from the end of eachfree induction decay (free induction decay (FID)) measurement multiplied by the squareroot of the fraction of the spectrum integrated from the total spectral width. This is doneto avoid the overestimation of error to the Fourier transformed spectrum. By multiplyingthe RMS noise with the square root of the fraction of the spectral width, we only accountfor the noise in the integrated area and not the entire spectrum. The data points were413.3. Data Interpretationthen fitted to a Gaussian curve using MATLAB (2013a, The MathWorks, Inc. , Natick,Massachusetts, United States). The self-diffusion coefficient was extracted from the fittedGaussian curve parameters. In addition, the natural logarithm of each data point and itserror was plotted against the magnetic field gradient strength squared to allow us to fitthe data to a linear version of Equation 2.4. The linearized data points were fitted to astraight line using linear regression analysis. Fig. 3.1 shows an example for the two typesof fits done.(a) (b)Figure 3.1: Relative signal intensityS(2\u03c4)S(0)vs. magnetic field gradient strength of calibrationsample 0.05 M TEAP in PC. The Bv coefficient was extracted by fitting the data to aGaussian curve with the known diffusion coefficient [1]. (a) shows the single PF\u20136 fluorinepeak intensity vs. magnetic field gradient strength. (b) shows the same data, plotted asthe natural logarithm of the signal intensity vs. magnetic field gradient strength squared tolinearize the Gaussian behaviour of the Stejskal-Tanner diffusion equation (Eq. 2.4). Bothfigures show excellent fits to a single diffusion coefficient.3.3.1 Temperature ControlThe probe was used previously for high temperature measurements, but before the workdone in this dissertation, no measurements had been done below ambient temperature. In423.3. Data Interpretationorder to perform experiments below room temperature, additional components were addedto the temperature control system. The probe was connected to a Bruker temperaturecontrol system, B-VT-1000 to allow for temperature control of the samples. A glass dewarin the probe acted as an insulator allowing hot or cold air to travel to the sample withoutharming the probe body. A thermocouple placed at the end of the dewar, just below thesample, measured the temperature of the air as it reached the sample. This thermocouplewas connected to the B-VT-1000 system and allowed for temperature feedback. To accountfor the difference between the air and the sample temperature, calibrations of the sampletemperature were done using a sample of pure methanol (\u2265 99.9% VWR, CAS 67-56-1) [91].Figure 3.2: An illustration of the temperature control set-up. The cold N2 gas, labelled asgrey arrows, flows to the probe through an insulating glass dewar. B-VT-1000 temperaturecontroller is connected to a thermocouple near the sample holder and controls both thecoil heater in the glass dewar and the heater in the LN2 dewar.For heating, room temperature air flowing through the glass dewar was heated by aheating coil inside the glass dewar (see illustration in Fig. 3.2). For cooling, an external433.4. Bruker Imaging 3-axes Gradient Setliquid nitrogen (LN2) dewar was added to the temperature control system. The boil offgas from the LN2 dewar was connected to the probe and flowed to the sample throughthe insulating glass dewar. In order to control the amount of nitrogen gas flowing tothe probe, a heater attachment was placed inside the LN2 dewar, controlling the amountof cold N2 gas to boil off. Both the LN2 heater and the coil heater in the glass dewarwere controlled by the B-VT-1000. This combination of controlled cooling and heatingextended the temperature range of the probe and allowed was able to maintain a constanttemperature of anywhere between 200 and 380 K.In order to avoid heating or cooling of the probe shield, an additional source of roomtemperature air was supplied to the area surrounding the gradient coil and sample. Adouble walled vent pipe was connected to the top of the probe to carry the hot or cold airout of the spectrometer in order to avoid heat transfer to and from the magnet.3.4 Bruker Imaging 3-axes Gradient SetAn external imaging gradient set made by Bruker allowed us to measure both self anddriven diffusion in multiple directions. This set-up was used in all eNMR measurements(see Section 3.5 and Chapter 5). This imaging set is made of a Micro2.5 gradient unit witha gradient sensitivity of 2.5 Gcm\u00b7A and a current amplifier system (Bruker BGU-II BAFPA40) made of three gradient amplifiers capable of 40 A each. This allowed for a maximal100 Gcm magnetic field gradient in each of the three-axes and allowed us to control the shapeof the magnetic field gradient pulses, expanding our set-up to more than just a sinusoidalpulse. Utilizing this gradient set allowed us to control the length, magnitude and directionof the magnetic field to high accuracy, but was limited to considerably lower magnetic fieldgradient strengths.443.5. eNMR Probe Construction and Validation3.5 eNMR Probe Construction and ValidationWhile preexisting equipment was available for self-diffusion measurements, electrophoreticNMR measurements required the design and construction of a new probe. In this section wewill go through the design, construction and validation of a sample holder which allows foreNMR measurements with minimal artifacts, at the expense of magnetic field homogeneity.3.5.1 IntroductionAs discussed in Section 2.2, electrophoretic NMR (eNMR) is a slowly developing methodthat was initially conceptualized by Packer in 1969 [64]. Even today, there is no indus-try standard for an eNMR probe design and a handful of labs around the world haveimplemented their own solutions to the common issues presented when trying to mea-sure electrophoretic mobility using magnetic resonance methods. The following sectiondescribes the experimental setup constructed and used in all eNMR experiments presentedin chapter 5. This setup has a few advantages when compared to other eNMR probes, es-pecially for measuring high conductivity samples such as RTILs at different concentrations.These advantages enable the usage of a relatively simple sample cell, sacrificing magneticfield homogeneity while overcoming most of the known issues of solution eNMR that aredescribed in detail in Section 2.2.2.3.5.2 MethodsIn order to overcome eNMR artifact flows and obtain accurate mobility measurements,a horizontal magnetic field gradient component of the Micro2.5 gradient set describedin Section 3.4 was utilized. The horizontal gradient, which is perpendicular to gravityand to the direction of the main magnetic field,B0, is also perpendicular to convectiveflows and bubbles that can cause artifacts. This enabled the construction of a probe453.5. eNMR Probe Construction and Validationwhich is simple, inexpensive, and free from artifacts at the cost of static magnetic fieldhomogeneity. A diagram of the probe can be seen in Fig. 3.3. In this simple design, twoinert electrodes (either carbon glass or Pd wire) were dipped in the electrolyte solution.The sample itself was placed in a cut section of a standard 5 mm NMR tube, providinga large fill factor. In order to withstand high polarity solvents (such as PC), the sampleholder was machined from polyoxymethylene which has excellent solvent resistance. Theelectrodes were positioned 3.4 cm from each other and kept at a relatively large distancefrom the RF coil (larger than half of the coil size) to minimize noise pick up. A comparisonof signal to noise (SNR) between scans with applied current and without has shown anaverage of 20% decrease in SNR when current was applied. It was concluded that thisslight decrease in SNR does not pose a significant issue. The electrodes were connectedto a relatively low 50 V power supply with a constant current regulator (\u223c1 mA) and anH-bridge, which allows control of the direction of current through the sample. This low,constant current allowed for steady state electrophoretic flow [58] with minimal heating,and the voltage sign switching, controlled by a micro-controller connected to the pulseprogrammer, made it possible to avoid the build up of charge at the electrodes whichwould stop the measurable flow. In order to ensure measurement of flow in the steadystate, the electric field was initiated 200 ms before the pulse sequence started and wascut off after the signal acquisition was done. External air cooling was added under thesample to minimize changes in \u00b5 due to resistive heating which would cause changes tothe sample\u2019s viscosity. The 3-axis gradient set (Bruker BGU BAFPA 40) had a maximumgradient strength of 100 Gcm along each axis and was used in a direction parallel to thedirection of the sample cell and electrophoretic motion.When voltage is applied, the ions in the sample migrate to the oppositely chargedelectrode. The constant drift velocity for the ions is reached when the drag force is equal463.5. eNMR Probe Construction and ValidationFigure 3.3: Diagram of our simple, horizontal home-build eNMR probe and sample holder.to the driving force, a very quick process which happens almost instantly (ns) [92]. Theions in the sample continue to flow as long as the electric field is not masked by the ionsin solution. As the electric field is applied, the ions begin to migrate in the direction ofthe oppositely charged electrode (Fig. 3.4). This flow in a direction perpendicular to thestatic magnetic field, B0, may give rise to an ionic Hall effect. This would result in aslight redistribution of the ions in a direction perpendicular to the measured flow, andthus have minimal effect on the measured signal. As the ions flow from one electrode tothe other, they accumulate near the electrodes and a concentration gradient grows in theopposite direction of the electrophoretic flow. When the total charge of the ion layer isequal to the charge at the electrode an equilibrium is reached. The diffusion resulting fromthe concentration gradient is equal to the electrophoretic migration; the total flow of ionswill stop [93]. This migration time constant, \u03c4eq is unique to each sample as it depends473.5. eNMR Probe Construction and ValidationFigure 3.4: Diagram of the charged particle distribution in the sample at the undesiredequilibrium state in which no electrophoretic flow occurs. q represents the charge on theelectrode. The electric field is masked by the layer of ions close to the electrode surface.on the sample size, ionic strength, and the applied electric field strength. This state ofequilibrium is avoided when the ions undergo electrochemical processes at the electrode ina process similar to that seen in an electrochemical cell [94]. In this scenario, the migratingions undergo a redox reaction at the electrode, charge accumulation is avoided, and theelectrophoretic migration continues.A constant current regulator was used to ensure that the speed of the ions remainsconstant for the duration of the experiment. This regulator will make adjustments to thevoltage if the resistance of the sample changes, resulting in a constant current flowingthrough the sample. This both helped to make sure no minor fluctuations of the ion flowoccurred during the experiment and helped us to confirm no equilibrium was reached.An external voltage measurement showed that after a rise time of \u223c 100 ms, the voltageremained constant, indicating no state of equilibrium was reached.In order to avoid the effects that charge accumulation might have on the experimentspresented in Section 3.6 and Chapter 5, two precautions were taken. One was to switch thedirection of the applied voltage in between acquisitions [85, 79, 95]. The other was to waita significant amount of time (at least 10 s) in between scans to allow for diffusion driven483.6. Data Interpretationby chemical gradient enough time to restore equilibrium [58, 68, 57]. This long delay alsohelps to reduce resistive heating (as discussed in Section 2.2.2). The pulse sequences canbe seen in Fig. 3.5.Figure 3.5: eNMR pulse sequence diagram. The polarity of the applied voltage is reversedbetween scans and the delay between scans was on the order of 10 s to allow the ions toreturn to equilibrium.3.6 Data InterpretationThe zeroth order phase was extracted from the data by plotting the phase of the freeinduction decay (FID) as a function of time and fitting to a line. The zeroth order phasewas taken as the intercept of the line at time zero. The uncertainty in the measured phasewas less than one degree. Together with the uncertainty in the voltage measurements, theaverage mobility error was \u2248 5 %.Results and DiscussionThis experimental setup and pulse sequence allowed us to measure the phase shift of theNMR signal in response to an externally applied voltage. The measured phase is dependent493.6. Data Interpretationon the ions\u2019 displacement \u03c6 \u221d \u03bdt and therefore on the magnitude and direction of theapplied current. An example of the resulting phase shift can be seen in Fig. 3.6. Fig. 3.6ashows the positive zeroth-order phase shift caused by the application of the required voltageto allow for a positive current of 1.06 mA. Fig. 3.6b shows the response of the NMR signalto the same magnitude of V but in the opposite direction. It is easy to see the inversionof direction in the zeroth-order phase shift, caused by the inversion in migration directionfor the NMR visible ion.The extent of the measured phase can be seen in Fig. 3.7 as the phase difference vs.the applied current. The measured phase showed an excellent linear correlation to theapplied current. The magnitude of the measured phase increases with an increase in cur-rent magnitude. When no current was applied, the measured phase remained unchanged.From the measured phase we can extract the distance the ions passed during the experi-ment. In combination with measured electric field, this distance allows us to extract theelectrophoretic mobility coefficient, \u00b5, using Eq. 2.7. Using this set-up we measured ionicmovement in the range of \u223c10-80\u00b5m and measured mobility constants in the range of \u223c8.5\u00b710\u201312 to 1.5\u00b710\u20138 mV\u00b7s . This large range of mobility is necessary in order to investigateconduction mechanisms of ion dense materials.3.6.1 ConclusionsThese measurements show that electrophoretic driven transport can be measured in iondense solutions with relatively high conductivity. The validation of this probe has shownthat by using a simple eNMR probe in combination with a horizontal magnetic field gradientset, it is possible to minimize the effects gravitational artifact flows on the measured signal.Using this eNMR set-up, we are able to separate flow that is driven by the external electricfield.503.6. Data Interpretation(a) (b)Figure 3.6: 19F-NMR spectra of 2.0 M LiTFSI in PC under applied voltage, performedat 22 \u25e6C. The constant current was kept at (a) +1.06 mA or (b) -1.06 mA. Notice theinversion in zeroth-order phase in response to the reverse direction of ionic electrophoreticflow. \u2206\u03c6 = 17.25\u25e6 and the calculated distance the ions have passed during the experimentwas \u223c39.8\u00b5m.Figure 3.7: Acquired zeroth order phase vs. constant current shows an expected linearrelation between flow and phase according to \u03c6 = \u03b3g\u2206\u03b4\u00b5E. The direction of the phasechanged as the direction of the applied current was changed. When no current was applied,the phase shift remained zero. Mobility was extracted from the linear fit and calculated tobe 2.49\u00b710\u20138 mV\u00b7s .51Chapter 4Diffusion of EMI-TFSI Dilutionsin Propylene Carbonate4.1 IntroductionIontronics are devices based on the conductivity of ions, in contrast to electron conductiv-ity which is used in electronics. Driven by energy storage needs, flammability concerns,and the potential for novel new devices, interest in ion dense ionic conductors continuesto increase [96, 97]. Ionic electroactive polymers (IEAPs) are a class of ionic-mechanicaltransducers composed of an ionic conducting membrane swollen with electrolyte. Theseare used to create actuators that can convert applied voltage to mechanical work and canalso work in reverse by behaving as sensors that can convert mechanical deformation tomeasurable voltage. This ability to act as an actuator and a sensor, together with theirgeneral softness, soundless operation, light weight, and low operating voltage are all prop-erties that have made IEAPs promising candidates for artificial muscles and as mechanicalstress sensors [98].The sensing and actuation behaviours are attributed to the newly termed \u201cpiezoioniceffect\u201d which has been previously observed, but its mechanism is still under investiga-tion [99, 100, 101, 102]. A suggested mechanism attributes the produced voltage to thedifferences in diffusive properties of oppositely charged ions. Upon pressure application,524.1. Introductionthe movement of the ions is not necessarily equal. If one ion moves faster than the counte-rion, temporary charge separation is created and this potential difference can be measured.In Fig. 4.1 a diagram of an IEAP based sensor device is shown. The mechanical motionof a lever arm can be measured using IEAP that is in direct contact with two conductingelectrodes. When the level arm moves and bends the IEAP, voltage is generated across theIEAP. The electrodes, which are connected to a potentiostat, can measure the generatedvoltage and quantify the extent of the arm motion. Research by Woehling et al. usingRTILs as potential electrolytes for actuators revealed an interesting phenomenon.As previously measured by Tokuda et al. [59], self-diffusion measurements of neat 1-ethyl-3-methylimidazolium-bis(trifluoromethylsulfonyl)imide (EMI-TFSI) displayed fasterdiffusion for the cation ( D+D++D\u2013 > 0.6) which is significantly smaller than the TFSI\u2013 an-ion. And indeed, film actuators built from an interpenatrating polymer network (IPN) ofpoly(ethylene oxide) (PEO) and nitrile butadiene rubber (NBR), swollen with neat EMI-TFSI have shown cationic behaviour. However, when actuators with different dilutions ofEMI-TFSI with propylene carbonate (PC) as a co-solvent were fabricated, the displace-ment direction of the actuator changed and began showing signs of anionic behaviour as thedirection of displacement reversed (see Fig. 4.2a). To confirm this inversion, open circuitvoltage measurements in response to applied strain were performed on the samples, andindeed an inversion in the voltage direction was observed (see Fig. 4.2b).It is believed that at the inversion point, the migration of the ions is equal, the sumof physical displacement in both directions is null. Interestingly, the maximum ionic con-ductivity was also observed at the inversion point (Fig. 4.3), suggesting that the maximumionic conductivity occurs when transport numbers of the two ions are equal. The migrationof the ions is expected to depend on the interactions between the ions. According to Dupontet al., neat imidazolium-based RTILs form a supramolecular crystal-like structure that is534.1. IntroductionFigure 4.1: Experimental setup for sensor characterization. Figures show the sensor withapplied stimulation (right) and without (left). The mechanical stimulation was appliedat 3 mm from the electrical contacts on the lower side using the Muscle Lever Arm 300C, which means that during bending movement (right), the sensor side connected to theworking electrode (WE) is contracted and the side connected to the counter electrode (CE)expanded. The copper electrodes are in direct contact with the IEAP and are connectedto a potentiostat in such a way that the WE is always the upper one. Diagram is adaptedfrom Woehling et al. [3].(a) (b)Figure 4.2: (a) Actuator mode strain response to applied potential of 2 V. Notice theinversion in strain direction around 2.5M. (b) Open circuit voltage for different [EMI-TFSI] in PC in response to 4 % strain. Notice the inversion in voltage direction around2.5M [3]544.1. IntroductionFigure 4.3: Ionic conductivity obtained by electrochemical impedance spectroscopy. Max-imum is measured at the same concentration as the actuation inversion point.based on hydrogen bonds (through the imidazolium hydrogens and anion\u2019s halide) [22, 103].With the addition of a co-solvent, the correlation length of the supramolecular structure isreduced; the structure breaks down into smaller fractions of associated ion species such astriple-ions, contact ion-pairs, solvent separated ion-pairs, and, eventually, free ions. Thelevel of separation is dependent on the polarity of the co-solvent as it masks the attractionbetween the ions [104, 105]. This suggested mechanism seems very intuitive and is knownfrom the similar behaviour of electrolyte solutions [28] but due to the relatively short listof techniques that are able to investigate individual ions in RTILs, most RTIL\/solventmixtures behaviour remains speculative.This chapter describes the self-diffusion measurements for the two ions of RTIL EMI-TFSI in varying concentration in PC solvent, both in IPN and in solution. The concen-trations investigated were chosen according to the inversion point seen in the preliminarysensor measurements, to assure measurements of samples which are dilute, concentrated,and close to the inversion point. These measurements help us better explain the effectdilution has on EMI-TFSI association in PC.554.2. MethodsFigure 4.4: Structural formula of RTIL EMI-TFSI and co-solvent PC4.2 MethodsHigh resolution 1H spectra of EMI-TFSI were collected using a high-resolution solutionprobe in a 400 MHz Varian Unity Inova spectrometer. For IPN samples NBR\/PEO (40\/60)rubber sheets were cut to 4\u00d720\u00d70.25 mm3 and soaked in the diluted solutions for a min-imum of 48 hours before measurements. Two sheets were placed between glass slides andinserted into an NMR tube with the sheets positioned either parallel or perpendicular tothe magnetic field gradient. PFG-STE experiments were performed at 22 \u25e6C using a home-built PFG probe [2] in a home-built NMR spectrometer [89] operating at 8.4 T, which areboth described in Chapter 3.To track the individual ions, 19F-NMR was used for TFSI\u2013 and 1H-NMR for EMI+.The home-built horizontal diffusion probe has limited B0 homogeneity but excels in highgradient strengths that can measure slow diffusion of samples with short relaxation times.In combination with the high resolution solution spectra of the RTIL, it was possible to usethe relatively well separated cation peaks around 7-10 ppm to track the motion of the ion,as well as the overlaying solvent peaks around 3 ppm to track the diffusion of PC. Diffusionmeasurements were made using the PFG-STE NMR pulse sequence (see Section 3.2 andFig. 2.2 for the complete pulse sequence). A gradient pulse of \u03b4=318\u00b5s was applied in564.3. Results and DiscussionFigure 4.5: Example 19F spectra of 3 M EMI-TFSI\/PC in PEO-NBR IPN. The singleTFSI\u2013 fluorine peak is shown at -79 ppm. Spectrum collected on the 360 MHz spectrometerusing a home-built horizontal diffusion PFG-NMR probe.varying strength from about g = 50 Gcm to 1000Gcm . For the samples soaked in NBR\/PEOsheets, the diffusion is attenuated by the medium and required higher gradient strength.Gradients between g = 100 Gcm to 2000Gcm were used. The diffusion time \u2206 was variedbetween 500 \u2013 700 ms. An acquisition delay of T=20 ms was used in order to allow eddycurrents to decay before acquisition.4.3 Results and DiscussionTo track the individual ions, 19F-NMR was used for TFSI\u2013 and 1H-NMR for EMI+. Theresulting spectra can be seen in Figs. 4.5 and 4.6, respectively. The lines in both spectraare generally broad due to the shorter relaxation times which are a result of the PEO-NBR polymer network [60]. In addition, the sample could not be well shimmed due toits geometry as it is made of individual films mounted inside a short horizontal section ofa 5 mm NMR tube. Having a single peak spectrum with no background, the TFSI\u2013 ionwas easily tracked by simply integrating the single peak. The same cannot be said for the574.3. Results and Discussion(a)(b)Figure 4.6: Example 1H spectra of 3 M EMI-TFSI in PC (a) shows a solution sampleand (b) shows the same sample, embedded in PEO-NBR IPN. (a) was acquired usinga high resolution solution probe in a Varian Unity Inova 400 MHz spectrometer. TheEMI+ protons are seen around 8 ppm and the PC protons around 3 and 0 ppm. (b)Spectrum collected on 360 MHz spectrometer using a home-built horizontal diffusion PFG-NMR probe and shows the EMI+ protons around 7 ppm and the PC protons around 3 and0 ppm. The background signal from the PEO-NBR IPN is present in the spectrum andmakes peak separation challenging.proton spectrum. The 1H NMR spectrum contains more peaks, which overlap with theproton background from the solvent and the polymer network. To help separate the peaksof the cation from the solvent and polymer background, a solution spectrum of 3 M EMI-TFSI in PC was collected. This spectrum can be seen in Fig. 4.6a. From this spectrum,which has an improved resolution, we can easily separate the EMI+ cation peaks fromthe solvent. By choosing to track the diffusion of the peak which is most separated fromthe background, around 8 ppm, we can separate the ion diffusion from the solvent and thepolymer.Self-diffusion coefficients (D) were measured at room temperature for a set of differentconcentrations of EMI-TFSI\/PC embedded in PEO-NBR IPN. The self-diffusion coeffi-584.3. Results and Discussion(a) (b)Figure 4.7: Example of a diffusion experiment for the anion, (TFSI\u2013) of 2.50 M EMI-TFSI\/PC in PEO-NBR IPN with a diffusion time \u2206 = 375 ms, performed at 22 \u25e6C.(a) shows the single TFSI\u2013 fluorine peak vs. gradient strength. The data is fitted to show aGaussian decay of the signal with an increase in gradient strength. (b) shows the same data,plotted as the natural logarithm of the signal intensity vs the gradient strength squaredto linearize the Gaussian behaviour of the Stejskal-Tanner diffusion equation. Both figuresshow excellent fits to a single diffusion coefficient of D = (4.71\u00b1 0.05)\u00d7 10\u20137 [cm2s ].cients were extracted by fitting the signal intensity vs. magnetic field gradient strength toa Gaussian curve as described in Eq. 2.4. To further validate the fit, the natural logarithmof the intensity was plotted against the magnetic field gradient strength squared and mod-elled with a straight line. A typical data set can be seen in Fig. 4.7 where the data waswell fitted to a single component Gaussian decay, indicating a single diffusion coefficientfor each ion. The measured values for the two ions were of the same order of magnitude astypical self-diffusion coefficients for EMI-TFSI solutions observed previously [21]. For IPNsswollen with solutions, the values were again on the same order of magnitude as similarexperiments done on EMI+ embedded in ionic polymer conductor network composites [106]and for TFSI\u2013 in LiTFSI-PEO complexes [107].Measurements were performed at two orientations and no notable differences in diffusioncoefficients were detected between samples that were measured parallel to the magnetic594.3. Results and Discussionfield gradient vs. perpendicular, indicating no structural anisotropy of the films. Theexperiment was repeated at different diffusion times for each concentration and the self-diffusion coefficients were averaged. Fig. 4.8 summarizes the self-diffusion coefficients ofEMI-TFSI\/PC with varying concentrations, both for solution samples and for IPNs swollenwith solution. Table 4.1 shows the ratio of measured D for the two ions (D+\/D\u2013). Theself-diffusion of the co-solvent PC is presented in Fig. 4.11 and will be discussed below.