Subramanian Udel 006

OPERANDO LIQUID-CELL ELECTRON MICROSCOPY OF THE

ELECTROCHEMICAL POLYMERIZATION OF

BEAM-SENSITIVE CONJUGATED POLYMERS

Vivek Subramanian

A dissertation submitted to the Faculty of the University of Delaware in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Materials Science and Engineering

Fall 2020

OPERANDO LIQUID-CELL ELECTRON MICROSCOPY OF THE

ELECTROCHEMICAL POLYMERIZATION OF

BEAM-SENSITIVE CONJUGATED POLYMERS

Vivek Subramanian

Approved: ______Darrin J. Pochan, Ph.D. Chair of the Department of Materials Science and Engineering

Approved: ______Levi T. Thompson, Ph.D. Dean of the College of Engineering

Approved: Louis F. Rossi, Ph.D. Vice Provost for Graduate & Professional Education and Dean of the Graduate College

I certify that I have read this dissertation and that in my opinion it meets the academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy.

Signed: David C. Martin, Ph.D. Professor in charge of dissertation

Signed: ______Darrin J. Pochan, Ph.D. Member of dissertation committee

Signed: ______Chaoying Ni, Ph.D. Member of dissertation committee

Signed: ______Karl S. Booksh, Ph.D. Member of dissertation committee ACKNOWLEDGEMENTS

First and foremost, I would like to thank you Dr. Martin for helping me evolve as a scientist during my time in the lab. I think you deserve this whole paragraph to yourself. I am grateful for your unwavering support and guidance throughout my Ph.D. Thanks for always having the answers to all my questions especially when I started out. I still vividly remember the days when you came with me to the Talos room to teach me low-dose electron microscopy. Those moments have made my grad school life much more unique and memorable. Honestly, I would not have been able to achieve whatever I have until this point without your continuous support throughout my Ph.D. In the future, I truly hope to see PEDOT being used in our bodies someday (maybe Elon Musk would help us do that) and that your relentless quest for playing with “expensive toys” never ends. Next, this thesis would definitely not have been complete without the consistent feedback from my Ph.D. committee. Dr. Ni, Dr. Pochan and Dr. Booksh (and Dr. Kloxin for being on my qualifying committee), a special thank you to all of you for being available from time to time and for coming up with valuable suggestions about my work. Those suggestions and questions challenged me and made me think about my dissertation from a totally different perspective. Jen, I also owe you a debt of gratitude for all the help you have done. After the senior students graduated, you were a great mentor to me. Thanks for the time you

iv invested in training me on the TEM and for all the suggestions and inputs you had in general on my experiments. Yong, thanks to you as well for helping me understand SEM and FIB. I would also like to thank my current lab members Shrirang, Peter, Quintin, Junghyun, Dr. Samadhan Nagane and Yuhang for their support. Likewise, I would also like to extend a vote of thanks to my previous group members Dr. Jinglin Liu, Dr. Bin Wei, Dr. Jing Qu and Dr. Chin-Chen Kuo for guiding me during the initial years of my Ph.D. Furthermore, Prof. Glenn Yap and Casey, thank you to both of you as well for teaching me the art of single crystal growth and of course for all the help with single crystal x-ray diffraction. Finally, this would not have been possible without the continuous emotional and moral support of my parents, relatives and friends both here in the USA, India and residing elsewhere in the world. Together, you all have made this journey so much fun for me and full of growth, happiness and scientific discoveries.

TABLE OF CONTENTS

LIST OF TABLES ...... x LIST OF FIGURES ...... xi ABSTRACT ...... xviii

Chapter

1 INTRODUCTION ...... 1

1.1 Motivation: Elucidating the mechanistic details of the electrochemical polymerization reaction of poly(3,4-thylenedioxythiophene)(PEDOT) ... 1 1.2 The Technique: Operando Liquid-cell Transmission Electron Microscopy ...... 2 1.3 Organization of the chapters ...... 4

1.3.1 Chapter 1: ...... 4 1.3.2 Chapter 2: ...... 4 1.3.3 Chapter 3: ...... 5 1.3.4 Chapter 4: ...... 5 1.3.5 Chapter 5: ...... 6 1.3.6 Chapter 6: ...... 6

REFERENCES ...... 8

2 MOLECULAR MOVIES: A REVIEW OF RECENT WORK, CURRENT CHALLENGES AND FUTURE OPPORTUNITIES IN LIQUID-PHASE TRANSMISSION ELECTRON MICROSCOPY (LPTEM) OF BEAM- SENSITIVE ORGANIC MATERIALS ...... 11

2.1 Introduction ...... 11 2.2 Considerations and challenges during imaging beam-sensitive materials in LPTEM ...... 14

2.2.1 Electron-water interactions ...... 14 2.2.2 Beam-specimen interactions ...... 16

2.2.3 The “bowing effect” of viewing windows ...... 19 2.2.4 Gas bubbles in the feed solution ...... 21 2.2.5 Dewetting due to hydrophobicity ...... 22 2.2.6 Irregular flow ...... 22 2.2.7 Finding the right focal plane ...... 23 2.2.8 Voltage spikes in the potentiostat ...... 24

2.3 Recent applications of LPTEM in Materials Science ...... 24

2.3.1 Soft materials ...... 24 2.3.2 Crystal Growth ...... 31 2.3.3 Batteries ...... 33 2.3.4 Metal Organic Frameworks (MOFs) ...... 36 2.3.5 Nanowires ...... 38 2.3.6 Nanoparticles ...... 39

2.4 Future Opportunities ...... 39

2.4.1 Machine Learning ...... 39 2.4.2 Vibrational spectroscopy in liquid-phase electron microscopy ... 40

REFERENCES ...... 42

3 IN-SITU TRANSMISSION ELECTRON MICROSCOPY (TEM) OF THE ELECTROCHEMICALLY-DRIVEN NUCLEATION, GROWTH, AND SOLIDIFICATION OF POLY(3,4-ETHYLENEDIOXYTHIOPHENE) (PEDOT) ...... 54

3.1 Introduction ...... 54 3.2 Materials and Methods ...... 58

3.2.1 Monomer solution: ...... 58 3.2.2 Liquid flow cell and electrochemistry chips: ...... 58 3.2.3 Transmission Electron Microscopy: ...... 59 3.2.4 Electrochemistry: ...... 59 3.2.5 Video recording: ...... 59 3.2.6 Optical Microscopy: ...... 59 3.2.7 Image analysis: ...... 59

3.3 Results and discussion ...... 60

3.3.1 Nucleation and growth of PEDOT ...... 60 3.3.2 Role of nucleation in determining the thicknesses ...... 65 3.3.3 Transitions from liquid-like oligomers to solid-like polymers.... 70

vii

3.3.4 Mechanistic details involved in the evolution of shapes and sizes during the formation of solid-like films and liquid-like clusters ...... 77 3.3.5 “Pinned-edge” mechanism of liquid-like droplets transforming into solid-like films ...... 80

3.4 Conclusions ...... 85

REFERENCES ...... 87

4 QUANTITATIVE ANALYSIS OF NANO-FIBRIL GROWTH DURING THE ELECTROCHEMICAL POLYMERIZATION OF PEDOT BY LIQUID CELL TEM ...... 94

4.1 Introduction ...... 94 4.2 Materials and Methods ...... 96 4.3 Results and discussion ...... 98 4.4 Conclusions ...... 112

REFERENCES ...... 115

5 MORPHOLOGY, MOLECULAR ORIENTATION AND SOLID-STATE CHARACTERIZATION OF 2,3-DIHYDROTHIENO[3,4-B] [1,4]DIOXINE-2-CARBOXYLIC ACID (EDOTACID) ...... 120

5.1 Introduction ...... 120 5.2 Experimental section ...... 123

5.2.1 Materials used...... 123 5.2.2 Single crystal growth...... 123 5.2.3 Single-crystal X-ray diffraction...... 124 5.2.4 Structure and Morphology Characterizations...... 124 5.2.5 Physical properties ...... 125 5.2.6 Molecular Simulations...... 126

5.3 Results and discussion ...... 126 5.4 Conclusions: ...... 137

REFERENCES ...... 139

6 CONCLUSIONS AND PROPOSED FUTURE STUDIES ...... 144

6.1 Conclusions ...... 144 6.2 Proposed Future Studies ...... 146

viii

6.2.1 Investigating the early-stage nucleation and growth of copolymers of Poly (EDOT-co-EDOT-acid) ...... 146 6.2.2 Comparing differences between the structure of semi- crystalline fibrillar poly (5,6-dimethoxyindole-2-carboxylic acid) (PDMICA) and fibrillar PEDOT with comparatively less order...... 148 6.2.3 Influence of nanoparticles on the nucleation and growth of PEDOT ...... 152

REFERENCES ...... 154

Appendix

A ADDITIONAL FIGURES ...... 155 B RIGHTS AND PERMISSIONS ...... 158

LIST OF TABLES

Table 3.1: Specific metrics that can be used to differentiate between liquid-like droplets and solid-like regions ...... 76

Table 5.1 Unit cell parameters of 2,3-dihydrothieno[3,4-b][1,4]dioxine-2- carboxylic acid...... 127

LIST OF FIGURES

Figure 2.1: The liquid to be imaged encapsulated between graphene sheets (adapted from Yuk et al. 2012)...... 11

Figure 2.2: Schematic capturing the central idea behind contemporary commercially available TEM sample holders. The liquid cell is held in the electron beam between two ultrathin silicon nitride membranes. In an electrochemical version of such a cell, an electrode is used to drive a chemical reaction...... 12

Figure 2.3: Schematics of devices used for imaging in liquids. (a) Regular device with a single rectangular viewing window. (b) Echips placed one on top of each other with the SiN windows perpendicular to each other. (c)The novel design as proposed and fabricated by Moser and colleagues which has multiple windows. (d) The Echips with multiple windows fabricated by Moser et.al placed on top of each other thus creating an array of viewing areas. (e) Echips similar to (c), but equipped with additional gold bars for use as focusing aids (f) Devices from (e) placed one on top of each other thus generating an array of viewing windows equipped with focusing aids(figure adapted from Moser et.al)...... 16

Figure 2.4: Phase contrast TEM images, corresponding FFTs and line profiles of individual spatial frequencies as a function of the electron dose(Keskin and de Jonge 2018)...... 17

Figure 2.5: Schematic of bowing or bulging of Silicon Nitride membranes under the influence of vacuum inside the TEM...... 19

Figure 2.6: LPTEM images of amorphous SiO2 nanoparticles at an internal pressure of (a)1bar; (b) 0.05 bar; (c) FFT of (a) and (b); (d) Radially averaged pixel intensities of the FFT...... 21

Figure 2.7: Schematic of chips equipped with flow (left) and static spacers (right). .. 22

Figure 2.8: Schematic and BFTEM image of transition states involved in micelle fragmentation in liquid (Early, Yager, and Lodge 2020)...... 24

Figure 2.9: Beam damage as observed by Lodge and colleagues (Early, Yager, and Lodge 2020) during their micelle fragmentation experiments...... 25

Figure 2.10: LPTEM images of butyl acrylate (BA) / Methyl Methacrylate (MMA) latex particles in liquid (L.Liu et.al, 2015)...... 26

Figure 2.11: Direct imaging of early stage electrodeposition of Poly(3,4- ethylenedioxythiophene) on glassy carbon working electrode using LPTEM (J. Liu et.al, 2015)...... 27

Figure 2.12: (a)Individual higher molecular weight clusters of PEDOT nucleating and growing from the working electrode as the electropolymerization reaction occurs. (b-g):Sequential Bright-field TEM images showing merging of oligomeric clusters followed by their deposition onto the working electrode and as observed during the electrodeposition of PEDOT on glassy carbon working electrode taken at (b)180s (c) 190s (d) 200s (e) 210s (f) 220s (g) 230s...... 28

Figure 2.13: (a-f) Bright-field TEM images that are part of a video taken during the initial, intermediate and later stages showing the nucleation of individual fibrils during the electrodeposition of PEDOT with a nanofibrillar morphology (g): Plot of average length, width and thickness profiles of 25 nanofibrils...... 30

Figure 2.14: Shape evolution of Pt nanocrystal as observed by Yuk and colleagues (Yuk et.al, 2012)...... 32

Figure 2.15: Crystals of flufenamic acid observed using LPTEM by Cookman et.al (Cookman et.al 2020)...... 33

Figure 2.16: High-angle annular dark field (HAADF) images of the charging and discharging cycles of lithium battery (Mehdi et.al 2015)...... 35

Figure 2.17: (a)Schematic of the experimental setup demonstrating electron diffraction tomography (EDT) of the specimen in the liquid cell (b) Reconstructed reciprocal lattice from EDT (Karakulina et.al 2018)...... 36

Figure 2.18: Snapshots of Metal Organic Nanotubes (MONTs) grown in LPTEM as reported by Vailonis et.al...... 36

Figure 2.19: Beam induced damage of MONTs as observed by Vailonis et.al...... 37

xii

Figure 2.20: (a)LPTEM images of the nanowire growth (b) Their growth rates as a function of time (c) The number of times their growth direction changes as a function of the growth rate...... 38

Figure 3.1: Schematic of a typical 3 - electrode system utilized for electrodeposition) conjugated polymers...... 55

Figure 3.2: The monomers first oxidize to form reactive radical cation species which then react to form dimers, trimers and higher order oligomers before finally depositing as the solid polymer product on the anode. There are two hydrogens removed from either side of the thiophene ring for every monomer reacted...... 56

Figure 3.3: Early stages of the electrochemical polymerization of PEDOT imaged for the first-time using liquid-phase TEM (J. Liu et al. 2015)...... 57

Figure 3.4: (a)Brightfield TEM (BFTEM) of the direct imaging of the electrodeposition of PEDOT on glassy carbon working electrode during 60s, 70s, 80s, 100s of the electrochemical polymerization reaction (b) Plot of the nucleation density as a function of time...... 63

Figure 3.5: (a) BFTEM of a polystyrene sphere (b) Schematic of the liquid-like oligomeric droplets inside the in-situ chamber...... 64

Figure 3.6: Plot of film thickness (t in µm) as a function of total charge density (c in C/cm2) showing that the nominally dense PEDOT films follow an empirical relationship (t=6.8c0.75). While some of the values of thicknesses are lesser than the expected values in the nucleation and growth regime presumably due to their discontinuous nature, the thicknesses of electrodeposited films onto gels or tissue scaffolds are much higher than expected...... 66

Figure 3.7: Images showing that with increasing time, the deposited material goes through a transition from an initially translucent phase (EDOT oligomers) to a final dark, solid product (PEDOT polymer)...... 68

Figure 3.8: Sequential Bright-field TEM images showing liquid-like clusters transitioning into solid-like films observed during the electrodeposition of PEDOT on glassy carbon working electrode taken at (a)180s (b) 190s (c) 200s (d) 210s (e) 220s (f) 230s L - Liquid-like; S - Solid-like...... 73

xiii

Figure 3.9: (a) Intensities of typical droplets and a solid area with time. As seen in the graph, the intensities of the droplets could increase or decrease during the reaction whereas the solid-like films constantly decreased in intensities, meaning they increased in mass thicknesses (b) Schematic of rough edges in the solid-like components as seen during the electrochemical polymerization...... 74

Figure 3.10: (a)Bright-field TEM images of liquid-like droplets and solid-like films observed during the deposition (b) Liquid-like droplets are shaded in red (horizontal lines); solid-like films are shaded in blue (tilted lines)...... 77

Figure 3.11: Plot of the transition in circularity with time of the droplets in Figure 3.8 while they transition from liquid to solid...... 79

Figure 3.12: (a) Plot of circularity and thickness of the transitioning droplets as a function of time (b) Schematic explaining the pinned-edge mechanism...... 82

Figure 3.13: (a)Yunker et al. describe the formation of relatively more uniform films (left) when anisotropic particles are deposited onto the glass substrate due to their loose packing as compared the coffee ring effect exhibited by spherical particles(right). They claim that the mobility of the anisotropic particles is reduced, and they resist the radially outward flow (Yunker et al. 2011) (b) Nellinmoottil et al. showed that the evaporation of a droplet containing non-motile bacteria(right) follows the coffee ring effect while the effect is suppressed in case of motile bacteria(left) where a relatively more uniform film is deposited on the surface (Nellimoottil et al. 2007)...... 83

Figure 3.14: Schematic relating the in-situ BFTEM images to the mechanism showing the nucleation and growth of liquid-like clusters transforming into solid-like films through coalescence...... 85

Figure 4.1: Scanning electron micrographs of PEDOT with (a) regular (rough and bumpy) and (b) nano-fibrillar morphologies...... 98

Figure 4.2: Bright-field TEM images of the glassy carbon working electrode (a)before and (b) after electrodeposition, (c) Transmitted light optical image of the same electrode after electrodeposition...... 100

Figure 4.3:(a) BFTEM of a polystyrene sphere (b) Schematic of the liquid-like oligomeric droplets inside the in-situ chamber...... 101

xiv

Figure 4.4: Sequential bright-field TEM images during the initial, intermediate and later stages showing the nucleation of individual fibrils during the electrodeposition of PEDOT with a nanofibrillar morphology...... 103

Figure 4.5: Bright field TEM images showing the edge of the working electrode (a)before, (b)after nucleation of PEDOT nano-fibrils and (c) Average length, width and thickness profiles of 25 nanofibrils (d) An estimate of the number of molecules of EDOT contained in a fibril of mean length, width and thickness at each time-stamp assuming that each fibril is an elliptic cylinder. The length of the fibril would correspond to the height of the cylinder and width and the thickness as the major and minor axes. Additionally, we have assumed that PEDOT packs itself into an orthorhombic unit cell with a = 1.4 nm, b = 0.68 nm and c = 0.78 nm and that there are 4 molecules of EDOT per unit cell (Martin et al. 2010)...... 104

Figure 4.6: Shows liquid-phase bright-field TEM images at varying time frames subtracted from each other using Fiji From top left to top right (a) 100s - 80s, (b) 110s - 80s, (c) 120s – 80s, from Bottom left to bottom right (d) 130s – 80s, (e) 140s – 80s, (f) 150s – 80s...... 107

Figure 4.7: (a) Intensity profile of a typical fibril along its length and a schematic of the fibril with lighter areas corresponding to thinner sections and darker areas corresponding to thicker sections (b) Intensity profile of a typical fibril along its width and a schematic of the fibril with lighter areas corresponding to thinner sections and darker areas corresponding to thicker sections(c) Bright-field TEM image of a cluster of fibrils(d) Intensity profile of a cluster of fibrils and a schematic of the cluster with lighter areas corresponding to thinner sections and darker areas corresponding to thicker sections...... 109

Figure 4.8: Schematic of the mechanism for the growth of nanofibrillar PEDOT proposed based on observations through liquid-cell electron microscopy...... 113

Figure 5.1:(a) Chemical structures of 3,4-ethylenedioxythiophene (EDOT) (b) carboxylic acid-functionalized 3,4-ethylenedioxythiophene (EDOTacid)...... 120

Figures 5.2: Polarized optical micrographs of solution-cast EDOT acid single crystals obtained with a full-wave red filter. The blue crystals have the higher refractive index direction running from the lower left to upper right, whereas the yellow crystals are oriented with the higher refractive index direction from lower right to upper left. This means the higher refractive index is parallel to the long axes of these needle shaped crystals...... 126

Figure 5.3: Projections of EDOTacid unit cell along the (a)[100] (b)[010] (c)[001] (d) [012] ((002) slice). Axes labels: red, green and blue colors correspond to a, b and c axes, respectively...... 128

Figure 5.4: (a) Indexed simulated powder X-ray diffraction pattern of the unit cell (CrystalMaker 10)) (b) X-ray powder diffraction pattern of the EDOTacid powder as received (c) X-ray pattern of a textured thin- film (taken in a symmetric reflection θ/2θ geometry) of EDOTacid deposited from acetone...... 130

Figure 5.5: (a) Low dose bright field TEM images of needle-shaped single crystals with high aspect ratio (b) Low dose bright field TEM image of a single crystal with low aspect ratio (c) ED pattern from the single crystal in Figure. 5.5(b) (d) Simulated ED pattern of the unit cell when viewed along the [001] direction (solid spots are allowed reflections, hollow boxes are forbidden)...... 131

Figures. 5.6: Electron diffraction patterns from a single crystal of EDOTacid at electron doses of (a) 0 e/A2 (0 mC/cm2), (b) 5.5 e/A2 (9 mC/cm2), and (c) 9 e/A2 (15 mC/cm2)...... 133

Figure 5.7: Intensity profiles of diffraction spots of (200) and (220) reflections with increasing electron doses. The total end point dose was estimated to be 9 e/A2 (15 mC/cm2) ...... 134

Figure 5.8: Images of the single crystals of EDOTacid as recorded from the scanning electron microscope (Auriga 60 CrossBeam FIB/SEM). The a and the b directions of a single crystal are shown...... 135

Figure 5.9: (a) SEM image of a crystal formed by sublimation of the EDOTacid powder (b) Simulated shape of the EDOTacid single crystal using BFDH calculations (Mercury 4.0.0)...... 136

xvi

Figure 6.1: Reflected light (left) and transmitted light (right) optical micrographs of an electrodeposited 50-50 P(EDOT-co-EDOT-acid) copolymer. The copolymer film showed a much higher nucleation density than homopolymer PEDOT...... 146

Figure 6.2: Top: Optical micrographs of nanofibrillar PEDOT-PAA (left) and PDMICA (right) electrochemically deposited onto interdigitated electrodes. Middle: SEM images of nanofibrillar PEDOT-PAA (left)and PDMICA (right). Bottom: GIWAXS patterns from PEDOT- PAA (left) and PDMICA (right). The PEDOT is relatively less ordered, with a d-spacing of ~ 1.3-1.4 nm. Whereas, PDMICA is semi-crystalline, with a strong intermolecular spacing near a d-spacing of ~ 1.4 nm...... 150

Figure 6.3: Quantitative analysis of the growth dynamics of nanofibrillar structures in PEDOT/PAA (left) and PDMICA (right). The PEDOT nanofibrils grew along their lengths and also their edges. The PDMICA nanofibrils were thinner and grew predominantly at their tips. PEDOT fibrils had a lot of variations in thickness along their length in width. Whereas the PDMICA nanofibrils were more uniform in width, but considerably dendritic or branched. A schematic of both types of nanofibrils based on these evidences has been shown. Lighter shades of blue correspond to lower thicknesses while darker shades correspond to thicker regions...... 151

Figure 6.4:(a) Schematic of the liquid-like and solid-like clusters interacting with the flowing metallic nanoparticles (b) Anisotropic gold nanorods (5- 10 nm wide) flowing in liquid as observed by liquid-cell TEM...... 152

Figure A1: Solubility calculations of EDOTacid as a function of pH ...... 155

Figure A2: Fourier Transform Infrared (FTIR) Spectra of EDOT and EDOTacid for comparison ...... 156

Figure A3: Thermogravimetric analysis (TGA) of EDOT acid showing a degradation onset of 146.7 oC ...... 156

Figure A4: AFM-IR spectra of a single of crystal of EDOTacid crystallized by sublimation of the powder ...... 157

Figure A5:(a) Low-dose high resolution electron micrograph of thin film of EDOT acid deposited on a carbon coated copper grid (b) FFT of (a) showing evidence for (200) spacings ...... 157

xvii

ABSTRACT

The conjugated polymer poly(3,4-ethylenedioxythiophene) (PEDOT) and its derivatives have received widespread recent interest due to their high chemical stability, biocompatibility, low oxidation potential and high conductivity. Their application in interfacing ionically conducting living tissue with electronically conducting metallic or semiconducting biomedical devices has been of particular interest due to their ability to provide a mechanically compliant interface and transport charge both ionically and electronically. PEDOT-coated electrodes typically have impedances around two to three orders of magnitude lower than the uncoated metal electrodes in the low-frequency region (<1 kHz) of particular interest for biological signals. Electrochemically deposited conjugated polymer films can show dramatic variations in morphology. Typically, the films have a somewhat lumpy surface structure, with details that depend on the deposition conditions and choice of solvent. They can also form relatively fibrillar, low density, open structures. They can also deposit around dissolvable templates, or onto gels or even living tissues if they are present in the reaction medium. However, the factors that determine the development of these structures during electrodeposition are not yet well established. This makes it difficult to design new systems and optimize device performance. Considering that the morphology of electrochemically-polymerized thin-films can be fine-tuned by controlling the early stage nucleation and growth of the oligomeric clusters, we have

xviii used operando liquid-cell low-dose Transmission Electron Microscopy to image and obtain a detailed understanding of the fundamental processes occurring at the electrode-solution interface, especially the evolution of the mobile oligomeric clusters that precede solid polymer film formation. During the deposition, we clearly see the early stages of the electrodeposition where the liquid-like oligomers (dark in contrast due to higher mass thickness) initially nucleate from the glassy carbon working electrode (lighter shade of grey compared to the oligomers) then merge, coalesce and increase in size and thickness before finally depositing onto the working electrode as a solid, stable and dark conjugated polymer film. Further, we used transmitted light optical microscopy to correlatively study this process which revealed the change in color of the translucent clusters to the dark polymer film caused due to the increase in conjugation length. Furthermore, using a macromolecular counter-ion, poly(acrylic acid) (PAA) during the electrodeposition facilitates the formation of highly anisotropic nano-fibrils of PEDOT. The nanofibrillar structure thereby reduces the overall impedances due to increase in the effective surface area available for charge transport. Operando liquid- cell electron microscopy has helped us observe their early stage growth in an oriented, dendritic manner at the solution-electrode interface, allowing us to quantitatively understand the nano-fibril formation process. Finally, we have also found substantial variations when the chemistry of the monomer is changed, including a dramatic increase in the nucleation density when a more hydrophilic, crystalline, carboxylic acid substituted EDOT (EDOT-acid) was used as a comonomer. These insights have proven to be of utmost importance while understanding the polymerization process thus helping us to design better systems and optimize device performance.

xix

Chapter 1

INTRODUCTION

1.1 Motivation: Elucidating the mechanistic details of the electrochemical polymerization reaction of poly(3,4-ethylenedioxythiophene)(PEDOT) Conducting polymer thin films are well known for a variety of applications including photovoltaics, organic electronics and chemical sensors (Zhu et al. 2004; Ikeda et al. 2016; Inal et al. 2018). Of the conducting polymers, PEDOT is particularly important due to its chemical and mechanical stability and high charge transport properties. It can be polymerized by a variety of methods including chemical oxidation, electrochemical deposition, radiation-induced polymerization, or chemical vapor deposition (Ghosh et al. 2016; Winther-Jensen and West 2004; Lock, Im, and Gleason 2006; Lattach et al. 2013). Among these methods, chemical oxidation (followed by spin-coating onto the electrode) and electrochemical deposition have been extensively used for modifying substrates and electrodes. A particular advantage of using electrochemical polymerization is its ability to coat selected areas or patterned surfaces (Mariani et al. 2011). Electrodeposition also requires charge transport between the working electrode and the depositing material, ensuring electrical contact between the substrate and the deposited polymer. Electrochemically fabricated conjugated polymer films can have varying morphologies ranging from relatively smooth, to lumpy, to relatively fibrillar, low density, open structures. They can also form microporous structures by depositing

1 around dissolvable templates. However, the factors that determine the development of these structures during electrodeposition are not yet well established. To be able to track the evolution of these structures in real-time is of utmost importance for understanding the mechanisms involved in the polymerization of PEDOT and its derivatives. To that end, Atomic Force Microscopy (AFM) was used by Ventosa et.al to image the early stage nucleation and growth mechanisms of PEDOT (Ventosa, Palacios, and Unwin 2008). They observed that at low driving force, 2D layer-by-layer growth governs electrodeposition. Nevertheless, the AFM studies had relatively limited field of view, and did not reveal information about the intermediate states or their interactions with one another. They also did not provide insight about relatively thick films of interest for most applications. This thesis work focuses on the use of operando liquid-cell low-dose transmission electron microscopy to elucidate the finer details of the PEDOT electrochemical deposition process.

1.2 The Technique: Operando Liquid-cell Transmission Electron Microscopy Before commercially available holders became widely available, sample preparation for the liquid-phase TEM (LPTEM) work was done by encapsulating the liquid between thin films such as graphene sheets (Yuk et al. 2012). However, recently, commercially available holders and microfabricated chips of various designs have gained a lot of attention and have been used more commonly. LPTEM studies have since been used for understanding the underlying mechanisms behind a variety of fundamental processes and reactions. This technique has proved to be a useful tool in studying nanomaterial nucleation and growth from solution (Evans et al. 2011), electrochemistry (Mehdi et al. 2015; Radisic et al. 2006), life sciences (Degen et al. 2012) and other applications. The primary purpose of LPTEM is to visualize the

2 processes occurring at the nanoscale in a thin enough liquid layer to let the electrons penetrate through at electron doses below the critical dose limits (in the case of organic and biological specimens). Major commercial players including Protochips, Inc. , Hummingbird Scientific LLC, and Dens Solutions B.V have been designing specialized holders capable of imaging the dynamically occurring processes by TEM. These holders each have some unique features, however they all use the same general principle. The liquid flows through an inlet, and is then directed into the region of interest. This region is usually a thin (500-1000 nm) section of liquid sandwiched between two microfabricated chips equipped with ultrathin (~10 nm) electron transparent membranes. For our experiments, we have generally used a commercially available electrochemical holder from Protochips, Inc. for performing the in-situ electrochemical liquid TEM experiments. Two microfabricated chips are used; the top chip has the electrochemical connections and the bottom chip (with a spacer of thickness ~ 500 nm) is used to define the liquid thickness in the beam direction. These chips are both equipped with ~50 nm thick SiN membranes. The dimensions of the window (SiN membrane) are 550 µm x 40 µm on the top chip and 300 µm x 90 µm on the bottom chip. The glassy carbon working electrode (WE) is about 20 µm wide. The 100 µm wide circular CE is located about 500 µm away from the WE. The CE surrounded the WE and the RE was located away from the WE. Both CE and RE are located off the SiN window. The holder is additionally equipped with two inlets and an outlet all of which were attached to polyetheretherketone (PEEK) tubings for flowing liquid through the holder. The monomer solution (sandwiched between the

3 two microfabricated chips) is passed through the holder at a certain flow rate using a syringe and microfluidic pump.

