Liquid-Phase Exfoliated Graphene for Supercapacitors · Liquid-Phase Exfoliated Graphene for Supercapacitors Der Naturwissenschaftlichen Fakult¨at der Friedrich ... graphene through

Liquid-Phase ExfoliatedGraphene for Supercapacitors

Der Naturwissenschaftlichen Fakultat der Friedrich-AlexanderUniversitat Erlangen-Nurnberg

zur Erlangung des Doktorgrades Dr. rer. nat.

vorgelegt von

Alexandra Schneideraus Erlangen

Als Dissertation genehmigt durch die Naturwissenschaftliche Fakultat derFriedrich-Alexander Universitat Erlangen-Nurnberg.

Tag der mundlichen Prufung: 21.02.2017

Vorsitzender des Promotionsorgans: Prof. Dr. Georg Kreimer

Gutachter:1. Prof. Dr. Dirk M. Guldi2. Prof. Dr. Andreas Hirsch

Die vorliegende Arbeit wurde in der Zeit von 01.12.2011 bis 31.03.2015 in denLaboratorien der Eckart GmbH, im Department of Chemistry im Arbeitskreis vonProf. Dr. Richard B. Kaner der University of California, Los Angeles und imDepartment Chemie und Pharmazie am Lehrstuhl fur Physikalische Chemie derFriedrich-Alexander-Universitat Erlangen-Nurnberg unter der Leitung von Prof.Dr. Dirk M. Guldi angefertigt. Die Ergebnisse, Meinungen und Schlüsse dieserDissertation sind nicht notwendigerweise die der Eckart GmbH.

Ich versichere hiermit, dass ich die vorliegende Arbeit selbststandig und nur mit Hilfeder angegebenen Hilfsmittel angefertigt habe.

Alexandra Schneider

”I am not young enough to know every-thing.”

Oscar Wilde

Abstract

Recently, supercapacitors have attracted great attention due to their high charge/dischargerate, long cycle life, outstanding power density, and no short circuit concerns. However,their low energy density keeps them from widespread application. This limitation can beovercome by employing new advanced electrode materials, such as graphene, which offershigh conductivity, high surface area, and a theoretical capacitance of about 550 F g−1.Nevertheless, the technical and industrial exploitation of these exceptional propertiesin supercapacitor devices strongly depends on the accessibility of high-quality few-layergraphene through processable, low cost, and scalable approaches.

This thesis surveys new advances in the liquid-phase exfoliation of graphite towards highlystable few-layer graphene dispersions, which can be easily processed into supercapacitorelectrodes. In particular, different stabilization approaches are explored, namely noncovalentfunctionalization in aqueous media, pretreatment of graphite by means of small intercalantmolecules, solvent-based graphene exfoliation, and cosolvency.

In the noncovalent functionalization approach, a comparative study on the liquid-phaseexfoliation of graphite by means of different amphiphilic aromatic stabilizers, such aspyrene, naphthalene, and perylene derivates, is presented. These stabilizers allow for adirect exfoliation and dispersion of graphene in aqueous medium due to π-π interactionsbetween their aromatic core and graphene. The efficiency of the exfoliation and dispersionof graphene is found to depend on the structure of the stabilizer, which includes themolecular weight, the nature of the hydrophilic part, and the binding motif. Importantly,high yields of single-layer graphene of up to 21% are accomplished with the use of anaphthalene sulfonic acid formaldehyde condensate, which is a common industrial additiveused for the formulation of pigments, paper, and dyes.

In an attempt to further increase the dispersion and stabilization efficiency of surfactants,a dispersion approach is applied, which involves the use of a small intercalant additionallyto the dispersant. In particular, a pronounced effect is observed upon the pretreatment ofgraphite with tetraalkylammonium salts, which impact the dispersibility and exfoliationof graphene by permeating into the graphite edges. This strategy enables single-layergraphene of up to 25%. Moreover, the stability of the resulting dispersions is remark-able. This is affected by the deprotonation of edge functionalities in alkaline solutions oftetraalkylammonium hydroxide, which provides additional electrostatic stabilization of thegraphene layers in solution.

Turning to solvent-based liquid-phase exfoliation, organic solvents and aqueous-organic-cosolvent mixtures are tested as exfoliation agents. While organic solvents enable surfactant-free dispersion and exfoliation of few- to multilayer graphene, cosolvent mixtures are foundto be unsuitable for the production of stable graphene dispersions. However, combiningboth approaches, that is, the pretreatment with intercalants and the cosolvency approach,graphite is successfully exfoliated in an ethanol/water mixture with a surface tension closeto 40 mN m−1 by employing tetraethylammonium hydroxide as an intercalant yieldingsingle-layer graphene of up to 15%.