(a) Electrolyte in solution (b) Electrolyte in IPNFigure 4.8: Self-diffusion measurements for EMI-TFSI dilutions in PC in solution (a) orembedded in IPN (b).To discuss the effect dilution has on EMI-TFSI diffusion in PC, we should first returnto the Sutherland equation (Eq. 1.4) and the Stokes-Einstein equation (Eq. 1.5) in order tosee what the theory predicts. In addition, the solution permittivity and Debye length will,in theory, affect self-diffusion. In general, the theory predicts two opposing effects; one ofthem suggests an increase and the other a decrease in D with increased concentration. Asmentioned before, these equations are mostly applicable in the low concentration limit, buthave shown to be relevant to higher concentrations as well [10].604.3. Results and DiscussionBoth the Sutherland and the Stokes-Einstein equations predict an inverse relation be-tween self-diffusion and electrolyte concentration. The factor most affected by ion concen-tration is the friction drag factor, f. Many factors contribute to the change in f, with themajor effect coming from changes in the viscosity of the solution. In a polar solvent, suchas PC, an order of magnitude increase in concentration typically increases the viscosity bya factor of approximately 30 [108]. The increased amount of solute ions leads to a morecrowded environment which results in more interactions between all types of ions. Molec-ular crowding is a term mostly used for biological macromolecules, but it has been shownto apply to electrolytes in high concentration solutions, causing decreased solvent activity,resulting in higher molecular interactions [109]. These interactions cause a dramatic in-crease in viscosity. In addition to the increase in viscosity, higher electrolyte concentrationand ion interactions are expected to increase the ionic pairing and other longer lastingion-ion associations, causing a larger effective size for each moving species, again causinga decrease in self-diffusion with an increase in concentration. The data in Fig. 4.8 clearlyshows a decrease in self-diffusion with an increase in concentration, but this decrease isonly about a fifth of what the theory predicts.The effect that might be causing the opposite trend comes from the fact that thesolution\u2019s permittivity is inversely proportional to its ionic strength. An increase in con-centration corresponds to an increase in ionic strength and an increase in screening whichdecreases the Debye length (Eq. 1.1). This decrease in the Debye length in turn decreasesthe ion-solvent interactions and the size of the solvation layer around each ion. This cancause the effective ion size to decrease, increasing the self-diffusion with an increase inconcentration. Yet, this effect remains small as it is countered by the increase in Bjerrumlength (Eq. 1.2), caused by the same increase in ionic strength which increases ion-ioninteractions. In general, it is safe to say that as ionic strength increases, the solvation layer614.3. Results and Discussionis replaced with ion-ion long range interactions.This combination of opposing effects leads to the general prediction that self-diffusionis expected to decrease as ion concentration increases. This prediction has been verified,as can be seen for example in the work of Anderson et al., where self-diffusion of neutrallarge molecules was measured with respect to solutions\u2019 ionic strength [12] and in the workof Coglitore et al. [110] where self-diffusion of gold nanoparticles and polystyrene was mea-sured with respect to the diffusing particle size. Fig. 4.8 shows that the anion\u2019s behaviourgenerally follows this expectation. The self-diffusion coefficient increases with a decreasein concentration, especially as it reaches 0.01 M, closer to the low concentration limit. Thesame cannot be said for the cation\u2019s self-diffusion as its concentration dependence displaysseveral characteristics which are initially surprising.An increased radius of a moving species is expected to hinder the self-diffusion coeffi-cient. From the ions\u2019 structure (see Fig. 4.4) we can see the anion is larger in size thanthe cation. According to this, one would expect the self-diffusion of the anion to be consis-tently smaller than that of the cation. Indeed we see that for high concentrations of RTIL,the cation diffuses faster than the anion. Yet, for low concentrations, it appears that theopposite is observed. Looking at Table 4.1 which shows D+\/D\u2013, we can see that for verydilute solutions, the cation diffuses slower than the anion.Electrolyteconcentration (M) Solution IPN0.01 0.56 0.911.00 2.01 1.502.50 2.67 1.693.00 2.64 1.98Table 4.1: Ratio of self-diffusion coefficientsD+D\u2013for EMI-TFSI in solution and IPNThis surprising behaviour can be explained by the charge distribution of the cation. At624.3. Results and Discussionlow concentrations, the EMI+ cation is heavily solvated due to its substantial electric dipolemoment [111] in addition to the dipole moment of the co-solvent. Because PC is an aproticpolar solvent with a relatively high dipole moment of 4.9 D [112], we expect a high affiliationbetween the solvent and the cation (see structures in Fig. 4.4) [113]. This relatively largesolvation layer causes the effective radius and mass of the solvated ion to hinder its self-diffusion. The effect of high solvation results in the unexpected diffusion ratio betweenthe ions. At low concentrations, the diffusion of the smaller cation is slower than that ofthe larger anion. This is again a result of the solvation layer around the EMI+ causing asignificantly higher effective size and mass and consequently lower diffusion coefficient thanTFSI\u2013, as the anion is relatively free of solvent because the symmetrical charge distributionin its dominant conformation results in a very small dipole moment[114].With an increase in concentration, we can see another atypical behaviour. The maximalself-diffusion of the cation is not at the lowest concentration. This is again a result of thelarge solvation layer around it, resulting in a large effective radius. With an increase inconcentration, the EMI+ sheds much of this solvation as the ionic strength of the solutionincreases, causing the maximal self-diffusion to be at a relative high concentration of 1 M.This effect can be seen in the IPN as well, but to a much smaller extent as the interactionof the ions with the polymer network reduces the solvation layer around the ions as EMI-TFSI is known to have high affinity to polymer networks with polar groups, specificallyto PEO where the PEO ethers coordinate both the EMI+ and TFSI\u2013 ions [115]. Thesehigh affinity interactions replace the ions\u2019 solvation layer with fixed groups on the polymer,resulting in an absorption of the RTIL in the PEO polymer network and a reduction inthe solvation layer size.At still higher concentration, both diffusion coefficients decrease as the solution viscosityincreases and ion interactions slow ion motion. The TFSI\u2013 diffusion coefficient decreases634.3. Results and Discussionslightly faster than that of the EMI+ as the RTIL concentration increases, as has beenobserved in some other RTILs [57].Another point of comparison between the self-diffusion data and the device behaviour isthe concentration at which an inversion between anionic and cationic behaviour is observed.Table 4.1, which shows the ratio of self-diffusion coefficients for the ions, shows this inversionbetween 0.01 M and 1 M. This inversion in D is expected to be seen in the device behaviour-both in actuation behaviour and in open circuit voltage measurements, as they both relyon differences in diffusion rates of the different ions. Yet, the concentration at which wesee the actuation inversion does not match that seen in self-diffusion measurements. Theinversion in device behaviour occurs at a higher concentration as can be seen in Fig. 4.2,around 2.5 M. This disagreement is an indication that there is a difference between thecharge carrying abilities of the ions to their general self-diffusion, i.e., it suggests thatthere are some species that diffuse, but do not carry charge. A direct comparison betweendevice behaviour and self-diffusion is not possible and direct measurements of mobilityare necessary. From this difference in inversions, we expect that correlated ion motionexists in the system and affects its conductivity. To further investigate this assumption,conductivities were estimated from our self-diffusion measurements.Using Eq. 1.16 the estimated ionic conductivity can be derived from self-diffusion mea-surements. These results are presented in Fig. 4.9. As has been discussed in Section 1.1.4,the derived conductivities from PFG-NMR measurements do not include the importantfactor of zeff and are therefore used in a qualitative manner, in comparison to directlymeasured conductivities.The estimated conductivity trend in Fig. 4.9 seems to agree with the directly measuredionic conductivity (Fig. 4.3). A maximum of both measured and estimated conductivitiesis observed around 2.5 M. The fact that the maximal conductivity is not at maximal or644.3. Results and Discussionminimal dilutions is common for RTIL dilutions in solvent. This is a result of the balancebetween two opposing factors; as the charge carrier concentration increases the conductivityincreases, however the increased concentration reduces their diffusivity.(a) Solution (b) IPNFigure 4.9: Estimated conductivity derived from self-diffusion measurements. Notice theconductivity for 0.01 M is lower by two orders of magnitude from the highest ionic conduc-tivity measured at 2.50 MAs suggested by Tokuda et al., the ionicity (or level of ionic association) can be calcu-lated for a total electrolyte by using the directly measured total ionic conductivity and theestimated total conductivity derived using PFG-NMR, self-diffusion measurements, andthe Nernst-Einstein relation (Eq. 1.15). This ratio of\u03c3NMR\u03c3directis presented in Fig. 4.10 andshows an interesting trend.For solution samples at low concentrations, this ratio starts from a number slightlyhigher than one, indicating that the estimated conductivity is slightly higher than themeasured conductivity. This observation indicates very few to no neutral aggregates andmight even indicate an additional conduction mechanism that does not rely solely on charge654.3. Results and DiscussionFigure 4.10: Ratio of directly measured conductivity \u03c3mes to the estimated \u03c3NMR. \u03c3NMR isestimated from NMR diffusion measurements using the Nernst-Einstein relation. This ratiois an estimation of the \u201cionicity\u201d of an ionic liquid, and shows a minimum at 2.50 M, thepoint of maximal total conductivity. Values are shown for both solution and IPN embeddedelectrolyte. Value of neat RTIL (self concentration of 3.90 M) is added for comparison.transport via the motion of the ions. This observation does not exist for IPN embeddedsamples, as even at low concentration, the local concentration in the IPNs is high enough tonot allow for the ions to be distant enough to avoid the formation of neutral ion aggregates.For both solution and IPN embedded samples, a trend as the concentration of RTIL in-creases can be identified. The ionicity of the solution decreases, indicating the formation ofneutral associations. This is expected from an increase in the molecular crowding found inthe solution and an increasing number of collisions between counterions. This phenomenonis common for many electrolyte solutions and has been observed before for both RTILs andsimpler electrolyte solutions [28, 116, 117]. Interestingly, an inversion in the trend can beseen for high concentrations of RTIL. The ionicity increases as the concentration reaches\u223c75% of the neat ionic liquid, at 2.5 M. A similar behaviour was seen by Hou et al. forthe same cation but different anion at lower concentrations using water as a solvent [57].664.3. Results and DiscussionThis inversion indicates that in high enough concentration, the ions tend to decrease inion-ion associations or shift to odd-numbered aggregates, either of which would result inincreased conductivity. This means that the addition of a solvent to RTIL initially firstpromotes ionic association and after further dilution, promotes dissociation. The crossoverbetween these two behaviours is seen at the same concentration as the maximum of ionicconductivity, again confirming the effect of ion-ion associations.Figure 4.11: Co-solvent PC diffusion measurements in a solution sample vs. RTIL concen-tration.The measured diffusion of the co-solvent is presented in Fig. 4.11. In agreement withpredictions for solvent self-diffusion [118], D decreases with increased ionic strength. Be-cause there is no inversion in the diffusion trend for the solvent, we can conclude thatthe inversions we see in the diffusion of the ion is a result of ion-ion associations and notgeneral effects which arise due to the viscosity.Because the solvation of the anion and the cation in PC is not the same, the effect ithas on each of the ions might not be identical. Both PC and TFSI\u2013 are electron donormolecules in comparison to EMI+ which is an electron acceptor. Also, if we look at the674.4. Conclusionsshape of PC (Fig. 4.4) we notice its negative charge distribution is concentrated on theoxygen on its less bulky side. This might indicate an easier association to its negative sideand therefore a heavier solvation layer on the cation. The charge and its location indicatethat the cation is more easily solvated with PC than the anion. This could explain whyTFSI\u2013 tends to aggregate more causing a reduced anion diffusion at higher concentration.Another possible interaction that affects the diffusion arises from the interactions ofthe ions with the polymer network which were mentioned before. The high affinity be-tween PEO and the RTIL indeed decreases the solvation layer around the ions, but mightalso decrease the ionic mobility of the solution by hindering the motion of the ions. Todistinguish these interactions from one another, further research is needed using differentco-solvents and different polymer compositions.4.4 ConclusionsSelf-diffusion measurements of EMI-TFSI dilutions in PC have been measured for bothRTIL ions in solutions and IPNs. These measurements help to shed some light on themechanisms behind the behaviour of actuators and sensors made of these materials.In both solution and swollen IPNs, the change from cationic to anionic behaviour thatwas seen previously in actuator devices can also be seen in the self-diffusion data. Aninversion in diffusion ratios between the ions depends on the RTIL concentration. Fromthis inversion, we confirm that the observed change in the devices\u2019 actuation behaviour is aresult of changes in their transport properties. This change in transport properties with re-spect to dilution shows that the extent and amount of ion-ion and ion-solvent associationsdepend on the sample concentration and helps confirm the suggested piezoionic mecha-nism. In addition, self-diffusion measurements show an expected disagreement betweendirectly measured conductivity and self-diffusion estimated conductivity. This disagree-684.4. Conclusionsment again suggests the existence of ion-ion and ion-solvent interactions and enables us toasses the\u201cionicity\u201d of each sample. The level of ionicity is an important factor in the un-derstanding of the conduction mechanism of these ion dense solutions and ion dense IPNs.The ionicity level measured is minimal at the concentration of maximal conductivity mea-sured in devices. This agreement helps to confirm the proposed Dupont mechanism, whichsuggests an increased fraction of ion associated species compared to neat RTIL.From the measurements of self-diffusion and conductivity, we conclude that ion-ionassociations exist in both solution and IPN devices, and they affect the conductivity of ionicconductors greatly. The dilution of RTIL can change the abundance and composition ofthese associated species and thus affect both the sign and magnitude of device conductivityand actuation response.69Chapter 5Ion Transport in Touch SensorElectrolytes5.1 IntroductionAs has been discussed in the previous chapter\u2019s introduction (Section 4.1), iontronics is agrowing field of interest, in part because of the potential for iontronic devices to operateboth as actuators and as sensors. In comparison to conventional electronics, iontronics havethe ability to interface with biological tissues as they do not provoke an immune responsefrom the body [119]. These properties make iontronics ideal candidates for artificial musclesand nerves.Potential devices rely on the mechanoelectrical transduction properties of ion denseconductors that have been shown in a number of ion-polymer systems such as hydrogels,polyelectrolytes, and ionic polymer materials [120, 121]. This ability to act as sensorsand actuators comes from the \u2018piezoionic effect\u2019 [100, 101, 99, 102] which was introduced inChapter 4. The piezoionic effect is defined in a similar way to the piezo-electric effect. Whenundergoing physical deformation, a measurable voltage is produced in the device. Thesuggested mechanism to this phenomenon involves pressure induced ionic redistributionof the electrolyte and depends on diffusion and conduction mechanisms of the ions in thesystem.705.1. IntroductionWhile it has been widely observed that diffusion and conduction mechanisms in ion-dense electrolytes differ from those of ideal solutions [58], detailed understanding of thephenomenon suffers from the small number of experimental techniques that allow accuratecharacterization. In solutions with concentrations higher than approximately half a molar,a shift from typical ionic conduction mechanisms occurs. Due to crowding effects, oppo-sitely charged ions tend to stick to each other, forming ion clusters which have both highermass and volume than a free ion and as a result decrease their self-diffusion coefficients. Inaddition, these clusters have a different total charge termed \u201ceffective charge\u201d (zeff). All ofthese affect the response of the ion cluster to an applied electric field. The clustering mightresult in charge neutral clusters of ions which do not respond to an external applied electricfield and cause the ions to lose their charge carrying abilities. Due to these differences insolvation and the formation of ion-associated structures, the electrophoretic mobility (\u00b5)and self-diffusion depart from the linear ratio predicted by the Nernst-Einstein relation(see Eq. 1.15 and Section 1.1.3 for further information).Although the mechanism behind the piezoionic phenomenon is not yet clear, both themagnitude and direction of this effect can be tuned using variations of ions, solvents, andconcentration. In this chapter we discuss the use of NMR as a method to investigate themechanism behind the piezoionic effect. Transport properties have been measured usingboth PFG-NMR and eNMR in a range of materials. We have measured both self-diffusion(D) and electrophoretic mobility (\u00b5) of ions in salt solutions of varying concentration in anattempt to investigate the difference between self and driven diffusion and to learn aboutthe interesting conduction mechanism of Li+ in these fascinating new materials.715.2. Methods5.2 Methods5.2.1 Sample PreparationElectrolyte solutions of varying concentration between 0.01 M to 3.00 M were made bydissolving bis(trifluoromethane)-sulfonimide lithium (LiTFSI) salt with PC as solvent. Themolecular structure of the ions is shown in Figs. 5.1 and 5.2, respectively. The mixtureswere then sonicated for approximately 10 minutes until completely dissolved.Polymer gels based on Poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP)co-polymer (Fig. 5.3) were synthesized with LiTFSI salt using a solution casting technique.A polymer stock solution was made by dissolving PVDF-HFP pellets in acetone at 7.5 %wt. Electrolyte solutions of varying concentration were then mixed with the host polymersolution at a ratio of 5 mL\/33.3 g (electrolyte\/polymer stock). To ensure complete mixing,the mixture was sealed and sonicated for 2 hours with occasional stirring. The mixture waspoured into a mould and covered with a perforated lid to allow gradual evaporation of theacetone at room temperature. The final polymer content was 30 wt% and the thickness ofthe films was approximately 200\u00b5m.Figure 5.1: Structures of Li+ and bis(trifluoromethane)-sulfonimide (TFSI\u2013) ions5.2.2 Viscosity MeasurementsViscosity measurements were done at room temperature (22 \u25e6C) using an Anton Paar MCR301 rheometer applied with a rotational probe head.725.2. MethodsFigure 5.2: Structure of Propylene Carbonate (PC)Figure 5.3: Poly(vinylidene fluoride-co-hexafluoropropylene) (PVDF-HFP) co-polymer5.2.3 Self-Diffusion MeasurementsSelf-diffusion measurements were performed at 22 \u25e6C using a home-built PFG probe [2] ina home-built NMR spectrometer [89] operating at 8.4 T (both are described in Chapter 3).To track the individual ions, 19F-NMR was used for TFSI\u2013 and 7Li-NMR for Li+. PFG-STE NMR pulse sequence was used (described in Section 3.2 and Fig. 2.2) with diffusionecho time, \u03c4 , of 0.5 ms to minimize the loss of signal due to spin-spin relaxation. A gradientpulse of \u03b4=318\u00b5s was applied in varying strength from about g = 50 Gcm to 1000Gcm forsolution samples, and about g = 100 Gcm to 2000Gcm for samples embedded in polymer films.The diffusion time \u2206 was varied between 350 \u2013 900 ms, depending on the sample\u2019s diffusionrate. An acquisition delay of T=10 ms was used in order to allow eddy currents to decaybefore acquisition.5.2.4 eNMR MeasurementseNMR mobility measurements of electrolyte samples were done at 22 \u25e6C using the home-built eNMR probe (described in detail in Section 3.5). A spin-echo eNMR pulse sequence735.2. Methods(Fig. 2.3) was used with varying echo times of between 25-300 ms. The diffusion time,\u2206, was set between 50 to 600 ms and the pulse gradient duration \u03b4 was varied between1-10 ms. In order to ensure measurement of flow in the steady state, the electric fieldwas initiated 200 ms before the pulse sequence started and was cut off after the signalacquisition was done. The voltage was applied to the sample through a constant currentregulator to achieve a current of 1.5 mA. Tracking the voltage applied to the sample usingan oscilloscope ensured that the current did not reach saturation and remained consistentthroughout the eNMR experiment time. The gradient strength, g, was kept relatively lowat a maximum of 34 Gcm in order to balance the loss of signal due to self-diffusion, andmaximize the phase shift due to the electrophoretic mobility of the ions. The time betweenthe scans was set to 10 s in order to allow the sample to return fully to equilibrium and inorder to avoid heat accumulation due to current flow in the sample.To measure the electric field applied during the eNMR experiment, four point voltagemeasurements were done outside the magnet. The measurements were done by applyingthe same current as was used in the eNMR experiments, using the pulse programmer tomaintain accurate duration as well. The AgCl electrodes used to measure the voltage wereplaced at known distances from each other via two holes drilled through the sample holdersides. This combination of known distance and measured voltage allowed us to calculatethe electric field with an uncertainty of less than 5%.In addition, attempts to measure the electrophoretic mobility in polymer electrolyteswere done using the same set-up. Some of the electrolyte solution was removed fromthe polymer samples by wiping gently with a Kimwipe, to encourage slight shrinkage. Thesamples were then cut and placed in the 5 mm NMR tube and re-soaked with the electrolytesolution while in the tube. As the samples swelled again, contact of the polymer with theglass was maximized. The electrodes were kept in contact with the electrolyte solution to745.2. Methodsconduct the current through the gel sample. The gel sample itself was longer than twicethe NMR coil length to ensure any motion detected was a result of ionic motion inside thegel sample and not the electrolyte solution.5.2.5 Conductivity and Piezoionic MeasurementsIonic conductivities of the polymer gels were determined by electrochemical impedancespectroscopy. Each sample was cut to a 2.5 cm by 1 cm rectangle and clamped between twostainless steel plates. Measurements were performed by sweeping a sinusoidal excitationvoltage from 1 MHz to 0.1 Hz. The resistance of each sample was recorded at high frequencywhere the metal\/polymer interface capacitance effectively shorted the interface impedance.For electrolyte solution samples, the bulk ionic conductivity was measured using acomputer-controlled Hewlett Packard 4192A LF impedance analyzer by applying 10 mV ACat 10 kHz. The measurements of the samples were carried out in a commercially availableconductivity cell with platinized platinum (black) electrode cells (TOA Electronics, CG-511B).Measurements of the piezoionic response were performed by subjecting the polymersto a mechanical deformation and recording both open circuit voltage and current responseas a function of amplitude and frequency of the mechanical perturbation. The deforma-tion was applied using a Bose ElectroForceR\u00a9 3000 series dynamic mechanical analyzer andthe voltage response was measured using a Metrohm AutotlabR\u00a9 PGSTAT101 potentio-stat\/galvanostat. The perturbation was load controlled with a maximum displacementof 2 mm. As pure PVDF-HFP can also show a piezoelectric response, similar tests wereperformed on PVDF-HFP films with no electrolyte concentration to confirm that the piezo-electric response was not a result of the polymer itself. The overall setup is presented indetail in Dobashi\u2019s MASc thesis [122].755.3. Results and Discussion5.3 Results and DiscussionTo initially characterize their piezoionic behaviour, the voltage response of PVDF-HFPco-polymer samples with varying concentrations of LiTFSI were measured in sensor mode.The general set-up and the results are presented in Figure 5.4. A general decrease ofproduced voltage can be seen with respect to increased concentration of salt. An inversionpoint can be seen around 2 M. This change of both magnitude and direction of voltageindicates a significant change in conduction and diffusion properties. This motivates furtherinvestigation with direct measurements of both self and driven diffusion for each individualion in these systems. Measurements were made for both electrolyte swollen polymer filmsand for the electrolyes alone.