1.3 Organization of the chapters

1.3.1 Chapter 1: This serves as an introduction chapter and as a guide to navigate through this thesis work.

1.3.2 Chapter 2: This chapter is a detailed overview of the literature (also including some of our own unpublished results) with a focus on LPTEM for beam-sensitive soft materials. It includes the considerations one must keep in mind and challenges that you can expect to encounter while performing these experiments on beam-sensitive samples. Additionally, the chapter also covers latest studies on other material classes like metal- organic frameworks, batteries and nanowires. The chapter ends with a brief discussion on ways in which LPTEM could evolve in the near future. We plan to submit this for consideration to ACS Nano.

1.3.3 Chapter 3: The previous chapter provides some useful technical background and literature that can be used as a segue into discussing the nature of the electrochemical polymerization reaction observed by LPTEM which is at the heart of this dissertation. Unlike the electrodeposition of metals, the electrodeposition of polymers involves a series of intermediate states as the molecular weight of the material increases. The liquid monomer thereby undergoes a series of rheological transitions as well from being viscous or viscoelastic before finally depositing as a solid polymer product on the working electrode. During this process, the transition from the liquid-like precursor state to the final solid film is subtle and complicated and involves substances with rheological properties that are between those of the precursor liquid and final solid products. Further, we consistently observed substantial differences between the dynamics of the liquid-like oligomer droplets and solid-like PEDOT films. This chapter thus aims to look at the mechanics of the process in detail by discussing dynamic events like the changing nucleation densities, coalescence, varying shapes and sizes, and thicknesses of the oligomeric clusters during the intermediate stages of the reaction. We are planning to submit this work for consideration to Macromolecules, ACS Publications.

1.3.4 Chapter 4: With the understanding we have already built around the technique and the polymerization process, we next we delve into understanding the evolution of nanofibrillar PEDOT. Previous studies had shown that this interesting nanofibrillar structure could form when certain polymeric counterions were added to the system, however the details of film growth were not clear. Real-time observations using

LPTEM have given us insights into the nature of the growth and development of the nanofibrils of PEDOT. This chapter is dedicated to elucidating the structural evolution of the nanofibrils through qualitative and quantitative analysis of the deposition at different time frames. We plan to submit this work for consideration to Nano Letters, ACS publications.

1.3.5 Chapter 5: (EDOTacid) is a comonomer for 3,4-ethylenedioxythiophene (EDOT) designed to introduce controlled amounts of hydrophilicity into the resulting copolymers (Povlich et al. 2013). It is also known that electropolymerized PEDOT- PEDOTacid copolymer films can be coupled with peptides resulting in films with increased bio-activity(Povlich et al. 2013). In this chapter, we investigate the crystal structure and morphology of EDOTacid and study its sensitivity to the electron beam. This chapter serves as a precursor to future studies involving the dramatic change in the nucleation density observed when EDOTacid is introduced as a comonomer. This chapter is already published in Crystal Growth and Design, ACS Publications.

1.3.6 Chapter 6:

In the final chapter, we discuss our conclusions from each of the above chapters (chapters 2-5). Using that as a platform, we talk about where we are headed in the future. Specifically, we discuss the dramatic variation in the nucleation density with EDOTacid is used as a copolymer, compare nanofibrils of a relatively more order polymer PDMICA and nanofibrils of PEDOT, and the effect of flowing nanoparticles during the electrochemical polymerization of PEDOT. We also show some

6 preliminary data backing up our hypotheses and ideas for experiments to be performed in the future.

REFERENCES

1. Chen, Fu-Chun, Jui-Yuan Chen, Ya-Hsuan Lin, Ming-Yu Kuo, Yung-Jung Hsu, and Wen-Wei Wu. 2019. “In Situ TEM Observation of Au–Cu2O Core–Shell Growth in Liquids.” Nanoscale 11 (21): 10486–92. https://doi.org/10.1039/C9NR00972H.

2. Chen, Fu-Chun, Jui-Yuan Chen, Ya-Hsuan Lin, Ming-Yu Kuo, Yung-Jung Hsu, and Wen-Wei Wu. 2019. “In Situ TEM Observation of Au–Cu2O Core–Shell Growth in Liquids.” Nanoscale 11 (21): 10486–92. https://doi.org/10.1039/C9NR00972H.

3. Degen, Katherine, Madeline Dukes, Justin R. Tanner, and Deborah F. Kelly. 2012. “The Development of Affinity Capture Devices—a Nanoscale Purification Platform for Biological in Situ Transmission Electron Microscopy.” RSC Advances 2 (6): 2408–12. https://doi.org/10.1039/C2RA01163H.

4. Evans, James E., Katherine L. Jungjohann, Nigel D. Browning, and Ilke Arslan. 2011. “Controlled Growth of Nanoparticles from Solution with In Situ Liquid Transmission Electron Microscopy.” Nano Letters 11 (7): 2809–13. https://doi.org/10.1021/nl201166k.

5. Ghosh, Srabanti, Natalie Amoin Kouame, Samy Remita, Laurence Ramos, Fabrice Goubard, Pierre-Henri Aubert, Alexandre Dazzi, Ariane Deniset- Besseau, and Hynd Remita. 2016. “Visible-Light Active Conducting Polymer Nanostructures with Superior Photocatalytic Activity.” Scientific Reports 5 (1). https://doi.org/10.1038/srep18002.

6. Ikeda, Natsumi, Tomoyuki Koganezawa, Daisuke Kajiya, and Ken-ichi Saitow. 2016. “Performance of Si/PEDOT:PSS Hybrid Solar Cell Controlled by PEDOT:PSS Film Nanostructure.” The Journal of Physical Chemistry C 120 (34): 19043–48. https://doi.org/10.1021/acs.jpcc.6b07101.

7. Inal, Sahika, Jonathan Rivnay, Andreea-Otilia Suiu, George G. Malliaras, and Iain McCulloch. 2018. “Conjugated Polymers in Bioelectronics.” Accounts of Chemical Research 51 (6): 1368–76. https://doi.org/10.1021/acs.accounts.7b00624.

8. Lattach, Youssef, Ariane Deniset-Besseau, Jean-Michel Guigner, and Samy Remita. 2013. “Radiation Chemistry as an Alternative Way for the Synthesis of PEDOT Conducting Polymers under ‘Soft’ Conditions.” Radiation Physics and Chemistry 82 (January): 44–53. https://doi.org/10.1016/j.radphyschem.2012.09.009.

9. Lock, John P., Sung Gap Im, and Karen K. Gleason. 2006. “Oxidative Chemical Vapor Deposition of Electrically Conducting Poly(3,4- Ethylenedioxythiophene) Films.” Macromolecules 39 (16): 5326–29. https://doi.org/10.1021/ma060113o.

10. Mariani, G., Y. Wang, P. S. Wong, R. B. Kaner, and D. L. Huffaker. 2011. “Electrochemical Polymerization of PEDOT on Catalyst-Free Patterned GaAs Nanopillars for High Efficiency Hybrid Photovoltaics 37th IEEE Photovoltaic Specialists Conference.” In 2011 37th IEEE Photovoltaic Specialists Conference, 002639–41. https://doi.org/10.1109/PVSC.2011.6186490.

11. Mehdi, B. L., J. Qian, E. Nasybulin, C. Park, D. A. Welch, R. Faller, H. Mehta, et al. 2015. “Observation and Quantification of Nanoscale Processes in Lithium Batteries by Operando Electrochemical (S)TEM.” Nano Letters 15 (3): 2168–73. https://doi.org/10.1021/acs.nanolett.5b00175.

12. Povlich, Laura K., Jae Cheol Cho, Michelle K. Leach, Joseph M. Corey, Jinsang Kim, and David C. Martin. 2013. “Synthesis, Copolymerization and Peptide-Modification of Carboxylic Acid-Functionalized 3,4- Ethylenedioxythiophene (EDOTacid) for Neural Electrode Interfaces.” Biochimica et Biophysica Acta - General Subjects 1830 (9): 4288–4293. https://doi.org/10.1016/j.bbagen.2012.10.017.

13. Radisic, Aleksandar, Philippe M. Vereecken, James B. Hannon, Peter C. Searson, and Frances M. Ross. 2006. “Quantifying Electrochemical Nucleation and Growth of Nanoscale Clusters Using Real-Time Kinetic Data.” Nano Letters 6 (2): 238–42. https://doi.org/10.1021/nl052175i.

14. Ventosa, Edgar, Jesús López Palacios, and Patrick R. Unwin. 2008. “Nucleation and Growth of Poly(3,4-Ethylenedioxythiophene) Thin Films on Highly Oriented Pyrolytic Graphite (HOPG) Electrodes.” Electrochemistry Communications 10 (11): 1752–55. https://doi.org/10.1016/j.elecom.2008.09.003.

15. Winther-Jensen, Bjørn, and Keld West. 2004. “Vapor-Phase Polymerization of 3,4-Ethylenedioxythiophene: A Route to Highly Conducting Polymer Surface Layers.” Macromolecules 37 (12): 4538–43. https://doi.org/10.1021/ma049864l.

16. Yuk, Jong Min, Jungwon Park, Peter Ercius, Kwanpyo Kim, Daniel J. Hellebusch, Michael F. Crommie, Jeong Yong Lee, A. Zettl, and A. Paul Alivisatos. 2012. “High-Resolution EM of Colloidal Nanocrystal Growth Using Graphene Liquid Cells.” Science 336 (6077): 61–64. https://doi.org/10.1126/science.1217654.

17. Zhu, Zheng-Tao, Jeffrey T. Mabeck, Changcheng Zhu, Nathaniel C. Cady, Carl A. Batt, and George G. Malliaras. 2004. “A Simple Poly(3,4-Ethylene Dioxythiophene)/Poly(Styrene Sulfonic Acid) Transistor for Glucose Sensing at Neutral PH.” Chemical Communications 0 (13): 1556–57. https://doi.org/10.1039/B403327M.

Chapter 2

MOLECULAR MOVIES: A REVIEW OF RECENT WORK, CURRENT CHALLENGES AND FUTURE OPPORTUNITIES IN LIQUID-PHASE TRANSMISSION ELECTRON MICROSCOPY (LPTEM) OF BEAM- SENSITIVE ORGANIC MATERIALS

2.1 Introduction

In principle, complicated phenomena are made up of individual, simpler processes. Identifying and characterizing these underlying processes makes it easier to better understand and control the phenomena. Operando or “in-situ” studies help us to directly image the important events, providing insights about how to more precisely tune the ultimate structure and therefore properties of the materials of interest. Direct imaging techniques such as Operando Atomic Force Microscopy or Operando Transmission Electron Microscopy are particularly effective because they provide local information about reactions that are difficult or impossible to obtain by more indirect techniques. In this review, we will discuss how Liquid Phase Transmission

Electron Microscopy Figure 2.1: The liquid to be imaged encapsulated (LPTEM) has given betweengraphene sheets (adapted from Yuk et al. 2012). us the power to

11 observe, monitor and control the individual processes and reactions over the last two decades. Since the use of LPTEM to study relatively stable inorganic materials has been discussed in some detail elsewhere (F.-C. Chen et al. 2019; Impagnatiello et al. 2020; Schneider et al. 2014; Chee et al. 2013), our focus will be on the use of these techniques to investigate more electron beam sensitive materials such as organic polymers and molecular crystals. The earliest TEM studies in liquids were done as early as 1944, and laid the groundwork for the LPTEM based work done today (Abrams and McBain 1944). Prior to commercially available holders becoming popular, sample preparation for the operando work was done in-house by encapsulating the liquid between thin films such as graphene sheets (Yuk et al. 2012) (Figure 2.1). In the past decade, commercially available holders and microfabricated chips of various designs have gained a lot of attention and have been used more commonly. LPTEM studies have since become the norm for understanding the underlying mechanisms behind a variety of fundamental processes and Figure 2.2: Schematic capturing the central idea behind contemporary commercially available TEM reactions. This sample holders. The liquid cell is held in the electron beam between two ultrathin silicon nitride membranes. technique has proved In an electrochemical version of such a cell, an electrode is used to drive a chemical reaction. to be a useful tool in studying nanomaterial nucleation and growth

12 from solution (Evans et al. 2011), electrochemistry (Mehdi et al. 2015; Radisic et al. 2006), life sciences (Degen et al. 2012) and other applications. The primary purpose of LPTEM is to visualize the processes occurring at the nanoscale in a thin enough liquid layer to let the electrons penetrate through at electron doses below the critical dose limits (in the case of organic and biological specimens). Major commercial players including Protochips, Inc. (Chen et al. 2019), Hummingbird Scientific LLC (Wu et al. 2014), and Dens Solutions B.V (Garza et al. 2016) have been actively involved in designing specialized holders to make it easier for researchers to study the dynamically occurring processes. These holders each have some unique features, however they all use the same general principle (Figure 2.2). The liquid flows through an inlet, and is then directed into the region of interest. This region is usually a thin (500-1000 nm) section of liquid sandwiched between two microfabricated chips equipped with ultrathin (~10 nm) electron transparent membranes. Although these commercially available designs have made in-situ experiments easier than they were earlier, they still require a non-trivial amount of optimization for each system. For instance, apart from protecting the sample from irradiation (Q. Chen et al., 2020), for beam-sensitive materials like organic polymers, proteins or other biological molecules, there are other important considerations while imaging in liquid (He, Shokuhfar, and Shahbazian-Yassar 2019). The liquid can undergo radiolysis when imaged under the electron beam. Schneider et.al (Schneider et al. 2014) showed that electron-water interactions during in-situ imaging can lead to various by products, most importantly gas bubbles. There may consequently be dewetting in the sample chamber, causing either skewed results or leading to failure of

13 the experiments. This makes it important to have careful control over the electron doses used during LPTEM.

2.2 Considerations and challenges during imaging beam-sensitive materials in LPTEM

Imaging inorganic materials through High Resolution Transmission Electron Microscopy (HRTEM) has become relatively routine and typically requires accumulated doses on the order of a few thousand e/Å2 (Yancey et al. 2018). These doses would not only destroy beam-sensitive materials like organic polymers, peptides or other biological materials within the irradiated area, but could also lead to beam- induced artifacts thus complicating data interpretation. In comparison, the total doses used for imaging soft materials are usually somewhere between 10 – 100 e/ Å2 with maximums up to a few hundreds of e/Å2 (Subramanian, Rowland, et al. 2019; Q. Chen et al., n.d.; Leijten et al. 2017). This makes it challenging for a microscopist to operate the microscope at low-dose conditions and still be able to get enough contrast and spatial resolution to image the specimen. A layer of liquid adds to the complexity due to its interactions with the electron beam. Here we discuss some of the most commonly experienced challenges and possible measures from literature and our own experience in dealing with beam sensitive specimens in LPTEM.

2.2.1 Electron-water interactions Schneider et. al studied the effect of electron beam on the radiolysis of water and developed a kinetic model to estimate the concentration of various radiolysis products including hydrogen and hydroxyl radicals, and hydronium ions (Schneider et al. 2014). These are particularly important in situations where the pH of the solution is

14 crucial to the user’s system. High doses can cause significant radiolysis of water leading to higher hydronium ion concentration in turn resulting in lower pH values within the irradiated regions. Other studies that have discussed this phenomenon give useful insights to reduce the damage (T. J. Woehl and Abellan 2017; Rehn and Jones 2018; Taylor J. Woehl et al. 2013; A. Sutter and W. Sutter 2017; Liao and Zheng 2016). For instance, Woehl et al. discussed different ways one can alter the solution (liquid) chemistry to mitigate the effect of radical formation. They described the experimental changes that they make to the solution, the characteristic effect of that change followed by the advantages and the disadvantages giving the reader well- organized insights that can be utilized in their experiments. In a similar fashion they also describe the use of low-dose, STEM probes instead of TEM probes and use of higher operating voltages to reduce the formation of radicals. In sum, it is critical to establish the dose sensitivity of a particular materials system first and then move on to doing the actual experiments under conditions where the flux of electrons is known and precisely controlled.

2.2.2 Beam-specimen interactions There have been a multitude of TEM studies in the recent past on polymers and other soft matter in liquid. This includes direct observation of the film formation process of latex films (L. Liu et al. 2015), the electrodeposition process of functionalized thiophenes (J. Liu et al. 2015), micelle fusion and fragmentation (Early, Yager, and Lodge 2020; Parent et al. 2017) and several other materials (Evans et al. 2012; Aplan et al. 2019). If the specimen is crystalline, electron beam sensitivity can be measured by the

total dose Jc required to amorphize the Figure 2.3: Schematics of devices used material by monitoring the decrease in for imaging in liquids. (a) Regular device with a single rectangular viewing intensity of particular Bragg reflections window. (b) Echips placed one on top of for a specific set of crystallographic each other with the SiN windows perpendicular to each other. (c)The novel planes (C. Liu et al. 2018). It has been design as proposed and fabricated by Moser and colleagues which has multiple shown that the sensitivity to electron windows. (d) The Echips with multiple windows fabricated by Moser et.al placed beam damage Jc of crystalline polymers on top of each other thus creating an correlates with their melting or array of viewing areas. (e) Echips similar to (c), but equipped with additional gold degradation temperature Tm or Td bars for use as focusing aids (f) Devices from (e) placed one on top of each other (Kumar and Adams 1990). The total thus generating an array of viewing windows equipped with focusing electron flux would then have to be kept aids(figure adapted from Moser et.al). under the damage limits to ensure that

16 the structure of the sample remains relatively intact. Alternatively, low-loss EELS can also be used to estimate the critical dose (Guo et al. 2015). Once the critical dose Jc is determined it is possible to to stay below the known damage limits. Techniques like blanking the beam and modulating the dose delivery using a pulsed laser have also been explored to mitigate the damage inflicted on the sample by the electron beam (Martin et al. 2005; VandenBussche and Flannigan 2019). Moser and colleagues recently described methods involving multiple areas of imaging that help understand the effect of cumulative electron dose. Although the experiments were performed on nanoparticles and biological materials, it can apply to an array of other beam- sensitive specimens. As shown in Figure 2.3, instead of having one traditional window, they fabricated devices with multiple windows. These devices were then placed over each other with the viewing SiN windows perpendicular to each other to sandwich a thin layer of liquid. This created an Figure 2.4: Phase contrast TEM images, array of viewing windows that allowed corresponding FFTs and line profiles of individual spatial frequencies as a the user to do multiple experiments on function of the electron dose(Keskin and de Jonge 2018). the same device. Additionally, the chips were equipped with gold bars that could be used as focusing aids while imaging.

If used at reasonable doses (below the critical damage dose limits), this solution gave the user a way to circumvent issues like limited viewing area and reusability of the chips/devices in case one of the viewing areas had been damaged. Kesin and colleagues recently showed a way to track the damage in beam- sensitive protein microtubules by using the spatial frequencies of the observed features in LPTEM (Keskin and de Jonge 2018). They argued that a graphene-based LPTEM imaging technique of the protein samples was better compared to cryoTEM of these samples, as determined by the damage sensitivity of the protein microtubules in both of the techniques. As shown in figure 2.4, they used the Fast Fourier Transform (FFT) of the panels, a,b,c and d to analyze the intensity of the spatial frequencies as a function of the electron dose/flux. Specifically, they chose particular spatial frequencies (0.1 nm-1, 0.15 nm-1 and 0.2 nm-1) corresponding to the features observed in real space and measured their intensities using line profiles as a function of the total electron dose used. The idea was to use the peaks in the reciprocal space to quantitatively measure the damage dose of the microtubules. This approach is similar to the method that many adopt to quantify the critical damage dose by measuring the fading intensities of the diffraction spots of highly ordered specimen (Subramanian, Rowland, et al. 2019; C. Liu et al. 2018). However, using FFTs of the features seen in the real space can be a useful way in instances where the diffraction intensity is hard to track either due to the specimen itself being weakly diffracting or if the liquid layer results in excess scattering and in turn affects the diffraction intensity. The key takeaway from this study was the applicability of the above-mentioned methodology even in liquids. It is definitely one of the powerful ways to track the damage of beam

18 sensitive specimen in liquids especially in cases where a diffraction pattern cannot be obtained.

2.2.3 The “bowing effect” of viewing windows The microfabricated chips used in commercial LPTEM sample holders are typically equipped with viewing windows made from electron transparent silicon nitride membranes. Their mechanical robustness is crucial for the success of in-situ experiments. Any cracks or tears in these membranes would cause a leak and unstable pressures inside the column. Therefore, utmost care must be taken to not damage them in any manner. Regardless of whether you are fabricating them yourselves or purchasing commercially available chips, the membranes must be one of the first things Figure 2.5: Schematic of bowing or bulging of Silicon Nitride membranes under the influence of that should be checked for vacuum inside the TEM. failures. When not subjected to vacuum, the membranes are flat thin-films. However, when in vacuum they may bulge or bow under low pressure as shown in Figure 2.5. This can cause ruptures at the edges leading to leaking of the liquid trapped inside. This is why the widths of the membranes are more important than lengths. The lesser the width, the more uniform

19 the liquid thicknesses will be within the viewing area. As a result, it would be easier to find the right focal plane . One implication of this is that the liquid thickness is not consistent throughout the viewing window. This has been discussed in detail and quantified to some degree in previous studies (Grogan and Bau 2010; Holtz et al. 2013; Dukes et al. 2013; de Jonge et al. 2010). This effect can also adversely affect the spatial resolution of the viewing area if there is too much bulging. Additionally, we have experienced that in some instances a combined effect of irregular flow, gas bubbles and wide windows cause the windows to bulge significantly, making the image too fuzzy or hazy to make reliable interpretations. This can be followed by a sudden squishing of the liquid resulting in either a thin layer of liquid or complete de-wetting of the viewing area. The fuzzy images are associated with the extra inelastic, low-angle scattering from electrons across thick liquid layers. Having narrower viewing windows, checking for good flow and creating many smaller viewing areas instead of one large window area (Dukes et al. 2013) have all been shown to be effective ways to mitigate this effect.

De Jonge and colleagues found another unique way to tackle this problem. Since the thickness of the liquid is controlled by the deformation of the silicon nitride membranes, they used a pressure controller to balance the pressure difference between the two electron transparent windows and thereby regulated the shape of the membranes. In this manner they had Figure 2.6: LPTEM images of amorphous good control over the liquid thickness SiO2 nanoparticles at an internal pressure of (a)1bar; (b) 0.05 bar; (c) FFT of (a) and (b); and associated spatial resolution. They (d) Radially averaged pixel intensities of the FFT. showed evidence both in real space (qualitative) and reciprocal space (quantitative) for a better resolution with their pressure controlling system as described in Figure 2.6. The total electron doses reported were ~10 e/Å2.

2.2.4 Gas bubbles in the feed solution As discussed earlier, gas bubbles can be caused by radiolysis of water by electrons thus causing experimental challenges like dewetting, irregular flow and fuzzy images. If electrochemistry is involved, there may also be electrolysis-induced bubble generation at the counter electrodes that can obstruct the liquid flow. These concerns can be mitigated by using low doses of electrons and staying within the

21 electrochemical window of the liquid. Sonicating the liquid before flowing it through the holder also helps to eliminate any inherent gasses dissolved within the solution.

2.2.5 Dewetting due to hydrophobicity De-wetting of the viewing chamber is a commonly faced challenge in LPTEM. This becomes more significant in electrochemical experiments since they require all of the electrodes to remain in intimate contact with the solution. Plasma cleaning the chip for 2-5 minutes or applying a coating of poly-L-lysine usually helps in making the surface more hydrophilic (de Jonge, Pfaff, and Peckys 2014). Doses as high as 200-

300 e/Å2 can also make the silicon nitride membrane hydrophobic and cause the liquid to dewet the surface. This again underscores the particular importance of controlling the total electron doses in LPTEM (Vailonis et al. 2019).

2.2.6 Irregular flow

Figure 2.7: Schematic of chips equipped with flow (left) and static spacers (right).

The in-situ sample stages are microfluidic devices can be operated in either of two modes, static or flow (Figure 2.7). In the static mode, a few nanoliters of the

22 liquid are placed directly on one of the hydrophilic chips and another chip is placed on top of the former so as to sandwich the liquid layer. The bottom chip has a spacer of defined thickness confining the liquid that does not allow the liquid to flow out. In contrast, in the flow mode, the liquid is under continuous flow from an inlet to an outlet port (Figure 2.2). If the liquid is flowing, it is recommended that the flow continue even while performing vacuum checks on the pumping station. This ensures that no leaks are present even during flow. Irregular flows are also a matter of concern since they can cause accumulation of liquid and thus lead to leakage by damaging the windows. Having a continuous ongoing flow set up the previous night before the experiment can help to maintain regular or continuous flow the following day. We have found that flow mode is typically preferred since there is always new influx of the liquid coming in and there are fewer worries about the liquid drying up or the specimen getting damaged. Additionally, too much liquid in static mode can lead to rupture of the windows.

2.2.7 Finding the right focal plane It is often difficult to find the right focus while imaging in liquid. As a starting point it is recommended that the user start focusing on the edges of the viewing window since, by definition, the liquid thickness is the lowest near the edges. Once a reasonable focus is found, the user can move to the region of interest (ROI) and start optimizing the focus. Other methods like fabricating chips with grid bars as discussed by Moser and colleagues are also desirable (Moser et al. 2018).

2.2.8 Voltage spikes in the potentiostat For experiments involving electrochemistry, in our experience, plugging in the potentiostat connections while the holder is in the microscope is not a good practice. Electrical fluctuations can cause spikes in voltages leading to bubbling and other issues. Thus, the safest approach is to connect the poteniostat to the holder about 30 minutes prior to starting the experiment.

2.3 Recent applications of LPTEM in Materials Science

2.3.1 Soft materials Polymers, peptides and other macromolecules have been extensively studied using LPTEM (Patterson, Proetto, and Gianneschi 2015; Aplan et al. 2019; Parent et al. 2017; J. Liu et al. 2015; Subramanian, Liu, et al. 2019; L. Liu et al. 2015; Touve et al. 2018; X. Figure 2.8: Schematic and BFTEM Chen et al. 2017). However, many image of transition states involved in micelle fragmentation in liquid (Early, questions still remain. As discussed in Yager, and Lodge 2020). the previous sections, soft materials need to be handled with care due to the beam damage that ensues from their interactions with the electron beam. Interesting recent work was done by Lodge and colleagues where they imaged the micelle- fragmentation using LPTEM as shown in Figure 2.8 (Early, Yager, and Lodge 2020). They observed initially spherical micelles transitioning into more “dumbbell” or

“peanut” shaped structures. This was followed by a final fragmentation step where the micelle core and corona separated into two smaller micelles. Although they did not discuss beam-sensitivity in their manuscript, their SI has some useful information (Figure 2.9) where they show images of damaged specimens. This is consistent with

Figure 2.9: Beam damage as observed by Lodge and colleagues (Early, Yager, and Lodge 2020) during their micelle fragmentation experiments. what we discussed about controlling the electron doses while performing liquid cell experiments. Polymer latex particles have been of interest to a large community of researchers and have been studied for many years. However, it is not until recent times that it has been possible to image these particles in liquid while in motion. Liu et.al reported a short study in 2015 of film forming latex particles with a range of particle sizes and glass transition temperatures using LPTEM (L. Liu et al. 2015)(Figure 2.10). They used a system from Protochips, Inc. equipped with a 500 nm spacer. They were able to identify particles displaying various amounts of contrast (arising from electron- dense elements) and core-shell structures in liquid. Besides, like most other LPTEM work, they discuss the difficulty and challenges in imaging such specimens. They concluded that there must be just enough liquid to wet the particles in the cell to be

25 able to achieve acceptable levels of spatial resolution. Too much liquid in the chamber proved to have a detrimental effect on the imaging quality due to excess scattering.

Figure 2.10: LPTEM images of butyl acrylate (BA) / Methyl Methacrylate (MMA) latex particles in liquid (L.Liu et.al, 2015).