Notably, all produced graphene samples show undisturbed sp2 carbon networks resultingfrom our non-oxidative production routes, which renders them high-quality few-layergraphene. Throughout the study, the stabilization approaches and exfoliation agents aretested with a constant and reproducible ultrasonication dispersion process. This allowsa quantitative comparison in terms of graphene concentration, exfoliation degree, andgraphene flake size. The dispersion efficiency of the stabilization methodologies and thequality of graphene in the form of single-, bi-, and few-layer graphene is comprehensivelycharacterized by absorption spectroscopy, statistical Raman analysis, transmission electronmicroscopy, and atomic force microscopy.

Finally, the prepared liquid-phase exfoliated graphene (LPEG) is employed as an electrodematerial in supercapacitor devices. The electrochemical characteristics and performanceof the supercapacitors are studied in depth by galvanostatic charging-discharging mea-surements, cyclic voltammetry, and electrochemical impedance spectroscopy. Featuringlarge ion-accessible surface area, high packing density, as well as efficient electron and iontransport pathways, the LPEG electrodes deliver a high gravimetric capacitance of up to202 F g−1 and a volumetric capacitance of 9 F cm−3 in aqueous electrolyte. Furthermore,it is shown that the devices are able to provide gravimetric and volumetric energy densitiesof up to 121 Wh kg−1 and up to 4 mWh cm−3 in ionic liquid electrolyte. Moreover, lowequivalent series resistances translate into high gravimetric and volumetric power densitiesof up to 48×105 W kg−1 and 346 W cm−3 in an ionic liquid electrolyte. Thus, these novelLPEG electrode materials enable to bridge the gap between batteries and conventionalsupercapacitors and, therefore, may pave the way for translating graphene technology froma scientific breakthrough to real-world applications.

Kurzzusammenfassung

Superkondensatoren haben aufgrund ihrer hohen Ladungs-/Entladungsrate, langen Lebens-dauer, hervorragenden Leistungsdichte und nicht vorhandenen Kurzschlussbedenken un-langst große Aufmerksamkeit erlangt. Allerdings halt ihre geringe Energiedichte sie vonder weitflachigen Anwendung ab. Diese Einschrankung kann durch fortgeschrittene Elek-trodenmaterialien uberwunden werden. Ein Beispiel hierfur ist Graphen, das eine hoheLeitfahigkeit, hohe spezifische Oberflache und eine theoretische Kapazitat von ungefahr550 F g−1 bietet. Jedoch hangt die technische und industrielle Nutzung dieser außergewohn-lichen Eigenschaften in Superkondensatoren stark von der Moglichkeit ab, qualitativ hoch-wertiges geringlagiges Graphen uber prozessierbare, preiswerte und skalierbare Methodenherzustellen.

Diese Doktorarbeit befasst sich mit der Flussigphasen-Exfolierung von Graphit hin zuhochstabilen geringlagigen Graphen-Dispersionen, welche leicht in Elektroden von Super-kondensatoren verarbeitet werden konnen. Insbesondere werden verschiedene Stabilisie-rungsansatze untersucht, und zwar die nichtkovalente Funktionalisierung in wassrigenMedien, die Vorbehandlung von Graphit mittels kleiner interkalierenden Molekulen, dieGraphen-Exfolierung auf Losungsmittelbasis und Kosolvenz.

Im nichtkovalenten Funktionalisierungsansatz wird eine vergleichende Studie uber dieFlussigphasen-Exfolierung von Graphit mittels verschiedener amphiphiler aromatischenStabilisatoren, wie Pyren-, Naphthalen- und Perylen-Derivaten, dargestellt. Diese Stabili-satoren ermoglichen aufgrund von π-π Wechselwirkungen zwischen ihrem aromatischenKern und Graphen eine direkte Exfolierung und Dispergierung von Graphen in wassrigenMedien. Die Effizienz der Exfolierung und Dispergierung von Graphen hat sich von derStruktur des Stabilisators abhangig erwiesen. Im Detail sind das Molekulargewicht, die Artdes hydrophilen Teils und das Bindungsmotiv von besonderer Bedeutung. Im Wesentlichenwird eine hohe Ausbeute an einlagigem Graphen von bis zu 21% durch die Verwendungeines Naphthalensulfonsaure-Formaldehyd-Kondensats, ein gebrauchliches Industrieadditiv,das zur Herstellung von Pigmenten, Papier und Farbstoffen verwendet wird, erreicht.