(a) (b)Figure 5.4: Electromechanical voltage response to controlled deformation of PVDF-HFPco-polymer prepared with varying LiTFSI salt concentration solutions. An inversion ofoutput voltage is seen between 1.50 and 2.00 M which implies an inversion in the dominantcharge carrier species. (b) a general diagram of the measurement set-up.765.3. Results and Discussion5.3.1 Self-Diffusion in Electrolyte SamplesExamples of self-diffusion measurements for both TFSI\u2013 and Li+ ions dissolved in PC arepresented in Figs. 5.5 and 5.6, respectively. The Gaussian decay of the signal intensity vs.gradient strength, as well as the linearized version of the same data, show excellent fits tosingle Gaussians, indicating the presence of a single diffusion coefficient.A summary of self-diffusion measurements for both ions in electrolyte solutions is pre-sented in Fig. 5.7. The results show an expected decreasing trend of self-diffusion as thesalt concentration increases. Viscosity measurements for the same set of samples can beseen in Fig. 5.8a and show a dramatic increase in viscosity with increasing salt concentra-tion which is consistent with calculations [123] and measurements seen in similar RTILs inwater [108]. The expected increase in both crowding effects and viscosity causes a decreaseof more than an order of magnitude in self-diffusion from highly diluted to concentratedsolutions. When the concentration of ions in solution is high, the distance between the ionsis reduced. When the distance between them is lower than the Debye length (\u03ba\u20131), electro-static interactions take place. Depending on the sign of the species\u2019 charge, electrostaticforces either repel or attract the neighbouring ions. The ions that are attracted result inclustered species of two or more ions. These clusters have a larger radius than individualions and diffuse slower, as expected from the Stokes-Einstein equation (Eq. 1.5).Unlike previous observations of concentrated LiBF4 in sulfolane which showed increasedself-diffusion of Li+ in comparison to BF\u20134 [124], the cation\u2019s self-diffusion remains consis-tently lower than that of the anion for all concentrations. This initially might seem surpris-ing due to the fact that the cation itself is significantly smaller than the anion (see Fig. 5.1).Yet, this can be explained by the larger and bulkier solvation layer surrounding the cation,resulting in a large effective total radius of the moving species. Previous calculations by Shiet al. [125] have shown that for concentrations lower than 2 M, the solvation layer formed775.3. Results and Discussionby PC solvent molecules around a Li+ ion consists of 4 solvent molecules, resulting in atetrahedral solvation structure and effective size of more than 10 A\u02da\u00d710 A\u02da\u00d710 A\u02da. In com-parison, due to its symmetrical shape, single hydrogen bonding site (the nitrogen [126]),and well distributed negative charge the TFSI\u2013 anion has lower interactions with the sol-vent molecules [114, 127]. As shown by density functional theory (DFT) calculations andRaman spectroscopy measurements, two conformers are common for the TFSI\u2013 anion. Thestaggered conformer is slightly more stable than the eclipsed, and rotation between the twois expected at room temperature (\u2206E = 2.3 \u2013 2.9 kJmol) [114]. This rotational motion, alongwith the fact that both conformers have a relatively delocalized charge, are the reason theanion struggles to make long-lasting interactions with solvent molecules, resulting in ananion that is almost bare of solvent. This leaves the anion with an estimated effective sizeof approximately 8 A\u02da\u00d74 A\u02da\u00d72 A\u02da [128, 129, 130] which is smaller than the effective size ofLi+-(PC)4[125, 131]. This difference in solvent-ion interactions between the two ions helpsexplain the difference in self-diffusion for the ions and rationalize the fact that the cationself-diffusion is consistently smaller than that of the anion.Another observation from the self-diffusion data in Fig. 5.7 is the general decreasingtrend in self-diffusion as the salt concentration increases. The self-diffusion coefficients forboth ions decrease and become similar. At 2.00 M, self-diffusion coefficients for both ionsare nearly identical and at 2.50 and 3.00 M the difference is eliminated. As the salt concen-tration increases, the ratio of solute to solvent molecules decreases, making it impossiblefor the cation to be solvated by 4 solvent molecules. As the concentration reaches 2.1 Mthe molar ratio of LiTFSI:PC is 1:4 and decreases further as the concentration increases.Shi et al. shows that for higher than 2 M, the dominant species is a planar Li+-(PC)3 andthe effective radius of such species is at least 3 times smaller than the tetrahedral Li+-(PC)4 (3 A\u02da\u00d710 A\u02da\u00d710 A\u02da [125]). In addition, this decrease in solvation layer makes ion-ion785.3. Results and Discussion(a) (b)Figure 5.5: Example of a diffusion experiment of TFSI\u2013 in 2 M LiTFSI\/PC with a diffusiontime \u2206 = 350 ms, performed at 22 \u25e6C. (a) shows TFSI\u2013 fluorine peak integral vs. magneticfield gradient strength. The data is fitted to a Gaussian decay function. (b) shows thesame data, plotted as the natural logarithm of the signal intensity vs. gradient strengthsquared to linearize the Gaussian. Both figures show excellent fits to a single diffusioncoefficient of D = (1.91\u00b1 0.07)\u00d7 10\u201311 [m2s ](a) (b)Figure 5.6: Example of a diffusion experiment for Li+ in 2 M LiTFSI\/PC with a diffusiontime \u2206 = 350 ms, performed at 22 \u25e6C. (a) shows the Li+ peak integral vs. gradient strength.The data is fitted to a Gaussian decay function. (b) shows the same data, plotted asthe natural logarithm of the signal intensity vs. gradient strength squared to linearizethe Gaussian. Both figures show excellent fits to a single diffusion coefficient of D =(1.59\u00b1 0.09)\u00d7 10\u201311 [m2s ]795.3. Results and DiscussionFigure 5.7: Self-diffusion measurements of LiTFSI vs. electrolyte concentration as measuredby PFG-NMR. Solid line shows D for the cation and dashed line for anion. The lines aremeant only to guide the eye.interactions more probable and the ions spend more time as contact ion pairs and solventseparated ion pairs. This ion pair formation has been suggested by Wang et al. [126] whohave done Raman spectroscopy experiments on the same system in varying concentrations.They interpreted their measurements as a potential formation of ion pairs in concentrationshigher than 2.5 M which can be validated by the data presented here.In order to separate the effects of ion-ion and ion-solvent interactions from the generalviscosity of the solution, the total dynamic viscosity of the solution, \u03b7, was measured.Results are presented, on a logarithmic scale, in Fig. 5.8a. As previously mentioned,the viscosity increases as the salt concentration of the solution increases, indicating againthe increased crowding effects and more ion-ion interactions. By combining self-diffusionwith viscosity measurements, we can use the Stokes-Einstein equation (Eq. 1.5) to extractan approximate effective radius (r) for each of the ions. The values for r are presented inFig. 5.8b. These calculations should be considered an approximation due to the underlyingassumptions involved in the Stokes-Einstein equation. The equation accounts for drag force805.3. Results and Discussion(a) (b)Figure 5.8: (a) Dynamic viscosity of the solution vs. salt concentration in PC measuredusing a rotational rheometer (MCR 301) (b) Estimated effective radius (r). The lines aremeant only to guide the eye.applied on rigid, spherical molecules with small Reynolds numbers. These assumptions donot apply accurately to this specific set of salt and solvent conditions and therefore thenumerical values should be considered an estimation. Furthermore, the factor b in thisequation, which accounts for the level of interactions between the moving species and theirenvironment, potentially changes with changes in concentration. Therefore, it is againimportant to interpret these numerical values with caution.Observing the graph in Fig. 5.8b, we can see that the effective radius (r) of the cationhas a strong initial decrease as concentration increases. This dramatic decrease in r is aresult of the increase in \u03b7 as well as changes to the b factor due to initial changes fromtetrahedrally solvated ions to trigonal. This decrease is followed by a more gradual decreasewhich becomes significant again around 2.00 M, by which point we expect the majority ofthe cations have had their solvation layer change from tetrahedral to planar. Looking atthe change in the anion\u2019s effective radius we see a slow, gradual decrease in r from a dilutesolution to about 2.50 M where there is a more drastic decrease in r and the effective radius815.3. Results and Discussionbecomes identical to that of the cation. Around this concentration, we expect to have amore significant amount of ion-ion associations. From our measurements, we see thatthese associated species decrease the measured diffusion coefficient significantly (\u223corderof magnitude), but do not increase the effective size to the same extent. This comes as asurprise because we can see the effect of ion-ion species in D, but not as much in r. Becauseof the method used to calculate r using \u03b7, we do not account for the gradual change in theb factor due to the ion clustering.To compare the self-diffusion data to the experimental sensor data, we needed a com-parable quantity. Conductivity was chosen as the most direct solution yet, to be able tocalculate conductivity from D, the traditional (and most likely wrong) assumption thatthe effective charge, zeff , is equal to the ions\u2019 valence had to be used (Eq. 1.16). Bothderived and directly measured conductivities are summarized in Fig. 5.9 to allow for aneasy comparison.For low concentrations, the derived conductivity from self-diffusion shows an excellentmatch to the directly measured conductivity. This means the assumption of zeff equalto valence charge is valid and the existence of ion-ion associated structures is minimaland insignificant. In highly diluted solutions, the distance between the ions is large enough(larger than the Bjerrum length) therefore the interactions between oppositely charged ionsare minimal and conductivity can be safely derived from self-diffusion measurements. Oncethe salt concentration exceeds 0.25 M, derived conductivities from self-diffusion consistentlyover estimate the ions\u2019 ability to carry charge. Because self-diffusion measurement is notable to distinguish between neutral and charged species, it falsely assumes that neutralspecies conduct charge. It is therefore expected that self-diffusion measurements will fail toaccount for the effect ion pairs, and other charge altering associations, have on conductivity.In an alternative approach, ion transport numbers were used to observe the charge825.3. Results and DiscussionFigure 5.9: Total conductivity derived from PFG-NMR self-diffusion measurements vs.directly measured conductivity for LiTFSI vs. salt concentration in PC. The lines aremeant only to guide the eye.carrying ratio between the ions. Derived from the ratio of ion conductivity to the totalconductivity, both derived from self-diffusion measurements, the transport numbers arepresented in Fig. 5.10. Unlike the inversion seen between 1.5 and 2 M in sensor modemeasurements (Fig. 5.4), we see no inversion of transport numbers from self-diffusion mea-surements in solution. This means that, according to the self-diffusion data, the anionremains the main charge carrier for all concentrations, contradicting the data shown bydeformation measurements. In addition to the known issue of zeff , this disagreement po-tentially arises because we compare the self-diffusion of electrolyte solutions to deformationmeasurements done on IEAP devices. To improve our comparison and look further intothis inconsistency, we have measured the self-diffusion of samples identical to the IEAPdevices and those are discussed in the next section.835.3. Results and DiscussionFigure 5.10: Transport number derived from self-diffusion measurements of LiTFSI vs.salt concentration in PC. Solid line for the cation and dashed line for anion. The lines aremeant only to guide the eye.5.3.2 Self-Diffusion in Polymer Electrolyte SamplesExamples of self-diffusion measurements for the PVDF-HFP co-polymer samples are pre-sented in Figs. 5.11 and 5.12 for TFSI\u2013 and Li+, respectively. The Gaussian decay of thesignal intensity vs. gradient strength and the linearized version of the same data show anexcellent fit to single Gaussians, indicating the presence of single diffusion coefficients. Asummary of the self-diffusion measurements for the IEAP samples is presented in Fig. 5.13a.These measurements are in better agreement with the deformation data as they show asimilar pattern of inversion as was seen for the sensor deformation experiment in Fig. 5.4.The sensor showed an inversion of output voltage in concentrations between 1.5 and 2.0 M,and self-diffusion measurements of the IEAP samples show an inversion at \u223c1 M. Thismight indicate that ion-polymer interactions indeed have a role in changing the chargecarrying abilities of the ions, but they are not identical since we are still only looking at845.3. Results and Discussion(a) (b)Figure 5.11: Example of a diffusion experiment for the anion, (TFSI\u2013) of 2 M LiTFSI\/PCin PVDF-HFP co-polymer with a diffusion time \u2206 = 350 ms, performed at 22 \u25e6C. (a)shows the single TFSI\u2013 fluorine peak vs. gradient strength. The data is fitted to show aGaussian decay of the signal with an increase in gradient strength. (b) shows the same data,plotted as the natural logarithm of the signal intensity vs the gradient strength squaredto linearize the Gaussian behaviour of the Stejskal-Tanner diffusion equation. Both figuresshow excellent fits to a single diffusion coefficient of D = (6.35\u00b1 0.09)\u00d7 10\u201311 [m2s ]self-diffusion, a property that does not take into account zeff of the ions.The self-diffusion of the anion in IEAPs shows a similar trend to that observed in theelectrolyte solutions (repeated as Fig. 5.13b). TFSI\u2013\u2019s self-diffusion coefficient consistentlydecreases with an increase in concentration. Unlike the anion, Li+ shows a different trend.The self-diffusion coefficient of the cation stays relatively constant with an increase inconcentration. This is potentially the result of a similar solvation effect to that which isseen in the electrolyte. However, when embedded in a polymer, solvation and effectiveradius of the cation have a more significant effect over the diffusion as the ion tries to movethrough the polymer network. As the concentration increases, the cation\u2019s solvation andeffective radius are reduced, making it easier for Li+ to move through the polymer. Dueto the small effect solvation has on the anion, it is not as heavily affected by interactionswith the polymer and the general trend of decreasing self-diffusion with increased salt855.3. Results and Discussion(a) (b)Figure 5.12: Example of a diffusion experiment for the anion, (Li+) of 2 M LiTFSI\/PCin PVDF-HFP co-polymer with a diffusion time \u2206 = 350 ms, performed at 22 \u25e6C. (a)shows the single Li+ fluorine peak vs. gradient strength. The data is fitted to show aGaussian decay of the signal with an increase in gradient strength. (b) shows the same data,plotted as the natural logarithm of the signal intensity vs the gradient strength squaredto linearize the Gaussian behaviour of the Stejskal-Tanner diffusion equation. Both figuresshow excellent fits to a single diffusion coefficient of D = (8.2\u00b1 0.1)\u00d7 10\u201311 [m2s ].concentration remains.The change in environment between IEAPs and solution results in the following sur-prising observation: For high concentrations (above 1.5 M), the self-diffusion of both ionsis higher than in solution. This is counterintuitive as we expect a viscous, harder to passthrough medium to hinder the diffusion coefficient of the ion, yet it appears that the intra-molecular interactions in solution are causing a stronger decrease in self-diffusion than theintroduction of a polymer mesh. This is potentially a result of supra-molecular aggregatesformed by the ions and the solvent as was seen by Dupont et al. and others [22, 103]in different RTILs and solvents. In these aggregates, the ions are arranged in a non-rigidstructure in which the ions\u2019 motion is correlated. A visual representation can be seen inFig. 1.3. We believe that in the case for LiTFSI salt, self-diffusion measurements sup-port similar supra-molecular structures. The salt forms large electrostatic aggregated ionic865.3. Results and Discussion(a) (b)Figure 5.13: (a) Self-diffusion measurements of LiTFSI vs. electrolyte concentration em-bedded in PVDF-HFP co-polymer as measured by PFG-NMR. Solid line shows D for thecation and dashed line for anion. (b) is a repeat of Fig. 5.7 to allow for an easy comparison.The lines are meant only to guide the eye.species in solution, but less so when embedded in a polymer network that interrupts thedevelopment of these large structures. These aggregates cause the general diffusion of bothions to be higher in a polymer network than in high concentration solutions.To again compare results of self-diffusion to conductivities, the estimated conductivitywas derived from self-diffusion and compared to directly measured ionic conductivity. Thiscomparison is presented in Fig. 5.14a. Again, we see an excellent match between conduc-tivities at low concentrations and as the concentration of ions increases the results diverge.This is again a result of the increased fraction of ion pairs that affect zeff , a quantity thatis not accounted for in our estimated conductivities. The divergence of the estimated con-ductivity from the directly measured ionic conductivity is greater than that seen for theelectrolyte. This is a result of the self-diffusion coefficients in high concentrations beingan order of magnitude higher than those of the equivalent concentrations samples withoutpolymer. This is an indication that the polymer might be acting as an obstacle, inhibit-ing the formation of especially large ion-ion associated structures, resulting in an average875.3. Results and Discussionsmaller species size that can diffuse faster.(a)(b)Figure 5.14: (a) Total conductivity of Gel samples derived from PFG-NMR self-diffusionmeasurements vs. directly measure conductivity for LiTFSI vs. salt concentration in PC.(b) Transport number derived from self-diffusion measurements. Solid line for the cationand dashed line for anion.It is clear to see that for both solution and IEAP, self-diffusion does not predict con-ductivity parameters well for concentrations higher than \u223c0.5 M. A more direct approachof mobility measurements is needed in order to make predictions or reach any conclusionsregarding conduction mechanisms in either electrolyte solutions or co-polymer electrolytesystems. These direct approach measurements are discussed below.5.3.3 Driven Diffusion in Solution Electrolyte SampleseNMR is a more direct approach to assess conductivity as it directly measures the elec-trophoretic mobility of an ion in solution in response to an applied electric field. The NMRsignal phase modulations are presented in Fig. 5.15a for the TFSI\u2013 anion and Fig. 5.15bfor the Li+ cation. In both Figs. the zeroth-order phase changes in response to the mag-nitude and direction of the applied, constant current through the samples. The direction885.3. Results and Discussionof the phase shift for the counterions is reverse in direction, indicating that the ions movein opposite directions in response to voltage. Mobility coefficients for the individual ionsin solution samples were extracted from the measured zeroth-order phase shift vs. currentflow through the sample.A summary of the mobility coefficients for varying concentrations of LiTFSI in PC ispresented in Fig. 5.16. It is clear to see that the more affected ion by the increase inconcentration is the larger of the two: TFSI\u2013. The mobility of the anion plummets byalmost 3 orders of magnitude between 0.10 and 3.00 M, similar to what was seen in self-diffusion. This similarity is an indication that for the anion, the conductivity mechanism isbased on the diffusivity of the ion. As discussed, in the Stokes-Einstein equation (Eq. 1.5),diffusion is inversely proportional to viscosity. With the increase in salt concentration, thetotal viscosity increases by a factor of 10 for each increase of 1.00 M. This effect explainsthe large decrease in mobility for the TFSI\u2013 anion. Considering this increase in viscosityshould also affect the Li+ cation, it is surprising to see that its mobility is hardly affectedby the concentration increase.With the initial increase in concentration from 0.10 to 0.50 M, the effects of solvationchanges on the cation can be seen in \u00b5 as we leave the infinitely diluted regime. Aspreviously observed and discussed, the number of solvating molecules decreases with anincrease in the solution\u2019s ionic strength [125, 126]. This decrease in r can explain thesmall increase in mobility observed between 0.25 and 1.00 M. Yet, with further increase inconcentration, the mobility remains more or less constant for Li+. This contradicts two ofour previous observations. We know the viscosity of solution increases by approximately afactor of 10 with the concentration increase from 2 M to 3 M. This has been both measured(Fig. 5.8a) and confirmed in the literature [123, 108]. Second, we measured a decreasein Ds for the same solution samples. The drop in measured self-diffusion for both ions895.3. Results and Discussion(a)(b)Figure 5.15: NMR spectra of 2.0 M LiTFSI in PC as a function of applied electric field.The zeroth-order phase of the peaks changes in response to the magnitude and directionof the applied electric field. (a) shows 19F-NMR of the TFSI\u2013 anion (b) shows 7Li-NMRspectra of the Li+ cation. All experiments were performed at 22 \u25e6C using our home-builteNMR probe and eNMR pulse sequence (described in detail in Section 3.5)means the diffusivity of both ions is affected in a similar manner, yet their mobility isnot. From this contradictory information, we must conclude there is a different conductionmechanism for Li+, one that relies on factors other than the ion\u2019s diffusivity and causes a905.3. Results and DiscussionFigure 5.16: Electrophoretic mobility measurements for LiTFSI vs. salt concentration inPC as measured by eNMR. Solid line shows D for the cation and dashed line for anion.divergence from the Nernst-Einstein relation (Eq. 1.15). We suggest that Li+\u2019s conductionmechanism follows a non-Stokesian charge transport that resembles the Grotthuss protonhopping mechanism in water [132, 133] and vacancy Li+ conduction that is seen in solidmaterials [134, 135].In acidic or basic conditions at high enough concentration, water molecules form ahydrogen bond network that can conduct the charge carried by a proton by what is termeda structural diffusion mechanism or Grotthuss proton hopping [136]. By forming a bondwith a proton in a neighbouring ion and breaking an existing bond in the hydronium ion,a proton can \u201cmove\u201d through the length a solution without any single ion required tomove the entire distance. Since the same bond that breaks also forms immediately, thereis no total energy difference and since the distances are small and the hydrogen bondingnetwork already exists, the energy barrier is small [132]. It is important to note, becausethis mechanism does not involve any movement of a nucleus, it can not be observed usingNMR methods. This mechanism, where a structural feature moves through the material,results in much higher conductance than predicted by the diffusivity of water.915.3. Results and DiscussionFor the solid case, it has been established that Li+ conduction becomes depletion basedin solids such as inorganic solid electrolytes [137, 138]. In such materials, the relativelymobile ion can hop from one site to another in the crystal structure. This mechanismrequires the existence of a hole; a vacancy of an energetically equivalent site to hop into.In solid vacancy conduction, the energy barrier to create a hole may be large as it involvesbreaking bonds that link an atom in the crystal structure to its neighbours [139].In our case, it is possible that Li+ conduction occurs by a mechanism having elementssimilar to both of these. The conduction itself which is similar to the vacancy basedconduction seen in the solid case, is made possible by aggregates that are formed in solution,similar to the Grotthuss case. In LiTFSI electrolyte solution, there is no solid crystal butthe proximity of the ions to each other in this concentrated solution creates large supra-molecular structure that resembles the hydrogen bonded network formed by water. Withan increased concentration of ions in solution, large ionic aggregates are formed due to theelectrostatic forces between the ions (Fig.1.3, [22]). When the distances between the ions aresmall enough, the Debye length of one ion overlaps, or nearly so, with a neighbouring ion.In this very viscous crowded case, the diffusion coefficients of the ions are small, as they arebound together in an ordered structure so diffusion requires overcoming the electrostaticforces between the ions. The measured self-diffusion in electrolyte solution agrees withthis assumption as was seen in Fig. 5.7 where the ions\u2019 self-diffusion is very small at highconcentrations. This arrangement, however, could allow fast vacancy migration similar tothat in the solid state.Previous observations [124, 140] in other solvents (e.g. sulfolane, succinonitrile) andwith other anions (e.g. BF\u20134, bis(fluorosulfonyl)amide) have shown that the self-diffusioncoefficient of Li+ becomes larger than that of the anion in concentrated solutions. In