Electrodeposition of polymers is a complex process that involves the oxidation of the initial monomer species to radical cations followed by dimerization, trimerization and formation of higher order oligomers before finally depositing onto the working electrode as a solid, stable and dark conjugated polymer film. The process involves a subtle interplay between a multitude of factors like the oxidation potential of the monomer, reaction kinetics, type of solvent and counter ions. Liu et.al (Figure 2.11) reported for the first time an operando study of the electropolymerization process of 3,4-ethylenedioxythiophene where they observed the oligomeric clusters nucleating then growing in size and thickness with time (J. Liu et al. 2015). The Martin lab has since continued to investigate the nature of the electropolymerization of functionalized thiophenes looking at the precipitation of the higher molecular weight droplets from the isotropic monomer solution and depositing onto the working electrode (Subramanian, Liu, et al. 2019). These studies were usually performed using

26 a Protochips, Inc. holder equipped with a 500 nm spacer chip. The total electron dose used for these experiments were ~ 6 e/A2 which were well below the critical dose for PEDOT ~ 60 e/A2. Figure 2.12. Shows the intermediate states involved in the electrodeposition of PEDOT as the molecular weight of the oligomeric species increases from being a monomer, to dimer, to higher molecular weight oligomers, and

Figure 2.11: Direct imaging of early stage electrodeposition of Poly(3,4- ethylenedioxythiophene) on glassy carbon working electrode using LPTEM (J. Liu et.al, 2015).

then finally to the deposited solid polymer. The transition from the liquid-like precursor state to the final solid film is subtle and involves substances with rheological properties that are between those of the precursor liquid and final solid products. Imaging this electrodeposition process is challenging due to the beam sensitivity of the conjugated polymer materials. However, these experiments show that with careful control of the doses, it is possible to observe the substantial differences between the dynamics of the liquid-like oligomer droplets and solid-like PEDOT films.

60s

20 µm 70s

80s

90s

(a)

180 s 190 s 200 s

(b) (c) (d)

(e) (f) (g)

Figure 2.13. is an example of quantitative observations growth of PEDOT nanofibrils grown using poly(acrylic acid) as a counterion (potentiostatic deposition ~ 1.2 V) that nucleated from the edge of the glassy carbon electrode. The growth was categorized into three different regimes as seen in Figure 2.13(g). First, in stage I there is was little to no growth, presumably here the process is dominated by nucleation as the oligomers and the counter-ions were attracted to the working electrode due to the

(a) (b) (c)

(d) (e) (f)

(g)

applied bias. This stage presumably exists since the EDOT oligomers have to attain a 30 certain molecular weight before it precipitates out of the aqueous solution. Also, insolubility of the higher molecular weight oligomers in the solution is one of the major factors driving the electrodeposition process. In the second stage II, the fibrils grew at a relatively fast rate and in the third III the growth rate along the length of the fibril slowed as they started to grow wider and thicker. These experiments have therefore reveals details of the nucleation and growth processes, and variations in the growth rates at different points in time during the process.

2.3.2 Crystal Growth The thermodynamics and kinetics of crystal growth have been studied in detail for a variety of materials systems (Davies and Hull 1976; R.-Y. Wang et al. 2017; Celik and Yildiz 2015; Wunderlich 2008). In recent years it has been possible to observe the crystallization phenomena directly as it occurs using LPTEM (Yuk et al. 2012; Nielsen et al. 2014; Zeng, Zheng, and Zheng 2017; Sutter et al. 2016; Zheng et al. 2009; Ren et al. 2020; T. Liu et al. 2020; J. Li et al. 2019). Such studies have helped us visualize the process clearer and have a better grasp of the subject matter.

In 2012, Yuk et.al reported the direct observation of colloidal nanocrystal

growth of platinum using LPTEM (Yuk et al. 2012)(Figure 2.14). Apart from the

Figure 2.14: Shape evolution of Pt nanocrystal as observed by Yuk and colleagues (Yuk et.al, 2012).

science itself, the uniqueness of the study lies in their sample preparation technique. They utilized a graphene liquid cell prepared in their laboratory to sandwich aliquots of liquid to observe the process in the TEM. Some of the key observations were the colloidal nanocrystals growing through coalescence along <111>-type directions and that twin boundaries formed during coalescence remain locked within the nanocrystal for the duration of their experiments. These observations and reflections could not be observed or appreciated in previous experiments done ex-situ. Most of the above studies discussed above were done in holders fabricated by either Protochips, Inc. or Hummingbird Scientific. A recent paper by Cookman et.al looked at crystal growth in a DENS solution holder(Figure 2.15). They induced the nucleation of organic flufenamic acid (FFA) crystals using electron irradiation and reported the direct visualization of crystals growing in shapes and morphologies that agreed with BFDH and attachment energy calculations. The flux rates reported in this study were of the order of 200-400 e/Å2s. Kumar and Adams established a relationship between the melting or degradation temperature and the electron beam

32 sensitivity of a material (Kumar and Adams 1990). Given that flufenamic acid melts at a temperature similar to polyethylene (~120 oC - 130oC), its thermal stability can be estimated to be around 1-3 e/Å2 (depending on the operating voltage of the microscope). This raises some concern while thinking about the consequences of such high doses (~100x beyond Jc) and increases the chances that significant damage was induced in that study. Further work needs to be done to confirm that the crystals formed retain the symmetry of the pristine materials. Despite these concerns the study nevertheless provides an example of the power of LPTEM for exploring the local mechanisms and dynamics of self-assembly of organic materials in the liquid state.

Figure 2.15: Crystals of flufenamic acid observed using LPTEM by Cookman et.al (Cookman et.al 2020).

2.3.3 Batteries Improved materials for energy storage devices (batteries) are a critical current need. Not only is the increasing the storage capacity crucial, but the storage devices must be safe to use for long periods of time. One of the best ways to improve battery performance is to directly monitor the structural changes during charging and discharging cycles and see if there are any undesirable observations that can be prevented. Along those lines, a plethora of studies have looked at batteries operating in-situ using LPTEM (Karakulina et al. 2018; Lutz et al. 2018; Mehdi et al. 2015;

Sacci et al. 2014; Xu et al., n.d.; Yang et al. 2018; Leenheer et al. 2015; Yuan et al. 2017). Most of the studies have tried to quantify and understand the electrically conductive dendrite growth of lithium as a measure to prevent degradation from occurring. Mehdi and colleagues reported the observation and quantification of lithium dendrites using liquid cell TEM (Mehdi et al. 2015). This was one of the first studies performed that looked at the operando monitoring of metallic dendrite formation. They also discuss the electric field distribution in the specific electrode geometry they used in the in-situ cell. This discussion is useful while reflecting about the implications for experiments involving different materials with a similar geometry. Figure 2.16 shows High-Angle Annular Dark Field (HAADF) images of the charging and discharging cycles. Interestingly, there is contrast reversal in these HAADF images due to lithium being a lighter element than the electrolyte that is flowing through the chamber and hence appears darker. In this study they found evidence for “dead lithium” after the charging and discharging cycles, i.e. lithium dendrites that were unattached to the electrode indicating irreversibility in the process which was not appreciated before. This and other similar studies have substantially improved our understanding of battery systems and their degradation mechanisms.

Figure 2.16: High-angle annular dark field (HAADF) images of the charging and discharging cycles of lithium battery (Mehdi et.al 2015).

Karakulina et.al performed a detailed study involving electron diffraction tomography (EDT) of lithium battery cathode materials to look at the changes in the unit cell during electrochemical cycling(Figure 2.17). During the charging and discharging cycles of a lithium-ion battery, the Li+ ions move back and forth between the anode and cathode. This can lead to significant changes in the cathode structure which needs to be monitored to precisely tune the properties of the battery. Karakulina and colleagues measured the changes that occur during these cycles using EDT and compared the obtained values with other techniques like x-ray and neutron scattering. The values obtained using EDT were comparable to the results obtained from x-ray and neutron scattering techniques thus demonstrating that it is indeed possible to obtain and monitor reliable electron diffraction data even through sections of liquid.

(a) (b)

2.3.4 Metal Organic Frameworks (MOFs)

Metal Organic Frameworks (MOFs) have a been a popular class of materials in

Figure 2.18: Snapshots of Metal Organic Nanotubes (MONTs) grown in LPTEM as reported by Vailonis et.al.

the recent past for a variety of applications including catalysis and gas storage (Y.-Z.

Chen et al. 2018; Shen et al. 2018; S. Wang et al. 2018; Lyu et al. 2020). In a recent study published in JACS, Vailonis et.al tracked the

growth of Metal Figure 2.19: Beam induced damage of MONTs as Organic Nanotubes observed by Vailonis et.al. (MONTs) using LPTEM (Vailonis et al. 2019)(Figure 2.18). These materials are presumably slightly more stable under the electron beam as compared other organic materials due to the presence of metals which can help with immobilizing some of the atoms and with thermal conduction. The challenge to control the dose still prevails due the risk of causing significant damage to the organic structures and radiolysis of the liquid. They report a total electron dose of < 10 e/Å2s which is fairly low to induce any significant beam damage to most liquids and beam-sensitive materials. However, since they showed that their specimen can undergo significant damage under relatively higher electron doses (Figure 2.19), it would have been better if they had a quantitative limit of how much exposure the material can handle using one of the methodologies as discussed in the earlier sections.

2.3.5 Nanowires Although there have been myriad of studies looking at nanowires for a variety of applications, it is not until recently that LPTEM has been utilized to track the growth profiles of these materials (Cheek et al. 2020; Kraus and de Jonge 2013; Panciera et al. 2020). This is important for getting insights into the elementary processes like coalescence, rate of crystallization or any other rate determining steps that govern nanowire formation phenomenon as a whole. This way, there would be more control over each process and in turn would lead Figure 2.20: (a)LPTEM images of the nanowire growth (b) Their growth rates as a to materials with superior properties. function of time (c) The number of times Cheek et.al synthesized germanium their growth direction changes as a function of the growth rate. nanowires using an electrochemical liquid−liquid−solid (ec-LLS) technique (Cheek et al. 2020). Using LPTEM they found that the outer surface of the liquid metal droplets was crucial in determining if the nanowire formation could occur. Further, the feed rate of Ge to the crystal limited the nanowire growth rate. Figure 2.20, again, shows the power of the LPTEM technique in observing processes like the growth rate and even the number of times the nanowires changed their growth direction.

2.3.6 Nanoparticles Nanoparticles are one of the most widely studied type of materials studied by LPTEM (Evans et al. 2011; Patterson, Proetto, and Gianneschi 2015; S. Wang et al. 2018; Bhattarai and Prozorov 2019; Chee et al. 2016; Colliex 2012; Arnould et al. 2018; Cepeda-Pérez and de Jonge 2019). Cepeda-Perez and de Jonge observed the dynamics of colloidal chitosan-coated gold nanoparticles (TCHIT-AuNP) and branched polyethylenimine (BPEI) coated AuNPs using in-situ scanning transmission electron microscopy with the help of an annular dark field detector. At lower electron rates (~1 e/Å2s) they observed that the AuNPs could form clusters, while at relatively higher dose rates (~ 6 e/Å2s) they could fragment. Finally, at even higher dose rates (~25 e/Å2s) they observed that these AuNPs moved very slowly as compared to Brownian motion in liquid despite the fact the NPs were not in contact with the silicon nitride membranes.

2.4 Future Opportunities

2.4.1 Machine Learning Materials Scientists have begun to employ artificial intelligence (AI) and machine learning to help reduce the manual rigor involved in quantification. Some examples include nanoparticles, defects and, intramolecular bonding in water (Patra et al. 2018; Ueno and Tanimura 2020; Timoshenko et al. 2019). It is only until recently that machine learning has started to be used to examine the dynamics of in-situ TEM (Yao et al. 2020; Chakraborty et al. 2020; X. Li et al. 2020; Oxley et al. 2020; Vasudevan et al. 2020). Using a statistical learning approach on a collection of frames

39 consisting of various transformations, Li et al. developed a certain reaction- convection-diffusion model to track spatial-temporal patterns in in-situ STEM videos of Pt nanoparticle formation and graphene contamination (X. Li et al. 2020). Although studies like these are beginning to emerge, in our opinion, what is limiting the further use of machine learning and AI in in-situ/operando TEM is the sheer complexity of the experimental procedure itself. It is difficult for a computer to understand what is going on when it looks at millions of frames capturing dynamics of droplets or particles. Extraordinary amounts of data need to be fed to the computer for it to be able to reliably quantify our observations. Also bear in mind that the experiments involve a subtle interplay between a large number of different hardware and software interfaces including the potentiostat or any other probe, the microscope, the camera and the video-recording system. However as familiarity with these approaches continues to grow, we anticipate that the microscope manufacturers will help to integrate these capabilities directly into the instrument systems themselves.

2.4.2 Vibrational spectroscopy in liquid-phase electron microscopy

Vibrational spectroscopy of materials inside the electron microscope has been made possible by recent improvements in the energy resolution of electron energy loss spectrometers (EELS) (Krivanek et al. 2014; Hachtel et al. 2019; Rez et al. 2016). Krivanek et al. showed that with the current energy resolution (~ 10 meV) of EELS systems, it is possible to obtain chemically-sensitive information complementary to infrared, Raman, and inelastic neutron scattering (Krivanek et al. 2014). They discussed examples of applications in inorganic and organic materials, including the

40 detection of hydrogen. They also showed that the vibrational signal can be used to map features at nanometer-level resolution. For instance, they demonstrated the difference in the observation of vibrational peaks due to hydrogen in TiH2 and in the epoxy resin. Since the hydrogen is mobile and bound weakly in TiH2, it results in the relatively low (for hydrogen) vibrational energy of 147 meV. They saw evidence for a 360 meV peak in epoxy (2,900 cm-1) which is a typical C–H stretch vibrational energy. To execute these experiments successfully, they cite three major developments that have played an important part: (1) monochromators that are able to reach the required energy resolution and that allow an atom-size probe to be formed; (2) ultra- bright cold field-emission electron guns and aberration-corrected optics; and (3) operating modes that reduce the tail of the intense zero loss peak (ZLP) in the EELS spectrum. To date there are have not yet been detailed studies of the vibrational spectroscopy liquid-phase electron microscopy by EELS, evidently due to the challenges of the liquid thickness, additional scattering and relatively small viewing areas. It will require cointinued collaborations between the materials researchers, instrumentation experts and the manufacturers to optimize the process of obtaining vibrational spectra in liquid-phase electron microscopy. In addition to the physical contrast, chemical signals from techniques like vibrational spectroscopy combined with the dynamics of the process would be a paradigm shift in the way we look at various phenomena and the methodologies we employ to understand them.

REFERENCES

1. Abrams, I. M., and J. W. McBain. 1944. “A Closed Cell for Electron Microscopy.” Journal of Applied Physics 15 (8): 607–9. https://doi.org/10.1063/1.1707475.

2. Aplan, Melissa P., Christopher Grieco, Youngmin Lee, Jason M. Munro, Wonho Lee, Jennifer L. Gray, Zach D. Seibers, et al. 2019. “Conjugated Block Copolymers as Model Systems to Examine Mechanisms of Charge Generation in Donor–Acceptor Materials.” Advanced Functional Materials 29 (1): 1804858. https://doi.org/10.1002/adfm.201804858.

3. Arnould, Amandine, Fanny Caputo, Maria Bacia, Isabelle Texier, Marie Escude, Delphine Boutry, Aurelien Auger, Romain Soulas, and Jean- Francois Damlencourt. 2018. “Contrast Enhancement for Lipid NanoParticles (LNPs) Characterization Using Transmission Electron Microscopy (TEM).” Microscopy and Microanalysis 24 (S1): 314–15. https://doi.org/10.1017/S1431927618002064.

4. A. Sutter, Eli, and Peter W. Sutter. 2017. “In Situ Liquid Cell Electron Microscopy of Ag–Au Galvanic Replacement Reactions.” Nanoscale 9 (3): 1271–78. https://doi.org/10.1039/C6NR07293C.

5. Bhattarai, Nabraj, and Tanya Prozorov. 2019. “Direct Observation of Early Stages of Growth of Multilayered DNA-Templated Au-Pd-Au Core-Shell Nanoparticles in Liquid Phase.” Frontiers in Bioengineering and Biotechnology 7 (February). https://doi.org/10.3389/fbioe.2019.00019.

6. Celik, F. A., and A. K. Yildiz. 2015. “Molecular Dynamics Simulation of Crystallization Kinetics and Homogenous Nucleation of Pt–Rh Alloy.” Journal of Non-Crystalline Solids 415 (May): 36–41. https://doi.org/10.1016/j.jnoncrysol.2015.02.011.

7. Cepeda-Pérez, Elisa, and Niels de Jonge. 2019. “Dynamics of Gold Nanoparticle Clusters Observed with Liquid-Phase Electron Microscopy.” Micron 117 (February): 68–75. https://doi.org/10.1016/j.micron.2018.11.006.

8. Chakraborty, M., M. Ziatdinov, O. Dyck, S. Jesse, A. D. White, and Sergei V. Kalinin. 2020. “Reconstruction of the Interatomic Forces from Dynamic Scanning Transmission Electron Microscopy Data.” Journal of Applied Physics 127 (22): 224301. https://doi.org/10.1063/5.0009413.

9. Chee, See Wee, Duane Loh, Zhaslan Baraissov, Paul Matsudaira, and Utkur Mirsaidov. 2016. “Hopping Diffusion of Gold Nanoparticles Observed with Liquid Cell TEM.” Microscopy and Microanalysis 22 (S3): 750–51. https://doi.org/10.1017/S1431927616004608.

10. Chee, See Wee, Frances M. Ross, David Duquette, and Robert Hull. 2013. “Studies of Corrosion of Al Thin Films Using Liquid Cell Transmission Electron Microscopy.” MRS Online Proceedings Library Archive 1525. https://doi.org/10.1557/opl.2013.558.

11. Cheek, Quintin, Eli Fahrenkrug, Sofiya Hlynchuk, Daan Hein Alsem, Norman J. Salmon, and Stephen Maldonado. 2020. “In Situ Transmission Electron Microscopy Measurements of Ge Nanowire Synthesis with Liquid Metal Nanodroplets in Water.” ACS Nano 14 (3): 2869–79. https://doi.org/10.1021/acsnano.9b06468.

12. Chen, Fu-Chun, Jui-Yuan Chen, Ya-Hsuan Lin, Ming-Yu Kuo, Yung-Jung Hsu, and Wen-Wei Wu. 2019. “In Situ TEM Observation of Au–Cu2O Core–Shell Growth in Liquids.” Nanoscale 11 (21): 10486–92. https://doi.org/10.1039/C9NR00972H.

13. Chen, Qiaoli, Christian Dwyer, Guan Sheng, Chongzhi Zhu, Xiaonian Li, Changlin Zheng, and Yihan Zhu. n.d. “Imaging Beam-Sensitive Materials by Electron Microscopy.” Advanced Materials n/a (n/a): 1907619. https://doi.org/10.1002/adma.201907619.

14. Chen, Xin, Chang Li, Xing Kong, Hongliang Cao, Hulian Wang, and Xiaoqin Zhou. 2017. “Direct Observation of Growth and Self-Assembly of Pt Nanoclusters in Water with the Aid of a Triblock Polymer Using in Situ Liquid Cell Transmission Electron Microscopy (TEM).” Chinese Journal of Chemistry 35 (8): 1278–83. https://doi.org/10.1002/cjoc.201700017.

15. Chen, Yu-Zhen, Rui Zhang, Long Jiao, and Hai-Long Jiang. 2018. “Metal– Organic Framework-Derived Porous Materials for Catalysis.” Coordination Chemistry Reviews 362 (May): 1–23. https://doi.org/10.1016/j.ccr.2018.02.008.

16. Colliex, Christian. 2012. “Watching Solution Growth of Nanoparticles in Graphene Cells.” Science 336 (6077): 44–45. https://doi.org/10.1126/science.1219835.

17. Davies, H. A., and J. B. Hull. 1976. “The Formation, Structure and Crystallization of Non-Crystalline Nickel Produced by Splat-Quenching.” Journal of Materials Science 11 (2): 215–23. https://doi.org/10.1007/BF00551430.

18. Degen, Katherine, Madeline Dukes, Justin R. Tanner, and Deborah F. Kelly. 2012. “The Development of Affinity Capture Devices—a Nanoscale Purification Platform for Biological in Situ Transmission Electron Microscopy.” RSC Advances 2 (6): 2408–12. https://doi.org/10.1039/C2RA01163H.

19. Dukes, M.J., K.L. Klein, R. Thomas, and J. Damiano. 2013. “Development of Reinforced Silicon Nitride Membranes for in Situ Liquid Electron Microscopy.” Microscopy and Microanalysis 19 (S2): 406–7. https://doi.org/10.1017/S1431927613004029.

20. Early, Julia T., Kevin G. Yager, and Timothy P. Lodge. 2020. “Direct Observation of Micelle Fragmentation via In Situ Liquid-Phase Transmission Electron Microscopy.” ACS Macro Letters 9 (5): 756–61. https://doi.org/10.1021/acsmacrolett.0c00273.

21. Evans, James E., Katherine L. Jungjohann, Nigel D. Browning, and Ilke Arslan. 2011. “Controlled Growth of Nanoparticles from Solution with In Situ Liquid Transmission Electron Microscopy.” Nano Letters 11 (7): 2809–13. https://doi.org/10.1021/nl201166k.

22. Evans, James E., Katherine L. Jungjohann, Peony C. K. Wong, Po-Lin Chiu, Gavin H. Dutrow, Ilke Arslan, and Nigel D. Browning. 2012. “Visualizing Macromolecular Complexes with in Situ Liquid Scanning Transmission Electron Microscopy.” Micron, In situ TEM, 43 (11): 1085– 90. https://doi.org/10.1016/j.micron.2012.01.018.

23. Garza, Héctor Hugo Pérez, Diederik Morsink, Jeff Xu, Justus Hermannsdörfer, Mariya Sholkina, Merijn Pen, Sander van Weperen, and Niels de Jonge. 2016. “The ‘Ocean’ System: Microfluidic Based System for in-Situ Analysis of Liquid Processes inside the TEM.” In European Microscopy Congress 2016: Proceedings, 217–18. American Cancer Society. https://doi.org/10.1002/9783527808465.EMC2016.6642.

24. Grogan, Joseph M., and Haim H. Bau. 2010. “The Nanoaquarium: A Platform for In Situ Transmission Electron Microscopy in Liquid Media.” Journal of Microelectromechanical Systems 19 (4): 885–94. https://doi.org/10.1109/JMEMS.2010.2051321.

25. Guo, Changhe, Frances I. Allen, Youngmin Lee, Thinh P. Le, Chengyu Song, Jim Ciston, Andrew M. Minor, and Enrique D. Gomez. 2015. “Probing Local Electronic Transitions in Organic Semiconductors through Energy-Loss Spectrum Imaging in the Transmission Electron Microscope.” Advanced Functional Materials 25 (38): 6071–6076. https://doi.org/10.1002/adfm.201502090.

26. Hachtel, Jordan A., Jingsong Huang, Ilja Popovs, Santa Jansone-Popova, Jong K. Keum, Jacek Jakowski, Tracy C. Lovejoy, Niklas Dellby, Ondrej L. Krivanek, and Juan Carlos Idrobo. 2019. “Identification of Site-Specific Isotopic Labels by Vibrational Spectroscopy in the Electron Microscope.” Science 363 (6426): 525–28. https://doi.org/10.1126/science.aav5845.

27. He, Kun, Tolou Shokuhfar, and Reza Shahbazian-Yassar. 2019. “Imaging of Soft Materials Using in Situ Liquid-Cell Transmission Electron Microscopy.” Journal of Physics: Condensed Matter 31 (10): 103001. https://doi.org/10.1088/1361-648X/aaf616.

28. Holtz, Megan E., Yingchao Yu, Jie Gao, Héctor D. Abruña, and David A. Muller. 2013. “In Situ Electron Energy-Loss Spectroscopy in Liquids.” Microscopy and Microanalysis 19 (4): 1027–35. https://doi.org/10.1017/S1431927613001505.

29. Impagnatiello, Andrea, Carolina Ferreira Cerqueira, Pierre-Eugène Coulon, Arnaud Morin, Sylvie Escribano, Laure Guetaz, Marie-Claude Clochard, and Giancarlo Rizza. 2020. “Degradation Mechanisms of Supported Pt Nanocatalysts in Proton Exchange Membrane Fuel Cells: An Operando Study through Liquid Cell Transmission Electron Microscopy.” ACS Applied Energy Materials, March. https://doi.org/10.1021/acsaem.9b02000.

30. Jonge, Niels de, Marina Pfaff, and Diana B. Peckys. 2014. “Chapter One - Practical Aspects of Transmission Electron Microscopy in Liquid.” In Advances in Imaging and Electron Physics, edited by Peter W. Hawkes, 186:1–37. Elsevier. https://doi.org/10.1016/B978-0-12-800264-3.00001-0.

31. Jonge, Niels de, Nicolas Poirier-Demers, Hendrix Demers, Diana B. Peckys, and Dominique Drouin. 2010. “Nanometer-Resolution Electron Microscopy through Micrometers-Thick Water Layers.” Ultramicroscopy 110 (9): 1114–19. https://doi.org/10.1016/j.ultramic.2010.04.001.

32. Karakulina, Olesia M., Arnaud Demortière, Walid Dachraoui, Artem M. Abakumov, and Joke Hadermann. 2018. “In Situ Electron Diffraction Tomography Using a Liquid-Electrochemical Transmission Electron Microscopy Cell for Crystal Structure Determination of Cathode Materials for Li-Ion Batteries.” Nano Letters 18 (10): 6286–91. https://doi.org/10.1021/acs.nanolett.8b02436.

33. Keskin, Sercan, and Niels de Jonge. 2018. “Reduced Radiation Damage in Transmission Electron Microscopy of Proteins in Graphene Liquid Cells.” Nano Letters 18 (12): 7435–40. https://doi.org/10.1021/acs.nanolett.8b02490.

34. Kraus, Tobias, and Niels de Jonge. 2013. “Dendritic Gold Nanowire Growth Observed in Liquid with Transmission Electron Microscopy.” Langmuir 29 (26): 8427–32. https://doi.org/10.1021/la401584z.

35. Krivanek, Ondrej L., Tracy C. Lovejoy, Niklas Dellby, Toshihiro Aoki, R. W. Carpenter, Peter Rez, Emmanuel Soignard, et al. 2014. “Vibrational Spectroscopy in the Electron Microscope.” Nature 514 (7521): 209–12. https://doi.org/10.1038/nature13870.

36. Kumar, Satish, and W. Wade Adams. 1990. “Electron Beam Damage in High Temperature Polymers.” Polymer 31 (1): 15–19. https://doi.org/10.1016/0032-3861(90)90341-U.

37. Leenheer, Andrew J., Katherine L. Jungjohann, Kevin R. Zavadil, John P. Sullivan, and C. Thomas Harris. 2015. “Lithium Electrodeposition Dynamics in Aprotic Electrolyte Observed in Situ via Transmission Electron Microscopy.” ACS Nano 9 (4): 4379–89. https://doi.org/10.1021/acsnano.5b00876.

38. Leijten, Zino J. W. A., Arthur D. A. Keizer, Gijsbertus de With, and Heiner Friedrich. 2017. “Quantitative Analysis of Electron Beam Damage in Organic Thin Films.” The Journal of Physical Chemistry C 121 (19): 10552–61. https://doi.org/10.1021/acs.jpcc.7b01749.

39. Li, Junjie, Yunping Li, Qiang Li, Zhongchang Wang, and Francis Leonard Deepak. 2019. “Atomic-Scale Dynamic Observation Reveals Temperature- Dependent Multistep Nucleation Pathways in Crystallization.” Nanoscale Horizons 4 (6): 1302–9. https://doi.org/10.1039/C9NH00308H.

40. Li, Xin, Ondrej Dyck, Raymond R. Unocic, Anton V. Ievlev, Stephen Jesse, and Sergei V. Kalinin. 2020. “Statistical Learning of Governing Equations of Dynamics from In-Situ Electron Microscopy Imaging Data.” Materials & Design 195 (October): 108973. https://doi.org/10.1016/j.matdes.2020.108973.

41. Liao, Hong-Gang, and Haimei Zheng. 2016. “Liquid Cell Transmission Electron Microscopy.” Annual Review of Physical Chemistry 67 (1): 719– 47. https://doi.org/10.1146/annurev-physchem-040215-112501.

42. Liu, Changhao, Isao Noda, David C. Martin, D. Bruce Chase, Chaoying Ni, and John F. Rabolt. 2018. “Growth of Anisotropic Single Crystals of a Random Copolymer, Poly[(R)-3-Hydroxybutyrate-Co-(R)-3- Hydroxyhexanoate] Driven by Cooperative –CH···O H-Bonding.” Polymer 154 (October): 111–18. https://doi.org/10.1016/j.polymer.2018.08.046.

43. Liu, Jinglin, Bin Wei, Jennifer D. Sloppy, Liangqi Ouyang, Chaoying Ni, and David C. Martin. 2015. “Direct Imaging of the Electrochemical Deposition of Poly(3,4-Ethylenedioxythiophene) by Transmission Electron Microscopy.” ACS Macro Letters 4 (9): 897–900. https://doi.org/10.1021/acsmacrolett.5b00479.

44. Liu, Lili, Yi Liu, Wenjun Wu, Christopher. M. Miller, and Elizabeth C. Dickey. 2015. “Visualization of Film-Forming Polymer Particles with a Liquid Cell Technique in a Transmission Electron Microscope.” The Analyst 140 (18): 6330–34. https://doi.org/10.1039/C5AN01067E.

45. Liu, Ting, Xiangyu Dou, Yonghui Xu, Yongjun Chen, and Yongsheng Han. 2020. “In Situ Investigation of Dynamic Silver Crystallization Driven by Chemical Reaction and Diffusion.” Research Article. Research. Science Partner Journal. February 10, 2020. https://doi.org/10.34133/2020/4370817.