In einem Versuch zur weiteren Steigerung der Dispergierungs- und Stabilisierungseffizienzvon Tensiden, wird ein Dispergierungssansatz angewendet, der die Verwendung eines kleinenInterkalanten zusatzlich zu dem Dispergiermittel beinhaltet. Eine ausgepragte Wirkungwird insbesondere bei der Vorbehandlung von Graphit mittels Tetraalkylammonium-salzen beobachtet, die die Dispergierung und Exfolierung von Graphen beeinflusst, in-dem die Molekule in die Graphitkanten eindringen und dadurch einlagiges Graphen vonbis zu 25% ermoglicht. Daruber hinaus ist die Stabilitat der erhaltenen Dispersionen

beachtlich. Dies wird durch die Deprotonierung der Kantenfunktionalitaten in alkali-schen Tetraalkylammoniumhydroxid-Losungen erreicht, die eine weitere elektrostatischeStabilisierung ermoglichen.

In Bezug auf die losungsmittelbasierte Flussigphasen-Exfolierung werden organische Losungs-mittel und wassrig-organische Kosolvenz-Mischungen als Exfolierungsmittel getestet. Wah-rend organische Losungsmittel die tensidfreie Dispergierung und Exfolierung von gering-und mehrlagigem Graphen ermoglichen, haben sich die Kosolvenz-Mischungen zur Her-stellung von stabilen Graphendispersionen als nicht geeignet erwiesen. Kombiniert manjedoch beide Ansatze, die Vorbehandlung mit Interkalanten und den Kosolvenz-Ansatz,dann wird Graphit erfolgreich in einer Ethanol/Wasser Kosolvenz-Mischung mit einerOberflachenspannung von nahezu 40 mN m−1 exfoliert, indem Tetraethylammoniumhy-droxid als Interkalant eingesetzt wird. Dies fuhrt zu einlagigem Graphen von bis zu15%.

Bemerkenswerterweise zeigen alle erzeugten Graphenproben unbeschadigte sp2 Kohlenstoff-Netzwerke, was durch unsere nichtoxidativen Produktionswege ermoglicht wird und dieProben als qualitativ hochwertige geringlagige Graphene auszeichnen. Im Laufe derStudie werden die Stabilisierungsansatze und Exfolierungsmittel mit einem konstanten undreproduzierbaren Ultraschall-Dispergierungsverfahren getestet, das einen quantitativenVergleich in Bezug auf Graphen-Konzentration, Exfolierungsgrad und die Flockengroßedes Graphens ermoglicht. Die Dispergierungseffizienz der Stabilisierungsmethoden unddie Qualitat von Graphen in Form von ein-, zwei-, und geringlagigem Graphen wirddurch Absorptionsspektroskopie, statistischer Raman-Analyse, Transmissionselektronen-mikroskopie und Rasterkraftmikroskopie umfassend charakterisiert.

Abschließend wird das hergestellte flussigphasen-exfolierte Graphen (LPEG) als Elektro-denmaterial im Superkondensator eingesetzt. Die elektrochemischen Eigenschaften unddie Leistung der Superkondensatoren werden im Detail mittels galvanostatischer Ladungs-Entladungs-Messungen, Cyclovoltammetrie und elektrochemischer Impedanzspektroskopieuntersucht. Aufgrund großer ionen-zuganglicher Oberflache, hoher Packungsdichte, sowieeffizienter Elektronen- und Ionentransportwege, liefern die LPEG Elektroden eine hohegravimetrische Kapazitat von bis zu 202 F g−1 und eine volumetrische Kapazitat von9 F cm−3 im wassrigen Elektrolyt. Daruber hinaus wird gezeigt, dass die Bauelementegravimetrische und volumetrische Energiedichten von bis zu 121 Wh kg−1 und bis zu4 Wh cm−3 in ionischen Flussigkeiten liefern konnen. Ferner fuhren niedrige aquivalenteSerienwiderstande zu hohen gravimetrischen und volumetrischen Leistungsdichten von biszu 48×105 W kg−1 und 346 W cm−3 in einer ionischen Flussigkeit. Somit ermoglichen diese

neuen LPEG Elektrodenmaterialien die Brucke zwischen Batterien und konventionellenSuperkondensatoren zu schlagen. Sie ebnen daher den Weg fur Graphen von einem wis-senschaftlichen Durchbruch zu einem breiten Einsatz in unterschiedlichen Anwendungen.