thissuggested mechanism, the solvent and anion create aggregates which are involved in bridg-925.3. Results and Discussioning coordination of Li+ that can be observed in increased self-diffusion coefficients. Theseaggregates are non-rigid associated species in which the motion of the ions is correlated,and they provide a scaffold for increased Li+ hopping [124]. This mechanism does notrequire external potential and is random in its direction. However, this increase in Li+self-diffusion is not observed in our case of LiTFSI dissolved in PC. Based on our observa-tions of increased Li+ mobility, we believe certain boundary conditions are necessary forLi+ hopping to occur and those are satisfied by applying an electric field.As an electric field is applied, the Li+ ion closest to the cathode moves toward it, inresponse to the potential difference, leaving a vacant space, or a hole, in the supra-molecularstructure, similar to the hole seen in solid Li ion conductivity. This newly formed vacancyallows for another Li+ ion to travel a small distance, perhaps even inside the same Debyesphere, to occupy it. This continues through the length of the solution, to the anode. Thecombination of transient movements of the Li+ leads to an effective conduction of the chargeover the long distance between the electrodes. Due to the high concentration of solution,Li+ can travel from one anion to the other, without ever leaving the electrostatic sphereand never having to overcome electrostatic forces. In the absence of the driving potential,no holes are injected and the transport remain Stokes-Einstein based. In this mechanismthe effects of higher viscosity, increased drag, and crowding are minimized as the ion doesnot need to diffuse large distances. This allows for faster current conduction that does notsolely rely on diffusivity. As this mechanism relies on the existence of supramolecular ionicaggregates, in future research, it should be possible to test it by studying the effects ofelevated temperatures, which would decrease the correlation length of the supramolecularstructure, breaking the structures down and inhibiting Li+ conduction.This divergence from diffusivity is emphasized when comparing the measured mobilityto measured self-diffusion. Although self-diffusion of the anion is higher than that of the935.3. Results and Discussioncation for all concentrations, the mobility data shows a crossover between 1.50 and 2.00 M.At low concentrations, the mobility of TFSI\u2013 is higher by an order of magnitude than theLi+ and at higher concentrations, the mobility of Li+ becomes higher than TFSI\u2013. Thesame effect can be seen in the transport numbers which are presented in Fig. 5.17. Thisshows that at low ionic concentrations, 90% of the current is carried by the TFSI\u2013 anionwhile at higher concentrations, the majority of the current (89%) is carried by the Li+making it practically a single ion conductor. This transport number shift from a currentwhich is TFSI\u2013 dominant to Li+ dominant is seen at around 2.00 M which explains theswitch in behaviour of the IEAP device at similar concentrations as well as the maximumin conductivity. Since this crossover is not seen in self-diffusion in the electrolyte, we canconclude that Li+\u2019s conduction does not follow the Nernst-Einstein relation and is notbased on self-diffusion.A very similar mechanism has been described independently by a research group inWuhan, China[4]. Using a computational approach, Gao et al. have developed a modeldescribing ion transport as an ion-vacancy coupled charge transfer reaction and can rea-sonably predict conductivities in lithium based electrolytes. This computational work andindependently developed model strongly support the lithium hopping mechanism that wasdescribed here based on our eNMR measurements.The effective charge, zeff, can be derived using the ratio of measured mobility and self-diffusion (Eq. 1.17) and is plotted in Fig. 5.18 for both ions. The effective charge of Li+ risessignificantly at concentrations higher than 1.00 M, up to the point of being close to 1. Thisis initially surprising, as we have come to expect a decrease in effective charge with higherconcentration of solute. Although this increase has been observed before with charge-delocalized ions like imidazolium-based cations [58], it is unexpected for small, sphericalcations such as Li+. The suggested depletion based conduction mechanism can explain945.3. Results and DiscussionFigure 5.17: Transport number derived from driven diffusion measurements (eNMR) vs.salt concentration in PC. solid line for the cation and dashed line for anion.this increase in zeff as it requires a crowded mesh of ions to occur. This mechanism is onlypossible if the TFSI\u2013 ions create a close enough cluster that allows for quick movementbetween one overlapping Debye length to the next and therefore it is expected to occurexclusively in high concentrations.To confirm the validity of our measurements, we compare eNMR derived conductivity todirectly measured conductivity. eNMR derived conductivity can be calculated by pluggingmobilities into Eq. 1.10. Using directly measured mobilities we arrive at the total ionicconductivity which includes the effects of zeff . Results are shown in Fig. 5.19. We can see aconsistently excellent match for all concentrations between directly measured conductivityto eNMR derived conductivity. This match proves that eNMR can successfully predict theconductivity of a sample to high accuracy, even for concentrated solutions. This means thateNMR indeed directly measures the motion of ionic species in response to an electric field.The improved match of eNMR measurements in comparison to self-diffusion indicates thatthere are clustering species that affect the conduction, especially for higher concentrations.955.3. Results and DiscussionFigure 5.18: Effective charge (zeff) derived from driven diffusion measurements (eNMR)vs. salt concentration in PC. Solid line for the cation and dashed line for anion.Because eNMR measures the displacement of ions under applied electric field, the measuredmotion goes through the same mechanism as in a typical conductivity measurement. Byperforming the eNMR experiment under the same chemical and physical conditions as thedesired device, one can resolve the individual ionic conductivities and transport numbersof any magnetically visible ion.5.3.4 Driven Diffusion in Polymer Electrolyte SamplesAttempts were made to measure mobility in IEAP samples. Successful mobility mea-surements in IEAPs would increase the range of possible devices to explore and allow usto investigate conduction mechanisms further, in different media. Sadly, measurementsof eNMR in LiTFSI swollen PVDF-HFP were not successful. Attempts done on gelledsamples left us with no measurable signal at the end of the pulse sequence due to thecombination of short relaxation and signal decay due to self-diffusion.As mentioned in Section 3.5.2, the eNMR pulse sequence does not use the stimulated965.3. Results and DiscussionFigure 5.19: Total conductivity of electrolyte samples derived from eNMR driven diffu-sion measurements vs. directly measured conductivity for LiTFSI vs. salt concentrationin PC. Graph shows excellent match between directly measured conductivity (dashed) toconductivity derived from eNMR measurements (solid).echo method, and therefore T2 relaxation occurs for the entire duration of the pulse se-quence. In addition, Brownian motion, which naturally occurs in the samples, furtherdecays the signal in accordance to the Stejskal-Tanner equation (Eq. 2.3). Self-diffusionwas in some cases faster than that seen in solution electrolyte samples, which in combi-nation with the shorter relaxation times seen in polymers, reduced the measurable signalgreatly. Attempts to minimize the effect of self-diffusion on the signal by shortening thediffusion time (max of \u2206 = 60 ms) and the magnetic field gradient strength (max of 3.4 Gcm)also reduces the measurable phase shift due to driven diffusion. In these attempts, somesignal was left at the end of the measurement, but no phase shift was detected, eliminatingour ability to calculate mobility coefficients.One more added difficulty was observed after the eNMR attempts were done. Thesample changed colour in addition to the formation of fine black particles near the elec-trode. These might be an indication of an electrochemical reaction. In our set-up, an975.4. Conclusionsinterface between the liquid electrolyte and the IEAPs exist and cannot be eliminated.This interface increases the total sample resistance and forces our constant current set-upto a higher voltage, well above the electrochemical window for all the components in thesystem (LiTFSI 4.4V [141] and PC -2 to +6V [142]). As the voltage exceeded the electro-chemical window, Li+ may have been reduced to Li metal, which appeared as black grainsnear the electrode, and PC was oxidized, changing the colour of the solvent to brown. Webelieve that the changes to the sample would reduce the quality of the collected data andadjustments are required in order to reduce the total electrical resistance of the sample.5.4 ConclusionsIn order to learn about the mechanism behind innovative piezoionic materials, self-diffusionmeasurements were collected for both ions in LiTFSI salts dissolved in PC, both in solutionand embedded in polymer. Comparing these results to directly measured conductivitiesresults in poor prediction of conductive properties from self-diffusion, especially at highconcentrations. The total conductivity estimated from self-diffusion has predicted consis-tently higher ionic conductivities, especially in high concentrations. This was expected aswe know that self-diffusion does not account for the reduced ability of correlated ions toconduct charge.In addition, measurements of the output voltage from the piezoionic material in sen-sor mode were in direct contradiction to transport numbers derived from self-diffusionmeasurements of solution samples. In IEAP samples on the other hand, an inversion oftransport numbers was measured, but at a lower concentration. We conclude from theseresults that self-diffusion is a poor predictor of conductivity properties and has a tendencyto overestimate the conduction ability of ions.Our simple, horizontal eNMR probe has successfully measured the ionic mobility for985.4. Conclusionsboth ions in some samples. Comparing conductivity estimated from in-situ eNMR mobil-ity measurements to directly measured conductivities has shown an excellent match. Inaddition, an inversion in transport numbers was measured at the same concentration aswas seen in the output voltage measurements of devices in sensor mode. This demonstratesthat eNMR can accurately predict conductive properties for NMR visible ions.This inversion in transport numbers is explained by combining self-diffusion with mo-bility measurements to extract zeff for each ion. This enabled us to see that lithium\u2019seffective charge increases with an increase in concentration, resulting in the Li+ carryingnearly 90% of the total current making it practically a single ion conductor. This resultcontradicts most diffusion based conductivity theory as Li+\u2019s conductivity is higher thanits diffusivity suggests. This increase implies that a different conduction mechanism takesplace.We suggest from this data that at high concentrations Li+ conduction is by a mechanismsimilar to vacancy based diffusion which is possible due to supramolecular structures formedby the anion and the solvent. In this suggested mechanism the lithium ions can hopfrom occupied to vacant sites in the supramolecular ionic structures, without the need toovercome electrostatic interactions and solution viscosity. This hopping mechanism allowsthe system to perform better as a charge carrier than predictions based on the Nernst-Einstein relation.99Chapter 6Chromophore Diffusion in OpticalFilter Devices6.1 IntroductionIn addition to applications as touch sensors, artificial muscles, and iontronics, ion dense gelsare the basis for some optical device applications. SWITCH Materials Inc. has developed acommercial optical filter for the automotive glazing industry utilizing electro-photochromicmolecules. These chromophore molecules are integrated into an optical film, which issubsequently encapsulated into car sunroofs. The optical film can modulate the lighttransmission of a car sunroof in response to both light and electric potential [143]. Bymodulating the solar gain through the sunroof, climate control within the vehicle is possiblewithout the need for active cooling or heating. This reduced load on the vehicle AC systemcan improve the vehicle\u2019s fuel economy. Unlike many other photochromic materials, thesefilms do not show thermochromic behaviour - expanding their applications to a large rangeof temperatures.The optical films are made by embedding chromophore molecules into a polymer gelelectrolyte, which is coated between two sheets of transparent conducting electrodes. Thepolymer gel electrolyte acts as a conductive scaffold allowing current to propagate to theentire film with little resistance, minimizing voltage drop across the film. The chromophore1006.1. Introductionmolecules selected for different applications are determined by the structure of the sub-stituents, represented as R in Scheme 6.1. The chromophore molecules inter-convert be-tween two isomers that have either high or low absorbance of visible light. Transitionbetween those states happens in response to either specific wavelengths of light or to ap-plied voltage [144]. The rate of conversion between these two states depends, in additionto other factors, on the diffusion rate of the chromophore molecules in the polymer gelelectrolytes. In particular, diffusion of the chromophore to the transparent conductingelectrode is a necessary step for the electrochromic dark-to-light state transition.light state dark state\u03bb=250-450 nm(structure dependent)\u03bb=~500-700 nm(structure dependent)+e-e-+Figure 6.1: Chromophore inter-conversion as a response to voltage or a photon of wave-length \u03bb for a common molecular structure [144]. The state on the right represents theclosed-ring isomer which acts as an optical filter. Figure adapted from SWITCH MaterialsInc. with permission.To understand the performance limitations of devices based on these chromophores1016.2. Methodswe collaborated with SWITCH Materials Inc. to measure the self-diffusion of a variety ofchromophores in different solvent compositions as a function of temperature. These mea-surements will guide the development of devices with increased usability in more extremeclimates.In order to measure self-diffusion in conditions as similar as possible to operating con-ditions, our PFG-STE measuring setup was upgraded. A cooling attachment was addedto allow for a larger range of temperatures and an external electric field was added to theprobe. This allowed us to observe the effect an electric field has on the diffusion of theseuncharged molecules, in an environment as similar as possible to that of a fully operationaldevice.6.2 MethodsThe system is composed of 10 wt% of chromophore molecules dissolved in a solution con-taining a combination of ester based solvents, a paramagnetic charge compensator, cata-lyst, and RTIL all embedded in a polymer network that was either crosslinked or left un-crosslinked. Fig. 6.1 shows the interconversion processes undergone by the chromophores,which differ in the structure of substituent R. Those discussed in this chapter are shownin Fig. 6.2. Sample preparation was done by the Research and Development team atSWITCH Materials Inc. by creating solutions, full formulation samples, or polymer filmsthat could be illuminated to either the closed or open isomer of the chromophore. Thedetailed composition of the chromophore samples cannot be divulged as this informationis protected under a non-disclosure agreement with SWITCH Materials Inc.High resolution solution sample NMR spectra for each chromophore dissolved in Rhodi-asolvR\u00a9IRIS\/butylene carbonate (BC) mixture were collected using 19F-NMR on 8.4 T and4.7 T home-built NMR spectrometers as well as a 9.4 T (400 MHz) Varian Unity Inova1026.2. Methods(a) Red chromophore S164 (b) Blue chromophore S158(c) Blue chromophore S109(d) Blue chromophore S304Figure 6.2: Structure of chromophore molecules investigated in this chapter [144]1036.2. Methodsspectrometer. Spectrometers are described in detail in Chapter 3. Experimental parame-ters, such as pulse widths and frequencies, were set in accordance width the nuclei beinginvestigated.In order to push the photochromic reaction to either open or closed states, samples wereilluminated with the appropriate wavelength of light. In order to activate the reaction tothe open, clear state the samples were illuminated for a minimum of 4 hours by a low-pressure sodium lamp (Philips 179753 SOX 55 W) emitting at wavelength \u03bd=589 nm. Topush the reaction to the closed, opaque state, a homemade UV LED light box containingthree 5 mm LEDs emitting UV light of \u03bd=380 nm was used for a minimum of 10 hours.Self-diffusion coefficients of the chromophores and solution components were measuredusing 19F PFG-STE NMR (full pulse sequence is presented in Fig. 2.2). Experiments useda home-built PFG probe [2] in a home-built NMR spectrometer [89] operating at 8.4 T.Temperature control was achieved using a Bruker B-VT-1000 temperature control unit.More information regarding the spectrometer and the setup can be found in Chapter 3.Diffusion time, \u2206, was varied between 350 and 700 ms, depending on the diffusion rateof the sample. A short gradient pulse of \u03b4=318\u00b5s was applied in varying strength fromapproximately g = 200 Gcm to 3000Gcm . This short pulse was necessary due to the short T2times typical for polymer gels [63]. A longitudinal acquisition delay of T=10 ms was usedin order to allow eddy currents from the magnetic field gradient pulses to decay beforeacquisition.Measurements to explore the effects electric field has on self-diffusion coefficients weremade using an apparatus shown in Fig. 6.3. The voltage applicator and the sample holdercan be seen in Fig. 6.3a. The gelled sample was pressed between two sheets of conductiveIndium Tin Oxide (ITO) covered with polyethylene terephthalate (PET) that both held thesample in place and were used to apply an electric potential across the film. This sample1046.2. Methods(a) (b)Figure 6.3: (a) Sample holder, showing electrodes and film, and (b) wiring diagram of gelledsample between ITO electrodes. Electrodes are shown in white, whilst the electromagneticcoil is in grey. Notice the electrode contacts are outside the RF coils.holder was inserted into the RF coil of the PFG-NMR probe and the same experimentalparameters were used as in the previous PFG-NMR experiments, other than the delay timewhich was extended to 10 s to minimize any accumulative resistive heating and avoid a stateof equilibrium from charge accumulation near the electrodes (as described in Section 3.5and Fig. 3.4). These conductive plates were connected to a 5 V source through a voltagedivider that can be seen in Fig. 6.3b. While one source was open, the other was closedallowing a directional switch of the applied voltage. To further minimize the probable effectof resistive heating, an approach of using an alternating pattern between applied voltageand non-applied voltage scans was employed. The resistor ratio determines the voltage onone plate while the other was connected to ground, and thus the voltage across the samplewas equal to Vout of the voltage divider according to Vplate = Vout \u00b7 R1R2+R1 . This voltageassumes the resistance of the device is significantly higher than that of the resistors. Thisassumption was confirmed by measuring the voltage across the sample. For a sample ofthickness of 200\u00b5m, the maximum electric field is 250 [ Vcm ]. The 5 V source was drivenby the NMR pulse programmer and therefore timed with sub-microsecond accuracy withrespect to the total pulse sequence.1056.3. Results and Discussion6.3 Results and Discussion(a) (b)Figure 6.4: (a) The general molecular structure of a blue chromophore in its open form(upper) and closed (lower). Carbons 3, 4, and 5 of the cyclopentene ring are labelled.Out of plane Fs are labelled with b and in plane Fs are labelled with a [144]. The rotationaround the single bond between the cyclopentene and the thiophene ring is marked with anarrow. (b) 1D 19F NMR spectra, N= 256 scans, of S158 for both open (light) state (upper)and partially closed (dark) state (bottom). Spectrum collected on 400 MHz spectrometerusing high resolution solution probe.High resolution 1D 19F solution spectra of chromophore S158 dissolved in RhodiasolvR\u00a9IRIS\/BC mixture can be seen in Fig. 6.4 alongside the general molecular structure of thechromophore molecule. The spectra show the chromophore in two states: 1) completelyopen when the bis(terthiophene)ring is in its open form and 2) partially closed whereboth the open and closed states of the bis(terthiophene) ring are present. In the open state1066.3. Results and Discussionspectrum (upper) two peaks are recognized. The upfield peak around -132 ppm is generallysharper and represents the two magnetically equivalent Fs on carbon 4 of the cyclopentenering, 4Fa and 4Fb. The downfield peak, around -112 ppm, represents the four magneticallyequivalent Fs on positions 3 and 5 of the cyclopentene ring. This line appears broad as aresult of the hindered rotation of the bond between the cyclopentene ring and the thiophenering (labelled with a rotation arrow on Fig. 6.4a). In the open state, this single C-C bondallows for rotation, but the geometry of the R arms hinders this rotation resulting in abroad line. This broadening is a result of the molecular motion being in the intermediatetime scale relative to the frequency difference between the chemical shifts in the variousconformations. For a detailed explanation regarding this broadening mechanism, see Part7 of Spin Dynamics by Malcolm H. Levitt [60].In the lower spectrum where the closed and open forms coexist, smaller sharp peaks canbe seen near the open state peaks. These peaks indicate the structural difference betweenthe closed and open forms. As the structure changes to form a closed bis(terthiophene) ring,the conjugated pi system becomes longer and all the peaks shift upfield [145]. In addition,the geminal fluorines on carbon 3 are no longer equivalent (3Fa 6= 3Fb) and are thereforecoupled to each other. The same happens to fluorines 5Fa and 5Fb as they are as well notequivalent in the closed state. This creates an AB pattern quartet of JAB = 280 Hz and\u2206\u03bdAB = 390 Hz that are seen as 4 sharp peaks that appear, in addition to the open statebroad peak, around -113.6 ppm. The rotation of the bond between the cyclopentene andthe thiophene ring is eliminated when the ring is closed and the lack of free rotation in thisbond forces the molecular symmetry to a C2 rotation about C4 of the cyclopentene ring.This means that in the closed state, the R arms are confined in an anti-parallel positionwhere R1s are facing above and below the plane of the cyclopentene ring. This symmetrymakes the opposing Fs on carbons 3 and 5 equivalent (3Fa = 5Fb and 5Fa = 3Fb).1076.3. Results and DiscussionThe small sharp peak that appears close to the open state peak, near -132 ppm is aresult of the closed-state geminal fluorines in position 4 which remain equivalent. The peakis shifted upfield, again, due to the change in the conjugated pi system. These fluorinesare coupled to the fluorines in the vicinal positions, but the F-C-C-F coupling is too small(\u223c2 Hz) to be resolved [146].This solution spectrum allows us to resolve the closed state peaks from the open state.Yet, a much lower resolution was achieved for the viscous uncrosslinked full formulationsamples and even more so for the crosslinked samples (linewidths in the order of \u223c500-1000 Hz or larger). From this, it was determined impossible to distinguish the dark and lightstate in a crosslinked sample. The linewidths were especially large for the red chromophore,S164, raising a question regarding the structure and dynamics of the molecules and led tofurther investigation described below.In the measurements of the red chromophore, S164, generally broader lines were ob-served with interesting shoulders around both of the major peaks (see Fig. 6.5). It wasinitially suspected that these shoulders might be a sign of molecular degradation due tooxygen exposure or cycling fatigue between the open and closed states. In order to examinethese shoulder peaks, samples with different amounts of air exposure and different num-ber of open\/closed cycles were examined. All samples showed the same spectral features.In an attempt to see whether solvent interactions were causing the shoulders, the chro-mophore was tested in a variety of solvents: butylene carbonate (BC), acetone, deuteratedchloroform and methanol. The shoulders appeared in all solvents except acetone.Higher resolution, variable temperature measurements of the liquid samples were takenusing the 400 MHz Varian Unity Inova spectrometer and can be seen in Fig. 6.7. Thesespectra are of the fully open state and surprisingly show an AB pattern similar to thatidentified in the closed state spectrum in previous measurements. This time, the sharp1086.3. Results and DiscussionFigure 6.5: 19F NMR measurements of S164 chromophore in the open state dissolvedin RhodiasolvR\u00a9 IRIS\/BC. The shoulders are visible around -102, -131, and -137 ppm.Spectrum collected on 360 MHz spectrometer using horizontal diffusion PFG-NMR probe.peaks appear around the central broad peak and are not shifted upfield. Recall that theupfield shift was a result of the change in length of the conjugated system in the closedring. In this case, there is no change in the conjugated system as the molecule is still inits open state and the length of the conjugated system did not change.As was mentioned before, the AB pattern in the closed state is a result of the inequiv-alence of fluorines 5Fa and 5Fb, as well as 3Fa and 3Fb, due to the rigid bond betweenthe cyclopentene and the thiophene ring. When the arms are free to move, they can spendtime in either of the two arm positions: parallel or anti-parallel (see Fig. 6.6). When thetransition between these arm positions occurs on an intermediate time scale, the broadlines that are a characteristic of the open state spectrum, appear as well (see Fig. 6.4).However, the spectra in Fig. 6.5 and 6.7 suggest that transitions between the parallel andanti-parallel states are hindered. When the arms face opposite directions (anti-parallel)they mimic the structure of a closed state molecule, resulting in the AB pattern split seenin the closed spectra. When facing the same direction (are parallel) the molecule has a Csmirror symmetry about carbon 4, where 4Fa and 4Fb are inequivalent and split by eachother. This symmetry makes fluorines 3Fa and 5Fb equivalent as well as 3Fb and 5Fa.1096.3. Results and DiscussionBoth are split by geminal J couplings as couplings from vicinal fluorines are too small toresolve.