46. Lutz, Lukas, Walid Dachraoui, Arnaud Demortière, Lee R. Johnson, Peter G. Bruce, Alexis Grimaud, and Jean-Marie Tarascon. 2018. “Operando Monitoring of the Solution-Mediated Discharge and Charge Processes in a Na–O2 Battery Using Liquid-Electrochemical Transmission Electron Microscopy.” Nano Letters 18 (2): 1280–89. https://doi.org/10.1021/acs.nanolett.7b04937.

47. Lyu, Jiafei, Xinyi Gong, Seung-Joon Lee, Karthikeyan Gnanasekaran, Xuan Zhang, Megan C. Wasson, Xingjie Wang, et al. 2020. “Phase Transitions in Metal–Organic Frameworks Directly Monitored through In Situ Variable Temperature Liquid-Cell Transmission Electron Microscopy and In Situ X-Ray Diffraction.” Journal of the American Chemical Society 142 (10): 4609–15. https://doi.org/10.1021/jacs.0c00542.

48. Martin, David C., Jihua Chen, Junyan Yang, Lawrence F. Drummy, and Christian Kübel. 2005. “High Resolution Electron Microscopy of Ordered Polymers and Organic Molecular Crystals: Recent Developments and Future Possibilities.” Journal of Polymer Science Part B: Polymer Physics 43 (14): 1749–78. https://doi.org/10.1002/polb.20419.

49. Mehdi, B. L., J. Qian, E. Nasybulin, C. Park, D. A. Welch, R. Faller, H. Mehta, et al. 2015. “Observation and Quantification of Nanoscale Processes in Lithium Batteries by Operando Electrochemical (S)TEM.” Nano Letters 15 (3): 2168–73. https://doi.org/10.1021/acs.nanolett.5b00175.

50. Moser, Trevor H., Hardeep Mehta, Chiwoo Park, Ryan T. Kelly, Tolou Shokuhfar, and James E. Evans. 2018. “The Role of Electron Irradiation History in Liquid Cell Transmission Electron Microscopy.” Science Advances 4 (4): eaaq1202. https://doi.org/10.1126/sciadv.aaq1202.

51. Nielsen, Michael H., Dongsheng Li, Hengzhong Zhang, Shaul Aloni, T. Yong-Jin Han, Cathrine Frandsen, Jong Seto, Jillian F. Banfield, Helmut Cölfen, and James J. De Yoreo. 2014. “Investigating Processes of Nanocrystal Formation and Transformation via Liquid Cell TEM.” Microscopy and Microanalysis: The Official Journal of Microscopy Society of America, Microbeam Analysis Society, Microscopical Society of Canada 20 (2): 425–36. https://doi.org/10.1017/S1431927614000294.

52. Oxley, Mark P., Junqi Yin, Nikolay Borodinov, Suhas Somnath, Maxim Ziatdinov, Andrew Lupini, Stephen Jesse, Rama K. Vasudevan, and Sergei V. Kalinin. 2020. “Deep Learning of Interface Structures from Simulated 4D STEM Data: Cation Intermixing vs. Roughening.” Machine Learning: Science and Technology. https://doi.org/10.1088/2632-2153/aba32d.

53. Panciera, Federico, Zhaslan Baraissov, Gilles Patriarche, Vladimir G. Dubrovskii, Frank Glas, Laurent Travers, Utkur Mirsaidov, and Jean- Christophe Harmand. 2020. “Phase Selection in Self-Catalyzed GaAs Nanowires.” Nano Letters 20 (3): 1669–75. https://doi.org/10.1021/acs.nanolett.9b04808.

54. Parent, Lucas R., Evangelos Bakalis, Abelardo Ramírez-Hernández, Jacquelin K. Kammeyer, Chiwoo Park, Juan de Pablo, Francesco Zerbetto, Joseph P. Patterson, and Nathan C. Gianneschi. 2017. “Directly Observing Micelle Fusion and Growth in Solution by Liquid-Cell Transmission Electron Microscopy.” Journal of the American Chemical Society 139 (47): 17140–51. https://doi.org/10.1021/jacs.7b09060.

55. Patra, Tarak K., Fu Zhang, Daniel S. Schulman, Henry Chan, Mathew J. Cherukara, Mauricio Terrones, Saptarshi Das, Badri Narayanan, and Subramanian K. R. S. Sankaranarayanan. 2018. “Defect Dynamics in 2-D MoS2 Probed by Using Machine Learning, Atomistic Simulations, and High-Resolution Microscopy.” ACS Nano 12 (8): 8006–16. https://doi.org/10.1021/acsnano.8b02844.

56. Patterson, Joseph P., Maria T. Proetto, and Nathan C. Gianneschi. 2015. “Soft Nanomaterials Analysed by in Situ Liquid TEM: Towards High Resolution Characterisation of Nanoparticles in Motion.” Perspectives in Science, Proceedings of the Beilstein Bozen Symposium 2014 – Chemistry and Time, 6 (December): 106–12. https://doi.org/10.1016/j.pisc.2015.10.003.

57. Radisic, Aleksandar, Philippe M. Vereecken, James B. Hannon, Peter C. Searson, and Frances M. Ross. 2006. “Quantifying Electrochemical Nucleation and Growth of Nanoscale Clusters Using Real-Time Kinetic Data.” Nano Letters 6 (2): 238–42. https://doi.org/10.1021/nl052175i.

58. Rehn, Sarah M., and Matthew R. Jones. 2018. “New Strategies for Probing Energy Systems with In Situ Liquid-Phase Transmission Electron Microscopy.” ACS Energy Letters 3 (6): 1269–78. https://doi.org/10.1021/acsenergylett.8b00527.

59. Ren, Amy, David Lu, Edward Wong, Matthew R. Hauwiller, A. Paul Alivisatos, and Gang Ren. 2020. “Real-Time Observation of Dynamic Structure of Liquid-Vapor Interface at Nanometer Resolution in Electron Irradiated Sodium Chloride Crystals.” Scientific Reports 10 (1): 8596. https://doi.org/10.1038/s41598-020-65274-9.

60. Rez, Peter, Toshihiro Aoki, Katia March, Dvir Gur, Ondrej L. Krivanek, Niklas Dellby, Tracy C. Lovejoy, Sharon G. Wolf, and Hagai Cohen. 2016. “Damage-Free Vibrational Spectroscopy of Biological Materials in the Electron Microscope.” Nature Communications 7 (1): 10945. https://doi.org/10.1038/ncomms10945.

61. Sacci, Robert L., Nancy J. Dudney, Karren L. More, and Raymond R. Unocic. 2014. “In Operando Transmission Electron Microscopy Imaging of SEI Formation and Structure in Li-Ion and Li-Metal Batteries.” Microscopy and Microanalysis 20 (S3): 1538–39. https://doi.org/10.1017/S1431927614009428.

62. Schneider, Nicholas M., Michael M. Norton, Brian J. Mendel, Joseph M. Grogan, Frances M. Ross, and Haim H. Bau. 2014. “Electron-Water Interactions and Implications for Liquid Cell Electron Microscopy.” Journal of Physical Chemistry C 118 (38): 22373–22382. https://doi.org/10.1021/jp507400n.

63. Shen, Kui, Lei Zhang, Xiaodong Chen, Lingmei Liu, Daliang Zhang, Yu Han, Junying Chen, et al. 2018. “Ordered Macro-Microporous Metal- Organic Framework Single Crystals.” Science 359 (6372): 206–10. https://doi.org/10.1126/science.aao3403.

64. Subramanian, Vivek, Jinglin Liu, Bin Wei, and David Charles Martin. 2019. “In-Situ Imaging of Electrochemical Polymerization of Functionalized Thiophenes Using Transmission Electron Microscopy.” ECS Meeting Abstracts MA2019-01 (47): 2258. https://doi.org/10.1149/MA2019-01/47/2258.

65. Subramanian, Vivek, Casey A Rowland, Glenn P. A. Yap, and David C. Martin. 2019. “Morphology, Molecular Orientation and Solid-State Characterization of 2,3-Dihydrothieno[3,4-b] [1,4]Dioxine-2-Carboxylic Acid (EDOTacid).” Crystal Growth & Design, September. https://doi.org/10.1021/acs.cgd.9b00613.

66. Sutter, Eli, Peter Sutter, Alexei V. Tkachenko, Roman Krahne, Joost de Graaf, Milena Arciniegas, and Liberato Manna. 2016. “In Situ Microscopy of the Self-Assembly of Branched Nanocrystals in Solution.” Nature Communications 7 (1): 11213. https://doi.org/10.1038/ncomms11213.

67. Timoshenko, Janis, Cody J. Wrasman, Mathilde Luneau, Tanya Shirman, Matteo Cargnello, Simon R. Bare, Joanna Aizenberg, Cynthia M. Friend, and Anatoly I. Frenkel. 2019. “Probing Atomic Distributions in Mono- and Bimetallic Nanoparticles by Supervised Machine Learning.” Nano Letters 19 (1): 520–29. https://doi.org/10.1021/acs.nanolett.8b04461.

68. Touve, Mollie A., C. Adrian Figg, Daniel B. Wright, Chiwoo Park, Joshua Cantlon, Brent S. Sumerlin, and Nathan C. Gianneschi. 2018. “Polymerization-Induced Self-Assembly of Micelles Observed by Liquid Cell Transmission Electron Microscopy.” ACS Central Science 4 (5): 543– 47. https://doi.org/10.1021/acscentsci.8b00148.

69. Ueno, Seiji, and Yoshitaka Tanimura. 2020. “Modeling Intermolecular and Intramolecular Modes of Liquid Water Using Multiple Heat Baths: Machine Learning Approach.” Journal of Chemical Theory and Computation 16 (4): 2099–2108. https://doi.org/10.1021/acs.jctc.9b01288.

70. Vailonis, Kristina M., Karthikeyan Gnanasekaran, Xian B. Powers, Nathan C. Gianneschi, and David M. Jenkins. 2019. “Elucidating the Growth of Metal–Organic Nanotubes Combining Isoreticular Synthesis with Liquid- Cell Transmission Electron Microscopy.” Journal of the American Chemical Society 141 (26): 10177–82. https://doi.org/10.1021/jacs.9b04586.

71. VandenBussche, Elisah J., and David J. Flannigan. 2019. “Reducing Radiation Damage in Soft Matter with Femtosecond-Timed Single- Electron Packets.” Nano Letters 19 (9): 6687–94. https://doi.org/10.1021/acs.nanolett.9b03074.

72. Vasudevan, Rama K., Maxim Ziatdinov, Lukas Vlcek, Anna N. Morozovska, Eugene A. Eliseev, Shi-Ze Yang, Yongji Gong, et al. 2020. “Investigating Phase Transitions from Local Crystallographic Analysis Based on Machine Learning of Atomic Environments.” ArXiv:2006.10001 [Cond-Mat], June. http://arxiv.org/abs/2006.10001.

73. Wang, Rui-Yang, Shu-Fen Zou, Bai-Yu Jiang, Bin Fan, Meng-Fei Hou, Biao Zuo, Xin-Ping Wang, Jun-Ting Xu, and Zhi-Qiang Fan. 2017. “A Generalized Avrami Equation for Crystallization Kinetics of Polymers with Concomitant Double Crystallization Processes.” Crystal Growth & Design 17 (11): 5908–17. https://doi.org/10.1021/acs.cgd.7b01016.

74. Wang, Shunzhi, C. Michael McGuirk, Andrea d’Aquino, Jarad A. Mason, and Chad A. Mirkin. 2018. “Metal–Organic Framework Nanoparticles.” Advanced Materials 30 (37): 1800202. https://doi.org/10.1002/adma.201800202.

75. Woehl, T. J., and P. Abellan. 2017. “Defining the radiation chemistry during liquid cell electron microscopy to enable visualization of nanomaterial growth and degradation dynamics.” Journal of Microscopy 265 (2): 135–47. https://doi.org/10.1111/jmi.12508.

76. Woehl, Taylor J., Katherine L. Jungjohann, James E. Evans, Ilke Arslan, William D. Ristenpart, and Nigel D. Browning. 2013. “Experimental Procedures to Mitigate Electron Beam Induced Artifacts during in Situ Fluid Imaging of Nanomaterials.” Ultramicroscopy, Frontiers of Electron Microscopy in Materials Science, 127 (April): 53–63. https://doi.org/10.1016/j.ultramic.2012.07.018.

77. Wu, Jianbo, Wenpei Gao, Hong Yang, and Jian-Min Zuo. 2014. “Imaging Shape-Dependent Corrosion Behavior of Pt Nanoparticles over Extended Time Using a Liquid Flow Cell and TEM.” Microscopy and Microanalysis. Cambridge University Press. August 2014. https://doi.org/10.1017/S1431927614009271.

78. Wunderlich, B. 2008. “Thermodynamics and Kinetics of Crystallization of Flexible Molecules.” Journal of Polymer Science Part B: Polymer Physics 46 (24): 2647–59. https://doi.org/10.1002/polb.21597.

79. Xu, Guoguang, Xinyi Zhang, Meinan Liu, Hongfei Li, Meizhen Zhao, Qingwen Li, Jinping Zhang, and Yuegang Zhang. n.d. “Multi-Step Phase Transitions of Mn3O4 During Galvanostatic Lithiation: An In Situ Transmission Electron Microscopic Investigation.” Small n/a (n/a): 1906499. https://doi.org/10.1002/smll.201906499.

80. Yancey, David F., Christian Kisielowski, Petra Specht, Steve Rozeveld, Joo Kang, and Peter Nickias. 2018. “Understanding Ziegler-Natta Catalyst Structure via Low-Dose Transmission Electron Microscopy.” Microscopy and Microanalysis 24 (S1): 1968–69. https://doi.org/10.1017/S1431927618010322.

81. Yang, Zhenzhong, Phuong-Vu Ong, Yang He, Le Wang, Mark E. Bowden, Wu Xu, Timothy C. Droubay, Chongmin Wang, Peter V. Sushko, and Yingge Du. 2018. “In Situ Transmission Electron Microscopy: Direct Visualization of Li Dendrite Effect on LiCoO2 Cathode by In Situ TEM (Small 52/2018).” Small 14 (52): 1870252. https://doi.org/10.1002/smll.201870252.

82. Yao, Lehan, Zihao Ou, Binbin Luo, Cong Xu, and Qian Chen. 2020. “Machine Learning to Reveal Nanoparticle Dynamics from Liquid-Phase TEM Videos.” ACS Central Science 6 (8): 1421–30. https://doi.org/10.1021/acscentsci.0c00430.

83. Yuan, Yifei, Khalil Amine, Jun Lu, and Reza Shahbazian-Yassar. 2017. “Understanding Materials Challenges for Rechargeable Ion Batteries with in Situ Transmission Electron Microscopy.” Nature Communications 8 (1): 15806. https://doi.org/10.1038/ncomms15806.

84. Yuk, Jong Min, Jungwon Park, Peter Ercius, Kwanpyo Kim, Daniel J. Hellebusch, Michael F. Crommie, Jeong Yong Lee, A. Zettl, and A. Paul Alivisatos. 2012. “High-Resolution EM of Colloidal Nanocrystal Growth Using Graphene Liquid Cells.” Science 336 (6077): 61–64. https://doi.org/10.1126/science.1217654.

85. Zeng, Zhiyuan, Wenjing Zheng, and Haimei Zheng. 2017. “Visualization of Colloidal Nanocrystal Formation and Electrode–Electrolyte Interfaces in Liquids Using TEM.” Accounts of Chemical Research 50 (8): 1808–17. https://doi.org/10.1021/acs.accounts.7b00161.

86. Zheng, H., R. K. Smith, Y.-w. Jun, C. Kisielowski, U. Dahmen, and A. P. Alivisatos. 2009. “Observation of Single Colloidal Platinum Nanocrystal Growth Trajectories.” Science 324 (5932): 1309–12. https://doi.org/10.1126/science.1172104.

Chapter 3

IN-SITU TRANSMISSION ELECTRON MICROSCOPY (TEM) OF THE ELECTROCHEMICALLY-DRIVEN NUCLEATION, GROWTH, AND SOLIDIFICATION OF POLY(3,4-ETHYLENEDIOXYTHIOPHENE) (PEDOT)

3.1 Introduction The conjugated polymer poly(3,4-ethylenedioxythiophene) (PEDOT) has received widespread recent interest due to its high chemical stability, biocompatibility, low oxidation potential and high conductivity (Martin et al. 2010; Groenendaal et al. 2000; Crispin et al. 2006). Its use in biomedical device-tissue interfaces has been of particular interest due to its ability to provide a mechanically compliant interface and transport charge both ionically and electronically (Martin 2015). PEDOT-coated electrodes typically have impedances around two to three orders of magnitude lower than uncoated flat metal electrodes in the low-frequency region (<1 kHz) of particular interest for biological signals (Koutsouras et al. 2017). Recent studies have also shown that using substituted EDOT monomers can have significant effects on the adhesion of the polymer film on inorganic substrates (Ouyang et al. 2017; B. Wei et al. 2015). Copolymers of PEDOT and a more hydrophilic comonomer (EDOTacid) have also been investigated to make PEDOT more biocompatible (Subramanian et al. 2019; Bhagwat, Kiick, and Martin 2014). A variety of methods can be employed to polymerize PEDOT, namely chemical oxidation, electrochemical deposition, radiation-induced polymerization, or chemical vapor deposition (Ghosh et al. 2016; Winther-Jensen and West 2004; Lock,

Im, and Gleason 2006; Lattach et al. 2014; 2013). Amongst these methods, chemical oxidation (followed by spin-coating onto the electrode) and electrochemical deposition have been extensively used for modifying substrates and electrodes. In a typical three- electrode electrochemical polymerization system (Figure 3.1), the working, the counter, and the reference electrodes are immersed in a monomer solution. The polymerization reaction takes place when the applied potential is higher than the oxidation potential of the monomer. The polymer is then deposited onto the working electrode (anode). Since electrodeposition requires charge transport between the working electrode and the Figure 3.1: Schematic of a typical 3 - electrode system depositing material, this method utilized for electrodeposition) conjugated polymers. ensures electrical contact between the substrate and the deposited polymer. An added advantage of this method is that it can also be useful for coating selected areas or coating patterned surfaces (Mariani et al. 2011). Recent studies have also shown that electrodeposition can also be used to deposit PEDOT from cuff electrodes around individual peripheral nerves (Murbach et al. 2018; Tong et al. 2018).

Electrodeposition or electroplating of inorganic metals has been employed for a variety of purposes. In electrodeposition of inorganic metals, there is an immediate first-order transition from an ionic liquid to a dense, solid, crystalline product. This makes it relatively straightforward to estimate or control the thicknesses of the materials deposited. On the contrary, the mechanism of deposition for conjugated polymers is more complicated than that of inorganic metal oxides or metals. During the polymerization, the monomers first oxidize to form reactive radical cation species (Sabouraud, Sadki, and Brodie 2000) that then react to form dimers, trimers and higher order oligomers near the solution-electrode interface(Figure 3.2). As the average molecular weight increases the film becomes insoluble in the solvent, leading to deposition of a conjugated polymer film on the working electrode.

Electrochemically deposited conjugated polymer films can show dramatic variations in morphology. Typically, the films have a somewhat lumpy surface structure, with details that depend on the deposition conditions and choice of solvent. They can also form relatively fibrillar, low density, open structures. They can also deposit around dissolvable templates, or onto gels or even living tissues if they are present in the reaction medium. Another complication is that the films are typically doped during the process, resulting in a residual positive charge on the molecule backbone that is balanced with a counter anion or dopant. Typical dopants include both

The electrodeposited films can have a wide array of morphologies ranging from being rough and bumpy to microporous (Martin et al. 2010) depending on the counterion, processing conditions or the chemistry of the monomer. Therefore, there is

Figure 3.3: Early stages of the electrochemical polymerization of PEDOT imaged for the first-time using liquid-phase TEM (J. Liu et al. 2015). a critical need to look at the early stages of the electrodeposition reaction to be able to better predict and tune the structure and the properties of the conjugated polymer films. To this end, we have used liquid-phase transmission electron microscopy to observe the nucleation and growth of the electrodeposition reaction of PEDOT and understand how the oligomers in the nascent stages evolve into a mature polymer film. Previously, J. Liu et al. showed that it is possible to observe the oligomeric clusters nucleating and growing from the glass carbon working electrode during the electrochemical polymerization of poly(3,4-ethylenedioxythiophene) (PEDOT)

(Figure 3.3) (J. Liu et al. 2015). However, the early stages of the process, particularly the evolution of the mobile oligomeric clusters, require a much more careful and detailed analyses to better understand the polymerization mechanisms. This work primarily focuses on quantifying and understanding the evolution of nucleation densities, thicknesses and shapes and sizes in detail during the electrochemical polymerization reaction.

3.2 Materials and Methods

3.2.1 Monomer solution: (3,4-ethylenedioxythiophene) (EDOT monomer), lithium perchlorate and poly(acrylic acid) (PAA) (Mw~1800 g/mol) were purchased from Sigma Aldrich. The aqueous solutions used were either 0.01M (3,4-ethylenedioxythiophene) and 0.1 M lithium perchlorate or 0.01M (3,4-ethylenedioxythiophene), 0.1 M lithium perchlorate and 0.5 wt% of PAA.

3.2.2 Liquid flow cell and electrochemistry chips: A commercially available electrochemical holder from Protochips, Inc. was used for performing the in-situ liquid TEM experiments. Two microfabricated chips were used; the top chip had the electrochemical connections and the bottom chip (with a spacer of thickness ~ 500 nm) was used to define the liquid thickness in the beam direction. These chips were both equipped with ~50 nm thick SiN membranes. The dimensions of the window (SiN membrane) were 550 µm x 40 µm on the top chip and

300 µm x 90 µm on the bottom chip. The glassy carbon working electrode (WE) was 20 µm wide. The 100 µm wide circular CE was located 500 µm away from the WE. The CE surrounded the WE and the RE was located away from the WE. Both CE and RE are located off the SiN window. The holder was additionally equipped with two inlets and an outlet all of which were attached to polyetheretherketone (PEEK) tubings for flowing liquid through the holder. The monomer solution (sandwiched between the two microfabricated chips) was passed through the holder at a flow rate of 3 ul/min using a syringe from Hamilton and microfluidic pump from Harvard apparatus.

3.2.3 Transmission Electron Microscopy: The in-situ experiments were performed on a 200 kV Thermofisher Scientific FEI Talos F200C in brightfield imaging mode. The total electron doses used for these experiments were around 6 - 12 e/A2 which were well below the critical dose for electron damage of the PEDOT (~ 60 e/A2).

3.2.4 Electrochemistry: The experiments were either conducted galvanostatically at ~150 nA or potentiostatically at a potential of 1.2 V using a Gamry Reference 600.

3.2.5 Video recording: The in-situ experiments were recorded using VeloxTM and screen recorder software.

3.2.6 Optical Microscopy: Optical micrographs were acquired with an Olympus BX60 BF microscope.

3.2.7 Image analysis: Quantitative analysis of the images were done using Fiji (ImageJ version 1.53a).

3.3 Results and discussion

3.3.1 Nucleation and growth of PEDOT The fundamental difference between electrodeposition of inorganic metals and organic polymers is that the latter does not involve an immediate phase transition from an ionic liquid solution to a solid, crystalline deposit like the former, but rather goes through a mobile viscoelastic oligomeric phase which then increases in molecular weight to become an anodic solid polymer deposit. One of the recent attempts to have a better understanding of the processes involved describes the mobile viscous oligomeric phase as the high density oligomeric region that develops before the formation of the final electrodeposit (Pontificia Universidad Católica de Chile, Facultad de Química, Departamento de Química Inorgánica, Laboratorio de Electroquímica de Polímeros (LEP), Vicuña Mackenna 4860, 7820436, Macul, Santiago, Chile and A. del Valle 2016). The researchers used current-time (i-t) transients to estimate the diffusion coefficients of EDOT in the presence of different supporting electrolytes which influences the nucleation of the high density oligomeric region at the electrode-solution interface and in turn changes the morphology of the final film. Another study of the early stage nucleation of PEDOT in acetonitrile on platinum discussed the effect of monomer concentration and polymerization potential on the nucleation and growth model (current density as a function of time) that they had developed (Randriamahazaka, Noël, and Chevrot 1999). Although these studies have helped us understand the nucleation mechanism of PEDOT better, they require fitting time resolved current data to models of the deposition process. In order to confirm what the actual mechanisms of film deposition

60 at the electrode are, it is necessary to have imaging techniques with the time and spatial resolution to actually monitor these events directly. Atomic Force Microscopy (AFM) was used by Ventosa et.al to image the early stage nucleation and growth mechanisms of PEDOT (Ventosa, Palacios, and Unwin

2008). Their system consisted of water as the solvent, LiClO4 as the counter-ion and Highly Oriented Pyrolytic Graphite (HOPG) as the substrate. They concluded that at low driving force, 2D layer-by-layer growth governs electrodeposition. However, these AFM studies had relatively limited field of view, and did not reveal information about the formation of droplets or their interactions with one another. They also did not provide insight about relatively thick films of interest for most applications. These examples are some of the efforts to have a basic understanding of the polymerization process. Here we have used low-dose liquid in-situ Transmission Electron Microscopy in this study to elucidate some the finer details involved in the process.

The electrochemical polymerization reaction was performed on a glassy carbon working electrode located on the electron transparent SiN window of the microfabricated chip. A smaller chip (spacer thickness ~ 500 nm) was used as the spacer to define the thickness of the liquid flowing inside liquid-cell. The solution containing the dissolved monomers and counter-ions was pumped through the liquid holder. This setup was then transferred onto a Gatan pumping station to ensure that the liquid-cell (especially the SiN membranes) could withstand the pressure inside the TEM column without the membranes/viewing windows disintegrating. Subsequently, several CV cycles were run to ensure complete wetting of the electrodes before loading the holder into the TEM.

These experiments have revealed previously unobserved mechanisms of electrochemical polymerization process of PEDOT like the nucleation of like the nucleation of the liquid-like droplets and their subsequent growth, merging, evaporation and eventual deposition onto the working electrode as a solid, stable and dark conjugated polymer film. Quantitative analysis of the bright field TEM images has provided us with detailed information about nucleation density, droplet area and film thickness as a function of time (total charge). We saw that the density of droplet nuclei increases with time and then reaches a maximum as they grow, eventually coalescing into a continuous film. The maximum nucleation densities observed on the glass carbon working electrode was reasonably consistent (~0.3-0.4 nuclei/um2) during the polymerization reaction. We also observed that some of the oligomeric nuclei dissolved back into the solution suggesting the existence of some degree of reversibility to the process which has never been observed or appreciated. This further underscores that point that the fundamental difference between electrodeposition of inorganic metals and organic polymers is that the latter does not involve an immediate phase transition from an isotropic liquid solution to a solid deposit like the former. On the contrary, the electrochemical polymerization reaction goes through a mobile viscoelastic oligomeric phase which then increases in molecular weight to become an anodic solid polymer deposit. This makes the observation of dissolution of certain oligomeric nuclei even more pertinent and crucial in many ways.

In Figure 3.4(a) we clearly see the early stages of the electrodeposition where the liquid-like oligomers (dark in contrast due to higher mass thickness) initially nucleate from the glassy carbon working electrode (lighter shade of grey compared to the oligomers) then merge, coalesce and increase in size and thickness before finally depositing onto the working electrode as a solid, stable and dark conjugated polymer film. Plotting the nucleation density with time (Figure 3.4(b)), we observed that the deposition occurs in two regimes. The first regime corresponds to an increase in nucleation density (the nucleation regime) and the second where the clusters start to merge and increase in size thus causing a decrease in the nucleation density.

60s

20 µm 70s

80s

90s

(a) (b)

Further, the increase in the densities and thicknesses of the individual clusters caused an overall increase in the mass-thickness of these clusters. We know that the ratio of the number of scattered electrons to the number of incident electrons and the product of the density ρ and the thickness t (mass thickness) are expected to follow the relationship I = Io exp (-Spρt) = Io exp (-t / Λt), (where Io is the incident electron intensity, Sp is the mass scattering cross-section, ρ is the density of the specimen, Λt is the total mean free path and t is its thickness) (Drummy, Yang, and Martin 2004).

First, we determined the mass scattering cross-section (Sp) using polystyrene spheres of known thicknesses (Figure 3.5(a) and Figure 3.5(b)). Then, we estimated the thicknesses of individual clusters using the calculated value of Sp. We then related the thicknesses to the applied charge density of these clusters.

(a) (b)

Figure 3.5: (a) BFTEM of a polystyrene sphere (b) Schematic of the liquid- like oligomeric droplets inside the in-situ chamber.

3.3.2 Role of nucleation in determining the thicknesses

In electrodeposition of inorganic metals, there is an immediate first-order transition from an ionic liquid to a solid, crystalline product. This makes it straightforward to estimate or control the thickness deposited on the electrode using Faraday’s law : t = ( ) ( ) ( ) where t is the thickness of the material deposited, Q is 𝑄𝑄 𝑀𝑀 1 the charge used for the𝐹𝐹 reaction,𝑛𝑛 𝜌𝜌𝜌𝜌 F is the Faraday’s constant, M is the molar mass of the material, n is the number of electrons transferred per ion, ρ is the density of the material and A is the area of the electrode (Gabe 1999). However, the electropolymerization of conjugated polymers is a much more complicated process than a simple first order phase transition from liquid to solid as observed by in-situ TEM and also appreciated by other groups using different techniques (Apperloo et al. 2002; Pontificia Universidad Católica de Chile, Facultad de Química, Departamento de Química Inorgánica, Laboratorio de Electroquímica de Polímeros (LEP), Vicuña Mackenna 4860, 7820436, Macul, Santiago, Chile and A. del Valle 2016; Sabouraud, Sadki, and Brodie 2000).