Acknowledgment

First and foremost I want to express my sincerest thanks and gratitude to Prof. Dr.Dirk M. Guldi as my research advisory. I would like to thank him for giving me theopportunity to explore this significant topic during my Ph.D. studies. I am deeply thankfulfor his advisement, encouragement, and guidance during the course of this research. Inparticular, I greatly acknowledge that he made it possible to accomplish this as an industrycollaboration, which gave me the chance to work on the things I am most passionate about- exploring today’s scientific questions and developing tomorrow’s product innovations.I am also very grateful for his encouragement and help to further develop my scientificabilities during my time at the UCLA.

Moreover, I am particularly grateful for the opportunity to work on this interesting andchallenging research topic within the Eckart GmbH, and the financial support, given byDr. Ulrich-Andreas (Uli) Hirth, Christian Wolfrum, Dr. Mark Stoll, and Dr.Wolfgang Schutt.

Especially, I would like to offer my special thanks to Dr. Uli Hirth as my supportivethesis mentor at Eckart. I am thankful for sharing his vast experience in chemistry, industry,and professional life with me. Thank you for encouraging me in bad and good times andfor allowing me to grow as a research scientist and young professional!

I would like to express my great appreciation to Prof. Richard Dr. B. Kaner, whoinvited me to spend a research stay in his group at the University of California, Los Angeles,and gave me the opportunity to gain deeper insights into the research topic of energystorage systems, especially supercapacitors. In this context, I am very thankful to Dr.Maher El-Kady, Dr. Lisa Wang, Jee-Young Hwang, and all other colleagues fortheir amazing hospitality, all the fruitful scientific discussions, friendly supervision, and formaking my stay at UCLA very enjoyable. For sure, I will never forget about the amazingtime I had in L.A.!

Moreover, my special thanks are extended to all collaborators within the Altana AG.I thank Dr. Bernd Gobelt and his team for additive synthesis and fruitful discussions onstabilizers and in situ polymerization techniques. I greatly thank Dr. Michael Berkei,Ninja Hanitzsch, and Tobias Tintenhoff from the nanotechnology platform for helpwith graphene coatings, close collaboration, and beneficial discussions concerning graphene,CNTs, and carbon allotropes. Dr. Robin von Hagen from the battery platform providedme with valuable advice with electrochemical measurements and with sharing his vastexperience on energy storage devices. Support given by Dr. Tina Radespiel from the

vi

biotechnology platform with freeze drying was greatly appreciated. Further, I acknowledgethe whole analytic team of Eckart. In particular, Dominik Doleschal for TGA,surface tension, and BET measurements, Jamin Bleisteiner for SEM imaging, BiancaBauer for zeta potential and DLS measurements, and Herbert Dummler for XRDmeasurements. Moreover, I thank Olga Isakin for AFM measurements of the referencesystem presented in Chapter 4.1.2. I wish to acknowledge the support provided by thewhole “Noventus” team during the last years. In particular, I want to thank AngelaHullin for the initial guidance and continuous collaboration during my doctorate.

Additionally, I would like to express my very great appreciation to my student traineeSebastian Hildebrand for his steady support with the liquid-phase exfoliation of graphite,specific surface area measurements by methylene blue absorption methods, and for the greatefforts he made within his practical semester study and throughout his student traineeship.Thank you for your incredible dedication, patience, all the beneficial discussions, andunexpected insights into mathematical philosophy.

The whole Research & Development team at Eckart deserves a huge THANK YOUfor their steady support, scientific advice, intense scientific discussion, and for the greattimes I spent with them at H307. Especially, I am grateful to my office colleagues SandraHubner, Tanja Preininger, Patrick Melerowicz, and later Dr. Christine Schillingfor the wonderful working atmosphere they created and for being always so patient andunderstanding. I am particularly thankful for the support given by Dr. Oliver Struckduring my studies, both by offering his laboratory and his scientific and personal advice.

Within the FAU, I want to thank Prof. Dr. Jana Zaumseil and her group for the supportwith Raman spectroscopy. In this regards, I want to acknowledge Manuel Schweiger,and, especially, Dr. Julia Schornbaum. Moreover, I would like to offer my specialthanks to Prof. Dr. Wolfgang Peukert and his group, particularly, Jonas Paul forsupport in all AFM-related aspects.