(a) (b)Figure 6.6: General structure of the open state chromophore in (a) parallel and (b) anti-parallel arm positions. The rotating bond is labelled with an arrow. The different armpositions change the symmetry of the molecule. The experiments suggest that a stericbarrier impedes transitions between these parallel and anti-parallel conformations, butpresumably significant rotational freedom within each conformation state exists.As the molecule rotates between the two arm positions, the broad shoulders of couplingsare formed and are clearly seen in Fig. 6.5. In the same spectrum, hints of the AB patternare seen on top of the downfield peak. When looking at the higher resolution spectrumat room temperature shown in Fig. 6.7, we can easily recognize the AB pattern peaks at -110 ppm. Due to the slow motion of the arms in a viscous solution, the two arm positions areseparated and are visible in the spectrum. As the solution is heated, transitions betweenthe parallel and anti-parallel states become more rapid, and the spectra coalesce into asingle time-averaged spectrum. Further heating sharpens the peaks even more. The factthat the shoulders are a result of limited molecular dynamics agrees with the fact thatthese shoulders are not present in samples dissolved in acetone, a low viscosity solvent(a factor of 10 lower than BC (0.3034 mPs vs. 2.98mPs [147, 148]), that allows rapid armmotion, similar to the high temperature spectra for more viscous solvents. These shoulderswere only observed for S164 and are assumed to be a result of the large steric hindrance to1106.3. Results and Discussionthe rotation the R arms cause. Further investigation of the matter was not deemed crucialfor a successful investigation of diffusion properties.Figure 6.7: S164 dissolved in RhodiasolvR\u00a9 IRIS\/BC mixture 19F NMR measurements vs.temperature. At room temperature an AB pattern can be seen around -110 ppm whichcoalesce to the main peak above 30 \u25e6C. The shoulder peaks disappear above 40 \u25e6C andthe peaks sharpen as temperature is increased further. Spectra collected on 400 MHzspectrometer using high resolution solution probe.One of the important first steps in our diffusion investigation was to assess the diffusionof both dark and light optical states. The generally poor resolution of samples in thegelled state made it impossible to separate the two isomers in crosslinked samples for allchromophores. In order to overcome this, the films of crosslinked samples were illuminatedin the glove box of SWITCH Materials Inc. with the appropriate light wavelength for anextended time (minimum of 24 hours) in order to ensure a maximal conversion of thechromophores to the desired state before the PFG-NMR measurements. A summary ofboth isomer diffusivities, for several chromophores, is presented in Table 6.1. The self-diffusion measurements show that self-diffusion of the two isomers are not dramaticallydifferent, but on closer inspection, the diffusion of the light state is generally faster than1116.3. Results and Discussionfor the dark state. As discussed, the steric packing of the molecule changes between thetwo states, and their effective radii are expected to be somewhat different and affect theirself-diffusions.Chromophore D-Open \u00d710\u20137 [cm2s ] D-Closed \u00d710\u20137 [cm2s ]S109 1.9 \u00b1 0.3 3.0 \u00b1 0.3S158 2.1 \u00b1 0.2 2.6 \u00b1 0.3S304 2.0 \u00b1 0.3 1.8 \u00b1 0.4S164 2.2 \u00b1 0.8 2.6 \u00b1 0.3Table 6.1: Self-diffusion coefficients at room temperature of blue chromophores S109, S158,and S304 and red chromophore S164 in full formulation crosslinked samples.When comparing the relative self-diffusion coefficients of blue chromophores, S109,S158, and S304 in the closed state to each other, we can see that S109, although it appearsto be larger, has the fastest diffusion coefficient, followed by S158 and S304. The effect ofthe diffusion coefficient of the blue chromophore on the transition time for optical filterswas analyzed and summarized in Table 6.2. The transition times, from dark to light state,of optical filters made with the different blue chromophore were measured at SWITCH Ma-terials Inc. The comparison between S158 and S304 was done in highly-crosslinked polymergel formulations. The comparison between S158 and S109 was done in uncrosslinked poly-mer gel formulations. In the uncrosslinked case, the optical filter made with S109 has amuch faster transition time than S158, consistent with the relative diffusion coefficientsmeasured by NMR. In the crosslinked formulation however, the optical filters with S158and S304 have almost identical transition times. This suggests that while diffusion of thechromophore may play a role in the transition speed, there are other factors involved.In order to measure diffusion coefficients under similar conditions to the expected op-erating conditions of devices, diffusion coefficients were measured at varying temperaturesbetween 235 K and 300 K. Results can be seen in Fig. 6.8. The general trend shows an1126.3. Results and DiscussionChromophore Formulation Transition time (s)S158 Crosslinked 69\u00b17S304 Crosslinked 65\u00b16S158 Uncrosslinked 32\u00b16S109 Uncrosslinked 21\u00b12Table 6.2: Transition time comparison of blue chromophores in crosslinked and samplesin full formulation that were left uncrosslinked. The transition time was between fullydarkened state to 87.5% faded.Figure 6.8: Self-diffusion measurements vs. temperature. Some of the chromophores haveshown an expected decrease in D with a decrease in temperature, yet others have shownan unexpected increase in temperature around 250 K.Arrhenius-like dependence on temperature, which indicates a Fickian diffusion behaviour,typical of polymers above their glass transition temperature (Tg). Two of the chromophores(S109 and S158) show a local maximum in diffusion as a function of temperature. This isan unexpected behaviour according to the Einstein-Smoluchowski-Sutherland relation (Eq.1136.3. Results and DiscussionFigure 6.9: Uncrosslinked full formulation sample S164 after being left at 250 K overnight.Some crystals can be seen in the sample.1.4). The increase in diffusion at 250 K might be explained by certain components of thesolution reaching their freezing point and precipitating. The composition of the remainingsolution is modified and has a lower viscosity allowing the chromophore to diffuse througha less viscous matrix. At the same temperature, a stronger decrease in the RTIL anionTFSI\u2013 self-diffusion is seen, perhaps an indication it is a component in the material precip-itating out of solution. Visual inspection of a full formulation liquid sample left overnightat 250 K revealed crystallization in the samples, but their composition was not identified.Fig. 6.9 shows the S164 full formulation solution sample with the unidentified crystals.More research is required in order to answer these questions.Another external condition required to fully mimic the operational conditions of thechromophore film is an applied electric field. The amount of the charged species in thesample does not exceed 1.5% w\/w. The chromophore itself remains uncharged most of thetime. It is charged briefly as it interconverts between the two isomers (see Scheme 6.1).Therefore, it was expected that the chromophore does not undergo electrophoretic diffusionand the motion should remain incoherent. It was predicted that applying an electric field tothe electroactive gel would affect the charge distribution of the polymer gel itself. This willcause a change in the interactions between the ions in solution and the polymer network,resulting in changes to self-diffusion coefficients, but will not create electrophoretic motionfor the chromophore molecules.1146.3. Results and Discussion(a) (b)Figure 6.10: Example of a diffusion experiment for chromophore S109 in full crosslinked for-mulation with a diffusion time \u2206 = 425 ms, performed at 22 \u25e6C. (a) shows the chromophorefluorine peak integration vs. gradient strength. The data is fitted to show a Gaussian decayof the signal with an increase in gradient strength. (b) shows the same data, plotted as thenatural logarithm of the signal intensity vs the gradient strength squared to linearize theGaussian behaviour of the Stejskal-Tanner diffusion equation. Both figures show excellentfits to a single diffusion coefficient of D = (1.9\u00b1 0.3)\u00d7 10\u20137 [cm2s ].Because the electric field does not directly affect the diffusion of the chromophore, itis expected that self-diffusion would remain unaltered, i.e. mostly thermally dependentand random (Brownian motion). This incoherent motion will create a decay in the signalwith a magnetic field gradient increase but not a phase modulation. More informationregarding the difference between coherent and incoherent diffusion and their effects on theNMR experiment can be found in Chapter 2.Figures 6.10 and 6.11 show an example of the diffusion measurement done on chro-mophore S109 in the open state with and without an external applied voltage. The voltageincreased the noise by about 20% as the electrodes were placed directly inside the RFcoil. The results are summarized in Table 6.3. Measurements of self-diffusion under anapplied voltage show hints of faster diffusion. The increase observed is about 5%, and isnot statistically significant. However, this small increase in self-diffusion is present in all1156.3. Results and Discussion(a) (b)Figure 6.11: Example of a diffusion experiment under applied voltage for chromophore S109in full crosslinked formulation with a diffusion time \u2206 = 425 ms, performed at 22 \u25e6C. (a)shows the chromophore fluorine peak integration vs. gradient strength. The data is fittedto show a Gaussian decay of the signal with an increase in gradient strength. (b) shows thesame data, plotted as the natural logarithm of the signal intensity vs the gradient strengthsquared to linearize the Gaussian behaviour of the Stejskal-Tanner diffusion equation. Bothfigures show excellent fits to a single diffusion coefficient of D = (2.0\u00b1 0.5)\u00d7 10\u20137 [cm2s ].three chromophores tested. This suggests that the chromophore molecules indeed diffusefaster through the polymer when voltage is applied.This increase in self-diffusion might be a result of small changes to the sample\u2019s tem-perature caused by resistive heating. The applied voltage to the sample produces heataccording to Joule\u2019s heating process (see Section 2.2.2) which might increase self-diffusion.Although the sample is embedded in a polymer gel (which is expected to minimize anyeffects of convection) self-diffusion coefficients depend on temperature both directly, and in-directly through viscosity (see Eq. 1.5). From extrapolations of our temperature dependentmeasurements (shown in Fig. 6.8), it is possible that such a small change in self-diffusionis a result of a temperature increase in the order of half a degree. Since precautions tominimize resistive heating were taken (by using a 10 s delay between scans in addition toan alternating pattern between applied voltage and non-applied voltage scans) we believe1166.3. Results and Discussionthe increase observed is not a result of resistive heating.Another possibility is that this observed increase in self-diffusion results from the redis-tribution of charge for both chromophore and solvent molecules. The delocalized charge ofthe pi system is easily polarized by an external electric field, resulting in an enhancement ofthe dipole moment of the molecule [149, 150]. This effect is especially prominent on largemolecules due to their higher polarizability and even more so on molecules that containhalogens due to their relatively high electronegativity [151].The change to the molecular dipole moment affects both the interaction of the chro-mophore with the polar solvent molecules and with the polymer itself. These new in-teractions can either increase or hinder diffusion, depending on many factors [152, 153].In this case, the slight increase can be explained by the relatively large size ratio of thechromophore to the solvent. As seen in previous calculations, large molecules surroundedby smaller solvents have a tendency to increase their self-diffusion with an increase indipole moment. The increased dipole moment causes a stronger interaction between thesolute and solvent, effectively increasing its solvation layer and creating a larger solute-solvent complex. In the case of large solute and small solvent, the shape of this complex ismore spherical in comparison to a less solvated solute. This spherical complex, althoughlarger, diffuses up to 24% faster in comparison to a complex of the same size with a pla-nar shape [154]. This mechanism may explain the slight increase in self-diffusion seen forchromophore molecules under an applied electric field. As the dipole moment of the chro-mophore is enhanced, the polar butylene carbonate (BC) solvent molecules increase theirsolvation, creating a large, more spherical complex with an increased self-diffusion throughthe polymer medium.1176.4. ConclusionsChromophore D (Control) \u00d710\u20137 [cm2s ] D (applied V) \u00d710\u20137 [cm2s ]S109 1.9 \u00b1 0.3 2.0 \u00b1 0.5S158 2.1 \u00b1 0.2 2.3 \u00b1 0.3S304 2.0 \u00b1 0.2 2.1 \u00b1 0.4Table 6.3: Self-diffusion coefficients at room temperature of blue chromophores S109, S158,and S304 with and without applied voltage. Samples were prepared in full formulation andcrosslinked as for an operating device.6.4 ConclusionsThe self-diffusion coefficients of chromophores under operating conditions of devices weremeasured using PFG-NMR. A total of four chromophores were examined in a vast variationof formulations and physical states (\u223c 30 samples), in order to optimize both the feasibilityof the measurements, and to provide experimental data to assist in formulation decisionsof future devices.The dynamics of the general chromophore were investigated as NMR spectra of somesamples showed unexpected shoulders. The collection of spectra in different solvents as wellas temperature dependent spectra leads us to attribute these features to hindered dynamicsin the molecules that depend on the different substituent groups and the differing molecularsymmetries arising from the different substituent configurations.The relation between diffusion and temperature has been investigated and showed ageneral decrease in self-diffusion with temperature decrease. It was observed that at 250 Kthe diffusion of the chromophore shows a non-Fickian behaviour and increases as the tem-perature decreases. It is likely that this is a result of some components precipitating outof solution, reducing the viscosity of the solution, causing the observed increase in D.Measurements of self-diffusion under applied voltage showed consistent hints of fasterdiffusion, though the change observed was not statistically significant. A large change is1186.4. Conclusionsnot expected because the chromophore itself is neutral most of the time. The small changein diffusivity may be due to changes in the sample\u2019s temperature due to resistive heatingor changes to the dipole moment of the chromophore itself in response to the electric field.The change in the molecular dipole moment could cause a change in the molecular shapeof the chromophore-solvent complex, resulting in a more spherical shape. This complexshape change would translate to a slight increase in self-diffusion with the application ofan electric field.The information provided by these measurements will be used in the determination offormulations that can be used for a wide range of operating temperatures. This increasedtemperature range will help and enhance the performance and usability of films used forautomotive glazing and help increase fuel economy by reducing the need for active coolingor heating in automobiles.119Chapter 7Vitrimers - Transport ofThermally Exchangeable MoleculesTransport measurements are of interest in many other classes of materials in addition toion dense materials. In this chapter we focus on a very different system, where similartechniques again help illuminate mechanisms that are the source of interesting materialproperties. We investigate the dynamics and transport of crosslinker molecules in largescale polymers, as they can be the key to more efficient polymer recycling.7.1 IntroductionThe introduction of plastics to the world enabled the development of an enormous array ofproducts with positive effects on hygiene, health, and the economy. Yet, the widespread useof plastics causes an ongoing problem of excessive plastic waste. Due to their availability,relatively low price, and simple manufacturing, cross-linked polymers are being disposedof in massive quantities creating a plastic waste crisis. Both bulk and microplastic wastein the oceans have potential health and environmental impacts. These have been wellresearched in recent years, and they are of growing public concern (e.g. the ever growingGreat Pacific Garbage Patch [155]). Better recycling of plastics to secondary materialsis one possible solution to the plastic pollution problem as currently only \u223c10% of plas-1207.1. Introductiontics manufactured between 1950 - 2015 have been recycled (\u223c600 Mt) while approximately4900 Mt were discarded to landfill [156].Commercially manufactured plastics are generally divided into two broad categories:thermoset and thermoplastic. Thermoset plastics (such as polyurethane) are irreversiblycrosslinked polymers and offer resilient, heat resistant end products that are non-remoldableor recyclable. Thermoplastics (such as nylon) are made of linear polymer chains and areless heat resistant and therefore can be more easily recycled as they can be melted downat high temperatures. Yet, they are less resistant to swelling or dissolution in chemicalsolvents and are not suitable for high temperature or high impact applications.Vitrimers are a recently developed class of thermoset polymers [157, 158] in which molec-ular cross-links are thermally activated and are exchangeable. This means that these cova-lently bonded crosslinkers can shift from one bonding site to another upon heating, allowingfor rearrangement of the whole network (see Scheme. 7.1). It is believed that as the systemis heated above what is termed its topological freezing point temperature, Tv, the covalentbonds exchange at a significant rate, and the crosslinked system is moldable and thereforecan be truly recycled to a new crosslinked polymer. Below Tv, the exchange reaction is slowand the vitrimer becomes rigid like a thermoset plastic [159]. The thermally labile natureof vitrimers allows them to be recast and recycled simply by heat and mechanical manip-ulation. These systems can have the strength and durability of crosslinked materials whilestill being recyclable over many cycles via cutting and compression molding above both Tvand the glass transition temperature, Tg. By changing the polymer backbone, additivesor processing conditions, vitrimers can satisfy specific needs of mechanical properties andbe made into a variety of applications [160].These crosslinked polymers can flow like linear polymers when heated above their Tvand Tg. It has been suggested that when heated to this malleable state, the number of cova-1217.2. MethodsScheme 7.1: General mechanism of thermally exchangeable covalent bonds in vitrimers.Red spheres represent moieties that can undergo reversible bond formation. The bondrearrangement leads to topological changes of the network itself.lent bonds in vitrimers remains constant but the rate of bond exchange increases [160, 161].This exchange can lead to a flow motion as the viscosity of the material decreases\u2013 similarto a viscous liquid that can flow with increased temperature. The rapid stress relaxationthese materials exhibit, has also been attributed to this dynamic bonding property [157].The Wolf group at UBC has suggested a new use for this rapid exchange reaction.By covalently attaching a tag molecule to vinylogous urethane linkages of a crosslinkedpolymer network (see Scheme 7.2), we can utilize the rapid bond exchange to allow the tagmolecule to diffuse throughout the matrix. This new mechanism for molecular movementmay provide greater control over the diffusion of additives through the polymer. In thischapter, the mechanism of this rapid bond exchange in polymers is investigated usingdiffusion NMR. By covalently attaching NMR visible molecules to the vinylogous urethanelinkages, we are able to directly track their motion and learn about some of the contributingfactors to this interesting molecular bond exchange.7.2 MethodsIn order to learn about the motion of the crosslinker, a set of vitrimer samples were madewith different combinations of covalently bonded crosslinkers and tag molecules. Samples1227.2. MethodsScheme 7.2: Vinylogous urethane exchange reaction of a pendant molecule covalently at-tached to a crosslinked polymer, resulting in a net movement of the polymer. Vinylogousurethane exchange is proposed to proceed through a tetrahedral intermediate which meansthe molecule is always attached to the polymer [160].were prepared by Taylor D. Wright in Dr. Michael Wolf\u2019s lab at UBC and the sample prepa-ration is described in detail in Appendix A. All polymers had the same backbone of poly-dimethylsiloxane (PDMS) random copolymer containing pendant propylamine groups tocreate PDMS-NH2 which can be seen in Fig. 7.1. The backbone was crosslinked with eitheran exchangeable vinylogous urethane crosslinker (VitCL - Fig. 7.2) or a non-exchangeablecrosslinker, triethylene glycol dimethacrylate (Fig. 7.3) which created a dynamic or staticnetwork [162]. To attach the tag in an exchangeable or a static way, the propylamine groupswere reacted with either 2,2,2-trifluoroethylacetoacetate or 2,2,2-trifluoroethyl methacry-late, respectively. The exchange reaction is based on a Michael addition reaction [163], inwhich the formed C-N bond tautomerizes between a single to a double C-N bond. Whilein the double bond state, the reaction can undergo the exchange. The exchangeable tagis presented in Fig. 7.4. With the non-exchangeable tag, shown in Fig. 7.5, the Michaelreaction forms a single C-N bond which is thermodynamicaly less reactive and does notallow exchange. Heating to approximately 60 \u25e6C was expected to bring the samples aboveTv and therefore allow for diffusion based on vinylogous urethane exchange, as can be seenin Scheme 7.2.1237.2. MethodsFigure 7.1: PDMS-NH2 polymer structureFigure 7.2: Exchangeable crosslinker 1,12-dodecane-bis-\u03b2-ketoenamine n-butylFigure 7.3: Non-exchangeable crosslinker triethylene glycol dimethacrylate.Figure 7.4: Exchangeable tag 2,2,2-Trifluoroethylacetoacetate. The 19F nuclei will be usedas the NMR visible tag for this molecule.Figure 7.5: Nonexchangeable tag 2,2,2-trifluoroethyl methacrylate. The 19F nuclei will beused as the NMR visible tag for this molecule.1247.2. MethodsIn addition to the exchangeable tag mentioned, an exchangeable, F labelled, blue light-absorbing NMR visible tag was prepared in order to allow for visual tracking of the tagmolecule as well as NMR. The synthetic steps are described in Appendix A and the resultingmolecule can be seen in Fig. 7.6.Figure 7.6: Orange coloured, 19F NMR visible exchangeable tag 4-trifluoromethyl-4\u2019-methylacetoacetate azobenzeneTo prepare the samples for the PFG-NMR experiment, PDMS-NH2 (1 g, 0.02 mmol),38 mg of exchangeable vinylogous urethane crosslinker or 23 mg of triethylene glycol dimeth-acrylate (0.08 mmol, 4 eq) or a non-exchange triethylene glycol dimethacrylate, p-toluenesu-lfonic acid (0.67 mg, 0.5 mol relative to moles of NH2, as a dilute solution in THF), 5 molarequivalents of an NMR visible tag, and 2.5 mL of THF were combined in a 20 mL vial andstirred using a magnetic stir bar for 5 minutes. The vial was degassed using a sonicatorfor 10 minutes. The solution was transferred to a silicone mold inside a glass desiccator,which was then evacuated. The system was placed under N2, covered in aluminum foil, andheated to 60 \u25e6C using an oil bath until the THF fully evaporated. The system was thenheated to 100 \u25e6C and kept at that temperature for 18 hours under N2. Polymer sampleswere removed from the warm molds and stored in a refrigerator when not in use.To prepare the samples to fit in the NMR tube, the polymer samples were cut intoapproximately 1.5 cm by 0.25 cm blocks using a razor blade and placed at the bottom ofa glass NMR tube. The sample was then attached to a Schlenk-line and evacuated under1257.2. Methodsvacuum and then backfilled using dry N2. This procedure was repeated twice. The tubewas then sealed under either vacuum or dynamic N2 using a butane torch.To create a heterogeneously loaded polymer sample, PDMS-NH2 polymer was cut into3 mm\u00d72 cm pieces. The lower 20% of the polymer was dipped in a solution of the orangecoloured product (see Fig. 7.6), creating a heterogeneously colour loaded polymer. Afterbeing soaked overnight, the polymer was rinsed and dried before being sealed under N2 inan NMR tube. The process and resulting polymer sample can be seen in Fig. 7.7.Figure 7.7: Sample heterogeneously loaded with 4-trifluoromethyl-4\u2019-methylacetoacetateazobenzene. Notice, the concentration of the orange colour varies through the length ofthe sample.In order to measure self-diffusion coefficients of the NMR visible tags, 19F or 1H PFG-STE NMR experiments were performed at varying temperatures using a home-built PFGprobe [2] in a home-built NMR spectrometer [89] operating at 8.4 T, which are describedin Chapter 3. Temperature control was achieved using a Bruker B-VT-1000 temperaturecontrol unit. A PFG-STE NMR pulse sequence was used, as described in Section 2.1 andpresented in Fig. 2.2. A gradient pulse of \u03b4=318\u00b5s was applied in varying strength fromabout g = 500 Gcm to 3640Gcm . The diffusion time \u2206 was varied between 700 and 2000 ms,according to the sample\u2019s diffusion rate and relaxation times. An acquisition delay ofT=10 ms was used in order to allow eddy currents to decay before acquisition.Self-diffusion measurements were fitted to a function similar to the Stejskal-Tanner1267.3. Results and Discussionequation (see Eq. 2.4) with an added weighting factor:S(g)S(0)= (1 \u2013 f) exp(\u20134pi2\u03b32g2\u03b42(\u2206 \u2013\u03b44)D)+ f, (7.1)where the weighting factor, (1-f ), accounts for some fraction, f, of the sample which doesnot participate in diffusion.Relaxation measurements were carried out by either using an inversion recovery pulsesequence to measure spin-lattice relaxation (T1) or a spin echo in order to measure spin-spinrelaxation (T2).7.3 Results and DiscussionTemperature [ \u25e6C] time under heat [hours] D\u00d710\u201312 [m2s ] f FigureRoom temperature 0 < 0.01 N\/A 7.8a40 4 3.1 0.03 7.8b50 6 4.5 0.12 7.8c60 8 6.5 0.50 7.8d70 10 < 0.01 N\/A 7.8e80 12 < 0.01 N\/A90 14 < 0.01 N\/A100 16 < 0.01 N\/A110 17 < 0.01 N\/A120 18 < 0.01 N\/A 7.8fTable 7.1: Self-diffusion coefficients of vitrimer sample PD1a vs. temperature. The mea-surements were taken consecutively, the first measurement being the one at room temper-ature.Initial PFG-NMR measurements of self-diffusion vs. temperature for a sample con-taining both an exchangeable crosslinker and an exchangeable tag, PD1a, can be seen inFig. 