The times during early stage nucleation and growth are vital and are extremely interesting periods to watch. However, there are few studies done in the nucleation and growth regime during the electrodeposition of PEDOT due to the complexity of the experiments. Terzi et.al studied the early stage growth of the electrodeposited thin films of thiophene derivatives on gold substrates. They found that EDOT molecules first get adsorbed onto the gold substrate before starting to polymerize (Terzi et al. 2011). Other nucleation and growth studies discussed the discontinuous nature of the films during the initial stages and the instantaneous and diffusion-controlled pathways (Ventosa, Palacios, and Unwin 2008; Arteaga et al. 2013). Our in-situ TEM studies

(Figure 3.4(a)) have confirmed the discontinuous nature of these films at the very early stages the deposition thus corroborating the findings of other groups. Presumably, the reported values of thicknesses in these studies show some amount of variation in the thicknesses for the same charge density due to the discontinuous nature of the deposited films during the earliest stages of deposition.

Compilation of all the data listed in Figure 3.6 gives us a collection of values of thicknesses of electrodeposited films of PEDOT as a function of charge density, with thicknesses ranging from about 1 nm to 1000 µm and charge densities from 10-5 C/cm2 to 10 C/cm2 as shown in Figure 3.6. The plot shows that the thicknesses of the films increase with increasing charge density as expected. The variation in the thickness values at low charge densities, i.e. during the nucleation and growth regime is due to the discontinuous nature of the film as stated earlier. During deposition onto gels or tissue scaffolds, the thicknesses obtained are significantly larger than nominally dense PEDOT films for similar charge densities (Kuo, 2017). The mesh- like structure in the gels resulting in lower densities could presumably be causing the thickness values to be higher than expected. From the plot, we can relate the nominal thicknesses (t in µm) of these films to the total charge density (c in C/cm2) transferred during the reaction using the empirical equation t=6.8c0.75. The fact that the scaling exponent is somewhat less than one is presumably due to the fact that the deposited films are not flat, but have relatively rough surfaces, as has been seen in by SEM, AFM, and Electrochemical Impedance Spectroscopy (EIS)(Cui et al. 2001). The electrodeposition of conjugated polymers clearly depends variety of factors like the type of solvent, monomer, counter-ion and

67 current efficiency. It also involves complex intermediate steps before completely solidifying and depositing on the working electrode as a stable polymer film as mentioned earlier.

Optical Microscopy

Furthermore, we used transmitted light optical microscopy to study this process which also revealed similar details like the nuclei forming, merging and finally forming a stable solid film on the glassy carbon working electrode. The additional information that we could extract was the change in color of the translucent clusters to the dark polymer film caused due to the increase in conjugation length.

Optical microscopy has helped us observe the dramatic change in light absorption that occurs as the conjugation length increases during the transition from translucent EDOT monomer, through oligomers, to the final dark solid-polymer product (Figure 3.7). As the reaction progressed and the molecular weight of the polymer increased we observed that the initially mobile liquid monomer solution went through a rheological transition to viscoelastic oligomers, and then to a solid polymer product that deposits on the transparent glassy carbon working electrode. During the merging events, the liquid-like EDOT oligomeric clusters were predominantly translucent, and also showed evidence for droplet break-up at certain instances. While the merging events happened more slowly (~ 10 seconds), break-up was relatively fast (~ 1-2 seconds). Additionally, as discussed previously we also observed evidence for dissolution, where a droplet completely went back into solution. However, these events were only seen for liquid-like droplets, whereas the solid-like films were stable, sessile and remained either constant or consistently increased in mass thickness. Further, the liquid-like oligomeric droplets had smoother edges, and were more circular compared to their solid-like counterparts. Since we require a detailed and quantitative analysis of differences between the liquid-like droplets and solid-like films, we have discussed the most commonly observed differentiating factors between the two components in the following sections.

3.3.3 Transitions from liquid-like oligomers to solid-like polymers During the polymerization reaction, we observed the mass thickness differences between the liquid-like droplets and the solid-like films during the deposition. The liquid-like droplets were relatively more mobile and dynamic as compared to the solid-like films. Some of the droplets completely dissolved back into the solution, while some of them decreased in mass-thickness and then deposited on the electrode as a solid film. In comparison, the solid-like films deposited were sessile, stable, and constantly increased in mass thickness (or sometimes constant, but never decreased) as the reaction proceeded. Liquid to solid transitions have been of interest to researchers focusing on a diverse aspects of materials ranging from proteins and polymers to polyelectrolytes (Wu et al. 2019; Weis et al. 2019; Hiamtup, Sirivat, and Jamieson 2006; Y. Liu et al. 2017; Patel et al. 2015). Other phase transitions like evaporation of liquid droplets have also been studied in detail in the recent past (Amini and Homsy 2017a; 2017b; Edwards et al. 2018; Kuznetsov, Feoktistov, and Orlova 2016; Edwards et al. 2018). Further, mechanisms of dissolution of sessile and dynamic droplets have been examined extensively by a wide variety of researchers (Chong et al. 2020; Curiotto et al. 2017; Encarnación Escobar et al. 2018; Jimidar 2016; Qian, Arends, and Zhang 2019; Jimidar 2016). Considering the mechanisms discussed in these studies and also keeping in mind that the physics involved in the electrochemical polymerization of the conjugated polymers is complicated, we discuss some mechanistic details of liquid to solid transitions that we observed during the process and subtle methods of differentiating the droplets from the stable films. The electrodeposition of polymers involves a series of intermediate states (as seen in the previous sections) as the molecular weight of the material increases from

70 monomer, to dimer, to longer oligomers, and then finally to polymers. During this process, the transition from the liquid-like precursor state to the final solid film is subtle and complicated and involves substances with rheological properties that are between those of the precursor liquid and final solid products. We consistently observed substantial differences between the dynamics of the liquid-like oligomer droplets and solid-like PEDOT films. Here, we describe some specific differences we have observed between the liquid-like oligomeric droplets and solid-like polymer films: Mobility: The liquid-like droplets were mobile, fluctuating and dynamic domains which changed in shapes and sizes frequently before merging and depositing onto the working electrode. In stark contrast, the solid-like films were sessile, stable regions, initially observed predominantly on the edges of the electrode which only increased in mass thickness with increasing charge density. This apparent “edge effect” has also been discussed by J.liu et.al, Cui et.al and Wei et.al in their respective works (J. Liu et al. 2015; Cui and Martin 2003; X. F. Wei and Grill 2009). Frames in Figure 3.8 show the differences between the fluctuating domains (red) and the stable solid-like films that constantly increase in the thickness (blue) (sometimes would remain constant). We could see how the shapes and sizes of the liquid-like clusters evolved during the reaction. Towards the later stages, the shapes of the liquid-like oligomers would remain constant for a while before transitioning onto becoming a solid followed by depositing onto the working electrode as a stable film. Specifically, we saw that liquid-like droplets had values of circularity that remained closer to one, representing their ability to minimize their surface free energy by locally reorganizing and rearranging the more mobile molecules within the droplet. The solid-like droplets

71 had circularities that became low, in some cases by having more than one droplet merge together and then solidify before significant subsequent molecular rearrangements could occur.

Circularity (C): Circularity is a dimensionless quantity which is defined as 4π (area) / (perimeter)2. C is defined so that perfectly circular droplets have a value of C = 1, and as they become more highly elongated the value of C drops accordingly.

Usually, the liquid-like fluctuating domains were initially quite circular, with values of C ~ 0.9. As the reaction proceeded, they started to change in size and shape,

180 s 190 s 200 s

(a) (b) (c)

(d) (e) (f)

with particularly dramatic changing occuring due to merging and coalescing of smaller oligomeric clusters to form larger ones. High C values (close to 1) usually

73 corresponded to liquid-like, fluctuating, mobile domains while clusters with lower C values (C<0.5-0.6) were predominantly solid-like, stable and sessile. Mass thickness: As seen in figure 3.9 the droplets would increase or decrease in bright field TEM intensity during the course of the reaction. The intensities of most of the droplets decreased (meaning increase in mass thickness) initially and then increased (meaning decrease in mass thickness due to dissolution of lower molecular weight species) before depositing onto the working electrode as a solid-like film. In some cases, the liquid-like clusters completely dissolve back into the solution. In stark

Droplet 1 Droplet 2 136 Solid area

134

132

130

Intensity (Arb) 128 a

126

124

180 190 200 210 220 230 Time (s) L

(a) (b)

contrast, the solid-like films either remained constant or steadily decreased in intensity (meaning increased in mass thickness) during the course of the reaction.

Roughness: As seen from the changes in shape from Figure 3.8 the liquid-like droplets undergo significant variations due to events like merging and coalescence before they transition into solid-like films. A schematic of the rough edges of a solid- like region is shown in Figure 3.9(b). Initially, the liquid-like droplets have smooth edges and circular. However, due to these merging events, the solid-like droplets become less circular have rough edges. On a length scale (L) ~ 2 - 3 µm we saw variations (a) of ~ 0.15-0.25 µm (~ 7-8 %) along the edges for the solid-like regions with rough edges, whereas the edges in the liquid-like droplets did not show as much variations (0.05-0.1 µm ~ 1-2% ) as the solid-like regions. Optical Absorbance: The mobile, liquid-like oligomeric droplets were optically translucent or light blue in color. While the solid-like films (dark regions in TEM due to mass-thickness) near the edge of the electrode in the optical micrograph (transmitted light) in figure 3.7 were dark blue in color due to the characteristic absorption of PEDOT. To summarize, the liquid-like clusters (usually from about 50 nm to 2-3 μm in size; larger droplets could also be observed in rare cases) would fluctuate and subsequently rearrange themselves by merging and coalescing within a span of 50 s – 100 s whereas the solid-like regions (towards the edge of the electrode and typically > 4-5 μm) had predominantly stable shapes over similar time frames. A specific example within a ~ 100 μm2 area has been shown in Figure 3.8. Next, merging events caused the liquid-like droplets (with circularities usually > 0.5-0.6) to become less circular with time as they become solid-like regions with rough edges (with circularities usually < 0.5-0.6). On length scales of L ~ 2 - 3 μm along the droplet edge we saw variations of ~ 0.15-0.25 μm (~ 7-8 %) along the edges for the solid-like regions with

75 rough edges, whereas the edges in the liquid-like droplets did not show as much variation (only 0.05-0.1 μm or 1-2% ). Also, we saw that the liquid-like droplets could decrease in thicknesses (up to ~ 50 nm- 70 nm in ~ 50 s) and in some cases even completely dissolve back into the solution. In contrast, the solid-like regions consistently increased in mass thicknesses. Finally, corresponding experiments by optical microscopy revealed that the liquid-like clusters were optically translucent (meaning the oligomers were not yet conjugated enough to efficiently absorb light), whereas the solid-like clusters were the dark blue color known to be characteristic of the PEDOT polymer. We have included these differences in Table 3.1. Furthermore, in Figure 3.10, we present an example where we have color coded the liquid-like droplets in red and solid-like films in blue by making qualitative and quantitative observations at different time frames in the video recording.

Table 3.1: Specific metrics that can be used to differentiate between liquid-like droplets and solid-like regions Metric Liquid-like Solid-like

Mobility/fluctuations Rearrange themselves by merging Predominantly stable within and coalescing within a span of similar time frames 50 s – 100 s Mass thickness differences Certain droplets decreased up to Consistently increased in mass ~50-70 nm in ~50 s, and thicknesses sometimes dissolved back into solution Roughness along the edges ~ 1-2 % (0.05-1µm variations over Variations up to ~ 7-8 % 3-4 µm) (0.15-0.25 µm over 3-4 µm) Optical absorbances Translucent Dark Blue

Size distribution ~ 50 nm – 2-3 µm ~ > 4-5 µm Larger droplets were also observed in rare cases Circularity ~ > 0.5 - 0.6 ~ < 0.5 - 0.6

(a) (b)

Figure 3.10: (a)Bright-field TEM images of liquid-like droplets and solid- like films observed during the deposition (b) Liquid-like droplets are shaded in red (horizontal lines); solid-like films are shaded in blue (tilted lines).

3.3.4 Mechanistic details involved in the evolution of shapes and sizes during the formation of solid-like films and liquid-like clusters

The change in the shapes and sizes with time of the clusters are important clues for understanding the electropolymerization mechanisms. Further, it is important to have quantitative estimates for the variations in the sizes and shapes. The majority of the liquid-like fluctuating domains are initially highly circular, with values of C ~ 0.8- 0.9. During the electrochemical polymerization reaction, the liquid-like domains start

77 out with changing sizes and shapes and as the reaction proceeds, they coalesce and start to become sessile and stable and become less and less circular as the films are more or less arbitrary in shape. As a result, their circularity decreases due to merging of smaller oligomeric clusters to form larger ones. Higher C values (closer to 1) usually corresponded to fluctuating domains while shapes with lower C values were predominantly solid-like, sessile and stable. We have chosen specific frames as representatives of the process as a whole to explain some of the mechanistic details involved in the process. The graph in Figure 3.11 gives a quantitative estimate.

• Circularity of liquid-like clusters . Circularity of solid-like clusters

Liquid like droplets

Liquid-like droplet transitioning into a solid-like deposit

Figure 3.11: Plot of the transition in circularity with time of the droplets in Figure 3.8 while they transition from liquid to solid.

Figure 3.11 shows how the circularity of the oligomeric clusters from figure 3.8 have changed during the electrochemical polymerization. These example frames provide two particularly important types of information that can help us understand the polymerization mechanism. First, how the clusters changed in shapes and sizes

79 during the polymerization and second, how the mass thicknesses of the clusters varied during their phase transition from liquid to solid. During early times of the reaction, we observed a larger number of data points, as compared to the later stages, each indicative of a droplet. Simultaneously, the values of circularity start to decrease as the reaction proceeds, indicating the merging and coalescing of the oligomeric droplets into bigger clusters thus becoming less circular.

3.3.5 “Pinned-edge” mechanism of liquid-like droplets transforming into solid- like films We can use the plot in Figure 3.11 to segue into discussions involving the change in the mass thicknesses while the phase transformations occur. The circularity values of the droplets transitioning from liquid to solid do not vary for a brief period of time in Figure 3.12. Based on our in-situ observations, initially, the clusters nucleate from the working electrode, grow in size and merge to form bigger droplets and lose their circularity. You can think of it as the core of the droplets extracting more and more monomers from the edges and the edges from the solution. Now, as the molecular weight increases and the droplets are ready to precipitate out of the solution the lower molecular weight species at the interface between the solution and solid-like clusters slowly start to dissolve back into the solution which explains the droplets reducing in thickness during this period as shown in Figure 3.12(a). This phenomenon of lower molecular weight species dissolving back into solution can proceed so far in some cases as to have the entire droplet go back into the solution. This process has some similarities to Ostwald ripening in metallic nanoparticles (Bezpalko 2011). However here the physics involved in electrodeposition of PEDOT is somewhat more complicated, since the ongoing polymerization reaction causes a local increase in the

80 concentration of the short-chain oligomers, long-chain oligomers followed by the final polymer product. Consider evaporation of any droplet containing solid particulates dissolved in it. In general, during the evaporation process, the liquid evaporates, while the solid particulates deposit on the surface. In this case there is a liquid-solid interface, a solid- air interface and a triple boundary interface. While the solid particulates deposit, either coffee ring effects are observed where there is deposition largely on the edges of the droplet (Wang et al. 2014; Anyfantakis and Baigl 2014; Weon and Je 2010) or the solid uniformly deposits on the surface. Although the phenomena we are concerned with for our electrochemical polymerization system are predominantly phase-transitions and dissolution of lower molecular weight species, our deposition reactions bear some similarities to the evaporation process. In our case, we see more of a uniform deposition throughout the shape of the droplet instead of just on the edges. This suppression of coffee ring effect

81 eventually leading to a disc-like films rather than a ring-like film has also been

Circularity Thickness 0.5

200

0.4

150 0.3

100

Circularity 0.2 Thickness (nm)

50 0.1

0.0 0 210 220 230 240 250 Time (s)

(a)

(b)

Figure 3.12: (a) Plot of circularity and thickness of the transitioning droplets as a function of time (b) Schematic explaining the pinned-edge mechanism.

82 discussed (Yunker et al. 2011). Yunker et al. showed that particles with anisotropic

(a)

(b)

Figure 3.13: (a)Yunker et al. describe the formation of relatively more uniform films (left) when anisotropic particles are deposited onto the glass substrate due to their loose packing as compared the coffee ring effect exhibited by spherical particles(right). They claim that the mobility of the anisotropic particles is reduced, and they resist the radially outward flow (Yunker et al. 2011) (b) Nellinmoottil et al. showed that the evaporation of a droplet containing nonmotile bacteria(right) follows the coffee ring effect while the effect is suppressed in case of motile bacteria(left) where a relatively more uniform film is deposited on the surface (Nellimoottil et al. 2007).

shapes tend to suppress this effect and form a disc-like, uniform film. This

83 phenomenon, they argued, was predominantly due to the loose packing of the anisotropic particles and as a result the capillary flow does not push them to deposit on the pinned edges. Comparing these phenomena to our case, our liquid-like droplets do not have isotropic shapes which presumably might be leading to deposition of uniform disc-like films instead of exhibiting a coffee ring effect. As a matter of fact, droplets containing motile bacteria have also been observed to form more uniform (disc-like) films as compared to nonmotile bacteria which follow the coffee ring effect (Nellimoottil et al. 2007). Figure 3.13 contains images adapted from the studies done to understand the suppression of the coffee ring effect by Nellimoottil et al. and Yunker et al. with which we can identify the differences between the ring-like and the disc-like deposition. The electrochemical polymerization process of conjugated polymers is a complicated process involving a variety of mechanisms and phenomena. We propose that the “pinned edge” mechanism is one of the more prevalent ways the solidification can take place where the shapes, as seen from the snapshots of the process, the circularity plots, and the boundaries of the droplets are more or less fixed during the solidification process. With fixed boundaries, their mass thicknesses decrease, indicative of the decrease in thicknesses as the lower molecular weight species from the interface dissolve back into solution and finally we have the polymerized product deposited onto the working electrode. As more and more oligomeric droplets nucleate, grow, coalesce and finally deposit we find a relatively uniform film of PEDOT deposited onto the working electrode.

3.4 Conclusions

Thus far, we have investigated the fundamental, unprecedented phase- transformations showing oligomeric clusters of PEDOT nucleating and growing in size and thickness before precipitating out and depositing onto the working electrode as a conjugated polymer film. The schematic in Figure 3.14 describes the mechanism

Mechanism of the electrochemical polymerization of PEDOT

Nuclei originating Nuclei merge to form higher from the working molecular weight oligomeric electrode clusters

Figure 3.14: Schematic relating the in-situ BFTEM images to the mechanism showing the nucleation and growth of liquid-like clusters transforming into solid- like films through coalescence. through which we understand the reaction to proceed. The electro-polymerization reaction, mainly driven by insolubility of the higher molecular weight oligomers is particularly interesting because the deposition proceeds from an initially highly mobile, isotropic liquid monomer precursor solution through viscoelastic EDOT oligomer intermediate states to a final solid PEDOT polymer film product. The local

85 mechanistic insights we have obtained have provided us with details about the electropolymerization induced liquid monomer to solid polymer phase transition. From detailed analyses from our in-situ TEM experiments, we have been able to associate specific observations with the local structural state and dynamics of the sample. Again, specifically, we have observed that initially liquid-like, mobile, liquid-like clusters of EDOT oligomers nucleate from the electrode surface which then further coalesce to form continuous, solid polymer PEDOT films. Additionally, we found that the typically rough and bumpy morphology of electrodeposited PEDOT is due to the phase transformations of these clusters of anisotropic shapes and sizes increasing and decreasing in thickness which then merge with each other and then grow onto each other. These insights have proven to be of utmost importance while understanding the polymerization process. Furthermore, the quantitative differences between the liquid- like droplets and the solid-like clusters provides information about the dynamics of molecular motion during this phase transition.

REFERENCES

1. Amini, Amirhossein, and G. M. Homsy. 2017a. “Evaporation of Liquid Droplets on Solid Substrates. I. Flat Substrate with Pinned or Moving Contact Line.” Physical Review Fluids 2 (4): 043603. https://doi.org/10.1103/PhysRevFluids.2.043603.

2. ———. 2017b. “Evaporation of Liquid Droplets on Solid Substrates. II. Periodic Substrates with Moving Contact Lines.” Physical Review Fluids 2 (4): 043604. https://doi.org/10.1103/PhysRevFluids.2.043604.

3. Anyfantakis, Manos, and Damien Baigl. 2014. “Dynamic Photocontrol of the Coffee-Ring Effect with Optically Tunable Particle Stickiness.” Angewandte Chemie International Edition 53 (51): 14077–81. https://doi.org/10.1002/anie.201406903.

4. Apperloo, Joke J., L. “Bert” Groenendaal, Hilde Verheyen, Manickam Jayakannan, René A. J. Janssen, Ahmed Dkhissi, David Beljonne, Roberto Lazzaroni, and Jean-Luc Brédas. 2002. “Optical and Redox Properties of a Series of 3,4‐Ethylenedioxythiophene Oligomers.” Chemistry – A European Journal 8 (10): 2384–96. https://doi.org/10.1002/1521- 3765(20020517)8:10<2384::AID-CHEM2384>3.0.CO;2-L.

5. Arteaga, G C, M C Arévalo, E Pastor, and G Louarn. 2013. “Nucleation and Growth Mechanism of Electro-Synthesized Poly(Pyrrole) on Steel.” Int. J. Electrochem. Sci. 8: 11.

6. Bhagwat, Nandita, Kristi L. Kiick, and David C. Martin. 2014. “Electrochemical Deposition and Characterization of Carboxylic Acid Functionalized PEDOT Copolymers.” Journal of Materials Research 29 (23): 2835–44. https://doi.org/10.1557/jmr.2014.314.

7. Chong, Kai Leong, Yanshen Li, Chong Shen Ng, Roberto Verzicco, and Detlef Lohse. 2020. “Convection-Dominated Dissolution for Single and Multiple Immersed Sessile Droplets.” Journal of Fluid Mechanics 892 (June). https://doi.org/10.1017/jfm.2020.175.

8. Crispin, X., F. L. E. Jakobsson, A. Crispin, P. C. M. Grim, P. Andersson, A. Volodin, C. van Haesendonck, M. Van der Auweraer, W. R. Salaneck, and M. Berggren. 2006. “The Origin of the High Conductivity of Poly(3,4- Ethylenedioxythiophene)−Poly(Styrenesulfonate) (PEDOT−PSS) Plastic Electrodes.” Chemistry of Materials 18 (18): 4354–60. https://doi.org/10.1021/cm061032+.

9. Cui, Xinyan, Jamille F. Hetke, James A. Wiler, David J. Anderson, and David C. Martin. 2001. “Electrochemical Deposition and Characterization of Conducting Polymer Polypyrrole/PSS on Multichannel Neural Probes.” Sensors and Actuators A: Physical 93 (1): 8–18. https://doi.org/10.1016/S0924-4247(01)00637-9.

10. Cui, Xinyan, and David C. Martin. 2003. “Electrochemical Deposition and Characterization of Poly(3,4-Ethylenedioxythiophene) on Neural Microelectrode Arrays.” Sensors and Actuators B: Chemical 89 (1): 92– 102. https://doi.org/10.1016/S0925-4005(02)00448-3.

11. Curiotto, Stefano, Frédéric Leroy, Fabien Cheynis, and Pierre Müller. 2017. “Surface-Dependent Scenarios for Dissolution-Driven Motion of Growing Droplets.” Scientific Reports 7 (1): 902. https://doi.org/10.1038/s41598-017-00886-2.

12. Drummy, Lawrence F., Junyan Yang, and David C. Martin. 2004. “Low- Voltage Electron Microscopy of Polymer and Organic Molecular Thin Films.” Ultramicroscopy 99 (4): 247–256. https://doi.org/10.1016/j.ultramic.2004.01.011.

13. Edwards, A. M. J., P. S. Atkinson, C. S. Cheung, H. Liang, D. J. Fairhurst, and F. F. Ouali. 2018. “Density-Driven Flows in Evaporating Binary Liquid Droplets.” Physical Review Letters 121 (18): 184501. https://doi.org/10.1103/PhysRevLett.121.184501.

14. Encarnación Escobar, José M., Erik Dietrich, Steve Arscott, Harold J. W. Zandvliet, Xuehua Zhang, and Detlef Lohse. 2018. “Zipping-Depinning: Dissolution of Droplets on Micropatterned Concentric Rings.” Langmuir 34 (19): 5396–5402. https://doi.org/10.1021/acs.langmuir.8b00256.

15. Gabe, D. R. 1999. “Coulometric Techniques for Surface Coatings.” Transactions of the IMF 77 (6): 213–17. https://doi.org/10.1080/00202967.1999.11871286.

16. Gentry, Stuart T., Shane F. Kendra, and Mark W. Bezpalko. 2011. “Ostwald Ripening in Metallic Nanoparticles: Stochastic Kinetics.” The Journal of Physical Chemistry C 115 (26): 12736–41. https://doi.org/10.1021/jp2009786.

17. Ghosh, Srabanti, Natalie Amoin Kouame, Samy Remita, Laurence Ramos, Fabrice Goubard, Pierre-Henri Aubert, Alexandre Dazzi, Ariane Deniset- Besseau, and Hynd Remita. 2016. “Visible-Light Active Conducting Polymer Nanostructures with Superior Photocatalytic Activity.” Scientific Reports 5 (1). https://doi.org/10.1038/srep18002.

18. Groenendaal, By L, Friedrich Jonas, Dieter Freitag, Harald Pielartzik, and John R Reynolds. 2000. “Its Derivatives : Past , Present , and Future **,” 481–494.

19. Hiamtup, Piyanoot, Anuvat Sirivat, and Alexander M. Jamieson. 2006. “Electrorheological Properties of Polyaniline Suspensions: Field-Induced Liquid to Solid Transition and Residual Gel Structure.” Journal of Colloid and Interface Science 295 (1): 270–78. https://doi.org/10.1016/j.jcis.2005.07.067.

20. Jimidar, I. S. M. 2016. “Dissolution Modes of Droplets on Patterned Surfaces.” Info:eu-repo/semantics/masterThesis. University of Twente. August 19, 2016. http://essay.utwente.nl/70801/.

21. Koutsouras, Dimitrios A., Paschalis Gkoupidenis, Clemens Stolz, Vivek Subramanian, George G. Malliaras, and David C. Martin. 2017. “Impedance Spectroscopy of Spin‐Cast and Electrochemically Deposited PEDOT:PSS Films on Microfabricated Electrodes with Various Areas.” ChemElectroChem 4 (9): 2321–27. https://doi.org/10.1002/celc.201700297.

22. Kuo, Chin-chen. 2017. “Optimizing the Performance of Neural Interface Devices with Hybrid Poly(3,4-Ethylene Dioxythiophene) (PEDOT).” Thesis, University of Delaware. https://udspace.udel.edu/handle/19716/21256.

23. Kuznetsov, G. V., D. V. Feoktistov, and E. G. Orlova. 2016. “Evaporation of Liquid Droplets from a Surface of Anodized Aluminum.” Thermophysics and Aeromechanics 23 (1): 17–22. https://doi.org/10.1134/S0869864316010029.

24. Lattach, Youssef, Cecilia Coletta, Srabanti Ghosh, and Samy Remita. 2014. “Radiation-Induced Synthesis of Nanostructured Conjugated Polymers in Aqueous Solution: Fundamental Effect of Oxidizing Species.” ChemPhysChem 15 (1): 208–18. https://doi.org/10.1002/cphc.201300915.

25. Lattach, Youssef, Ariane Deniset-Besseau, Jean-Michel Guigner, and Samy Remita. 2013. “Radiation Chemistry as an Alternative Way for the Synthesis of PEDOT Conducting Polymers under ‘Soft’ Conditions.” Radiation Physics and Chemistry 82 (January): 44–53. https://doi.org/10.1016/j.radphyschem.2012.09.009.

26. Liu, Jinglin, Bin Wei, Jennifer D. Sloppy, Liangqi Ouyang, Chaoying Ni, and David C. Martin. 2015. “Direct Imaging of the Electrochemical Deposition of Poly(3,4-Ethylenedioxythiophene) by Transmission Electron Microscopy.” ACS Macro Letters 4 (9): 897–900. https://doi.org/10.1021/acsmacrolett.5b00479.

27. Liu, Yalin, Brian Momani, H. Henning Winter, and Sarah L. Perry. 2017. “Rheological Characterization of Liquid-to-Solid Transitions in Bulk Polyelectrolyte Complexes.” Soft Matter 13 (40): 7332–40. https://doi.org/10.1039/C7SM01285C.

28. Lock, John P., Sung Gap Im, and Karen K. Gleason. 2006. “Oxidative Chemical Vapor Deposition of Electrically Conducting Poly(3,4- Ethylenedioxythiophene) Films.” Macromolecules 39 (16): 5326–29. https://doi.org/10.1021/ma060113o.