7.8 and the numerical values extracted from this data are in Table 7.1, which sum-1277.3. Results and Discussion(a) T= 22 \u25e6C (b) T= 40 \u25e6C(c) T= 50 \u25e6C (d) T= 60 \u25e6C(e) T= 70 \u25e6C (f) T= 120 \u25e6CFigure 7.8: PFG-STE NMR signal intensity vs. magnetic field gradient strength of samplePD1a as it is heated from room temperature to 120 \u25e6C. Solid line shows the Gaussian fitand dashed line shows the weighting factor f. Numerical values are provided in Table 7.1.1287.3. Results and Discussionmarizes the results. Starting from room temperature in Fig. 7.8(a) we see no measurablediffusion. When the sample was heated at 40 \u25e6C for 3 hours the measurements show thetypical Gaussian decay that can be seen in Fig. 7.8(b), which is fitted with a self-diffusioncoefficient of 3.1\u00d710\u201312 m2s . An added dashed line represents the weighting factor men-tioned in Eq. 7.1. This factor, f, shows the fraction of the sample which did not participatein self-diffusion and can be seen in the figures as the amount of signal which does notpresent the Gaussian decay expected from a PFG-NMR experiment of a diffusing species.Fig. 7.8(c) shows the signal after two more hours of heating to 50 \u25e6C. Here the extent ofthe weighting factor increases, and a larger portion of the sample does not show diffu-sive behaviour. The portion that does diffuse, shows a 50% higher diffusion coefficient.Fig. 7.8(d) shows the signal decay vs. gradient strength for the sample after being heatedfor another two hours at 60 \u25e6C. The diffusion coefficient measured increases by another45%, but the weighting factor nearly tripled to approximately 50% of the sample. As thesample was heated further to 70 \u25e6C, after spending 10 accumulative hours under heating,the self-diffusion coefficients measured for PD1a decayed until it was no longer possibleto measure in the existing set-up. Further heating up to 120 \u25e6C did not regenerate thediffusion and it remained undetectable. These measurements raised two related questions:why does only a fraction of the sample contribute to the diffusion, and why does the frac-tion change with time? In order to answer these questions, further investigations wereundertaken, discussed below.In order to confirm that oxidation was not the cause of this loss of diffusivity, theexperiments were repeated with samples open to air, sealed under vacuum, and sealedunder inert N2 gas. The same decrease in the diffusing fraction (increase in f ) and decreasein self-diffusion were observed under all conditions.To assess whether the loss of diffusivity factor was a result of time spent at temper-1297.3. Results and Discussion(a) (b)Figure 7.9: STEPFG-NMR experimental results for D of Sample PD1b, performed at 60 \u25e6C.(a) shows the diffusion measurements as the sample reached temperature. The solid lineshows the Gaussian fit and the dashed line shows the weighting factor f. D was measuredat 3.2\u00d710\u201312 m2s with a weighting factor f = 0.23. (b) shows the same experiment and thesame sample after being kept at the same temperature for 10 hours.ature or a result of increased temperature, the experiment was repeated at the highesttemperature diffusion was measured. A new piece of the same sample, PD1b, was heatedto 60 \u25e6C. The diffusion was measured a few minutes after reaching temperature and tenhours after being kept at the same temperature. The results can be seen in Fig. 7.9 wherethe typical decay of the NMR signal with increase in the magnetic field gradient can beseen, as expected of a diffusing sample. Data analysis showed an intial diffusion coefficientof 3.2\u00d710\u201312 m2s with a weighting factor f = 0.23. Measurements taken after ten hours ofheating do not show the Gaussian decay of the signal, indicating an unmeasurable dif-fusion rate (<10\u201314 m2s ). Decay of diffusion due to time spent under heat was observedin many repeated experiments, following the same sample preparation procedures and itwas concluded that self-diffusion coefficients decay with exposure to heat. It is thereforeimportant to note that the self-diffusion coefficients were changing during the PFG-NMRexperiments. Due to the time required for NMR signal averaging, the shortest experiment1307.3. Results and Discussionpossible was one hour. This signal averaging results in an averaged self-diffusion coefficientover the experiment, limiting the quantitative conclusions that can be drawn. By compar-ing the effect of different compositions, conditions, and stimuli on the presence or absenceof diffusion we are able to learn about the conditions that lead to or eliminate diffusion.In general, diffusion of the tag indicates either movement of the polymer network itselfas a whole or exchange reactions transferring the tag between polymer chains. To ensuremeasurements do not simply represent the polymer backbone motion, the experiment wasrepeated with a non-exchangeable tag. The results are summarized in Table 7.2. SamplesPD1 and PD2, which were made with an exchangeable tag (2,2,2- Trifluoroethylacetoac-etate) were found to have initial self-diffusion coefficients on the order of 10\u201312 m2s at 60\u25e6C,while sample PD3, which has no exchangeable tag, showed no measurable self-diffusion atany temperature. The measured self-diffusion coefficients of samples PD1 and PD2 arecomparable to the diffusion rates of penetrant n-heptane through a crosslinked polyamideat similar temperatures [164]. This supports the conclusion that diffusion in these samplesoccurred as a result of the exchangeable tag and not of the bulk movement of the polymerchains, as the diffusion rate depends on the presence of the exchangeable tag.Sample Exchangeable CL Exchangeable tag D\u00d710\u201312 [m2s ] fPD1 3 3 2.5 0.02PD2 7 3 3.1 0.01PD3 7 7 < 0.01 N\/ATable 7.2: Initial self-diffusion coefficients at 60 \u25e6C of NMR visible tags in PDMS backbonevitrimers.T1 and T2 measurements were taken at different temperatures to determine whethermolecular dynamics can shed some light regarding the mechanism of action. They arepresented in Fig. 7.10. The changes in relaxation time with temperature follow the expected1317.3. Results and Discussiondynamics of a polymer chain above its Tg, as they show a linear relation between relaxationtimes and temperature. This indicates that there is no major change in the dynamics thatoccur in the polymer in this range of temperatures.(a) (b)Figure 7.10: Relaxation times of sample PD1 vs temperature. (a) Spin-lattice relaxationtimes (T1) as measured using an inversion-recovery pulse sequence. (b) Spin-spin relaxationtimes (T2) as measured using a spin-echo pulse sequence. Both graphs show a quasi-linearrelation between temperature and relaxation times, as expected from a polymer chainabove Tg.As the exchange reaction evidently decays as samples are annealed, more investigationwas necessary to find its triggering mechanism. In a vinylogous urethane exchange reaction,the total number of bonds in the system does not change, therefore the system\u2019s potentialenergy remains the same. Unless an external potential drives it (compression or internalstrain), the enthalpy, \u2206H, of this exchange reaction is expected to be zero and thereforecannot be the driving force. Rovigatti et al. compare this system\u2019s dynamics to \u201ca strollon the flat ground state potential energy surface\u201d [165]. The only remaining driving force isentropy, as the system explores the enthalpy equivalent states, it will maximize its entropyby allowing each of the covalent bonds to be surrounded by as many possible sites foran exchange reaction. One of the early tests, done by the Wolf group, of these materials1327.3. Results and Discussionfound qualitatively that fluorescent dye soaked to one edge of the material seemed tomigrate approximately one mm per 5 hours, indicating a diffusion coefficient in the orderof 10\u201310 m2s . This thought process and the qualitative data drove us to repeat the diffusionexperiments using a sample with a driving force in the form of a chemical potential gradient.A heterogeneously loaded sample with tags being unevenly loaded in the polymer will havea gradient of concentration that may trigger the exchange reaction. This chemical potentialgradient was expected to drive the system to a measurable self-diffusion of the tag in itsattempt to increase entropy and reach a lower energy state. Results of this experiment canbe seen in Table 7.3.Time under heat [hours] D\u00d710\u201312 [m2s ] f0 < 0.01 N\/A2 1.5 0.234 2.5 0.186 < 0.01 N\/A8 < 0.01 N\/A10 < 0.01 N\/A12 < 0.01 N\/ATable 7.3: Self-diffusion coefficients of heterogeneously loaded vitrimer sample at 60 \u25e6C vs.time. The first measurement (time 0) was measured before the heater was turned on as acontrol.The results show again that diffusion decays over the course of several hours underisothermal annealing, to the point of no measurable diffusion. The induced diffusion wassmall, but still measurable, until it decayed entirely after six hours of exposure to heat.When the sample was removed from the NMR, it looked visibly very similar to the samplein its original state. No significantly visible diffusion of the coloured tag had occurredand the sample still had a very visible colour gradient and therefore still had a chemicalpotential gradient. It seems that chemical potential gradient is not enough to overcome1337.3. Results and Discussionthe energy barrier involved in creating the tetrahedral transition state that is required forthe exchange process (see Scheme 7.2).In an attempt to create other scenarios to produce the exchange process, a stress re-laxation mechanism in polymers was considered. In general, amorphous polymers heatedabove their Tg can have individual chains diffuse and flow in a similar way to a viscous liq-uid [166]. This means that any deformation or stress from applied force is relaxed throughthe rearrangement of the polymer chains, similar to how a liquid responds to mechanicalchanges by breaking and creating intermolecular associations [167, 168]. Polymers con-taining small-molecule penetrants can exhibit self-diffusion of the penetrant on the sameorder of magnitude as was measured here. This diffusion behaves as a stress relievingmechanism by creating different local penetrant concentrations in different areas of thepolymer, affecting the swelling and local mechanical properties of certain sections of thepolymer [169, 164]. It is possible that the exchange mechanism is enabled by an effectcalled \u2018stress dependant thermal activation\u2019, which uses the exchange to reduce stress inthe sample [170, 171]. This diffusion is dependant primarily on penetrant size, as the inter-actions between the polymer and the penetrants are weak. These interactions are similarto solute solvent interactions, and therefore present a size dependency for diffusion thatresembles the Stokes-Einstein equation (Eq. 1.5).We suggest that crosslinked vitrimers present covalent bond diffusion only when thesample is under mechanical stress. When the system is not in mechanical equilibrium,its initial state is higher in free energy and therefore the energy barrier for the exchangereaction is smaller and hence can be thermally overcome.To test this hypothesis, three subsequent measurements of the same sample were carriedout, in four different mechanical conditions. The steps are portrayed in Fig. 7.11 to simplifythe explanation of the process, and the numerical results are summarized in Table 7.4.1347.3. Results and DiscussionSample PD1c was made according to the exact same procedure as PD1a, cut to size, andpacked in a sealed NMR tube. The sample in this state has residual mechanical stress asa result of post-polymerization cooling [172]. This stress is portrayed in Fig. 7.11b as bentlines which represent mechanical stress. The sample was then heated to 60 \u25e6C and resultedin a measurable self-diffusion coefficient of 3.3\u00b710\u201312 m2s (see Table 7.4). The sample washeld at 60 \u25e6C overnight, after which no self-diffusion was measured. We suggest that thetime spent at 60 \u25e6C allowed the sample to go to a lower energy state via covalent bonddiffusion we measured in the previous step. This mechanically relaxed state is representedin Fig. 7.11c as straight lines.The sample was then removed from the NMR tube and mechanically cut and com-pressed to introduce mechanical stress to the sample. This step is portrayed in Fig.7.5dand 7.5e as well as shown in a picture in Fig. 7.12. The temperature was once againincreased to 60 \u25e6C and measurements of the compressed sample showed an initial diffu-sion coefficient of 1.5\u00b710\u201312 m2s . This measured diffusion decayed after about six hours ofexposure to 60 \u25e6C. This disappearance and reappearance of self-diffusion in response tophysical manipulation of the sample demonstrated that the diffusion in the sample couldbe stimulated by the application of mechanical stress.We propose that in both the original sample and the compressed sample, stress waspresent from either post-polymerization cooling of the sample or from the active compress-ing. The diffusing fraction of the sample is the fraction experiencing mechanical stress.Based on the regenerated diffusion after an external stress application and the lack of dif-fusion after the sample was allowed to return to equilibrium, it appears that the presenceof mechanical stress in the sample is the key criterion needed for the exchange to occurspontaneously in the polymer.1357.3. Results and DiscussionFigure 7.11: Explanation of the mechanical stress in the sample (a) when cast (b) postpolymerization and cooling (c) after heating (d) when cut (e) cut and compressed (f) afterheating. Bent lines represent the existence of mechanical stress and straight lines the lackof.Sample\u2019s physical state D\u00d710\u201312 [m2s ] f corresponds to Fig.:Cut from original polymer cast 3.5 0.78 7.5bKept at temperature overnight < 0.01 N\/A 7.5cGround and compressed manually 1.5 0.88 7.5eKept at temperature for 6 hours < 0.01 N\/A 7.5fTable 7.4: Self-diffusion coefficients at 60 \u25e6C of vitrimer sample PD1c with different statesof mechanical stress. The measurements were taken in the order presented on the samesample.Since the diffusion coefficients measured are similar to those of small molecules in densepolymers [173], another potential mechanism should be considered. It is possible that the1367.3. Results and Discussion(a) (b)Figure 7.12: Sample PG1 (a) as originally packed in a sealed NMR tube and (b) afterbeing opened and manually cut and compressed to introduce mechanical stress.self-diffusion measured is that of tag molecules released in response to mechanical stress bya mechano-chemical process [174, 175]. These released molecules diffuse before attachingagain to the polymer network, reforming the crosslinked polymer in its new configuration.The fact that no self-diffusion was observed when the polymer was crosslinked with a non-exchangeable crosslinker (see Table 7.2) does make this mechanism less likely. However,additional control experiments would be beneficial to our understanding of the mechanismbehind this diffusion regeneration.There is an interesting parallel with the piezoionic materials investigated in chapter 5.In both systems, mechanical stimulus leads to diffusion. There, a mechanical deformationcreates a non-equilibrium ion distribution, giving rise to gradually diminishing ionic motionand an associated voltage. Here, built-in mechanical strain is gradually relieved by covalentrearrangement of crosslinkers. Similarly, equilibrium self-diffusion does not predict thestimulus response in either system.1377.4. Conclusions7.4 ConclusionsThe investigated vinylogous urethane exchange reaction in a crosslinked polymer networkwas observed using 19F PFG-NMR. These measurements have shown that these tags, whichare stationary at room temperature, diffuse for a short period but abruptly stop after beingexposed to heat for a few hours. Further heating and the creation of a chemical potentialgradient do not appear to result in increased diffusion. Diffusion was only regeneratedby application of mechanical stress. We therefore propose that the exchange reaction is aresult of a thermally activated stress relaxation mechanism and is initiated by mechanicalstress in addition to heat which is required to overcome the energy barrier. Further controlexperiments are necessary to eliminate other possible mechanisms that can potentiallyexplain this observed behaviour. This suggested mechanism introduces opportunities forrecycling materials made of vitrimers and also shows the physical limitations of using suchmaterials under physical load in combination with heat.138Chapter 8Conclusions and Future Work8.1 ConclusionsIn this dissertation, transport properties were investigated in a variety of systems in orderto increase our understanding of the effects transport have on reaction rates, conductivities,and mechanical behaviour observed in innovative materials. We have used both existingand newly developed NMR methods to directly measure both driven and self-diffusionin-situ while investigating the effects concentration and environment have on them. Bymeasuring both driven and self-diffusion in a variety of materials using nuclear magneticresonance, we explain some of the transport mechanisms that are behind unique materialbehaviour.To shed light on the piezoionic behaviour of devices made from EMI-TFSI dilutionin PC, self-diffusion measurements were performed on RTIL ions in solutions and IPNs.The inversion observed in self-diffusion was similar to that seen in piezoionic behaviourproviding experimental evidence for a previously suggested mechanism. This work helpsclarify the ambiguities on the origin of the piezoionic effect and confirms that it is a resultof differences in diffusivity of oppositely charged ions. The differences observed betweendirectly measured conductivities and those derived by the Nernst-Einstein relation fromself-diffusion suggest that the size and composition of ionic aggregates depend on RTILconcentration and that these aggregated species greatly affect the charge carrying abilities1398.1. Conclusionsof the ions.A simple, low cost, eNMR probe was developed in order to directly measure the elec-trophoretic mobility of magnetically visible ions. Using this probe in conjunction withPFG-NMR measurements allowed us to investigate the unique behaviour of artificial mus-cle and nerve piezoionic materials. From these measurements, we conclude that in lowconcentrations the majority of the current is carried by the anion and in high concentra-tions, Li+ becomes the dominant carrier. This inversion contradicts the predictions of theNernst-Einstein equation. We suggest a new conduction mechanism for Li+ that relieson supramolecular structures formed by the ions and the solvent. The relatively smallLi+ ion can hop between vacant sites in these structures avoiding the hindering effects ofa crowded, viscous solution. A similar conduction model has been computationally de-rived by Gao et al. [4], providing more support for the existence and importance of thisconduction mechanism in concentrated solutions. By hopping between vacant sites in theconcentrated solution, Li+ can conduct charge better than its diffusivity suggests, changingour understanding of the relation between diffusion and ionic conductivity.The self-diffusion coefficients of four chromophores were measured in operating condi-tions of optical filters used in the automotive industry. The measurements were done in avast variety of formulations and physical states in order to optimize both the feasibility ofthe measurements and provide experimental data to assist in formulation decisions of futuredevices. The dynamics of chromophore molecules were investigated as well as the relationbetween chromophore diffusion and temperature. A general decrease in self-diffusion withtemperature decrease was observed. Yet, at 250 K the diffusion of the chromophore showeda non-Fickian behaviour and increases as the temperature decreases. This is likely a re-sult of some components precipitating out of solution resulting in a reduction of the totalviscosity of the solution. The information provided by these measurements has provided1408.1. Conclusionsexperimental data that assists in the determination of device formulations. This researchhelps improve the usability of these optical filters and increase fuel economy by reducingthe need for active cooling or heating in automobiles.A vinylogous urethane exchange reaction in vitrimers was investigated using PFG-NMR. In these samples, no diffusion was observed after the initial heating of the samples.Only when physical stress was introduced, diffusion regenerated. From these measure-ments, we propose a thermally activated stress relaxation mechanism which is initiated bymechanical stress. This work changed what we know about vitrimers recycling by showingthe necessary conditions to initiate the crosslinker exchange reaction. We learn the physicallimitation for usage of these vitrimers as they cannot be used in temperatures higher than60 \u25e6C while under physical load, as that will initiate the crosslinker exchange, forcing themto re-shape. In addition, this suggested mechanism introduces recycling opportunities formaterials made of vitrimers under the same conditions. This research has shown both thelimitations and the recycling opportunities for these materials.The investigation of these systems has demonstrated that transport is a key componentin their unique behaviours. This research changed what we know about these systems andproved, again, that classical description does not provide accurate predictions for transportin dense material. The new mechanisms suggested in this dissertation have shown that, inconcentrated media, diffusivity does not follow the same mechanism as in diluted solutions.The suggested mechanisms provide a different mental images that explains the experimentalevidence. From this work, it is clear to see that experimental data measured in-situ isnecessary to guide the development of innovative materials. With our new, simple, low-cost eNMR probe future scientists can continue and explore the diffusivity and conductionof dense materials.Although the research done in this dissertation has answered some questions, new1418.2. Recommended Future Workquestions are also raised. Further research is still necessary for the complete investigationof transport properties in dense media. The methods used in this dissertation have beentested and refined so that future researchers can utilize them, and continue the work andexpand it to further materials and types of media.8.2 Recommended Future Work8.2.1 eNMR Probe Construction and ValidationThe newly constructed eNMR probe was proven successful in directly measuring elec-trophoretic mobility coefficients for solution samples of both low and high conductivitiesand concentrations. Further validation work should be done on a variety of salts, com-paring the mobilities measured to known conductivities. In addition, more experimentsshould be done using the existing eNMR probe on samples based on polymers in orderto investigate the conduction mechanisms in polymer based materials. Our attempts tomake eNMR measurements in polymers were not successful; we believe this was due tohigh resistance of the solvent-polymer interface, which may be reduced by increasing thecontact area between the electrode and the polymer. This can be done by either finning thepolymer to increase the surface area of contact with the solution or by connecting the elec-trode directly to the polymer sample. If necessary, T2 times of the mobile species mightbe increased by employing a deuterated solvent which would reduce the intermolecularcontribution of the probe nucleus\u2019 relaxation.8.2.2 Diffusion of EMI-TFSI Dilutions in Propylene CarbonateThe self-diffusion measurements of EMI-TFSI dilutions have been successfully measuredand answered some of the question regarding the piezoionic effect seen in devices made1428.2. Recommended Future Workof these IEAPs. In order to fully investigate the transport mechanism and conductivityattenuating interactions, ionic mobility constants should be measured directly for the ex-isting set of liquid samples, using the eNMR set-up that is described in Chapter 3. Withfurther adjustment of the existing eNMR pulse sequences and sample holder, we believe itwill be possible to measure mobility for the set of IPN embedded electrolytes as well.By obtaining the effective charge (zeff) for the individual ions from the combination ofeNMR and PFG-NMR insight on the aggregation of RTILs as a function of dilution wouldbe gained. This will provide a more direct explanation of the observed phenomena inaddition to providing us with a better understanding of conduction in ion dense solutions.By measuring the mobility for samples with different solvents and polymer compositionswe will be able to achieve better understanding of ion-solvent and ion-polymer interactions,expanding our understanding of transport in ion dense media. This will make it possibleto better design future electrochemical devices with specific transport properties.8.2.3 Ion Transport in Touch Sensor ElectrolytesBoth self-diffusion and electrophoretic mobility have been measured for a set of differentconcentration LiTFSI salt solutions in PC that exhibit the pieozionic effect. The datashowed that for lithium, the effective charge, zeff , increases with an increase in concentra-tion. Together with self-diffusion data, we suggest a potential Li+ conduction mechanismthat explains the interesting behaviour of the ions in high concentrations. The mechanismis based on Li+ hopping between holes, created by an external application of voltage. Inorder to confirm the suggested Li+ conduction mechanism, more experiments should bedone at elevated temperature. Increasing the sample\u2019s temperature should decrease theion-ion aggregates necessary for conduction via this suggested mechanism. This mech-anism can be further validated if indeed Li+\u2019s mobility matches that predicted by the1438.2. Recommended Future WorkNernst-Einstein relation at elevated temperature. Additionally, proximity measurementsmay help confirm the existence of ionic aggregates in solutions and learn of their structure.Using the nuclear Overhauser effect (NOE) may assist in learning the proximity betweenthe anion and solvent and revealing the ratio between them to create the necessary supra-molecular structure. Raman spectroscopy might help provide experimental evidence of thesupra-molecular structures and their composition in solution (as done for solvation layermeasurements of Li+ by Shi et al. [125]). Molecular dynamics simulations can providehelpful insight (as shown by Gao et al. [4]) and assist in learning the minimal conditionsnecessary for the existence of this conduction mechanism.These experiments should be expanded to more lithium salts, especially those contain-ing a differently sized counterion. It would be interesting to see what is the minimal ionsize difference ratio necessary for the depletion conduction mechanism to occur.8.2.4 Chromophore Diffusion in Optical Filter DevicesIn order to fully examine the effect of an electric field, more eNMR experiments are needed.To confirm that the transport of the chromophores occurs as we suspect, transport mea-surements of the only magnetically visible ion in the sample, TFSI\u2013, could be performed.By comparing the measured mobility to relative device fading time, the effect an appliedelectric field has on chromophore diffusion can be examined.The time it takes for the charge to accumulate at the electrodes, shielding the electricfield on the sample can be measured using a constant-time eNMR pulse sequence. In thispulse sequence, the total duration of the experiment is fixed, but the time during whichthe electric field is applied, \u03c4, can be modified to measure the time migration constant, \u03c4eq(the time constant for charge accumulation near the electrodes that results in electric fieldshielding). Currently, devices are operated by switching the direction of the electric field1448.2. Recommended Future Workat a specific frequency to avoid this charge accumulation at the electrodes. By comparing\u03c4eq to the switching time, the transport mechanism of the chromophore molecules could bemore fully understood. In the case where \u03c4eq is shorter than the switching time used in theoperation of these devices, the switching time should be reduced. If after such a reduction,no measurable difference in fading time device is detected, it would be possible to concludethat the diffusion of the chromophore molecules is not affected by the application of anelectric field.8.2.5 Vitrimers - Transport of Thermally Exchangeable MoleculesIn this project, we have successfully confirmed that NMR measured diffusion is triggeredwhen the sample is physically manipulated, but diffusion shuts down after initial stressis relieved. Some questions still remain unanswered in regards to the type of stress thattriggers the diffusion.First, to confirm the effects seen with NMR, rheological studies are required. Thesemeasurements are currently in progress with the help of Tanya Tomkovic of Savvas Hatzikir-iakos\u2019 group in Chemical and Biological Engineering of UBC.To test the possible free crosslinker molecule explanation, a control experiment could bedone. An NMR visible penetrant molecule of similar size and polarity to the used crosslink-ers should be introduced to the system, but with no ability to bond to the network. If themeasured diffusion of such molecule is similar to that measured in our experiments, theconclusions reached should be reconsidered, as it is possible the original measurements haveonly observed the crosslinker molecules in a temporary state, when they are disconnectedfrom the polymer network.An NMR stress applicator can be used to attempt and trigger the diffusion in-situ. Anexisting set-up already exists in the lab and could easily be adapted to this purpose. In order1458.3. Outlookto combine the stress applicator with self-diffusion measurements, the stress applicatorrequires some adjustments to physically fit in the magnetic field gradient set. Once thisset-up is adjusted to fit in the gradient set, it will be possible to apply different types ofphysical stress on the sample in different temperatures and measure diffusion immediatelyafter. It will be very interesting to see which types of loading scenario trigger the diffusionand which do not. In addition, it will be interesting to see whether applying the stress athigh temperature does or does not trigger the diffusion.8.3 OutlookThe investigations presented in this dissertation demonstrated the importance of transportto our understanding of material behaviour and lay the ground work for future researchutilizing NMR methodology. The mechanisms suggested from this work show that diffu-sivity in concentrated media follows unique mechanisms that are not present in dilutedsolutions. Future NMR based research on additional salts and RTILs can help progressthe innovation of new materials with applications such as batteries, artificial muscles andnerves, and recyclable polymers, to list just a few possibilities. The eNMR probe systempresented here has demonstrated that accurate ionic conductivity of NMR visible ions canbe measured directly, using a relatively simple, cost-effective, and easily manufactured de-vice. By measuring diffusive parameters in-situ, the mechanisms underlying novel materialproperties can be explored in order to advance material science and technology.146Bibliography[1] C. H. Tso, \u201cNMR study of PF6\u2013 doped polypyrrole: a potential candidate for arti-ficial muscles,\u201d MSc, The University of British Columbia, 2010.[2] A. L. Michan, \u201cNuclear magnetic resonance characterization of solid polymer elec-trolyte materials,\u201d MASc, The University of British Columbia, 2012.[3] V. Woehling, G. T. M. Nguyen, C. Plesse, Y. Petel, Y. Dobashi, J. D. W. Madden,C. A. Michal, and F. Vidal, \u201cStudy of the piezoionic effect and influence of electrolytein conducting polymer based soft strain sensors,\u201d Multifunctional Materials, vol. 2,p. 045002, Dec. 2019.[4] Y. Gao, J. Huang, Y. Liu, J. Yan, B. Mao, and S. Chen, \u201cIon-vacancy coupled chargetransfer model for ion transport in concentrated solutions,\u201d Science China Chemistry,vol. 62, pp. 515 \u2013 520, Apr. 2019.[5] T. Wright, T. Tomkovic, S. G. Hatzikiriakos, and M. O. Wolf, \u201cPhotoactivated heal-able vitrimeric copolymers,\u201d Macromolecules, vol. 52, no. 1, pp. 36 \u2013 42, 2019.[6] A. Einstein, \u201cOn the theory of the Brownian movement,\u201d Ann. Phys, vol. 19, no. 4,pp. 371 \u2013 381, 1906.[7] A. Fick, \u201cUeber diffusion,\u201d Annalen der Physik, vol. 170, no. 1, pp. 59 \u2013 86, 1855.147Bibliography[8] M. Sharma and S. Yashonath, \u201cSize dependence of solute diffusivity and Stokes-Einstein relationship: effect of van der Waals interaction,\u201d Diffusion Fundamentals,vol. 6, pp. 35 \u2013 41, 2007.[9] G. Peskir, \u201cOn the diffusion coefficient: The Einstein relation and beyond,\u201d Stochas-tic Models, vol. 19, no. 3, pp. 383 \u2013 405, 2003.[10] J. T. Edward, \u201cMolecular volumes and the Stokes-Einstein equation,\u201d Journal ofChemical Education, vol. 47, no. 4, p. 261, 1970.[11] J. Rauch and W. Ko\u00a8hler, \u201cCollective and thermal diffusion in dilute, semidilute, andconcentrated solutions of polystyrene in toluene,\u201d The Journal of Chemical Physics,vol. 119, no. 22, pp. 11977 \u2013 11988, 2003.[12] J. L. Anderson, F. Rauh, and A. Morales, \u201cParticle diffusion as a function of con-centration and ionic strength,\u201d The Journal of Physical Chemistry, vol. 82, no. 5,pp. 608 \u2013 616, 1978.[13] B. Jachimska, M. Wasilewska, and Z. Adamczyk, \u201cCharacterization of globular pro-tein solutions by dynamic light scattering, electrophoretic mobility, and viscositymeasurements,\u201d Langmuir, vol. 24, no. 13, pp. 6866 \u2013 6872, 2008.[14] Z. Ne\u00b4meth, L. Erdei, and A. Kolics, \u201cA new approach to studying ion transportin corrosion protective coatings using an in situ radiotracer method,\u201d CorrosionScience, vol. 37, no. 7, pp. 1163 \u2013 1166, 1995.[15] T. Welton, \u201cRoom-temperature ionic liquids: solvents for synthesis and catalysis,\u201dChemical Reviews, vol. 99, no. 8, pp. 2071 \u2013 2084, 1999.148Bibliography[16] E. T. Fox, E. Paillard, O. Borodin, and W. A. Henderson, \u201cPhysicochemical prop-erties of binary ionic liquid-aprotic solvent electrolyte mixtures,\u201d The Journal ofPhysical Chemistry C, vol. 117, no. 1, pp. 78 \u2013 84, 2013.[17] K. Yuyama, G. Masuda, H. Yoshida, and T. Sato, \u201cIonic liquids containing thetetrafluoroborate anion have the best performance and stability for electric doublelayer capacitor applications,\u201d Journal of Power Sources, vol. 162, no. 2, pp. 1401 \u20131408, 2006.[18] R. D. Rogers, \u201cMaterials science: Reflections on ionic liquids,\u201d Nature, vol. 447,no. 7147, pp. 917 \u2013 918, 2007.[19] H. Weinga\u00a8rtner, P. Sasisanker, C. Daguenet, P. J. Dyson, I. Krossing, J. M. Slattery,and T. Schubert, \u201cThe dielectric response of room-temperature ionic liquids: Effectof cation variation,\u201d The Journal of Physical Chemistry B, vol. 111, no. 18, pp. 4775\u2013 4780, 2007.[20] K. Ueno, H. Tokuda, and M. Watanabe, \u201cIonicity in ionic liquids: correlationwith ionic structure and physicochemical properties,\u201d Physical Chemistry ChemicalPhysics, vol. 12, pp. 1649 \u2013 1658, 2010.[21] H. Tokuda, S. Tsuzuki, M. A. B. H. Susan, K. Hayamizu, and M. Watanabe, \u201cHowionic are room-temperature ionic liquids? an indicator of the physicochemical prop-erties,\u201d The Journal of Physical Chemistry B, vol. 110, no. 39, pp. 19593 \u2013 19600,2006.[22] J. Dupont, \u201cOn the solid, liquid and solution structural organization of imidazoliumionic liquids,\u201d Journal of the Brazilian Chemical Society, vol. 15, pp. 341 \u2013 350, June2004.149Bibliography[23] R. W. Jones and F. Mohling, \u201cAccuracy of theories of the primitive model of ionicsolutions,\u201d The Journal of Physical Chemistry, vol. 75, no. 25, pp. 3790 \u2013 3796, 1971.[24] H. L. Friedman, Ionic Solution Theory: based on Cluster expansion methods, vol. 3.Interscience Publishers, 1962.[25] R. Robinson and R. Stokes, Electrolyte Solutions. Dover Publications, 2nd revised ed.,1951 (2002).[26] V. Gutmann, \u201cEmpirical parameters for donor and acceptor properties of solvents,\u201dElectrochimica Acta, vol. 21, pp. 661 \u2013 670, 1976.[27] R. M. Adar, T. Markovich, and D. Andelman, \u201cBjerrum pairs in ionic solutions: APoisson-Boltzmann approach,\u201d The Journal of Chemical Physics, vol. 146, no. 19,p. 194904, 2017.[28] Y. Marcus and G. Hefter, \u201cIon pairing,\u201d Chemical Reviews, vol. 106, no. 11, pp. 4585\u2013 4621, 2006.[29] J. Lacour and D. Moraleda, \u201cChiral anion-mediated asymmetric ion pairing chem-istry,\u201d Chem. Commun., pp. 7073 \u2013 7089, 2009.[30] P. V. Wright, \u201cElectrical conductivity in ionic complexes of poly(ethylene oxide),\u201dBritish Polymer Journal, vol. 7, no. 5, pp. 319 \u2013 327, 1975.[31] M. Armand, \u201cPolymer solid electrolytes - an overview,\u201d Solid State Ionics, vol. 910,Part 2, pp. 745 \u2013 754, 1983.[32] G. Ehrenstein and R. Theriault, Polymeric Materials: Structure, Properties, Appli-cations. Hanser, 2001.150Bibliography[33] S. Lankalapalli and V. M. Kolapalli, \u201cPolyelectrolyte complexes: a review of their ap-plicability in drug delivery technology,\u201d Indian Journal of Pharmaceutical Sciences,vol. 71, no. 5, pp. 481 \u2013 487, 2009.[34] Y. Roichman, Charge Transport in Conjugated Polymers. PhD, Technion - IsraelInstitute of Technology, 2004.[35] A. D. McNaught and A. W. Blackwell, Compendium of Chemical Terminology. Sci-entific Publications, 2nd ed., 1997.[36] R. P. Feynman, R. B. Leighton, and M. L. Sands, The Feynman Lectures on Physics,vol. 1. Addison-Wesley Pub. Co, 1963.[37] W. S. Price, NMR Studies of Translational Motion: Principles and Applications.Cambridge University Press, 2009.[38] B. Amsden, \u201cSolute diffusion within hydrogels. Mechanisms and models,\u201d Macro-molecules, vol. 31, no. 23, pp. 8382 \u2013 8395, 1998.[39] L. Masaro and X. Zhu, \u201cPhysical models of diffusion for polymer solutions, gels andsolids,\u201d Progress in Polymer Science, vol. 24, no. 5, pp. 731 \u2013 775, 1999.[40] L. Johansson, C. Elvingson, and J. E. Loefroth, \u201cDiffusion and interaction in gels andsolutions. 3. Theoretical results on the obstruction effect,\u201d Macromolecules, vol. 24,no. 22, pp. 6024 \u2013 6029, 1991.[41] L. Johansson and J. Lo\u00a8froth, \u201cDiffusion and interaction in gels and solutions. 4. Hardsphere Brownian dynamics simulations,\u201d The Journal of Chemical Physics, vol. 98,no. 9, pp. 7471 \u2013 7479, 1993.151Bibliography[42] M. Golmohamadi and K. J. Wilkinson, \u201cDiffusion of ions in a calcium alginatehydrogel-structure is the primary factor controlling diffusion,\u201d Carbohydrate Poly-mers, vol. 94, no. 1, pp. 82 \u2013 87, 2013.[43] M. H. Rausch, L. Hopf, A. Heller, A. Leipertz, and A. P. Fro\u00a8ba, \u201cBinary diffu-sion coefficients for mixtures of ionic liquids [EMIM][N(CN)2], [EMIM][NTf2], and[HMIM][NTf2] with acetone and ethanol by dynamic light scattering (DLS),\u201d TheJournal of Physical Chemistry B, vol. 117, no. 8, pp. 2429 \u2013 2437, 2013.[44] S. M. Patil, D. A. Keire, and K. Chen, \u201cComparison of NMR and dynamic lightscattering for measuring diffusion coefficients of formulated insulin: Implicationsfor particle size distribution measurements in drug products,\u201d The AAPS Journal,vol. 19, pp. 1760 \u2013 1766, Nov. 2017.[45] M. Holz, \u201cField-assisted diffusion studied by electrophoretic NMR,\u201d in Diffusion inCondensed Matter (P. Heitjans and J. Ka\u00a8rger, eds.), ch. 17, pp. 717 \u2013 742, BerlinHeidelberg: Springer-Verlag, 2005.[46] G. E. Murch, \u201cThe haven ratio in fast ionic conductors,\u201d Solid State Ionics, vol. 7,no. 3, pp. 177 \u2013 198, 1982.[47] N. A. Stolwijk, J. Ko\u00a8sters, M. Wiencierz, and M. Scho\u00a8nhoff, \u201cOn the extraction ofion association data and transference numbers from ionic diffusivity and conductivitydata in polymer electrolytes,\u201d Electrochimica Acta, vol. 102, pp. 451 \u2013 458, 2013.[48] A. R. Cooper, \u201cThe haven ratio in terms of the self-diffusion of charge,\u201d in Transportin Nonstoichiometric Compounds (G. Simkovich and V. S. Stubican, eds.), pp. 127\u2013 137, Boston, MA: Springer US, 1985.152Bibliography[49] J. E. Bostwick, C. J. Zanelotti, C. Iacob, A. G. Korovich, L. A. Madsen, and R. H.Colby, \u201cIon transport and mechanical properties of non-crystallizable molecular ioniccomposite electrolytes,\u201d Macromolecules, vol. 53, no. 4, pp. 1405 \u2013 1414, 2020.[50] A. W. Adamson, \u201cChapter four - chemical thermodynamics. The first law of ther-modynamics,\u201d in A Textbook of Physical Chemistry, pp. 101 \u2013 144, Academic Press,2nd ed., 1979.[51] H. Okuzaki, S. Takagi, F. Hishiki, and R. Tanigawa, \u201cIonic liq-uid\/polyurethane\/PEDOT:PSS composites for electro-active polymer actuators,\u201dSensors and Actuators B: Chemical, vol. 194, pp. 59 \u2013 63, 2014.[52] J. Vatamanu, Z. Hu, D. Bedrov, C. Perez, and Y. Gogotsi, \u201cIncreasing energy storagein electrochemical capacitors with ionic liquid electrolytes and nanostructured carbonelectrodes,\u201d The Journal of Physical Chemistry Letters, vol. 4, no. 17, pp. 2829 \u20132837, 2013.[53] T. Sato, G. Masuda, and K. Takagi, \u201cElectrochemical properties of novel ionic liquidsfor electric double layer capacitor applications,\u201d Electrochimica Acta, vol. 49, no. 21,pp. 3603 \u2013 3611, 2004.[54] P. G. Bruce and C. A. Vincent, \u201cEffect of ion association on transport in polymerelectrolytes,\u201d Faraday Discussions of the Chemical Society, vol. 88, pp. 43 \u2013 54, 1989.[55] H. Fritz and A. Kuhn, \u201cComparative determination of effective transport numbersin solid lithium electrolytes,\u201d Journal of Power Sources, vol. 41, no. 3, pp. 253 \u2013 261,1993.[56] A. Noda, K. Hayamizu, and M. Watanabe, \u201cPulsed-gradient spin echo 1H and19F NMR ionic diffusion coefficient, viscosity, and ionic conductivity of non-153Bibliographychloroaluminate room\u2013temperature ionic liquids,\u201d The Journal of Physical ChemistryB, vol. 105, no. 20, pp. 4603 \u2013 4610, 2001.[57] J. Hou, Z. Zhang, and L. A. Madsen, \u201cCation\/anion associations in ionic liquidsmodulated by hydration and ionic medium,\u201d The Journal of Physical Chemistry B,vol. 115, no. 16, pp. 4576 \u2013 4582, 2011.[58] Z. Zhang and L. A. Madsen, \u201cObservation of separate cation and anion elec-trophoretic mobilities in pure ionic liquids,\u201d The Journal of Chemical Physics,vol. 140, no. 8, 2014.[59] H. Tokuda, K. Hayamizu, K. Ishii, M. A. B. H. Susan, and M. Watanabe, \u201cPhysic-ochemical properties and structures of room temperature ionic liquids. 2. Variationof alkyl chain length in imidazolium cation,\u201d The Journal of Physical Chemistry B,vol. 109, no. 13, pp. 6103 \u2013 6110, 2005.[60] M. H. Levitt, Spin Dynamics: Basics of Nuclear Magnetic Resonance. Wiley, 2nd ed.,2008.[61] E. O. Stejskal and J. E. Tanner, \u201cSpin diffusion measurements: spin echoes inthe presence of a time-dependent field gradient,\u201d The Journal of Chemical Physics,vol. 42, no. 1, pp. 288 \u2013 292, 1965.[62] T. R. Saarinen and C. S. Johnson, \u201cImaging of transient magnetization gratings inNMR analogies with laser-induced gratings and applications to diffusion and flow,\u201dJournal of Magnetic Resonance, vol. 78, no. 2, pp. 257 \u2013 270, 1988.[63] I. Solomon, \u201cRelaxation processes in a system of two spins,\u201d Physical Review, vol. 99,pp. 559 \u2013 565, July 1955.154Bibliography[64] K. J. Packer, \u201cThe study of slow coherent molecular motion by pulsed nuclear mag-netic resonance,\u201d Molecular Physics, vol. 17, no. 4, pp. 355 \u2013 368, 1969.[65] M. Holz and C. Mu\u00a8ller, \u201cDirect measurement of single ionic drift velocities in elec-trolyte solutions. an nmr method,\u201d Berichte der Bunsengesellschaft fu\u00a8r physikalischeChemie, vol. 86, no. 2, pp. 141 \u2013 147, 1982.[66] M. Holz, \u201cElectrophoretic NMR,\u201d Chemical Society Reviews, vol. 23, no. 3, pp. 165\u2013 174, 1994.[67] P&L Scientific I.S., 2013 (accessed February 1, 2015). http:\/\/plscientific.se\/.[68] F. Hallberg, Molecular Interactions Studied by Electrophoretic and Diffusion NMR.PhD, KTH Royal Institute of Technology, 2010.[69] F. Hallberg, I. Furo\u00b4, and P. Stilbs, \u201cIon pairing in ethanol\/water solution probedby electrophoretic and diffusion NMR,\u201d Journal of the American Chemical Society,vol. 131, no. 39, pp. 13900 \u2013 13901, 2009.[70] Q. He, Y. Liu, H. Sun, and E. Li, \u201cCapillary array electrophoretic NMR of proteins inbiological buffer solutions,\u201d Journal of Magnetic Resonance, vol. 141, no. 2, pp. 355\u2013 359, 1999.[71] T. R. Saarinen and C. S. Johnson, \u201cHigh-resolution electrophoretic NMR,\u201d Journalof the American Chemical Society, vol. 110, no. 10, pp. 3332 \u2013 3333, 1988.[72] U. Bo\u00a8hme and U. Scheler, \u201cHydrodynamic size and electrophoretic mobility ofpoly (styrene sulfonate) versus molecular weight,\u201d Macromolecular Chemistry andPhysics, vol. 208, no. 19-20, pp. 2254 \u2013 2257, 2007.155Bibliography[73] F. Hallberg, I. Furo\u00b4, P. V. Yushmanov, and P. Stilbs, \u201cSensitive and robust elec-trophoretic NMR: Instrumentation and experiments,\u201d Journal of Magnetic Reso-nance, vol. 192, no. 1, pp. 69 \u2013 77, 2008.[74] I. Furo\u00b4, \u201cAssociation of molecules and ions as observed by electrokinetic and diffusionNMR.\u201d https:\/\/www.euract-nmr.kit.edu\/downloads\/Furo.pdf, Jan. 2010.[75] E. Pettersson, I. Furo\u00b4, and P. Stilbs, \u201cOn experimental aspects of electrophoreticNMR,\u201d Concepts in Magnetic Resonance Part A, vol. 22, no. 2, pp. 61 \u2013 68, 2004.[76] U. Bo\u00a8hme and U. Scheler, \u201cEffective charge of poly(styrenesulfonate) and ionicstrengthan electrophoresis NMR investigation,\u201d Colloids and Surfaces A: Physico-chemical and Engineering Aspects, vol. 222, no. 1, pp. 35 \u2013 40, 2003.[77] W. S. Price, F. Hallberg, and P. Stilbs, \u201cA PGSE diffusion and electrophoretic NMRstudy of Cs+ and Na+ dynamics in aqueous crown ether systems,\u201d Magnetic Reso-nance in Chemistry, vol. 45, no. 2, pp. 152 \u2013 156, 2007.[78] M. Holz, O. Lucas, and C. Mu\u00a8ller, \u201cNMR in the presence of an electric current. Si-multaneous measurements of ionic mobilities, transference numbers, and self-diffusioncoefficients using an NMR pulsed-gradient experiment,\u201d Journal of Magnetic Reso-nance (1969), vol. 58, no. 2, pp. 294 \u2013 305, 1984.[79] M. Gouverneur, J. Kopp, L. van Wu\u00a8llen, and M. Scho\u00a8nhoff, \u201cDirect determinationof ionic transference numbers in ionic liquids by electrophoretic NMR,\u201d Phys. Chem.Chem. Phys., vol. 17, pp. 30680 \u2013 30686, 2015.[80] U. Scheler, Electrophoretic NMR. * Update based on original article by CharlesS. Johnson, Encyclopedia of Magnetic Resonance, 1996, John Wiley & Sons Ltd.American Cancer Society, Mar. 2012.156Bibliography[81] B. Manz, P. Stilbs, B. Joensson, O. Soederman, and P. T. Callaghan, \u201cNMR imagingof the time evolution of electroosmotic flow in a capillary,\u201d The Journal of PhysicalChemistry, vol. 99, no. 29, pp. 11297 \u2013 11301, 1995.[82] L. Cser, G. To\u00a8ro\u00a8k, G. Krexner, M. Prem, and I. Sharkov, \u201cNeutron holographic studyof palladium hydride,\u201d Applied physics letters, vol. 85, no. 7, pp. 1149 \u2013 1151, 2004.[83] M. Vlach, J. C\u02c7\u00b4\u0131z\u02c7ek, O. Melikhova, V. Kodetova, I. Stul\u00b4\u0131kova\u00b4, and I. Procha\u00b4zka,\u201cHydrogen trapping at defects in pd and thermally activated desorption,\u201d Defectand Diffusion Forum, vol. 365, pp. 36 \u2013 41, July 2015.[84] M. Moini, P. Cao, and A. J. Bard, \u201cHydroquinone as a buffer additive for suppres-sion of bubbles formed by electrochemical oxidation of the CE buffer at the out-let electrode in capillary electrophoresis\/electrospray ionization-mass spectrometry,\u201dAnalytical Chemistry, vol. 71, no. 8, pp. 1658 \u2013 1661, 1999.[85] Q. He and Z. Wei, \u201cConvection compensated electrophoretic NMR,\u201d Journal of Mag-netic Resonance, vol. 150, no. 2, pp. 126 \u2013 131, 2001.[86] M. Bielejewski, M. Giesecke, and I. Furo\u00b4, \u201cOn electrophoretic NMR. Exploring highconductivity samples,\u201d Journal of Magnetic Resonance, vol. 243, pp. 17 \u2013 24, 2014.[87] H. Y. Carr and E. M. Purcell, \u201cEffects of diffusion on free precession in nuclearmagnetic resonance experiments,\u201d Phys. Rev., vol. 94, pp. 630 \u2013 638, May 1954.[88] S. Meiboom and D. Gill, \u201cModified spin-echo method for measuring nuclear relax-ation times,\u201d Review of Scientific Instruments, vol. 29, no. 8, pp. 688 \u2013 691, 1958.[89] C. A. Michal, K. Broughton, and E. Hansen, \u201cA high performance digital receiverfor home-built nuclear magnetic resonance spectrometers,\u201d Review of Scientific In-struments, vol. 73, pp. 453 \u2013 458, 2002.157Bibliography[90] C. H. Tso, J. D. Madden, and C. A. Michal, \u201cAn NMR study of ions in polypyrrole,\u201dSynthetic Metals, vol. 157, no. 1012, pp. 460 \u2013 466, 2007.[91] C. Ammann, P. Meier, and A. E. Merbach, \u201cA simple multinuclear NMR thermome-ter,\u201d Journal of Magnetic Resonance (1969), vol. 46, no. 2, pp. 319 \u2013 21, 1982.[92] C. R. Cantor and P. Schimmel, Biophysical Chemistry: Part II: Techniques for thestudy of biological structure and function. W.H Freeman, 1980.[93] M. Schmuck and M. Z. Bazant, \u201cHomogenization of the Poisson-Nernst-Planck equa-tions for ion transport in charged porous media,\u201d SIAM Journal on Applied Mathe-matics, vol. 75, no. 3, pp. 1369 \u2013 1401, 2015.[94] P. Atkins, J. De Paula, and J. Keeler, Atkins\u2019 Physical Chemistry. Oxford UniversityPress, 2018.[95] M. Gouverneur, F. Schmidt, and M. Scho\u00a8nhoff, \u201cNegative effective Li transferencenumbers in Li salt\/ionic liquid mixtures: does Li drift in the \u201cwrong\u201d direction?,\u201dPhys. Chem. Chem. Phys., vol. 20, pp. 7470 \u2013 7478, 2018.[96] M. J. Park, I. Choi, J. Hong, and O. Kim, \u201cPolymer electrolytes integrated withionic liquids for future electrochemical devices,\u201d Journal of Applied Polymer Science,vol. 129, no. 5, pp. 2363 \u2013 2376, 2013.[97] S. Z. Bisri, S. Shimizu, M. Nakano, and Y. Iwasa, \u201cEndeavor of iontronics: From fun-damentals to applications of ion-controlled electronics,\u201d Advanced Materials, vol. 29,no. 25, pp. 1607054(1 \u2013 48), 2017.[98] L. Chang, Y. Liu, Q. Yang, L. Yu, J. Liu, Z. Zhu, P. Lu, Y. Wu, and Y. Hu,\u201cIonic electroactive polymers used in bionic robots: A review,\u201d Journal of BionicEngineering, vol. 15, pp. 765 \u2013 782, Sept. 2018.158Bibliography[99] Y. Liu, Y. Hu, J. Zhao, G. Wu, X. Tao, and W. Chen, \u201cSelf-powered piezoionic strainsensor toward the monitoring of human activities,\u201d Small, vol. 12, no. 36, pp. 5074 \u20135080, 2016.[100] M. S. U. Sarwar, Y. Dobashi, E. F. Scabeni Glitz, M. Farajollahi, S. Mirab-basi, S. Naficy, G. M. Spinks, and J. D. W. Madden, \u201cTransparent and conformal\u2018PiezoIonic\u2019 touch sensor,\u201d Proceedings of SPIE, vol. 9430, pp. 943026 \u2013 943029, 2015.[101] Y. Dobashi, G. Allegretto, M. S. Sarwar, E. Cretu, and J. D. Madden, \u201cMechanoionictransduction of solid polymer electrolytes and potential applications,\u201d MRS Ad-vances, vol. 1, pp. 63 \u2013 68, 2016.[102] Y. Dobashi, A. Fannir, M. Farajollahi, A. Mahmoudzadeh, A. Usgaocar, D. Yao,G. T. Nguyen, C. Plesse, F. Vidal, and J. D. Madden, \u201cIon transport in polymercomposites with non-uniform distributions of electronic conductors,\u201d ElectrochimicaActa, vol. 247, pp. 149 \u2013 162, 2017.[103] J. Hunger, A. Stoppa, R. Buchner, and G. Hefter, \u201cFrom ionic liquid to elec-trolyte solution: Dynamics of 1-N -butyl-3-N -methylimidazolium tetrafluorobo-rate\/dichloromethane mixtures,\u201d The Journal of Physical Chemistry B, vol. 112,no. 41, pp. 12913 \u2013 12919, 2008.[104] K. Fumino, V. Fossog, P. Stange, K. Wittler, W. Polet, R. Hempelmann, and R. Lud-wig, \u201cIon pairing in protic ionic liquids probed by far-infrared spectroscopy: Effectsof solvent polarity and temperature,\u201d ChemPhysChem, vol. 15, no. 12, pp. 2604 \u20132609, 2014.159Bibliography[105] T. Vogl, S. Menne, and A. Balducci, \u201cMixtures of protic ionic liquids and propylenecarbonate as advanced electrolytes for lithium-ion batteries,\u201d Physical ChemistryChemical Physics, vol. 16, pp. 25014 \u2013 25023, 2014.