29. Mariani, G., Y. Wang, P. S. Wong, R. B. Kaner, and D. L. Huffaker. 2011. “Electrochemical Polymerization of PEDOT on Catalyst-Free Patterned GaAs Nanopillars for High Efficiency Hybrid Photovoltaics 37th IEEE Photovoltaic Specialists Conference.” In 2011 37th IEEE Photovoltaic Specialists Conference, 002639–41. https://doi.org/10.1109/PVSC.2011.6186490.

30. Martin, David C. 2015. “Molecular Design, Synthesis, and Characterization of Conjugated Polymers for Interfacing Electronic Biomedical Devices with Living Tissue.” MRS Communications 5 (2): 131–153. https://doi.org/10.1557/mrc.2015.17.

31. Martin, David C., Jinghang Wu, Charles M. Shaw, Zachary King, Sarah A. Spanninga, Sarah Richardson-Burns, Jeffrey Hendricks, and Junyan Yang. 2010. “The Morphology of Poly(3,4-Ethylenedioxythiophene).” Polymer Reviews 50 (3): 340–384. https://doi.org/10.1080/15583724.2010.495440.

32. Murbach, Jamie M., Seth Currlin, Adrienne Widener, Yuxin Tong, Shrirang Chhatre, Vivek Subramanian, David C. Martin, Blake N. Johnson, and Kevin J. Otto. 2018. “In Situ Electrochemical Polymerization of Poly(3,4-Ethylenedioxythiophene) (PEDOT) for Peripheral Nerve Interfaces.” MRS Communications 8 (3): 1043–49. https://doi.org/10.1557/mrc.2018.138.

33. Nellimoottil, Tittu Thomas, Pinjala Nagaraju Rao, Siddhartha Sankar Ghosh, and Arun Chattopadhyay. 2007. “Evaporation-Induced Patterns from Droplets Containing Motile and Nonmotile Bacteria.” Langmuir 23 (17): 8655–58. https://doi.org/10.1021/la7006205.

34. Ouyang, Liangqi, Bin Wei, Chin-chen Kuo, Sheevangi Pathak, Brendan Farrell, and David C. Martin. 2017. “Enhanced PEDOT Adhesion on Solid Substrates with Electrografted P(EDOT-NH 2 ).” Science Advances 3 (3): e1600448. https://doi.org/10.1126/sciadv.1600448.

35. Patel, Avinash, Hyun O. Lee, Louise Jawerth, Shovamayee Maharana, Marcus Jahnel, Marco Y. Hein, Stoyno Stoynov, et al. 2015. “A Liquid-to- Solid Phase Transition of the ALS Protein FUS Accelerated by Disease Mutation.” Cell 162 (5): 1066–77. https://doi.org/10.1016/j.cell.2015.07.047.

36. Pontificia Universidad Católica de Chile, Facultad de Química, Departamento de Química Inorgánica, Laboratorio de Electroquímica de Polímeros (LEP), Vicuña Mackenna 4860, 7820436, Macul, Santiago, Chile, and M A. del Valle. 2016. “Influence of the Supporting Electrolyte on the Electrochemical Polymerization of 3,4-Ethylenedioxythiophene. Effect on p- and n-Doping/Undoping, Conductivity and Morphology.” International Journal of Electrochemical Science, August, 7048–65. https://doi.org/10.20964/2016.08.46.

37. Qian, Jiasheng, Gilmar F. Arends, and Xuehua Zhang. 2019. “Surface Nanodroplets: Formation, Dissolution, and Applications.” Langmuir 35 (39): 12583–96. https://doi.org/10.1021/acs.langmuir.9b01051.

38. Randriamahazaka, H, V Noël, and C Chevrot. 1999. “Nucleation and Growth of Poly(3,4-Ethylenedioxythiophene) in Acetonitrile on Platinum under Potentiostatic Conditions.” Journal of Electroanalytical Chemistry 472 (2): 103–11. https://doi.org/10.1016/S0022-0728(99)00258-2.

39. Sabouraud, Guillaume, Saïd Sadki, and Nancy Brodie. 2000. “The Mechanisms of Pyrrole Electropolymerization.” Chemical Society Reviews 29 (5): 283–93. https://doi.org/10.1039/a807124a.

40. Subramanian, Vivek, Casey A Rowland, Glenn P. A. Yap, and David C. Martin. 2019. “Morphology, Molecular Orientation and Solid-State Characterization of 2,3-Dihydrothieno[3,4-b] [1,4]Dioxine-2-Carboxylic Acid (EDOTacid).” Crystal Growth & Design, September. https://doi.org/10.1021/acs.cgd.9b00613.

41. Terzi, Fabio, Luca Pasquali, Monica Montecchi, Stefano Nannarone, Antti Viinikanoja, Timo Ääritalo, Mikko Salomäki, Jukka Lukkari, Bryan P. Doyle, and Renato Seeber. 2011. “New Insights on the Interaction between Thiophene Derivatives and Au Surfaces. the Case of 3,4- Ethylenedioxythiophene and the Relevant Polymer.” Journal of Physical Chemistry C 115 (36): 17836–17844. https://doi.org/10.1021/jp203219b.

42. Tong, Yuxin, Jamie M. Murbach, Vivek Subramanian, Shrirang Chhatre, Francisco Delgado, David C. Martin, Kevin J. Otto, Mario Romero-Ortega, and Blake N. Johnson. 2018. “A Hybrid 3D Printing and Robotic-Assisted Embedding Approach for Design and Fabrication of Nerve Cuffs with Integrated Locking Mechanisms.” MRS Advances, April, 1–8. https://doi.org/10.1557/adv.2018.378.

43. Ventosa, Edgar, Jesús López Palacios, and Patrick R. Unwin. 2008. “Nucleation and Growth of Poly(3,4-Ethylenedioxythiophene) Thin Films on Highly Oriented Pyrolytic Graphite (HOPG) Electrodes.” Electrochemistry Communications 10 (11): 1752–55. https://doi.org/10.1016/j.elecom.2008.09.003.

44. Wang, Weidong, Yongguang Yin, Zhiqiang Tan, and Jingfu Liu. 2014. “Coffee-Ring Effect-Based Simultaneous SERS Substrate Fabrication and Analyte Enrichment for Trace Analysis.” Nanoscale 6 (16): 9588–93. https://doi.org/10.1039/C4NR03198A.

45. Wei, Bin, Jinglin Liu, Liangqi Ouyang, Chin Chen Kuo, and David C. Martin. 2015. “Significant Enhancement of PEDOT Thin Film Adhesion to Inorganic Solid Substrates with EDOT-Acid.” ACS Applied Materials and Interfaces 7 (28): 15388–15394. https://doi.org/10.1021/acsami.5b03350.

46. Wei, Xuefeng F., and Warren M. Grill. 2009. “Analysis of High-Perimeter Planar Electrodes for Efficient Neural Stimulation.” Frontiers in Neuroengineering 2. https://doi.org/10.3389/neuro.16.015.2009.

47. Weis, Philipp, Andreas Hess, Gunnar Kircher, Shilin Huang, Günter K. Auernhammer, Kaloian Koynov, Hans-Jürgen Butt, and Si Wu. 2019. “Effects of Spacers on Photoinduced Reversible Solid-to-Liquid Transitions of Azobenzene-Containing Polymers.” Chemistry – A European Journal 25 (46): 10946–53. https://doi.org/10.1002/chem.201902273.

48. Weon, Byung Mook, and Jung Ho Je. 2010. “Capillary Force Repels Coffee-Ring Effect.” Physical Review E 82 (1): 015305. https://doi.org/10.1103/PhysRevE.82.015305.

49. Winther-Jensen, Bjørn, and Keld West. 2004. “Vapor-Phase Polymerization of 3,4-Ethylenedioxythiophene: A Route to Highly Conducting Polymer Surface Layers.” Macromolecules 37 (12): 4538–43. https://doi.org/10.1021/ma049864l.

50. Wu, Haoran, Matthew Genovese, Kevin Ton, and Keryn Lian. 2019. “A Comparative Study of Activated Carbons from Liquid to Solid Polymer Electrolytes for Electrochemical Capacitors.” Journal of The Electrochemical Society 166 (6): A821. https://doi.org/10.1149/2.0141906jes.

51. Yunker, Peter J., Tim Still, Matthew A. Lohr, and A. G. Yodh. 2011. “Suppression of the Coffee-Ring Effect by Shape-Dependent Capillary Interactions.” Nature 476 (7360): 308–11. https://doi.org/10.1038/nature10344.

Chapter 4

QUANTITATIVE ANALYSIS OF NANO-FIBRIL GROWTH DURING THE ELECTROCHEMICAL POLYMERIZATION OF PEDOT BY LIQUID CELL TEM

4.1 Introduction Poly(3,4-ethylenedioxythiophene) (PEDOT) and its derivatives have become increasingly promising candidates in recent years for interfacing ionically conducting living tissue with metallic or semiconducting biomedical devices (Inal et al. 2018; Richardson-Burns et al. 2007; Rivnay, Owens, and Malliaras 2014; Quirós-Solano et al. 2016; Murbach et al. 2018). The ability to conduct charge both electronically and ionically, combined with the relatively soft mechanical properties of these organic polymers as compared to rigid, inorganic metals or semiconductors, makes them effective for bio-interfacing applications (Martin 2015). These versatile conjugated polymers have also been used in solar cells, anti-static coatings and organic electronic devices (Adam et al. 2016; Wang et al. 2012; Liang et al. 2013). Electrochemistry has been widely used for materials synthesis, energy storage, and energy conversion (Chen et al. 2012; Nitta et al. 2015; Gottesfeld et al. 2018). Electrochemical deposition has been of particular interest for fabricating conjugated polymers due to its reproducibility and site-specificity (Wei et al. 2017; Ouyang et al. 2017; Koutsouras et al. 2017). The details involved in the early nucleation and growth stages of these processes are particularly important for precisely tuning the

94 morphology and properties of the resulting materials. However, probing the dynamic processes at the solid-liquid interface has been experimentally difficult. Liquid cell transmission electron microscopy has recently been proven to be an excellent method for investigating the electrochemical deposition process of inorganic materials, because of its outstanding temporal and spatial resolutions (Hauwiller et al. 2018; Ross 2015; Jonge and Ross 2011; Khelfa et al. 2019). Our group has shown that with careful control of beam current and microscope operating parameters, it is possible to monitor the electrochemical deposition of PEDOT and other conjugated polymers (Liu et al. 2015; Subramanian et al. 2019) Details of the processing conditions are known to have a significant influence on the uniformity, conductivity, morphology and other properties of electrochemically deposited conjugated polymer films (Martin et al. 2010; Zhou et al. 2010; Yang and Martin 2004). For example, it was shown by Yang et. al that using a macromolecular counter-ion, poly (acrylic acid) (PAA), during electrodeposition facilitates the formation of highly anisotropic nano-fibrils of PEDOT. These nanofibrils significantly increase the effective surface area of the film, resulting in a substantial reduction in the impedance, particularly at low temporal frequencies (Yang, Lipkin, and Martin 2007). It was hypothesized that the PAA slowed the lateral growth of PEDOT, resulting in the nano-fibrillar morphology. However, the lack of direct, high- resolution observations during growth meant that the details of this proposed mechanism were only speculative, based solely on examinations of film morphology by scanning electron microscopy after the growth had been completed. Direct observations of growth phenomena in situ, particularly during the earliest stages of the process, are therefore key to fully understanding and controlling

95 the fibril formation. Although there have been some detailed studies discussing the physical, chemical and electrical properties of PEDOT, only a few studies discussed the early stage nucleation and growth processes during the anodic electropolymerization of PEDOT (Morvant and Reynolds 1998; Zhou et al. 2010; Randriamahazaka, Noël, and Chevrot 1999; Ventosa, Palacios, and Unwin 2008). Therefore, there remained a critical need to observe the process in-situ and better understand the various phenomena occurring during these early stages for elucidating the mechanistic details involved. Here, we have electrochemically fabricated nanofibrils of PEDOT using poly(acrylic acid) with lithium perchlorate as the counter-ions, and have been able to monitor the reaction as it occurs using low dose, operando liquid cell transmission electron microscopy. This has enabled us to make quantitative observations about the PEDOT nanofibril formation, including quantitative measurements of growth rate in the different directions parallel and perpendicular to the primary film growth direction. Finally, we have also been able to quantify the nature of their microstructural evolution through careful quantitative observations of the evolution of their intensity profiles.

4.2 Materials and Methods a) Monomer solution: (3,4-ethylenedioxythiophene) (EDOT monomer), lithium perchlorate and poly(acrylic acid) (PAA) (Mw~1800 g/mol) were purchased from Sigma Aldrich. An aqueous solution of 0.01M (3,4-ethylenedioxythiophene), 0.1 M lithium perchlorate and 0.5 wt% of PAA was used as the monomer/ feed solution for fabricating nano-fibrils of PEDOT.

96 b) Liquid flow cell and electrochemistry chips: A commercially available electrochemical holder from Protochips, Inc. was used for performing the in-situ liquid TEM experiments. Two microfabricated chips were used; the top chip had the electrochemical connections and the bottom chip (with a spacer of thickness ~ 500 nm) was used to define the liquid thickness in the beam direction. These chips were both equipped with ~50 nm thick SiN membranes. The dimensions of the windows (SiN membranes) were 550 µm x 40 µm on the top chip and 300 µm x 90 µm on the bottom chip. The glassy carbon working electrode (WE) was 20 µm wide. The 100 µm wide circular CE was located 500 µm away from the WE. The CE surrounded the WE and the RE was located away from the WE. Both CE and RE were located off the SiN window. The holder was additionally equipped with two inlets and an outlet all of which were attached to polyetheretherketone (PEEK) tubings for flowing liquid through the holder. The monomer solution (sandwiched between the two microfabricated chips) was passed through the holder at a flow rate of 3 ul/min using a syringe from Hamilton and microfluidic pump from Harvard apparatus. c) Transmission Electron Microscopy: The in-situ experiments were performed on a 200 kV Thermofisher Scientific FEI Talos F200C in brightfield imaging mode. The total electron doses we used for these experiments were around 6-12 e/A2 which were well below the critical dose for electron damage of the PEDOT (60 e/A2). d) Electrochemistry: A potentiostatic deposition of 1.2 V was carried out for fabricating the nanofibrils of PEDOT using a Gamry Reference 600. e) Video recording: The in-situ experiments were recorded using VeloxTM software. f) Optical Microscopy: Optical micrographs were acquired with an Olympus BX60 BF microscope. g) Image analysis: Quantitative analysis of the images were done using Fiji (ImageJ version 1.53a).

(a) (b)

Figure 4.1: Scanning electron micrographs of PEDOT with (a) regular (rough and bumpy) and (b) nano-fibrillar morphologies.

4.3 Results and discussion Electrochemical polymerization is a commonly used strategy for coating conjugated polymers like PEDOT on electrodes where site-selectivity and reproducibility are desired (Koutsouras et al. 2017; Cui and Martin 2003). Electropolymerized PEDOT can exhibit a wide array of morphologies from being rough and bumpy to a micro-porous structure merely by altering the processing conditions, using different counter-ions or introducing secondary components (Martin et al. 2010; Yang and Martin 2004). Figure 4.1(a) shows SEM images of the typical bumpy or lumpy surface structure of PEDOT, and Figure 4.1(b) shows the corresponding nanofibrillar morphology that occurs when PAA is added to the

98 reaction mixture. The lumps in Figure 4.1(a) range from being about 80-90 nm to several hundreds of nanometers in diameter. The electrochemically fabricated PEDOT nanofibril structures were about 100 –500 nm in diameter and approximately 1-2 µm in length. However, the nature of the fibril formation process was not known. Thus, to have a better control over the morphological development and therefore the properties it is necessary to observe and understand the microstructural evolution during the early stages of the reaction. Consequently, in this study, we have focused on growing nano-fibrils of PEDOT in an electrochemical in-situ TEM stage developed by Protochips, Inc. and have observed the process as it occurs during the electropolymerization reaction. Using in-situ transmission electron microscopy, we have observed the nucleation of these dendritic fibrils at the solution-electrode interface, allowing us to better understand the nano-fibril formation mechanism through observation and quantification of the processes occurring at the nano-scale. The in-situ electrodeposition was performed on a glassy carbon working electrode located on the electron transparent SiN window of the microfabricated chip. A smaller chip was used as the spacer (spacer thickness ~ 500 nm) for confining the liquid flowing between the two chips. An aqueous solution containing 0.01 M EDOT monomer, 0.1M lithium perchlorate dopant and 0.5 wt% of poly(acrylic acid) was used as the feed solution which was passed between the two chips through the in-situ holder. The in-situ setup was then transferred onto a Gatan pumping station to ensure that the chamber could withstand the pressures required inside the TEM chamber without the liquid leaking from the SiN membrane. Subsequently, several CV cycles were run to ensure complete wetting of the electrodes before loading the holder into the TEM. Figures 4.2(a) and 4.2(b) show images of the glassy carbon working

99 electrode before and after a potentiostatic electrochemical polymerization at 1.2 V. Figure 4.2(c) is a transmitted light optical image of the same electrode after deposition which shows the optically absorptive nature of PEDOT that is dark blue in areas of the electrode corresponding to mass thickness observed in brightfield images post deposition. Further, droplets (or certain regions of the electrode) which had mass thickness in the TEM were completely translucent optically indicating that those

(a) (b) (c)

regions/droplets had not yet become fully conjugated and thus still transmit light, but scatter electrons (thus had mass thickness in TEM).

100

Using the ratio of the number of scattered electrons to the number of incident electrons and the product of the density ρ and the thickness t (mass thickness) are expected to follow the relationship I = Io exp (-Spρt) = Io exp (-t / Λt), (where Io is the incident electron intensity, Sp is the mass scattering cross-section, ρ is the density of

(a) (b)

Figure 4.3:(a) BFTEM of a polystyrene sphere (b) Schematic of the liquid- like oligomeric droplets inside the in-situ chamber.

the specimen, Λt is the total mean free path and t is its thickness) (Drummy, Yang, and

Martin 2004). First, we determined the mass scattering cross-section (Sp) using polystyrene spheres of known thicknesses (Figure 4.3(a) and Figure 4.3(b)) Then, we estimated the thicknesses of individual nanofibrils using the calculated value of Sp. We then related the thicknesses to the applied charge density of these clusters.

Further, electrochemical polymerization involves a series of intermediate states as the molecular weight of the material increases from monomer, to dimer, to longer oligomers, and then finally to polymers (Liu et al. 2015; Subramanian et al. 2019).

101

During this process, the transition from the liquid-like precursor state to the final solid film is subtle and complicated and involves substances with rheological properties that are between those of the precursor liquid and final solid products. The nucleation and growth dynamics of the oligomeric clusters involved in the electro-polymerization reaction usually resulted in a fairly rough and bumpy surface morphology as shown in Figure 4.1(a). The addition of PAA changed the bumpy morphology into a nanofibrillar morphology. This makes it important and interesting to observe and understand how the nanofibrils nucleate in their initial stages. In addition, being able to make quantitative observations about the growth of the fibrils in real -time makes the in-situ study an essential and a powerful part of the understanding and visualizing the process for a better control over the structure and properties of the deposited conjugated polymer. As stated earlier, Yang et al. showed that it is possible to completely alter the morphology of PEDOT by introducing an additional component, poly(acrylic acid) (PAA). However, in these initial studies it was not possible to track the formation of the fibrils as they precipitated out and grew from the isotropic monomer solution. (Yang, Lipkin, and Martin 2007). Recently, we have been using operando liquid-cell transmission electron microscopy to answer fundamental questions about the nature of the electrochemical polymerization reactions and elucidate the finer details involved in the process. Using this technique, we have now imaged the nucleation and growth of

102

(a) (b) (c)

(d) (e) (f)

the electrochemical fabrication of nanofibrils of PEDOT during the very early stages of their precipitation followed by deposition. We observed individual fibrils nucleating from the edge of the working electrode as shown in Figure 4.4. They predominantly

103 grew along the length of the fibril initially, then increased in width and thickness during the later stages. Here, we quantify and discuss these observations in detail.

(a) (b)

Number of EDOT molecules 70

10 Number of EDOT molecules (x(10^8))

0 0 20 40 60 80 100 120 140 160 180 Time (s) (c) (d)

104

Quantitative observations of the lengths, widths, and thicknesses of nanofibrils that nucleated in a similar manner to the ones shown in Figure 4.4 are plotted as a function of time in Figure 4.5. The nanofibrils were about 1 - 2 µm in length, 0.5 – 0.7 µm wide and 200 - 250 nm in thickness towards the later stages of the reaction. Based on the trends we observed, we have categorized the growth process into three different stages. In the first (I) stage, the monomers and the counter-ions were attracted to the working electrode due to the applied bias. There was no visible fibril-growth since PEDOT has to attain a certain molecular weight before it precipitates out of the aqueous solution. This is because insolubility of the higher molecular weight oligomers in the solution is one of the major factors driving the electrodeposition process. In the second (II) stage, the fibrils grew at a relatively fast rate, with growth predominant along their length The approximate growth velocities along the length, width and thicknesses were ~30-35 nm/s, ~5-7 nm/s, and ~0.5-0.6 nm/s respectively. In the third (III) stage, the growth rate along the length of the fibrils was lower as they started to grow wider and thicker. In this stage the velocity along the length was considerably lower than the previous stage ~ 7-8 nm/s. The growth rates along the widths were ~ 10-12 nm/s initially which plateaued later. Likewise, the velocities along the thicknesses were ~ 1-2 nm/s. Similar transitions have also been observed in the dendrite growth during electrodeposition of lithium in the TEM (Kushima et al. 2017). Further, Figure 4.5(d) gives an estimate of the number of molecules of EDOT contained in a fibril of mean length, width and thickness at each timestamp assuming that each fibril is an elliptic cylinder. The length of the fibril would correspond to the

105 height of the cylinder and width and the thickness as the major and minor axes. Additionally, we have assumed that PEDOT packs itself into an orthorhombic unit cell with a = 1.4 nm, b = 0.68 nm and c = 0.78 nm with 4 molecules of EDOT per unit cell (Martin et al. 2010). Following the trends in the above plots, the latter stages show involvement of factors like decrease in the electric field strength and decrease in monomer concentration away from the electrode that contribute towards the decrease in the growth velocities of the fibrils. To get a better idea of the more subtle mechanistic details involved in the process we used frame-to-frame comparisons of image intensity. To that end, we compared the bright-field TEM images of the fibrils at later time frames with an image at a former time frame as a reference point. An example is shown in Figure 4.6, where we chose the frame at 80s as the reference point. Next, we selected images at subsequent time frames and subtracted their intensities from the intensities of the image at 80s. The processed images are shown in Figure 4.6. Comparisons of Figure 4.4 and Figure 4.6 gave us an idea of the microstructural evolution during the early stages of the nanofibril growth. Specifically, we saw that during the initial times of the process (until ~ 70-80s) the central part of the fibrils grow and not too much material gets deposited into the interior parts of the fibrils after the initial deposition. Instead,

106 the fibrils grow mostly from the edges with the understructure established by the centre of the nanofibrils. These results thus suggest that as the nanofibrils grow, more and more monomers and lower-molecular weight oligomers are pulled into the nanofibrillar structure as their growth becomes more edge-dominated. In other words,

(a) (b) (c)

(d) (e) (f)

the interior part of the fibril acts as the working electrode and transfers electrons to the exterior solution where the monomers and the lower-molecular weight species are reacting and thus are more highly concentrated. As the molecular weight of the liquid-

107 like species increase, they become more solid-like and finally become a part of the entire fibril. Essentially, once they become solid-like and merge with the fibril their individual identity as liquid-dissolved monomers or lower-molecular weight oligomeric species ceases to exist, instead they become part of the entire solidified fibril.

Quantitative analysis of the microstructure of the nanofibrils through intensity profiles

If it is true that the growth of the fibrils mainly occurs through edges at any point of time, the thicknesses of the fibrils along their widths would not be constant. Meaning, the interior parts would be relatively thick and mature while the edges would be nascent and thin. Then, the addition of monomers and lower-molecular weight oligomers would make those edges thicker while thinner sections would build on the newly developed structure. We can quantitatively determine the thickness of the films and nanofibrils by measuring the bright field image intensity as a function of position in the TEM images. A higher intensity would correspond to lower mass- thickness, meaning thinner sections, while lower measured intensity would relate to

108

(a) (b)

higher mass-thickness. Figures 4.7(a) and 4.7(b) quantitatively describe the thickness variation of a typical fibril along the length and width. First, Figure 7(a) shows how the brightfield TEM image intensity (inversely related to mass-thickness) varies along

109 the length of a typical fibril. The horizontal axis is the distance from the base of the fibril. The vertical axis is the intensity of the pixels as measured using Fiji. From the plot, we can see that the intensity of the fibrils along the length is relatively constant from the base until about the centre of the fibril, then gradually starts to increase. This increase in intensity corresponds to a decrease in mass-thickness. Another typical intensity profile along the width of a fibril is shown in Figure 4.7(b). Based on these quantitative observations, we have proposed a schematic of a typical fibril in Figures 4.7(a) and 4.7(b).

The differences in mass-thicknesses were much more apparent in some nucleation sites where the fibrils grew in “clusters” as shown in Figure 4.7(c). These clusters presumably correspond to due to the competing bumpy morphology of PEDOT. They are areas where electrodeposited PEDOT nucleates and grows in a way that it is typical with when PAA counter-ions are absent. Although they are clumped or clustered together, their exterior is rough and fibrillar. The interesting aspect of these clusters is that they seem to grow in a similar fashion as the individual fibrils themselves do, in that the core matures first. Then the exterior fibrillar edges, pull in more and more monomers and lower-molecular weight species, which are initially in their incipient stages and mature towards the later stages of the reaction. This can be observed in Figure 4.7(c) which shows the contrast differences between the core and the shell of a typical cluster originating from the mass-thickness differences. A quantitative intensity profile of the cluster and a schematic based on the intensity profile is shown in Figure 4.7(d). Again, the higher the image intensity, lower the mass thickness and vice-versa.

110

Yang et al. proposed a mechanism by which they thought the nanofibril formation occurs. They speculated that the electrode will be initially covered with a layer of PAA that adsorbs from solution. The available spots on the electrode surface would then initiate the polymerization of PEDOT thus promoting preferential growth upward toward the solution. All along, as the fibrils grow out into fully formed nanofibrils, they would remain coated with a layer of PAA that continues to prevent lateral growth. While these speculations are physically reasonable, our recent observations from operando TEM have helped us develop a slightly different mechanism that agrees with our more recent data. Interestingly, PEDOT fails to form nanofibrils in the absence of PAA in the monomer mixture. So, it is true that PAA, being insulating due to lack of conjugation in its backbone, interferes with the charge transport between the electrode and PEDOT thus leaving only limited areas on the electrode for PEDOT to nucleate. Therefore, PAA initially helps prevent lateral growth on the electrode and promotes growth along the perpendicular direction. However, we did not find any evidence for a layer of PAA coated over the nanofibrils in the bright-field TEM (BFTEM) images. It is typical to see mass thickness contrast in BFTEM occurring due to density variations which was not the case here. Now consider the nucleated spots of PEDOT that have grown out or upward due the prevention of lateral growth by PAA. During the very initial stages, this almost behaves like a textured film with peaks and valleys. Fibrils made of PEDOT molecules are completely unaware of who/what is transferring them electrons to allow them to grow. Thus, when the fibrils grow they would grow as if the understructure established by the nucleated spots are the working electrodes. Further, the charge densities on the peaks of the fibrils would be the highest since they almost taper off

111 near the end as seen in the BFTEM and SEM images. This high charge density attracts the monomers and oligomers and the fibrils quickly grow along the length. The kinetics of the growth also corroborates this proposition because the fibrils grew relatively fast along the longitudinal direction initially. Once the fibrils attain a certain length, other factors like their distance from the electrode and monomer concentration start playing a bigger role. Also, now the entire surface of the fibrils (still rough and bumpy, so with varying charge densities) is available for charge transport thus allowing them to grow wider and thicker. Since the surface of the fibrils themselves had varying charge densities because of being rough (Figure 4.1(b)), they branched during their intermediate stages of growth. To reiterate, the kinetics of the growth are also in complete agreement with the mechanistic details proposed above.