[106] Y. Liu, S. Liu, J. Lin, D. Wang, V. Jain, R. Montazami, J. R. Heflin, J. Li, L. Mad-sen, and Q. M. Zhang, \u201cIon transport and storage of ionic liquids in ionic polymerconductor network composites,\u201d Applied Physics Letters, vol. 96, no. 22, p. 223503,2010.[107] W. Gorecki, M. Jeannin, E. Belorizky, C. Roux, and M. Armand, \u201cPhysical prop-erties of solid polymer electrolyte PEO(LiTFSI) complexes,\u201d Journal of Physics:Condensed Matter, vol. 7, pp. 6823 \u2013 6832, Aug. 1995.[108] Q. Cao, X. Lu, X. Wu, Y. Guo, L. Xu, and W. Fang, \u201cDensity, viscosity, and conduc-tivity of binary mixtures of the ionic liquid N -(2-hydroxyethyl)piperazinium propi-onate with water, methanol, or ethanol,\u201d Journal of Chemical & Engineering Data,vol. 60, no. 3, pp. 455 \u2013 463, 2015.[109] J. Xie, Z. Liang, and Y.-C. Lu, \u201cMolecular crowding electrolytes for high-voltageaqueous batteries,\u201d Nature Materials, Apr. 2020.[110] D. Coglitore, S. P. Edwardson, P. Macko, E. A. Patterson, and M. Whelan, \u201cTran-sition from fractional to classical Stokes-Einstein behaviour in simple fluids,\u201d RoyalSociety Open Science, vol. 4, no. 12, p. 170507, 2017.[111] K. Fujii, Y. Soejima, Y. Kyoshoin, S. Fukuda, R. Kanzaki, Y. Umebayashi, T. Yam-aguchi, S. Ishiguro, and T. Takamuku, \u201cLiquid structure of room-temperature ionicliquid, 1-ethyl-3-methylimidazolium bis-(trifluoromethanesulfonyl) imide,\u201d The Jour-nal of Physical Chemistry B, vol. 112, no. 14, pp. 4329 \u2013 4336, 2008.160Bibliography[112] W. M. Haynes, CRC Handbook of Chemistry and Physics, 92nd Edition. CRC Press,2011.[113] M. Kohagen, M. Brehm, Y. Lingscheid, R. Giernoth, J. Sangoro, F. Kremer, S. Nau-mov, C. Iacob, J. Krger, R. Valiullin, and B. Kirchner, \u201cHow hydrogen bonds influ-ence the mobility of imidazolium-based ionic liquids. A combined theoretical and ex-perimental study of 1-n-butyl-3-methylimidazolium bromide,\u201d The Journal of Phys-ical Chemistry B, vol. 115, no. 51, pp. 15280 \u2013 15288, 2011.[114] K. Fujii, T. Fujimori, T. Takamuku, R. Kanzaki, Y. Umebayashi, and S. Ishiguro,\u201cConformational equilibrium of bis(trifluoromethanesulfonyl) imid anion of a room-temperature ionic liquid: Raman spectroscopic study and DFT calculations,\u201d TheJournal of Physical Chemistry B, vol. 110, no. 16, pp. 8179 \u2013 8183, 2006.[115] L. J. Goujon, A. Khaldi, A. Maziz, C. Plesse, G. T. M. Nguyen, P.-H. Aubert,F. Vidal, C. Chevrot, and D. Teyssie\u00b4, \u201cFlexible solid polymer electrolytes basedon nitrile butadiene rubber\/poly(ethylene oxide) interpenetrating polymer networkscontaining either LiTFSI or EMITFSI,\u201d Macromolecules, vol. 44, no. 24, pp. 9683 \u20139691, 2011.[116] A. Jarosik, S. Krajewski, A. Lewandowski, and P. Radzimski, \u201cConductivity of ionicliquids in mixtures,\u201d Journal of Molecular Liquids, vol. 123, pp. 43 \u2013 50, Jan. 2006.[117] R. M. Fuoss and C. A. Kraus, \u201cProperties of electrolytic solutions. IV. The conduc-tance minimum and the formation of triple ions due to the action of Coulomb forces,\u201dJournal of the American Chemical Society, vol. 55, no. 6, pp. 2387 \u2013 2399, 1933.161Bibliography[118] D. W. McCall and D. C. Douglass, \u201cThe effect of ions on the self-diffusion of water.Ionic concentration dependence,\u201d The Journal of Physical Chemistry, vol. 69, no. 6,pp. 2001 \u2013 2011, 1965.[119] H. Palza, P. A. Zapata, and C. Angulo-Pineda, \u201cElectroactive smart polymers forbiomedical applications,\u201d Materials, vol. 12, no. 2, 2019.[120] P. G. de Gennes, K. Okumura, M. Shahinpoor, and K. J. Kim, \u201cMechanoelectriceffects in ionic gels,\u201d Europhysics Letters (EPL), vol. 50, pp. 513 \u2013 518, May 2000.[121] S. M. Villa, V. M. Mazzola, T. Santaniello, E. Locatelli, M. Maturi, L. Miglior-ini, I. Monaco, C. Lenardi, M. Comes Franchini, and P. Milani, \u201cSoftpiezoionic\/piezoelectric nanocomposites based on ionogel\/BaTiO3 nanoparticles forlow frequency and directional discriminative pressure sensing,\u201d ACS Macro Letters,vol. 8, no. 4, pp. 414 \u2013 420, 2019.[122] Y. Dobashi, \u201cCharacterization of ionic polymers: towards applications as soft sensorsin medicine,\u201d MASc, University of British Columbia, 2016.[123] K. Paduszyn\u00b4ski and U. Doman\u00b4ska, \u201cViscosity of ionic liquids: An extensive databaseand a new group contribution model based on a feed-forward artificial neural net-work,\u201d Journal of Chemical Information and Modeling, vol. 54, no. 5, pp. 1311 \u20131324, 2014.[124] K. Dokko, D. Watanabe, Y. Ugata, M. L. Thomas, S. Tsuzuki, W. Shinoda,K. Hashimoto, K. Ueno, Y. Umebayashi, and M. Watanabe, \u201cDirect evidence for Liion hopping conduction in highly concentrated sulfolane-based liquid electrolytes,\u201dThe Journal of Physical Chemistry B, vol. 122, no. 47, pp. 10736 \u2013 10745, 2018.162Bibliography[125] P. Shi, M. Lin, H. Zheng, X. He, Z. Xue, H. Xiang, and C. Chen, \u201cEffect of propylenecarbonate-Li+ solvation structures on graphite exfoliation and its application in Li-ion batteries,\u201d Electrochimica Acta, vol. 247, pp. 12 \u2013 18, 2017.[126] Z. Wang, W. Gao, X. Huang, Y. Mo, and L. Chen, \u201cSpectroscopic studies on inter-actions and microstructures in propylene carbonate\u2013LiTFSI electrolytes,\u201d Journal ofRaman Spectroscopy, vol. 32, no. 11, pp. 900 \u2013 905, 2001.[127] A. Martinelli, A. Matic, P. Jacobsson, L. Bo\u00a8 rjesson, A. Fernicola,and B. Scrosati, \u201cPhase behavior and ionic conductivity in lithiumbis(trifluoromethanesulfonyl)imide-doped ionic liquids of the pyrrolidinium cationand bis(trifluoromethanesulfonyl)imide anion,\u201d The Journal of Physical ChemistryB, vol. 113, no. 32, pp. 11247 \u2013 11251, 2009.[128] P. Meister, V. Siozios, J. Reiter, S. Klamor, S. Rothermel, O. Fromm, H.-W. Meyer,M. Winter, and T. Placke, \u201cDual-ion cells based on the electrochemical intercalationof asymmetric fluorosulfonyl-(trifluoromethanesulfonyl) imide anions into graphite,\u201dElectrochimica Acta, vol. 130, pp. 625 \u2013 633, 2014.[129] X. Qi, B. Blizanac, A. DuPasquier, P. Meister, T. Placke, M. Oljaca, J. Li, andM. Winter, \u201cInvestigation of PF6\u2013 and TFSI\u2013 anion intercalation into graphitizedcarbon blacks and its influence on high voltage lithium ion batteries,\u201d Phys. Chem.Chem. Phys., vol. 16, pp. 25306 \u2013 25313, 2014.[130] R.-S. Ku\u00a8hnel and A. Balducci, \u201cLithium ion transport and solvation in N -butyl-N -methylpyrrolidinium bis(trifluoromethanesulfonyl)imidepropylene carbonate mix-tures,\u201d The Journal of Physical Chemistry C, vol. 118, no. 11, pp. 5742 \u2013 5748,2014.163Bibliography[131] Y. Ikezawa and T. Ariga, \u201cIn situ FTIR spectra at the Cu electrode\/propylenecarbonate solution interface,\u201d Electrochimica Acta, vol. 52, no. 7, pp. 2710 \u2013 2715,2007.[132] E. Gileadi and E. Kirowa-Eisner, \u201cElectrolytic conductivity\u2013the hopping mechanismof the proton and beyond,\u201d Electrochimica Acta, vol. 51, no. 27, pp. 6003 \u2013 6011,2006.[133] O. Markovitch and N. Agmon, \u201cStructure and energetics of the hydronium hydrationshells,\u201d The Journal of Physical Chemistry A, vol. 111, no. 12, pp. 2253 \u2013 2256, 2007.[134] M. Park, X. Zhang, M. Chung, G. B. Less, and A. M. Sastry, \u201cA review of conductionphenomena in Li-ion batteries,\u201d Journal of Power Sources, vol. 195, no. 24, pp. 7904\u2013 7929, 2010.[135] M. Schumaker and R. MacKinnon, \u201cA simple model for multi\u2013ion permeation. Single-vacancy conduction in a simple pore model,\u201d Biophysical Journal, vol. 58, no. 4,pp. 975 \u2013 984, 1990.[136] K.-D. Kreuer, \u201cProton conductivity: Materials and applications,\u201d Chemistry of Ma-terials, vol. 8, no. 3, pp. 610 \u2013 641, 1996.[137] Y. Noda, K. Nakano, H. Takeda, M. Kotobuki, L. Lu, and M. Nakayama, \u201cCom-putational and experimental investigation of the electrochemical stability and Li-ionconduction mechanism of LiZr2(PO4)3,\u201d Chemistry of Materials, vol. 29, no. 21,pp. 8983 \u2013 8991, 2017.[138] O. Gerbig, R. Merkle, and J. Maier, \u201cElectron and ion transport in Li2O2,\u201d AdvancedMaterials, vol. 25, no. 22, pp. 3129 \u2013 3133, 2013.164Bibliography[139] H. Mehrer, Diffusion in Solids: Fundamentals, Methods, Materials, Diffusion-Controlled Processes. Springer Series in Solid-State Sciences, Springer Berlin Hei-delberg, 2007.[140] Y. Ugata, M. L. Thomas, T. Mandai, K. Ueno, K. Dokko, and M. Watanabe, \u201cLi-ionhopping conduction in highly concentrated lithium bis(fluorosulfonyl)amide\/dinitrileliquid electrolytes,\u201d Phys. Chem. Chem. Phys., vol. 21, pp. 9759 \u2013 9768, 2019.[141] Y.-S. Hu, H. Li, X. Huang, and L. Chen, \u201cNovel room temperature molten saltelectrolyte based on LiTFSI and acetamide for lithium batteries,\u201d ElectrochemistryCommunications, vol. 6, pp. 28 \u2013 32, Jan. 2004.[142] S. Campbell, C. Bowes, and R. McMillan, \u201cThe electrochemical behaviour of tetrahy-drofuran and propylene carbonate without added electrolyte,\u201d Journal of Electroan-alytical Chemistry and Interfacial Electrochemistry, vol. 284, no. 1, pp. 195 \u2013 204,1990.[143] A. Peters and N. R. Branda, \u201cElectrochromism in photochromic dithienylcy-clopentenes,\u201d Journal of the American Chemical Society, vol. 125, no. 12, pp. 3404\u2013 3405, 2003.[144] N. Branda, G. Brenmer, J. Finden, S. Gauthier, B. Gillon, A. Koutsandreas,V. Marshman, M. Pilapil, J. Sargent, J. Daniel, et al., \u201cSwitching materials, andcompositions and methods for making same,\u201d 2013. Patent WO2013152425 A1.[145] A. Peters, R. McDonald, and N. R. Branda, \u201cRegulating pi-piconjugated pathwaysusing a photochromic 1,2-dithienylcyclopentene,\u201d Chem. Commun., pp. 2274 \u2013 2275,2002.165Bibliography[146] J. E. D. Bene, P. F. Provasi, I. Alkorta, and J. Elguero, \u201cResolving an apparentdiscrepancy between theory and experiment: spin-spin coupling constants for FCCF,\u201dMagnetic Resonance in Chemistry, vol. 46, no. 11, pp. 1003 \u2013 1006, 2008.[147] M. D. Mohammadi and M. Hamzehloo, \u201cDensities, viscosities, and refractive indicesof binary and ternary mixtures of methanol, acetone, and chloroform at temperaturesfrom (298.15 - 318.15) K and ambient pressure,\u201d Fluid Phase Equilibria, vol. 483,pp. 14 \u2013 30, 2019.[148] L. H. Hess and A. Balducci, \u201c1,2-butylene carbonate as solvent for EDLCs,\u201d Elec-trochimica Acta, vol. 281, pp. 437 \u2013 444, 2018.[149] J. Mazurkiewicz and P. Tomasik, \u201cEffect of external electric field upon charge dis-tribution, energy and dipole moment of selected monosaccharide molecules,\u201d NaturalScience, vol. 04, Jan. 2012.[150] A. Ghoufi, K. Benhamed, L. Boukli-Hacene, and G. Maurin, \u201cElectrically inducedbreathing of the MIL-53(Cr) metal-organic framework,\u201d ACS Central Science, vol. 3,no. 5, pp. 394 \u2013 398, 2017.[151] S. Namsani and A. O. Yazaydin, \u201cElectric field induced rotation of halogenatedorganic linkers in isoreticular metalorganic frameworks for nanofluidic applications,\u201dMol. Syst. Des. Eng., vol. 3, pp. 951 \u2013 958, 2018.[152] A. Hirao and H. Nishizawa, \u201cEffect of dipoles on carrier drift and diffusion of molec-ularly doped polymers,\u201d Phys. Rev. B, vol. 56, pp. R2904 \u2013 R2907, Aug. 1997.[153] S. H. Northrup, J. D. Smith, J. O. Boles, and J. C. L. Reynolds, \u201cThe effect of dipolemoment on diffusion controlled bimolecular reaction rates,\u201d The Journal of ChemicalPhysics, vol. 84, no. 10, pp. 5536 \u2013 5544, 1986.166Bibliography[154] T. C. Chan, H. T. Li, and K. Y. Li, \u201cEffects of shapes of solute molecules on diffusion:A study of dependences on solute size, solvent, and temperature,\u201d The Journal ofPhysical Chemistry B, vol. 119, no. 51, pp. 15718 \u2013 15728, 2015.[155] L. Lebreton, B. Slat, F. Ferrari, B. Sainte-Rose, J. Aitken, R. Marthouse, S. Hajbane,S. Cunsolo, A. Schwarz, A. Levivier, K. Noble, P. Debeljak, H. Maral, R. Schoeneich-Argent, R. Brambini, and J. Reisser, \u201cEvidence that the great pacific garbage patchis rapidly accumulating plastic,\u201d Scientific Reports, vol. 8, no. 1, p. 4666, 2018.[156] R. Geyer, J. R. Jambeck, and K. L. Law, \u201cProduction, use, and fate of all plasticsever made,\u201d Science Advances, vol. 3, no. 7, 2017.[157] D. Montarnal, M. Capelot, F. Tournilhac, and L. Leibler, \u201cSilica-like malleable ma-terials from permanent organic networks,\u201d Science, vol. 334, no. 6058, pp. 965 \u2013 968,2011.[158] M. Capelot, D. Montarnal, F. Tournilhac, and L. Leibler, \u201cMetal-catalyzed trans-esterification for healing and assembling of thermosets,\u201d Journal of the AmericanChemical Society, vol. 134, no. 18, pp. 7664 \u2013 7667, 2012.[159] Y. Yang, S. Zhang, X. Zhang, L. Gao, Y. Wei, and Y. Ji, \u201cDetecting topology freezingtransition temperature of vitrimers by AIE luminogens,\u201d Nature Communications,vol. 10, no. 1, p. 3165, 2019.[160] W. Denissen, M. Droesbeke, R. Nicolay\u00a8, L. Leibler, J. M. Winne, and F. E. Du Prez,\u201cChemical control of the viscoelastic properties of vinylogous urethane vitrimers,\u201dNature communications, vol. 8, pp. 1 \u2013 7, Mar. 2017.167Bibliography[161] J. Peng, X. He, K. Wang, W. Tan, Y. Wang, and Y. Liu, \u201cNoninvasive monitoring ofintracellular ph change induced by drug stimulation using silica nanoparticle sensors,\u201dAnalytical and Bioanalytical Chemistry, vol. 388, no. 3, pp. 645 \u2013 654, 2007.[162] T. Stukenbroeker, W. Wang, J. M. Winne, F. E. Du Prez, R. Nicolay\u00a8, and L. Leibler,\u201cPolydimethylsiloxane quenchable vitrimers,\u201d Polym. Chem., vol. 8, pp. 6590 \u2013 6593,2017.[163] R. Baruah, A. Kumar, R. R. Ujjwal, S. Kedia, A. Ranjan, and U. Ojha, \u201cRecy-clable thermosets based on dynamic amidation and aza-Michael addition chemistry,\u201dMacromolecules, vol. 49, no. 20, pp. 7814 \u2013 7824, 2016.[164] K. S. Kwan, C. N. Subramaniam, and T. C. Ward, \u201cEffect of penetrant size andshape on its transport through a thermoset adhesive: I. n-alkanes,\u201d Polymer, vol. 44,no. 10, pp. 3061 \u2013 3069, 2003.[165] L. Rovigatti, G. Nava, T. Bellini, and F. Sciortino, \u201cSelf-dynamics and collectiveswap-driven dynamics in a particle model for vitrimers,\u201d Macromolecules, vol. 51,no. 3, pp. 1232 \u2013 1241, 2018.[166] R. G. C. Arridge, An Introduction to Polymer Mechanics. Taylor and Francis, 1985.[167] S. Blonski, W. Brostow, and J. Kubat, \u201cMolecular dynamics simulations of stressrelaxation in metals and polymers,\u201d Phys. Rev. B, vol. 49, pp. 6494 \u2013 6500, 1994.[168] G. Zhu, H. Rezvantalab, E. Hajizadeh, X. Wang, and R. G. Larson, \u201cStress-gradient-induced polymer migration: Perturbation theory and comparisons to stochastic sim-ulations,\u201d Journal of Rheology, vol. 60, no. 2, pp. 327 \u2013 343, 2016.168[169] W. Hong, X. Zhao, J. Zhou, and Z. Suo, \u201cA theory of coupled diffusion and largedeformation in polymeric gels,\u201d Journal of the Mechanics and Physics of Solids,vol. 56, no. 5, pp. 1779 \u2013 1793, 2008.[170] J. R. White, \u201cOn internal stress and activation volume in polymers,\u201d Journal ofMaterials Science, vol. 16, pp. 3249 \u2013 3262, Dec. 1981.[171] W. Brostow and R. Corneliussen, Failure of Plastics. Hanser Pub., 1986.[172] M. J. C. Plancha, Characterisation and modelling of multivalent polymer electrolytes.Woodhead Publishing, 2010.[173] F. Rittig, J. Ka\u00a8rger, C. M. Papadakis, G. Fleischer, P. S\u00b4te\u02c7pa\u00b4nek, and K. Almdal,\u201cSelf-diffusion investigations on a series of PEP-PDMS diblock copolymers with dif-ferent morphologies by pulsed field gradient NMR,\u201d Phys. Chem. Chem. Phys., vol. 1,pp. 3923 \u2013 3931, 1999.[174] W. H. Binder, \u201cThe labile chemical bond: A perspective on mechanochemistry inpolymers,\u201d Polymer, vol. 202, p. 122639, 2020.[175] X. Hu, T. Zeng, C. C. Husic, and M. J. Robb, \u201cMechanically triggered small moleculerelease from a masked furfuryl carbonate,\u201d Journal of the American Chemical Society,vol. 141, no. 38, pp. 15018 \u2013 15023, 2019. PMID: 31515994.169Appendix AVitrimer Sample PreparationVitrimers samples were prepared by Taylor D. Wright of Dr. Michael Wolf\u2019s lab at UBCin order to measure the diffusion of NMR visible tag molecules that are attached to thecrosslinker of the vitrimer samples. The preparation involved four main steps:1. Preparation of an exchangeable NMR visible tags 2,2,2-Trifluoroethylacetoacetate.Figure A.1: 2,2,2-Trifluoroethylacetoacetate.Trifluoroethanol (3.7 mL, 50 mmol), m-xylene, and 2,2,6-Trimethyl-1,3-dioxin-4-one (4.7mL, 33 mmol) were added to an open vial and heated to 150 \u25e6C while stirring. The reactionwas kept at 150 \u25e6C for thirty minutes, and then cooled to room temperature. The mixturewas transferred to a 50 mL round-bottom flask and dried under vacuum. The crude oilwas purified using column chromatography on silica gel with 3:1 petroleum ether to diethylether as the mobile phase.2. Preparation of vinylogous urethane crosslinker (VitCL).Figure A.2: 1,12-dodecane-bis-\u03b2-ketoenamine n-butyl170Appendix A. Vitrimer Sample Preparation1,12-dodecane-bis-\u03b2-ketoenamine n-butyl was prepared by condensing 1,12-dodecanebisac-etoacetate [5] with n-butylamine. 1 g, 2.7 mmol and 0.8 mL, 8.1 mmol, respectively, wereadded to 7 mL of THF in a 25 mL round bottom flask along with an excess of anhydrousmagnesium sulfate. The system was equipped with a magnetic stir bar and condenser andallowed to reflux for three hours. The mixture was cooled, vacuum filtered, and rinsedwith DCM. The eluent was dried under vacuum and then extracted using DCM and water.The organic layer was dried over anhydrous magnesium sulfate, filtered, and then dried toa pale clear oil.3. Preparation of an exchangeable coloured, NMR visible tags.Figure A.3: 4-trifluoromethyl-4\u2019-methylacetoacetate azobenzene4-trifluoromethyl-4\u2019-methylacetoacetate azobenzene an F labelled, blue light-absorbing dyewas prepared by a three-step synthesis with a starting material 3-(Trifluoromethyl)aniline.A schematics of this sythesis can be seen in scheme A.1. 3-(Trifluoromethyl)aniline wasreacted with 1 molar equivalent of Oxone to give the nitroso compound. To this, 0.9equivalents of 4-aminobenzyl alcohol was added under acidic conditions and reacted atroom temperature overnight. The resulting azobenzene molecule was purified, and thenreacted with 1.1 molar equivalents of the acetone diketene adduct in m-xylenes at 130 \u25e6Cfor 30 minutes. The product was purified using silica gel chromatography to afford thefluorine labellled acetoacetate azobenzene.4. Preparation of polymer network.PDMS-NH2 (1 g, 0.02 mmol), 38 mg of exchangeable vinylogous urethane crosslinker or171Appendix A. Vitrimer Sample PreparationScheme A.1: Synthesis steps for the preparation of fluorine labelled acetoacetateazobenzene. This molecule functions as an exchangeable tag that is both NMR visible(due to the presence of a CF3 group) and is both visible to the naked eye (due to theazobenzene group blue light absorbing properties). Synthesis steps were planned andperformed by collaborator Taylor D. Wright of UBC.Figure A.4: PDMS-NH2 polymer structure23 mg of triethylene glycol dimethacrylate (0.08 mmol, 4 eq) or a non-exchangeatriethyleneglycol dimethacrylate, p-toluenesulfonic acid (0.67 mg, 0.5 mol relative to moles of NH2, asa dilute solution in THF), 5 molar equivalents of an NMR visible tag, and 2.5 mL of THFwere combined in a 20 mL vial and stirred using a magnetic stir bar for 5 minutes. Themixture was found to gel rapidly if the bisacetoacetate was used directly, and as such thevinylogous urethane VitCL compound was used. The vial was degassed using a sonicatorfor 10 minutes, and the solution transferred to a silicone mold inside of a glass desiccator.The system was placed under N2, covered in aluminum foil, and heated to 60\u25e6C using anoil bath until the THF had fully evaporated. The system was then heated at 100 \u25e6C for 18hours under N2.Polymer samples were removed from the warm molds and stored in a refrigerator whennot in use.172","attrs":{"lang":"en","ns":"http:\/\/www.w3.org\/2009\/08\/skos-reference\/skos.html#note","classmap":"oc:AnnotationContainer"},"iri":"http:\/\/www.w3.org\/2009\/08\/skos-reference\/skos.html#note","explain":"Simple Knowledge Organisation System; Notes are used to provide information relating to SKOS concepts. There is no restriction on the nature of this information, e.g., it could be plain text, hypertext, or an image; it could be a definition, information about the scope of a concept, editorial information, or any other type of information."}],"Genre":[{"label":"Genre","value":"Thesis\/Dissertation","attrs":{"lang":"en","ns":"http:\/\/www.europeana.eu\/schemas\/edm\/hasType","classmap":"dpla:SourceResource","property":"edm:hasType"},"iri":"http:\/\/www.europeana.eu\/schemas\/edm\/hasType","explain":"A Europeana Data Model Property; This property relates a resource with the concepts it belongs to in a suitable type system such as MIME or any thesaurus that captures categories of objects in a given field. It does NOT capture aboutness"}],"GraduationDate":[{"label":"Graduation Date","value":"2020-11","attrs":{"lang":"en","ns":"http:\/\/vivoweb.org\/ontology\/core#dateIssued","classmap":"vivo:DateTimeValue","property":"vivo:dateIssued"},"iri":"http:\/\/vivoweb.org\/ontology\/core#dateIssued","explain":"VIVO-ISF Ontology V1.6 Property; Date Optional Time Value, DateTime+Timezone Preferred "}],"IsShownAt":[{"label":"DOI","value":"10.14288\/1.0392849","attrs":{"lang":"en","ns":"http:\/\/www.europeana.eu\/schemas\/edm\/isShownAt","classmap":"edm:WebResource","property":"edm:isShownAt"},"iri":"http:\/\/www.europeana.eu\/schemas\/edm\/isShownAt","explain":"A Europeana Data Model Property; An unambiguous URL reference to the digital object on the provider\u2019s website in its full information context."}],"Language":[{"label":"Language","value":"eng","attrs":{"lang":"en","ns":"http:\/\/purl.org\/dc\/terms\/language","classmap":"dpla:SourceResource","property":"dcterms:language"},"iri":"http:\/\/purl.org\/dc\/terms\/language","explain":"A Dublin Core Terms Property; A language of the resource.; Recommended best practice is to use a controlled vocabulary such as RFC 4646 [RFC4646]."}],"Program":[{"label":"Program (Theses)","value":"Chemistry","attrs":{"lang":"en","ns":"https:\/\/open.library.ubc.ca\/terms#degreeDiscipline","classmap":"oc:ThesisDescription","property":"oc:degreeDiscipline"},"iri":"https:\/\/open.library.ubc.ca\/terms#degreeDiscipline","explain":"UBC Open Collections Metadata Components; Local Field; Indicates the program for which the degree was granted."}],"Provider":[{"label":"Provider","value":"Vancouver : University of British Columbia Library","attrs":{"lang":"en","ns":"http:\/\/www.europeana.eu\/schemas\/edm\/provider","classmap":"ore:Aggregation","property":"edm:provider"},"iri":"http:\/\/www.europeana.eu\/schemas\/edm\/provider","explain":"A Europeana Data Model Property; The name or identifier of the organization who delivers data directly to an aggregation service (e.g. Europeana)"}],"Publisher":[{"label":"Publisher","value":"University of British Columbia","attrs":{"lang":"en","ns":"http:\/\/purl.org\/dc\/terms\/publisher","classmap":"dpla:SourceResource","property":"dcterms:publisher"},"iri":"http:\/\/purl.org\/dc\/terms\/publisher","explain":"A Dublin Core Terms Property; An entity responsible for making the resource available.; Examples of a Publisher include a person, an organization, or a service."}],"Rights":[{"label":"Rights","value":"Attribution-ShareAlike 4.0 International","attrs":{"lang":"*","ns":"http:\/\/purl.org\/dc\/terms\/rights","classmap":"edm:WebResource","property":"dcterms:rights"},"iri":"http:\/\/purl.org\/dc\/terms\/rights","explain":"A Dublin Core Terms Property; Information about rights held in and over the resource.; Typically, rights information includes a statement about various property rights associated with the resource, including intellectual property rights."}],"RightsURI":[{"label":"Rights URI","value":"http:\/\/creativecommons.org\/licenses\/by-sa\/4.0\/","attrs":{"lang":"*","ns":"https:\/\/open.library.ubc.ca\/terms#rightsURI","classmap":"oc:PublicationDescription","property":"oc:rightsURI"},"iri":"https:\/\/open.library.ubc.ca\/terms#rightsURI","explain":"UBC Open Collections Metadata Components; Local Field; Indicates the Creative Commons license url."}],"ScholarlyLevel":[{"label":"Scholarly Level","value":"Graduate","attrs":{"lang":"en","ns":"https:\/\/open.library.ubc.ca\/terms#scholarLevel","classmap":"oc:PublicationDescription","property":"oc:scholarLevel"},"iri":"https:\/\/open.library.ubc.ca\/terms#scholarLevel","explain":"UBC Open Collections Metadata Components; Local Field; Identifies the scholarly level of the author(s)\/creator(s)."}],"Title":[{"label":"Title ","value":"Transport in ion dense media and recyclable polymers","attrs":{"lang":"en","ns":"http:\/\/purl.org\/dc\/terms\/title","classmap":"dpla:SourceResource","property":"dcterms:title"},"iri":"http:\/\/purl.org\/dc\/terms\/title","explain":"A Dublin Core Terms Property; The name given to the resource."}],"Type":[{"label":"Type","value":"Text","attrs":{"lang":"en","ns":"http:\/\/purl.org\/dc\/terms\/type","classmap":"dpla:SourceResource","property":"dcterms:type"},"iri":"http:\/\/purl.org\/dc\/terms\/type","explain":"A Dublin Core Terms Property; The nature or genre of the resource.; Recommended best practice is to use a controlled vocabulary such as the DCMI Type Vocabulary [DCMITYPE]. To describe the file format, physical medium, or dimensions of the resource, use the Format element."}],"URI":[{"label":"URI","value":"http:\/\/hdl.handle.net\/2429\/75578","attrs":{"lang":"en","ns":"https:\/\/open.library.ubc.ca\/terms#identifierURI","classmap":"oc:PublicationDescription","property":"oc:identifierURI"},"iri":"https:\/\/open.library.ubc.ca\/terms#identifierURI","explain":"UBC Open Collections Metadata Components; Local Field; Indicates the handle for item record."}],"SortDate":[{"label":"Sort Date","value":"2020-12-31 AD","attrs":{"lang":"en","ns":"http:\/\/purl.org\/dc\/terms\/date","classmap":"oc:InternalResource","property":"dcterms:date"},"iri":"http:\/\/purl.org\/dc\/terms\/date","explain":"A Dublin Core Elements Property; A point or period of time associated with an event in the lifecycle of the resource.; Date may be used to express temporal information at any level of granularity. Recommended best practice is to use an encoding scheme, such as the W3CDTF profile of ISO 8601 [W3CDTF].; A point or period of time associated with an event in the lifecycle of the resource.; Date may be used to express temporal information at any level of granularity. Recommended best practice is to use an encoding scheme, such as the W3CDTF profile of ISO 8601 [W3CDTF]."}]}