4.4 Conclusions

Thus far, we have imaged the direct electrochemical fabrication of nanofibrils PEDOT via operando liquid-cell transmission electron microscopy. We saw that the growth occurs in three different regimes. In the first (I) stage, the monomers and the counter-ions were attracted to the working electrode due to the applied bias. There was no visible fibril-growth since PEDOT has to attain a certain molecular weight before it precipitates out of the aqueous solution. Second, where the fibrils grew at a relatively fast rate along their lengths compared to their widths and thicknesses. Lastly, the third regime, where the growth rate along the length of the fibrils was relatively low as they started to grow wider and thicker. We also estimated the growth velocities of the

112 fibrils along their lengths, widths and thicknesses. Additionally, we approximated the number of EDOT molecules that might be present in a fibril of average length, width and thickness at each timestamp. For this calculation, we assumed that PEDOT packs itself into an orthorhombic unit cell with the unit cell dimensions as a = 1.4 nm, b = 0.68 nm, and c = 0.78 nm and that 4 molecules of EDOT occupy the unit cell. We believe that factors like decrease in electric field strength and decrease in monomer concentration away from the electrode additionally contribute towards decrease in the growth velocities of the fibrils during the later stages of the polymerization process. Next, we looked at the nature of the morphological evolution of the nanofibrils. We could observe that during the initial stages of the process the central part of the fibrils grow. Not a lot of material got deposited into the interior parts of the

Figure 4.8: Schematic of the mechanism for the growth of nanofibrillar PEDOT proposed based on observations through liquid-cell electron microscopy. fibrils after the initial deposition. Further, these nanofibrils grew mostly from the

113 edges with the understructure established by the centre of the nanofibrils. From intensity profile measurements, we determined that they grew through their outer edges(presumably due to monomers and lower-molecular weight oligomers being attracted towards the central structure of the fibrils). A schematic of a typical fibril is shown in Figure 4.8. We could think of it as the interior parts of the fibril functioning as the working electrode and transferring electrons to the exterior part, a region where the monomers and the lower-molecular weight species are fairly concentrated. Particularly, the different phenomena and mechanisms revealed through transmission electron microscopy have given us a significantly improved understanding about the formation of fibrils during the electro-polymerization of PEDOT:PAA.

114

REFERENCES

1. Adam, Getachew, Martin Kaltenbrunner, Eric Daniel Głowacki, Dogukan Hazar Apaydin, Matthew Schuette White, Herwig Heilbrunner, Sekai Tombe, et al. 2016. “Solution Processed Perovskite Solar Cells Using Highly Conductive PEDOT:PSS Interfacial Layer.” Solar Energy Materials and Solar Cells 157 (December): 318–25. https://doi.org/10.1016/j.solmat.2016.05.011.

2. Chen, Kaiwu, Libin Chen, Yunqiang Chen, Hua Bai, and Lei Li. 2012. “Three-Dimensional Porous Graphene-Based Composite Materials: Electrochemical Synthesis and Application.” Journal of Materials Chemistry 22 (39): 20968–76. https://doi.org/10.1039/C2JM34816K.

3. Cui, Xinyan, and David C. Martin. 2003. “Electrochemical Deposition and Characterization of Poly(3,4-Ethylenedioxythiophene) on Neural Microelectrode Arrays.” Sensors and Actuators B: Chemical 89 (1): 92– 102. https://doi.org/10.1016/S0925-4005(02)00448-3.

4. Gottesfeld, Shimshon, Dario R. Dekel, Miles Page, Chulsung Bae, Yushan Yan, Piotr Zelenay, and Yu Seung Kim. 2018. “Anion Exchange Membrane Fuel Cells: Current Status and Remaining Challenges.” Journal of Power Sources 375 (January): 170–84. https://doi.org/10.1016/j.jpowsour.2017.08.010.

5. Hauwiller, Matthew R., Xiaowei Zhang, Wen-I Liang, Chung-Hua Chiu, Qian Zhang, Wenjing Zheng, Colin Ophus, et al. 2018. “Dynamics of Nanoscale Dendrite Formation in Solution Growth Revealed Through in Situ Liquid Cell Electron Microscopy.” Nano Letters 18 (10): 6427–33. https://doi.org/10.1021/acs.nanolett.8b02819.

6. Inal, Sahika, Jonathan Rivnay, Andreea-Otilia Suiu, George G. Malliaras, and Iain McCulloch. 2018. “Conjugated Polymers in Bioelectronics.” Accounts of Chemical Research 51 (6): 1368–76. https://doi.org/10.1021/acs.accounts.7b00624.

115

7. Jonge, Niels de, and Frances M. Ross. 2011. “Electron Microscopy of Specimens in Liquid.” Nature Nanotechnology 6 (11): 695–704. https://doi.org/10.1038/nnano.2011.161.

8. Khelfa, Abdelali, Caroline Byun, Jaysen Nelayah, Guillaume Wang, Christian Ricolleau, and Damien Alloyeau. 2019. “Structural Analysis of Single Nanoparticles in Liquid by Low-Dose STEM Nanodiffraction.” Micron 116 (January): 30–35. https://doi.org/10.1016/j.micron.2018.09.008.

9. Koutsouras, Dimitrios A., Paschalis Gkoupidenis, Clemens Stolz, Vivek Subramanian, George G. Malliaras, and David C. Martin. 2017. “Impedance Spectroscopy of Spin-Cast and Electrochemically Deposited PEDOT:PSS Films on Microfabricated Electrodes with Various Areas.” ChemElectroChem 4 (9): 2321–27. https://doi.org/10.1002/celc.201700297.

10. Kushima, Akihiro, Kang Pyo So, Cong Su, Peng Bai, Nariaki Kuriyama, Takanori Maebashi, Yoshiya Fujiwara, Martin Z. Bazant, and Ju Li. 2017. “Liquid Cell Transmission Electron Microscopy Observation of Lithium Metal Growth and Dissolution: Root Growth, Dead Lithium and Lithium Flotsams.” Nano Energy 32 (February): 271–79. https://doi.org/10.1016/j.nanoen.2016.12.001.

11. Liang, Jiajie, Lu Li, Xiaofan Niu, Zhibin Yu, and Qibing Pei. 2013. “Elastomeric Polymer Light-Emitting Devices and Displays.” Nature Photonics 7 (10): 817–24. https://doi.org/10.1038/nphoton.2013.242.

12. Liu, Jinglin, Bin Wei, Jennifer D. Sloppy, Liangqi Ouyang, Chaoying Ni, and David C. Martin. 2015. “Direct Imaging of the Electrochemical Deposition of Poly(3,4-Ethylenedioxythiophene) by Transmission Electron Microscopy.” ACS Macro Letters 4 (9): 897–900. https://doi.org/10.1021/acsmacrolett.5b00479.

13. Martin, David C. 2015. “Molecular Design, Synthesis, and Characterization of Conjugated Polymers for Interfacing Electronic Biomedical Devices with Living Tissue.” MRS Communications 5 (2): 131–53. https://doi.org/10.1557/mrc.2015.17.

14. Martin, David C., Jinghang Wu, Charles M. Shaw, Zachary King, Sarah A. Spanninga, Sarah Richardson-Burns, Jeffrey Hendricks, and Junyan Yang. 2010. “The Morphology of Poly(3,4-Ethylenedioxythiophene).” Polymer Reviews 50 (3): 340–84. https://doi.org/10.1080/15583724.2010.495440.

116

15. Morvant, Mark C., and John R. Reynolds. 1998. “In Situ Conductivity Studies of Poly(3,4-Ethylenedioxythiophene).” Synthetic Metals 92 (1): 57–61. https://doi.org/10.1016/S0379-6779(98)80023-4.

16. Murbach, Jamie M., Seth Currlin, Adrienne Widener, Yuxin Tong, Shrirang Chhatre, Vivek Subramanian, David C. Martin, Blake N. Johnson, and Kevin J. Otto. 2018. “In Situ Electrochemical Polymerization of Poly(3,4-Ethylenedioxythiophene) (PEDOT) for Peripheral Nerve Interfaces.” MRS Communications 8 (3): 1043–49. https://doi.org/10.1557/mrc.2018.138.

17. Nitta, Naoki, Feixiang Wu, Jung Tae Lee, and Gleb Yushin. 2015. “Li-Ion Battery Materials: Present and Future.” Materials Today 18 (5): 252–64. https://doi.org/10.1016/j.mattod.2014.10.040.

18. Ouyang, Liangqi, Rylie Green, Kathleen E. Feldman, and David C. Martin. 2011. “Direct Local Polymerization of Poly(3,4-Ethylenedioxythiophene) in Rat Cortex.” Progress in Brain Research 194: 263–71. https://doi.org/10.1016/B978-0-444-53815-4.00001-7.

19. Ouyang, Liangqi, Crystal L. Shaw, Chin-Chen Kuo, Amy L. Griffin, and David C. Martin. 2014. “In Vivo Polymerization of Poly(3,4- Ethylenedioxythiophene) in the Living Rat Hippocampus Does Not Cause a Significant Loss of Performance in a Delayed Alternation Task.” Journal of Neural Engineering 11 (2): 026005. https://doi.org/10.1088/1741- 2560/11/2/026005.

20. Ouyang, Liangqi, Bin Wei, Chin-chen Kuo, Sheevangi Pathak, Brendan Farrell, and David C. Martin. 2017. “Enhanced PEDOT Adhesion on Solid Substrates with Electrografted P(EDOT-NH2).” Science Advances 3 (3): e1600448. https://doi.org/10.1126/sciadv.1600448.

21. Quirós-Solano, W. F., N. Gaio, C. Silvestri, G. Pandraud, and P. M. Sarro. 2016. “PEDOT:PSS: A Conductive and Flexible Polymer for Sensor Integration in Organ-on-Chip Platforms.” Procedia Engineering, Proceedings of the 30th anniversary Eurosensors Conference – Eurosensors 2016, 4-7. Sepember 2016, Budapest, Hungary, 168 (January): 1184–87. https://doi.org/10.1016/j.proeng.2016.11.401.

22. Randriamahazaka, H, V Noël, and C Chevrot. 1999. “Nucleation and Growth of Poly(3,4-Ethylenedioxythiophene) in Acetonitrile on Platinum under Potentiostatic Conditions.” Journal of Electroanalytical Chemistry 472 (2): 103–11. https://doi.org/10.1016/S0022-0728(99)00258-2.

117

23. Richardson-Burns, Sarah M., Jeffrey L. Hendricks, Brian Foster, Laura K. Povlich, Dong-Hwan Kim, and David C. Martin. 2007. “Polymerization of the Conducting Polymer Poly(3,4-Ethylenedioxythiophene) (PEDOT) around Living Neural Cells.” Biomaterials 28 (8): 1539–52. https://doi.org/10.1016/j.biomaterials.2006.11.026.

24. Rivnay, Jonathan, Róisín M. Owens, and George G. Malliaras. 2014. “The Rise of Organic Bioelectronics.” Chemistry of Materials 26 (1): 679–85. https://doi.org/10.1021/cm4022003.

25. Ross, F. M. 2015. “Opportunities and Challenges in Liquid Cell Electron Microscopy.” Science 350 (6267): aaa9886–aaa9886. https://doi.org/10.1126/science.aaa9886.

26. Ventosa, Edgar, Jesús López Palacios, and Patrick R. Unwin. 2008. “Nucleation and Growth of Poly(3,4-Ethylenedioxythiophene) Thin Films on Highly Oriented Pyrolytic Graphite (HOPG) Electrodes.” Electrochemistry Communications 10 (11): 1752–55. https://doi.org/10.1016/j.elecom.2008.09.003.

27. Wang, Jikui, Guofeng Cai, Xudong Zhu, and Xiaping Zhou. 2012. “Oxidative Chemical Polymerization of 3, 4-Ethylenedioxythiophene and Its Applications in Antistatic Coatings.” Journal of Applied Polymer Science 124 (1): 109–15. https://doi.org/10.1002/app.35045.

28. Wei, Bin, Jinglin Liu, Liangqi Ouyang, and David C. Martin. 2017. “POSS-ProDOT Crosslinking of PEDOT.” Journal of Materials Chemistry B 5 (25): 5019–26. https://doi.org/10.1039/C7TB00598A.

29. Wei, Xuefeng F., and Warren M. Grill. 2009. “Analysis of High-Perimeter Planar Electrodes for Efficient Neural Stimulation.” Frontiers in Neuroengineering 2. https://doi.org/10.3389/neuro.16.015.2009.

30. Yang, Junyan, Karen Lipkin, and David C. Martin. 2007. “Electrochemical Fabrication of Conducting Polymer Poly(3,4-Ethylenedioxythiophene) (PEDOT) Nanofibrils on Microfabricated Neural Prosthetic Devices.” Journal of Biomaterials Science, Polymer Edition 18 (8): 1075–89. https://doi.org/10.1163/156856207781494359.

31. Yang, Junyan, and David C Martin. 2004. “Microporous Conducting Polymers on Neural Microelectrode Arrays: II. Physical Characterization.” Sensors and Actuators A: Physical 113 (2): 204–11. https://doi.org/10.1016/j.sna.2004.02.029.

118

32. Zhou, Cuifeng, Zongwen Liu, Xusheng Du, and Simon P. Ringer. 2010. “Electrodeposited PEDOT Films on ITO with a Flower-like Hierarchical Structure.” Synthetic Metals 160 (15): 1636–41. https://doi.org/10.1016/j.synthmet.2010.05.033.

119

Chapter 5

MORPHOLOGY, MOLECULAR ORIENTATION AND SOLID-STATE CHARACTERIZATION OF 2,3-DIHYDROTHIENO[3,4-B] [1,4]DIOXINE-2- CARBOXYLIC ACID (EDOTACID)

This work has been published in Crystal Growth and Design (DOI: 10.1021/acs.cgd.9b00613) . Reprinted with permission from MORPHOLOGY,

5.1 Introduction

Conjugated polymers have been investigated for a variety of applications including electrochromic displays, O photovoltaics and chemical OH sensors(Kawahara et al. 2012; Ikeda et O O O O al. 2016; Zhu et al. 2004). They have been of recent interest for interfacing H H H H S S biomedical devices with living tissue (a) (b) because of their ability to conduct charge Figure 5.1:(a) Chemical structures of 3,4- ethylenedioxythiophene (EDOT) both electronically and ionically (Martin (b) carboxylic acid-functionalized 3,4- ethylenedioxythiophene (EDOTacid). 2015; Rivnay et al. 2016). Poly(3,4- ethylenedioxythiophene) (PEDOT) has

120 received particular scientific and commercial attention because of its low oxidation potential, high chemical and thermal stability, and high conductivity(Groenendaal et al. 2000; Crispin et al. 2006). The lack of hydrophilic side groups makes it difficult for the relatively hydrophobic PEDOT to interact with living tissues. 2,3-dihydrothieno[3,4- b][1,4]dioxine-2-carboxylic acid (EDOTacid) (Figure 5.1) is a comonomer for 3,4- ethylenedioxythiophene (EDOT) designed to introduce controlled amounts of hydrophilicity into the resulting copolymers (Povlich et al. 2013). Electropolymerized PEDOT-PEDOTacid copolymer films can be coupled with peptides resulting in films with increased bio-activity(Povlich et al. 2013). These copolymers have a more pleated, open surface texture and higher surface energies than the PEDOT homopolymer(Bhagwat, Kiick, and Martin 2014). The EDOTacid monomer also has been shown to serve as an adhesion promoter for electropolymerized films of PEDOT (Wei et al. 2015). Studies aimed at taking advantage of the carboxylic substitutions in conjugated polymers to increase biological activity and thereby optimizing performances of devices like bio-sensors have also been published (Mouffouk and Higgins 2006; Lee et al. 2006; Kwon et al. 2016). Although the EDOT monomer (Figure 5.1(a)) is a liquid at room temperature, EDOTacid is a crystalline solid. Crystalline organic, small molecules such as pentacene have been used to fabricate thin-film transistors with relatively high mobilities (Newman et al. 2005). EDOT oligomers have also been synthesized and used to study the influence of molecular weight on electronic structure and optical properties (Apperloo et al. 2002). In this study, we show that EDOTacid single crystals have a preferred orientation when deposited onto flat substrates. This can be

121 utilized for other applications like forming textured conducting films for improving the order in homopolymer PEDOT coatings. Thus, this paper focuses to studying properties like crystal structure, shape, electron stability and thermal stability of EDOTacid crystals to understand its fundamental properties and interactions, giving us insights into improving order in PEDOT polymers and copolymers. We anticipated that since EDOTacid is reasonably similar to benzoic acid, it might also form hydrogen-bonded dimers in the solid-state. However, since EDOTacid also has a chiral center at the carbon where the carboxylic acid is attached, there will necessarily be enantiomers with opposite left and right handedness present in the as-synthesized racemic mixture. The manner of packing of these enantiomers in the solid-state was not known. The purpose of this paper is to understand the structure-property relationships of EDOT-acid molecular crystals using a variety of characterization techniques including optical and electron microscopy, thermal analysis and X-ray diffraction. Since crystallinity in P(EDOT-co-EDOTacid) copolymers have not yet been thoroughly investigated, this work will also provide some of the necessary groundwork for this purpose. Considering that this would require careful tuning of the pH or using a mixture of solvents to control the precipitation rate of the resulting copolymers to crystallize them, this will be part of a future study. Transmission electron microscopy (TEM) and electron diffraction (ED) have been widely used for the determination of crystal structures and morphology (Shaw et al. 2013; Chen, Martin, and Anthony 2007). While X-rays do not cause much damage to the crystallinity in organic materials, the high energy electrons of the TEM cause significant damage (Libera and Egerton 2010; Martin et al. 2010). This makes it a far

122 more challenging task to image organic materials using TEM. The damage caused by the electron beam can be quantified by estimating the critical electron doses using techniques like ED and Electron Energy Loss Spectroscopy (EELS) (Guo et al. 2015; Leijten et al. 2017). These estimates can then be used to optimize the electron dose used for imaging beam sensitive crystals. Here, we have used X-ray diffraction, TEM, ED, SEM, polarized optical microscopy, thermal analysis and molecular simulations to investigate the crystal structure and morphology of EDOTacid and study its interactions with the electron beam.

5.2 Experimental section

5.2.1 Materials used. EDOTacid powder was purchased from Tractus-chemistry and Shanghai Seebio Biotech, LLC and used as received. Dichloromethane (DCM) and acetone were purchased from Thermo Fisher Scientific.

5.2.2 Single crystal growth. For performing unit cell structure determinations, relatively large single crystals of EDOTacid (from 10s to 100s of microns in size) were grown by controlled evaporation of a dichloromethane (DCM) solution over a period of about 3 weeks.

123

5.2.3 Single-crystal X-ray diffraction. A crystal of 428 x 110 x 32 µm was obtained by slow evaporation of a dilute (0.0035 wt%) solution of EDOTacid in DCM. Crystals were mounted using viscous oil onto a plastic mesh and cooled to the data collection temperature. Data were collected on a Bruker-AXS APEX II DUO CCD diffractometer with Cu-Kα radiation (λ = 1.54178 Å) focused with Goebel mirrors. Unit cell parameters were obtained from 36 data frames, 0.5º ω, from three different sections of the Ewald sphere (Apex 3). The data were treated with multi-scan absorption corrections. The structure was solved using intrinsic phasing methods and refined with full-matrix, least-squares procedures on F2 (G. M. Sheldrick 2015; George M. Sheldrick 2015). Non-hydrogen atoms were refined with anisotropic displacement parameters. The carboxylic H-atom was located from the electron density difference map and was allowed to refine freely. All other H-atoms were treated as idealized contributions with geometrically calculated positions and with U iso equal to 1.2 U eq of the attached atom. Atomic scattering factors are contained in the SHELXTL program library (George M. Sheldrick 2015). The structure has been deposited at the Cambridge Structural Database under CCDC 1895659.

5.2.4 Structure and Morphology Characterizations. Transmission electron micrographs and electron diffraction patterns were obtained using a ThermoFisher Talos F200C Scanning Transmission Electron Microscope (STEM) at 200 kV. Polarized optical micrographs were acquired with a Nikon-Eclipse L200 using a full-wave red filter.

124

Scanning electron microscopy was performed using a Zeiss Auriga 60 Focused Ion Beam-Scanning Electron Microscope (FIB-SEM) operating at 3 kV. Powder X-ray diffraction was performed on Bruker D8 diffractometer in a symmetric θ/2θ reflection geometry using monochromated Cu-Kα radiation (λ=1.54 Å) with 2θ ranging from 5o to 70o. Atomic force microscopy with infrared spectroscopy (AFM-IR) was performed on a Bruker Nano-IR2 (Figure A4).

5.2.5 Physical properties Infrared experiments were conducted using a Thermo Nicolet 670 Nexus FT- IR spectrometer with a DTGS detector operating in attenuated total reflection (ATR) mode using a Specac Golden Gate ATR accessory. The spectra were obtained by averaging 128 scans from 600 cm-1 to 4000 cm-1 (Figure A2). Thermo gravimetric analysis was performed on TA Discovery TGA (Figure A3).

125

5.2.6 Molecular Simulations. Crystal Maker 10 software was used for performing diffraction pattern

(a) (b)

Freidel Donner Harkey (BFDH) calculations(Macrae et al. 2006).

5.3 Results and discussion EDOTacid can be readily crystallized by sublimation, solution casting, or slow growth from a polar solvent. Controlled diffusion of a non-polar solvent into a polar solvent can be used to induce slow precipitation, since the molecule is more soluble in polar solvents as compared to non-polar solvents. EDOTacid is expected to be more soluble in water at high pH, due to the presence of the carboxylic acid functional group in the molecule. Highly birefringent single crystals of EDOTacid were readily observed by solution casting from solvents such as acetone, as seen in the polarized

126 optical micrographs in Figures 5.2(a) and 5.2(b). Images obtained with a polarized light and a full wave red filter indicated that the higher refractive index direction was parallel to the long axes of the needle shaped crystals. Slowly solution-grown single crystals could be formed with overall dimensions of several hundred microns. Long needle crystals were also readily formed by sublimation. The unit cell of EDOTacid single crystals obtained by slow evaporation of the solvent was estimated from the single crystal X-ray diffraction as follows:

Table 5.1 Unit cell parameters of 2,3-dihydrothieno[3,4-b][1,4]dioxine-2- carboxylic acid.

Molecular Formula C7H6O4S Molecular Weight 186.18 g/mol Space group Pbca Lattice system Orthorhombic a 1.014 nm b 0.699 nm c 2.139 nm Density 1.63 g/cm3 Temperature 200(2)K Z 8 µ 3.596/mm

R1 [I>2σ] 0.0782

wR2 0.1873

127

The systematic absences in the diffraction data were uniquely consistent with Pbca symmetry (space group No. 61). It was also found that the compound was arranged into H-bonded pairs around an inversion center. The atomic positions of EDOTacid molecules are shown in Figure 5.3 in the [100], [010] and [001] viewing directions. There are 8 molecules of EDOTacid per unit cell, resulting in an estimated density of 1.63 g/cm3. This value is slightly higher than the density of unsubstituted EDOT (1.42 g/cm3) which is a liquid at room temperature. EDOTacid molecules pack

via the formation of hydrogen bonded dimers between the carboxylic acid groups in a manner similar to that seen in benzoic acid (BA) molecules (Sim, Robertson, and Goodwin 1955; Feld et al. 1981). BA packs in a monoclinic

unit-cell (P21/c) where the hydrogen bonded dimers are arranged in a herringbone fashion (Sim, Robertson, and Goodwin 1955; Feld et al. 1981; Dubey,

(a) (b)

Figure 5.3: Projections of EDOTacid unit cell along the (a)[100] (b)[010] (c)[001] (d) [012] ((002) slice). Axes labels: red, green and blue colors correspond to a, b and c axes, respectively. �

128

Pavan, and Desiraju 2012). The herringbone packing can be seen in both EDOTacid and BA crystals, corresponding to the [012] and [104] directions, respectively [27]. While the atoms in the BA dimer are all essentially� in� the same plane, for EDOTacid some of the atoms lie out of the plane of the thiophene ring. The crystal structure has the EDOTacid molecules paired into dimers, with each pair consisting of one left- handed and one right-handed enantiomer. These enantiomers are paired via hydrogen bonding which is evident when viewed along the [010] direction (Figure 5.3(b)). The packing of the enantiomers alternates along both the a and c directions with the dimer axis oriented at an angle which is smallest with respect to the b direction. This results in highly anisotropic shapes and mechanical properties, as is observed from the characteristic cracking in the single crystals (Figure 5.8). The herringbone-like packing motif seen here is quite common in other small molecules such as pentacene and benzoic acid (Tang et al. 2017; Holmbäck and Rasmuson 1999; Drummy et al. 2006). However, what is somewhat different here is that the herringbone crystal packing simultaneously propagates in two different lateral directions, leading to an overall orthogonal, rather than monoclinic crystal symmetry. BA molecules also forms platelet-like crystals which have been observed experimentally and calculated computationally (Holmbäck and Rasmuson 1999; Liang et al. 2017).

129

(a) (b)

Figure 5.4: (a) Indexed simulated powder X-ray diffraction pattern of the unit cell (CrystalMaker 10)) (b) X-ray powder diffraction pattern of the EDOTacid powder as received (c) X- ray pattern of a textured thin-film (taken in a symmetric reflection θ/2θ geometry) of EDOTacid deposited from acetone.

(c)

130

The positions and relative intensities of the peaks as well as absences due to the symmetry in the simulated pattern (CrystalMaker 10) were fully consistent with the experimental powder pattern (Figure 5.4). Thin films formed by solution casting

(a) (b)

131 on flat substrates were observed to be highly crystallographically textured by X-ray diffraction measurements. The X-ray pattern of such a textured film taken in a symmetric reflection (θ/2θ) geometry is shown in Figure 5.4(c). which shows predominant peaks from the (002), (004) and (006) planes, indicating that these are preferentially aligned parallel to the substrate.

Low-dose TEM and ED have been extensively used recently for structure determination of organic small molecules, biological molecules and proteins(Hattne et al. 2018; Nannenga and Gonen 2014; Nannenga, Bu, and Shi 2018; S. Liu et al. 2017). Electron microscopy makes it possible to examine crystals several orders of magnitude smaller than those needed for traditional crystallographic methods (S. Liu et al. 2017). In addition to using single crystal x-ray diffraction, we used low-dose TEM and ED to study these beam-sensitive slowly grown single crystals from DCM. Some of the crystals were elongated needles with high aspect ratios, while others were more equiaxed. Bright field TEM of both elongated and equiaxed crystals are shown in Figures 5.5(a) and 5.5(b). Contrast variations typical of bend contours were also observed in the crystals. Being sensitive to the electron beam, the bend contours would move continuously through the crystallite before disappearing completely after the sample lost its crystallinity. The ED data were also consistent with the calculated crystal structure. Figure 5.5(c) shows the electron diffraction pattern from the single crystal in Figure 5.5(b).

132

Analysis of ED patterns like those shown in Figure 5.5(b) showed that the preferred contact plane was (001), corresponding to a preferred zone axis of [001]. The patterns calculated from ED also were consistent with those calculated from unit cell determined from single crystal X-ray diffraction. Figure 5.5(d) shows the

(a) (b) (c)

Figures. 5.6: Electron diffraction patterns from a single crystal of EDOTacid at electron doses of (a) 0 e/A2 (0 mC/cm2), (b) 5.5 e/A2 (9 mC/cm2), and (c) 9 e/A2 (15 mC/cm2).

simulated electron diffraction patterns as viewed from the [001] direction. The absences are due to the glide plane and screw axis symmetries characteristic of the Pbca space group. All the absences were observed as expected from the simulations except for the odd (0k0) absences. These on axis absences are violated in the ED presumably due to double diffraction. The sensitivity of these crystals to the electron beam caused the Bragg reflections to fade as the diffraction patterns were being recorded (Figure 5.6). The total end-point dose (TEPD) is defined by the electron dose needed for the amorphization these crystals, and was estimated to be 9 e/A2 (15 mC/cm2). This value

133

is of the same order of the TEPD of polyethylene which has been reported to be around ~ 2 e/A2 at 120 kV. This TEPD value is also in agreement with the thermal stability of EDOTacid (~146oC – Figure A3). The correlation between the thermal stability of an organic specimen (degradation temperature or melting point) and its electron-beam stability was investigated by Kumar and Adams in 1990(Kumar and Adams 1990).Since the beam-induced damage is attributed to selective bond breakage and their subsequent inability to recombine, Kumar and Adams had an intuitive argument that a specimen with higher thermal stability will require more energy for damage. Additionally, at lower specimen temperatures, more energy (electron dose) is required for damaging the specimen (Leijten et al. 2017). This can then be related to Kumar and Adams’ argument that at lower temperatures, owing to reduced molecular mobility, the probability of recombination of a broken bond is higher. Figure 5.7 shows the evolution of the intensity profiles of the (200) and the (220) reflections as a function of 220 spacing 200 spacing electron dose. There were a 300 couple of consistent observations

200 during these experiments. The

Intensity% higher order Bragg reflections 100 faded first followed by the lower

0 0 5 10 15 order reflections indicating that Dose (mC/cm2) higher amounts of electron doses Figure 5.7: Intensity profiles of diffraction spots of are required for the (200) and (220) reflections with increasing electron doses. The total end point dose was estimated to be amorphization of the lower order 9 e/A2 (15 mC/cm2) (larger d-spacing) reflections.

134

Meaning, short-range order is destroyed, while long-range order still exists. This intermediate state with the long-range order can be compared to a block copolymer consisting of ordered blocks. This state however is not indestructible and the crystals amorphize after prolonged exposure. Second, we found evidence for a modest increase in intensity and increase in peak width (decrease in effective crystallite size) at intermediate doses. This decrease in effective crystallite size may be due to the bigger crystallites breaking up into smaller ones under continued flux. Similar intermediate states during electron beam irradiation have been seen in other organic and materials like 2,6-bis-octyl-pentathienoacene (C8) and copolymers of poly(vinylidene fluoride)(Wu, Shaw, and Martin 2012; Lovinger 1985). Since the dimer axis is aligned closest to the b direction or the [010] direction, we theorize that it is easier for the EDOTacid single crystals to cleave in the perpendicular directions, thereby causing cracks along directions almost parallel to the b direction. This is what we observed experimentally using SEM (Figure 5.8) which

(a) (b) (c)

Figure 5.8: Images of the single crystals of EDOTacid as recorded from the scanning electron microscope (Auriga 60 CrossBeam FIB/SEM). The a and the b directions of a single crystal are shown.

135 showed roughly rectangular, flat platelets with pronounced cracking parallel to their long axes near the platelet ends. The size and shape of these single crystals varied somewhat with the solvent and the processing conditions. From ED we know that the EDOTacid crystals go down with their (001) planes preferentially parallel to the substrate (or equivalently, with the [001] direction perpendicular to the substrate) and that the growth rate is highest along the b direction [010]. These cracks reveal the local anisotropy of physical properties that results from this molecular arrangement

(Figure 5.3). Single crystals of EDOTacid were also grown by sublimation. Crystals grown in this manner were faceted and needle shaped and usually larger in all dimensions as compared to the flat, platelet-like morphology that we observe in single crystals from solution. One such crystal is shown in the scanning electron micrograph in Figure 5.9(a). The needle-shaped crystals have thin layers/steps stacked on top of each other in AFM micrographs (Figure A4) with faceting observed primarily on the {002},

(a) (b)

Figure 5.9: (a) SEM image of a crystal formed by sublimation of the EDOTacid powder (b) Simulated shape of the EDOTacid single crystal using BFDH calculations (Mercury 4.0.0).

136

{200}and {220} type surfaces, and occasionally on {111} as well. We compared these observations with the simulated shapes predicted by the Bravais-Friedel-Donnay- Harker (BFDH) method using Mercury 4.0.0 software. The BFDH method uses the symmetry and shape of the unit cell to estimate crystal growth faces and their growth rates, but does not account for details in local molecular packing and intermolecular energetics. The overall shape and platelet motif of the EDOTacid crystals was reproduced well (Figure 5.9(b)). However, the BFDH method also predicted the existence of some of the faces of {102} type instead of {200} type.

5.4 Conclusions: We have examined the physical properties and have provided a detailed structural characterization of the solid-state packing of EDOTacid. This monomer has already shown the ability to introduce hydrophilicity and bio-activity in P(EDOT-co- EDOTacid) copolymers[8][9]. This study additionally shows that the carboxylic acid groups can potentially be used to improve the crystallinity of these conjugated polymers. However, this would require careful control of the pH or use of a combination of solvents to control the precipitation rate and thereby crystallize the polymer chains and will be part of a future study. We determined the crystal structure based on the data from single crystal X-ray diffraction and corroborating information from TEM, ED and powder X-ray diffraction. In crystals of EDOTacid the monomers arranged into hydrogen-bonded dimers, with the two molecules in each dimer having opposite handedness. The dimers are tilted by ~60 degrees with respect to the c-axis. The dimers are oriented closest to the b direction, resulting in anisotropic physical properties in the single

137 crystals. This was observed in crystals formed by precipitation from solution which were predominantly flat platelets, with cracks parallel to the b direction. As compared to the solution-grown single crystals, crystals grown by sublimation were much more elongated and needle-shaped with faceting observed primarily on the {002}, {200}, {220}, and {111} type planes. Our research group has been investigating the nucleation and growth of electrodeposition of polythiophenes, particularly PEDOT and its derivatives to optimize properties of the final polymer (J. Liu et al. 2015). For us to be able to perform similar reactions with EDOTacid in the TEM and understand its growth mechanisms, we need to have an estimate of its electron beam stability. Consequently, we obtained the electron-beam stability to be ~9 e/A2 (15 mC/cm2) which is consistent with its thermal stability of ~146 oC. In like manner, we hope our physical characterization of EDOTacid presented promotes better utility in current PEDOT applications and spur further new applications.

138

REFERENCES

1. Apex3. Bruker AXS Inc., Madison, WI, USA (2015).

2. Apperloo, Joke J., L. “Bert” Groenendaal, Hilde Verheyen, Manickam Jayakannan, René A. J. Janssen, Ahmed Dkhissi, David Beljonne, Roberto Lazzaroni, and Jean-Luc Brédas. 2002. “Optical and Redox Properties of a Series of 3,4‐Ethylenedioxythiophene Oligomers.” Chemistry – A European Journal 8 (10): 2384–96. https://doi.org/10.1002/1521- 3765(20020517)8:10<2384::AID-CHEM2384>3.0.CO;2-L.

3. Bhagwat, Nandita, Kristi L. Kiick, and David C. Martin. 2014. “Electrochemical Deposition and Characterization of Carboxylic Acid Functionalized PEDOT Copolymers.” Journal of Materials Research 29 (23): 2835–44. https://doi.org/10.1557/jmr.2014.314.

4. Chen, Jihua, David C. Martin, and John E. Anthony. 2007. “Morphology and Molecular Orientation of Thin-Film Bis(Triisopropylsilylethynyl) Pentacene.” Journal of Materials Research 22 (06): 1701–9. https://doi.org/10.1557/JMR.2007.0220.

5. Crispin, X., F. L. E. Jakobsson, A. Crispin, P. C. M. Grim, P. Andersson, A. Volodin, C. van Haesendonck, M. Van der Auweraer, W. R. Salaneck, and M. Berggren. 2006. “The Origin of the High Conductivity of Poly(3,4- Ethylenedioxythiophene)−Poly(Styrenesulfonate) (PEDOT−PSS) Plastic Electrodes.” Chemistry of Materials 18 (18): 4354–60. https://doi.org/10.1021/cm061032+.

6. Drummy, Lawrence F., Paul K. Miska, David Alberts, Nuram Lee, and David C. Martin. 2006. “Imaging of Crystal Morphology and Molecular Simulations of Surface Energies in Pentacene Thin Films.” The Journal of Physical Chemistry B 110 (12): 6066–71. https://doi.org/10.1021/jp054951g.

7. Dubey, Ritesh, Mysore S. Pavan, and Gautam R. Desiraju. 2012. “Structural Landscape of Benzoic Acid: Using Experimental Crystal Structures of Fluorobenzoic Acids as a Probe.” Chemical Communications 48 (72): 9020–22. https://doi.org/10.1039/C2CC33811D.

139

8. Feld, R., M.S. Lehmann, K.W. Muir, and J.C. Speakman. 1981. “The Crystal Structure of Benzoic Acid: A Redetermination with X-Rays at Room Temperature.” Zeitschrift Fuer Kristallographie 157 (3/4): 215–31.

9. Groenendaal, By L, Friedrich Jonas, Dieter Freitag, Harald Pielartzik, and John R Reynolds. 2000. “Its Derivatives : Past , Present , and Future **,” 481–494.

10. Guo, Changhe, Frances I. Allen, Youngmin Lee, Thinh P. Le, Chengyu Song, Jim Ciston, Andrew M. Minor, and Enrique D. Gomez. 2015. “Probing Local Electronic Transitions in Organic Semiconductors through Energy-Loss Spectrum Imaging in the Transmission Electron Microscope.” Advanced Functional Materials 25 (38): 6071–6076. https://doi.org/10.1002/adfm.201502090.

11. Hattne, Johan, Dan Shi, Calina Glynn, Chih-Te Zee, Marcus Gallagher- Jones, Michael W. Martynowycz, Jose A. Rodriguez, and Tamir Gonen. 2018. “Analysis of Global and Site-Specific Radiation Damage in Cryo- EM.” Structure 26 (5): 759-766.e4. https://doi.org/10.1016/j.str.2018.03.021.

12. Holmbäck, X, and Å. C Rasmuson. 1999. “Size and Morphology of Benzoic Acid Crystals Produced by Drowning-out Crystallisation.” Journal of Crystal Growth 198–199 (March): 780–88. https://doi.org/10.1016/S0022-0248(98)01027-6.

13. Ikeda, Natsumi, Tomoyuki Koganezawa, Daisuke Kajiya, and Ken-ichi Saitow. 2016. “Performance of Si/PEDOT:PSS Hybrid Solar Cell Controlled by PEDOT:PSS Film Nanostructure.” The Journal of Physical Chemistry C 120 (34): 19043–48. https://doi.org/10.1021/acs.jpcc.6b07101.

14. Kawahara, Jun, Peter Andersson Ersman, Isak Engquist, and Magnus Berggren. 2012. “Improving the Color Switch Contrast in PEDOT:PSS- Based Electrochromic Displays.” Organic Electronics 13 (3): 469–74. https://doi.org/10.1016/j.orgel.2011.12.007.

15. Kumar, Satish, and W. Wade Adams. 1990. “Electron Beam Damage in High Temperature Polymers.” Polymer 31 (1): 15–19. https://doi.org/10.1016/0032-3861(90)90341-U.

140

16. Kwon, Oh Seok, Chul Soon Park, Seon Joo Park, Seonmyeong Noh, Saerona Kim, Hye Jeong Kong, Joonwon Bae, Chang-Soo Lee, and Hyeonseok Yoon. 2016. “Carboxylic Acid-Functionalized Conducting- Polymer Nanotubes as Highly Sensitive Nerve-Agent Chemiresistors.” Scientific Reports 6 (September): 33724. https://doi.org/10.1038/srep33724.

17. Lee, Joo-Woon, Francisco Serna, Jonathan Nickels, and Christine E. Schmidt. 2006. “Carboxylic Acid-Functionalized Conductive Polypyrrole as a Bioactive Platform for Cell Adhesion.” Biomacromolecules 7 (6): 1692–95. https://doi.org/10.1021/bm060220q.

18. Leijten, Zino J. W. A., Arthur D. A. Keizer, Gijsbertus de With, and Heiner Friedrich. 2017. “Quantitative Analysis of Electron Beam Damage in Organic Thin Films.” The Journal of Physical Chemistry C 121 (19): 10552–61. https://doi.org/10.1021/acs.jpcc.7b01749.

19. Liang, Zuozhong, Min Zhang, Fei Wu, Jian-Feng Chen, Chunyu Xue, and Hong Zhao. 2017. “Supersaturation Controlled Morphology and Aspect Ratio Changes of Benzoic Acid Crystals.” Computers & Chemical Engineering 99 (April): 296–303. https://doi.org/10.1016/j.compchemeng.2017.01.010.

20. Libera, Matthew R., and Ray F. Egerton. 2010. “Advances in the Transmission Electron Microscopy of Polymers.” Polymer Reviews 50 (3): 321–39. https://doi.org/10.1080/15583724.2010.493256.

21. Liu, Jinglin, Bin Wei, Jennifer D. Sloppy, Liangqi Ouyang, Chaoying Ni, and David C. Martin. 2015. “Direct Imaging of the Electrochemical Deposition of Poly(3,4-Ethylenedioxythiophene) by Transmission Electron Microscopy.” ACS Macro Letters 4 (9): 897–900. https://doi.org/10.1021/acsmacrolett.5b00479.

22. Liu, Shian, Johan Hattne, Francis E. Reyes, Silvia Sanchez‐Martinez, M. Jason de la Cruz, Dan Shi, and Tamir Gonen. 2017. “Atomic Resolution Structure Determination by the Cryo-EM Method MicroED.” Protein Science 26 (1): 8–15. https://doi.org/10.1002/pro.2989.

23. Lovinger, Andrew J. 1985. “Polymorphic Transformations in Ferroelectric Copolymers of Vinylidene Fluoride Induced by Electron Irradiation.” Macromolecules 18 (5): 910–18. https://doi.org/10.1021/ma00147a016.

141

24. Macrae, C. F., P. R. Edgington, P. McCabe, E. Pidcock, G. P. Shields, R. Taylor, M. Towler, and J. van de Streek. 2006. “Mercury: Visualization and Analysis of Crystal Structures.” Journal of Applied Crystallography 39 (3): 453–57. https://doi.org/10.1107/S002188980600731X.

25. Martin, David C. 2015. “Molecular Design, Synthesis, and Characterization of Conjugated Polymers for Interfacing Electronic Biomedical Devices with Living Tissue.” MRS Communications 5 (2): 131–53. https://doi.org/10.1557/mrc.2015.17.

26. Martin, David C., Jinghang Wu, Charles M. Shaw, Zachary King, Sarah A. Spanninga, Sarah Richardson-Burns, Jeffrey Hendricks, and Junyan Yang. 2010. “The Morphology of Poly(3,4-Ethylenedioxythiophene).” Polymer Reviews 50 (3): 340–84. https://doi.org/10.1080/15583724.2010.495440.

27. Mouffouk, Fouzi, and Simon J. Higgins. 2006. “Oligonucleotide– Functionalised Poly(3,4-Ethylenedioxythiophene)-Coated Microelectrodes Which Show Selective Electrochemical Response to Hybridisation.” Electrochemistry Communications 8 (2): 317–22. https://doi.org/10.1016/j.elecom.2005.11.027.

28. Nannenga, Brent L., Guanhong Bu, and Dan Shi. 2018. “The Evolution and the Advantages of MicroED.” Frontiers in Molecular Biosciences 5 (December). https://doi.org/10.3389/fmolb.2018.00114.

29. Nannenga, Brent L., and Tamir Gonen. 2014. “Protein Structure Determination by MicroED.” Current Opinion in Structural Biology 27 (August): 24–31. https://doi.org/10.1016/j.sbi.2014.03.004.

30. Newman, Christopher R., Reid J. Chesterfield, Matthew J. Panzer, and C. Daniel Frisbie. 2005. “High Mobility Top-Gated Pentacene Thin-Film Transistors.” Journal of Applied Physics 98 (8): 084506. https://doi.org/10.1063/1.2076429.

31. Palmer, D. C. (2014). CrystalMaker 10. CrystalMaker Software Ltd, Begbroke, Oxfordshire, England

32. Povlich, Laura K., Jae Cheol Cho, Michelle K. Leach, Joseph M. Corey, Jinsang Kim, and David C. Martin. 2013. “Synthesis, Copolymerization and Peptide-Modification of Carboxylic Acid-Functionalized 3,4- Ethylenedioxythiophene (EDOTacid) for Neural Electrode Interfaces.” Biochimica et Biophysica Acta - General Subjects 1830 (9): 4288–4293. https://doi.org/10.1016/j.bbagen.2012.10.017.

142

33. Rivnay, Jonathan, Sahika Inal, Brian A. Collins, Michele Sessolo, Eleni Stavrinidou, Xenofon Strakosas, Christopher Tassone, Dean M. Delongchamp, and George G. Malliaras. 2016. “Structural Control of Mixed Ionic and Electronic Transport in Conducting Polymers.” Nature Communications 7 (April): 11287. https://doi.org/10.1038/ncomms11287.

34. Shaw, Charles M., Xinnan Zhang, Lidaris San Miguel, Adam J. Matzger, and David C. Martin. 2013. “Synthesis and Structure of α-Substituted Pentathienoacenes.” Journal of Materials Chemistry C 1 (23): 3686. https://doi.org/10.1039/c3tc30144c.

35. Sheldrick, G. M. 2015. “SHELXT – Integrated Space-Group and Crystal- Structure Determination.” Acta Crystallographica Section A: Foundations and Advances 71 (1): 3–8. https://doi.org/10.1107/S2053273314026370.

36. Sheldrick, George M. 2015. “Crystal Structure Refinement with SHELXL.” Acta Crystallographica Section C Structural Chemistry 71 (1): 3–8. https://doi.org/10.1107/S2053229614024218.

37. Sim, G. A., J. M. Robertson, and T. H. Goodwin. 1955. “The Crystal and Molecular Structure of Benzoic Acid.” Acta Crystallographica 8 (3): 157– 64. https://doi.org/10.1107/S0365110X55000601.

38. Tang, Weiwei, Mingtao Zhang, Huaping Mo, Junbo Gong, Jingkang Wang, and Tonglei Li. 2017. “Higher-Order Self-Assembly of Benzoic Acid in Solution.” Crystal Growth & Design 17 (10): 5049–53. https://doi.org/10.1021/acs.cgd.7b01078.

39. Wei, Bin, Jinglin Liu, Liangqi Ouyang, Chin Chen Kuo, and David C. Martin. 2015. “Significant Enhancement of PEDOT Thin Film Adhesion to Inorganic Solid Substrates with EDOT-Acid.” ACS Applied Materials and Interfaces 7 (28): 15388–15394. https://doi.org/10.1021/acsami.5b03350.

40. Wu, J., C. M. Shaw, and D. C. Martin. 2012. “2.19 - Electron Microscopy of Organic Materials: An Overview and Review of Recent Developments.” In Polymer Science: A Comprehensive Reference, edited by Krzysztof Matyjaszewski and Martin Möller, 509–25. Amsterdam: Elsevier. https://doi.org/10.1016/B978-0-444-53349-4.00041-8.

41. Zhu, Zheng-Tao, Jeffrey T. Mabeck, Changcheng Zhu, Nathaniel C. Cady, Carl A. Batt, and George G. Malliaras. 2004. “A Simple Poly(3,4-Ethylene Dioxythiophene)/Poly(Styrene Sulfonic Acid) Transistor for Glucose Sensing at Neutral PH.” Chemical Communications 0 (13): 1556–57. https://doi.org/10.1039/B403327M.

143

Chapter 6

CONCLUSIONS AND PROPOSED FUTURE STUDIES

6.1 Conclusions In chapter two we looked at a detailed overview of the literature with a focus on LPTEM for beam-sensitive soft materials (also including other material classes like metal-organic framework, nanowires and batteries). We described the considerations and challenges that are usually associated while performing liquid TEM experiments on beam-sensitive samples. This chapter summarized the recent studies in operando liquid-cell TEM thereby putting this thesis work into context. Next, we quantitatively observed the fundamental, unprecedented phase- transformations showing oligomeric clusters of PEDOT nucleating and growing in size and thickness (with a consistent nucleation density of ~ 0.3-0.4 N/µm2)before precipitating out and depositing onto the working electrode as a conjugated polymer film. We saw that the electrochemical polymerization is mainly driven by insolubility of the higher molecular weight oligomers. Further, we determined that the initially highly mobile, isotropic liquid monomer precursor solution goes through a viscoelastic oligomeric intermediate state to a final solid PEDOT polymer product. Correlating results from optical and electron microscopy we found that the liquid-like components would undergo coalescence, breakup or even complete dissolution while the solid-like components were stable and mostly increased in thicknesses. The mechanistic insights we obtained from this work have provided us with details about the electrochemical polymerization reaction would certainly help in understanding and

144 later optimizing performances of the devices that utilize electrodeposited PEDOT coatings. Further, in the subsequent chapter we imaged the electrochemical fabrication of nanofibrils PEDOT via operando liquid-cell transmission electron microscopy and saw that the growth occurs in three different regimes. We observed the nature of the morphological evolution of the nanofibrils and determined that during the initial stages of the process the central part of the fibrils grow. Not a lot of material got deposited into the interior parts of the fibrils after the initial deposition. Additionally, we estimated the growth velocities in different directions and approximated the number of EDOT molecules that might be present in a fibril of average length, width and thickness at each timestamp. For this calculation, we assumed that PEDOT packs itself into an orthorhombic unit cell with the unit cell dimensions as a = 1.4 nm, b = 0.68 nm, and c = 0.78 nm and that 4 molecules of EDOT occupy the unit cell. Finally, we saw that the nanofibrils grew mostly from the edges with the understructure established by the centre of the nanofibrils. We also estimated the growth velocities of the fibrils along their lengths, widths and thicknesses. In the final chapter, we estimated a crystal structure for EDOTacid, a hydrophilic, crystalline comonomer for PEDOT. In crystals of EDOTacid the monomers arranged into hydrogen-bonded dimers, with the two molecules in each dimer having opposite handedness. The dimers are tilted by ~60 degrees with respect to the c-axis. Additionally, we estimated the electron-beam stability to be ~ 9 e/A2 (~ 15 e/A2) which was consistent with its thermal stability of ~146-150 oC. These studies will serve as a primer to future studies discussed in the next section.

145

6.2 Proposed Future Studies

6.2.1 Investigating the early-stage nucleation and growth of copolymers of Poly (EDOT-co-EDOT-acid) In the previous chapter, we saw that the EDOTacid self-assembles into an orthorhombic unit-cell with its organization driven by the hydrogen-bonding of the carboxylic acid handles present on the functionalized thiophene monomer. We also saw that the electron-beam sensitivity values of these structures were ~ 9 e/A2. Considering Kumar and Adams relation, these values were in agreement with their thermal stability of ~146-150oC(Kumar and Adams 1990). After studying the physical properties of the monomer, especially the thermal properties and beam- sensitivity, as a natural progressive step, we have started to investigate the nucleation

EDOT-acid) using operando methods.

146

In chapter one we saw that homopolymers of PEDOT have a nucleation density of ~0.3 N/µm2 when electrochemically synthesized from water, with perchlorate counter-ion on a glassy carbon working electrode microfabricated by Protochips, Inc. However, optically we saw dramatic differences in the nucleation density (Figure 6.1) when EDOTacid was used as the comonomer. Particularly, in this case, the nucleation density was so high that it was almost impossible to distinguish individual nuclei on the glassy carbon working electrode. From Figure 6.1, we see that the electrodeposited copolymer film is uniform and dark, indicating that the oligomeric clusters formed initially were substantially smaller than the resolution of the optical microscope (~1 µm). This is in completely different with the relatively low nucleation density and much larger initial cluster sizes we observed for homopolymer PEDOT. These studies have thus allowed us to segue into our future studies using TEM to reveal details about the earliest stages of the process, and the similarities and differences from homopolymer PEDOT. Furthermore, the EDOT-acid rich copolymers are also known to have higher surface energies. Meaning, the oligomers would be much more soluble in the reaction than they were in the homopolymer case. Moreover, Bhagwat et al. observed that the copolymers of PEDOT and PEDOTacid have a “pleated” surface texture(Bhagwat, Kiick, and Martin 2014). We thus anticipate that TEM would allow us to learn more about the early-stages of the formation of these structures and understand the nucleation and growth of the copolymers in detail.

147

6.2.2 Comparing differences between the structure of semi-crystalline fibrillar poly (5,6-dimethoxyindole-2-carboxylic acid) (PDMICA) and fibrillar PEDOT with comparatively less order. As we have discussed in Chapter 4, PEDOT can organize itself into a fibrillar morphology in the presence of poly(acrylic acid) (PAA) as a co-dopant or a counterion. However, the structure of the nanofibrils of PEDOT have comparatively much less order than some of the highly crystalline or semi-crystalline polymers. Therefore, we have looked into the growth dynamics of a different conjugated polymer that is known to exhibit a fair amount of order in the solid state (PDMICA)(Povlich et al. 2010). Furthermore, since PEDOT:PAA as well as PDMICA have the ability to form nanofibers, we have made comparisons between them particularly by using operando LPTEM. The insights from these experiments in combination with corroborative data from GIWAXS, SEM and optical microscopy have given us an initial, well-informed guess about the differences in mechanisms of nanofibril formation during electrodeposion of PEDOT and PDMICA. Figure 6.2 shows the optical micrographs (top left – PEDOT:PAA; top right - PDMICA), SEM (middle left – PEDOT:PAA; middle right – PDMICA) and GIWAXS(bottom left PEDOT:PAA and bottom right – PDMICA). Optically, we saw that the PEDOT films are dark blue and uniform, whereas the PDMICA is more green in color. SEM showed us that there was a lot of variation in the widths of PEDOT nanofibrils (~100-500 nm). While, PDMICA fibers were consistently ~ 100 nm. Finally, and most importantly, the PEDOT films did not have as much order in their structure as the films of PDMICA which were relatively much more ordered as seen from GIWAXS. Next, we made comparisons between the early stages of growth between PEDOT and PDMICA nanofibers using operando LPTEM. As shown in the previous

148 chapters, we have quantified the intensity contrast in the bright field images to estimate their corresponding mass thicknesses.Figure 6.3 shows the results of the quantification and side-by-side comparison of the growth dynamics of PEDOT and PDMICA nanofibrils. The plot in Figure 6.3 of individual polymer nanofibrils as a function of electrodeposition time shows that the PEDOT-PAA nanofibrils were comparatively wider and vary in diameter more dramatically along their length. It is also worth noting that the PDMICA fibers grew at a much faster rate compared to the PEDOT nanofibrils. More interestingly, the PDMICA nanofibers are much more branched and dendritic as compared to the PEDOT fibers. Further, the PDMICA nanofibrils are smaller in diameter and grow out more uniformly thick at least in their nascent stages. Thus far, we have seen interesting differences between the morphology PEDOT and PDMICA nanofibrils using operando TEM studies. Going into the future, we plan to delve deeper into answering fundamental questions like what induces order in PDMICA filaments? Why do they grow faster? Why are they much more dendritic in nature and uniform in thickness? Further operando liquid-cell TEM studies are required to elucidate the mechanistic details in their polymerization process and answer these fundamental questions.

149

Figure 6.2: Top: Optical micrographs of nanofibrillar PEDOT-PAA (left) and PDMICA (right) electrochemically deposited onto interdigitated electrodes. Middle: SEM images of nanofibrillar PEDOT-PAA (left)and PDMICA (right). Bottom: GIWAXS patterns from PEDOT-PAA (left) and PDMICA (right). The PEDOT is relatively less ordered, with a d-spacing of ~ 1.3-1.4 nm. Whereas, PDMICA is semi-crystalline, with a strong intermolecular spacing near a d-spacing of ~ 1.4 nm.

150

151

6.2.3 Influence of nanoparticles on the nucleation and growth of PEDOT

(a)

(b)

Figure 6.4:(a) Schematic of the liquid-like and solid-like clusters interacting with the flowing metallic nanoparticles (b) Anisotropic gold nanorods (5-10 nm wide) flowing in liquid as observed by liquid-cell TEM.

In chapter 3, we saw that the electrochemical polymerization of PEDOT occurs via precipitation of the lower molecular-weight oligomeric clusters followed by subsequent solidification and deposition onto the working electrode. We expect

152 changes in the rheological properties as the deposited material as it transitions from the initial isotropic monomer solution, to liquid-like viscous viscoelastic oligomeric droplets and finally to the solid PEDOT polymer product. To that end, we think we can exploit the rheological differences by flowing metallic nanoparticles while the electrochemical polymerization occurs. Specifically, we plan to look for variations in the interactions when the metallic particulates encounter liquid-like oligomers and solid-like polymers(Figure 6.4). We anticipate that the interactions of particles with the low molecular weight oligomeric clusters or with the final solid polymer would not be as strong. Nevertheless, the viscoelastic intermediates states would have strong interactions with these nanoparticles. We are specifically interested filtering out components in these intermediate stages that are viscoelastic in nature. These interactions can well alter the nucleation and growth dynamics of the electrochemical deposition which we intend to investigate further. These studies will thereby help us understand the polymerization dynamics in much more detail and thus can also give us insights into the rheological properties of the oligomeric clusters.

153

REFERENCES

1. Bhagwat, Nandita, Kristi L. Kiick, and David C. Martin. 2014. “Electrochemical Deposition and Characterization of Carboxylic Acid Functionalized PEDOT Copolymers.” Journal of Materials Research 29 (23): 2835–44. https://doi.org/10.1557/jmr.2014.314.

2. Kumar, Satish, and W. Wade Adams. 1990. “Electron Beam Damage in High Temperature Polymers.” Polymer 31 (1): 15–19. https://doi.org/10.1016/0032-3861(90)90341-U.

3. Povlich, Laura K., Jason Le, Jinsang Kim, and David C. Martin. 2010. “Poly(5,6-Dimethoxyindole-2-Carboxylic Acid) (PDMICA): A Melanin- Like Polymer with Unique Electrochromic and Structural Properties.” Macromolecules 43 (8): 3770–74. https://doi.org/10.1021/ma9023558.

154

Appendix A

This work has been published in Crystal Growth and Design (DOI: 10.1021/acs.cgd.9b00613) . Reprinted with permission from MORPHOLOGY, MOLECULAR ORIENTATION AND SOLID-STATE CHARACTERIZATION OF 2,3-DIHYDROTHIENO[3,4-B] [1,4]DIOXINE-2-CARBOXYLIC ACID (EDOTACID). Copyright 2019 American Chemical Society

ADDITIONAL FIGURES

1. Chemicalize was used for solubility calculations of EDOTacid, https://chemicalize.com/, developed by ChemAxon.

Figure A1: Solubility calculations of EDOTacid as a function of pH

155

2. Infrared experiments were conducted using a Thermo Nicolet

670 Nexus FT-IR spectrometer

EDOTacid EDOT 1.2

1.0 0.8 0.6 0.4

Absorption (Arbitrary units) 0.2

0.0

0 1000 2000 3000 4000 Wavenumber (cm-1)

Figure A2: Fourier Transform Infrared (FTIR) Spectra of EDOT and EDOTacid for comparison

3. Thermo gravimetric analysis was performed using TA Discovery TGA

Figure A3: Thermogravimetric analysis (TGA) of EDOT acid showing a degradation onset of 146.7 oC

156

4. Atomic Force Micrograph – InfraRed (AFM-IR) Spectra from the crystals obtained by sublimation using Bruker Nano-IR2

Figure A4: AFM-IR spectra of a single of crystal of EDOTacid crystallized by sublimation of the powder

5. Low-dose high resolution image of EDOTacid film on the Talos F200C

(a) (b)

Figure A5:(a) Low-dose high resolution electron micrograph of thin film of EDOT acid deposited on a carbon coated copper grid (b) FFT of (a) showing evidence for (200) spacings

157

Appendix B

RIGHTS AND PERMISSIONS

158

159

160

161

162

163

164

165

166