Exceptionally, I want to deeply thank all the group members of the Guldi group forall the helpful discussions, friendly assistance, and constant support with analytics. Inparticular, I am grateful to Georgios Katsoukis for sharing his extensive experience andknowledge about crucial synthetic and analytical procedures, for valuable support withdata analysis, and for the nice time in the lab (and office). Moreover, I wish to acknowledgeAlexandra Roth and Leonie Wibmer for productive discussions and support withTEM and AFM imaging.

Huge THANKS goes to my friends and, especially, Pan Katsukis. Thank you forsupporting me during the last years and reminding me, whenever it was necessary, that,

vii

indeed, there is a world outside of graphene. Moreover, I want to express my appreciation,gratitude, and thanks to my parents, and my whole family, for their steady support inall circumstances of life.

It has been extraordinary three and a half years for me, both academically and personally.It was not an easy ride, but I have learned priceless and countless lessons, not only limitedto a subject of science but which makes me who I am today and in the future. I thankeveryone who has joined this journey.

viii

Contents

List of Abbreviations xii

1 Introduction and Motivation 1

I Theoretical Background 4

2 From Graphite to Graphene 52.1 Introduction to Graphene . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1.1 The Atomic Structure and Electronic Properties of Graphene . 52.1.2 Properties and Applications of Graphene-Based Materials . . . 8

2.2 Production of Graphene-Based Materials . . . . . . . . . . . . . . . . 122.2.1 Graphene Synthesis by Oxidation of Graphite . . . . . . . . . 132.2.2 Liquid-Phase Exfoliation of Graphite . . . . . . . . . . . . . . 15

Ultrasonication-Assisted Exfoliation . . . . . . . . . . . . . . . 15Graphene Dispersions in Organic Solvents . . . . . . . . . . . 16Graphene Dispersions in Surfactant Solutions . . . . . . . . . 16

2.3 Evaluation of Graphene Dispersion . . . . . . . . . . . . . . . . . . . 202.3.1 Determination of the Graphene Concentration . . . . . . . . . 212.3.2 Microscopic Techniques . . . . . . . . . . . . . . . . . . . . . . 21

Transmission Electron Microscopy . . . . . . . . . . . . . . . . 21Atomic Force Microscopy . . . . . . . . . . . . . . . . . . . . . 22

2.3.3 Raman Spectroscopy . . . . . . . . . . . . . . . . . . . . . . . 23

3 Supercapacitors 283.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283.2 The Electrical Double-Layer . . . . . . . . . . . . . . . . . . . . . . . 293.3 Operating Principles of Supercapacitors . . . . . . . . . . . . . . . . . 313.4 The Performance of Supercapacitors . . . . . . . . . . . . . . . . . . . 343.5 Evaluation of Supercapacitors by Electrochemical Measurements . . . 36

3.5.1 Cyclic Voltammetry . . . . . . . . . . . . . . . . . . . . . . . 373.5.2 Galvanostatic Cycling . . . . . . . . . . . . . . . . . . . . . . 393.5.3 Electrochemical Impedance Spectroscopy . . . . . . . . . . . . 42

ix

3.6 Carbon-based Electrodes for Supercapacitors . . . . . . . . . . . . . . 443.6.1 Activated Carbon . . . . . . . . . . . . . . . . . . . . . . . . . 453.6.2 Graphene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

II Objective 51

III Results and Discussion 55

4 Liquid-Phase Exfoliation of Graphite in Different Media 564.1 Establishment of a Dispersion Procedure . . . . . . . . . . . . . . . . 56

4.1.1 Optimization of the Dispersion Conditions . . . . . . . . . . . 574.1.2 Characterization of Graphene Dispersions in Hexadecyltri-

methylammonium Bromide as an Internal Reference . . . . . . 634.2 Graphene Dispersion by Aromatic Amphiphiles . . . . . . . . . . . . 68

4.2.1 Overview of Graphene Dispersability . . . . . . . . . . . . . . 714.2.2 Exfoliation Ability of Aromatic Amphiphiles . . . . . . . . . . 78

4.3 Pretreatment of Graphite as a Dispersion Enhancer . . . . . . . . . . 894.3.1 Influence of Tetraalkylammonium Salts on Graphene Dis-

persibility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 904.3.2 Influence of an Auxiliary Pretreatment Step on the Exfoliation

Degree . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 984.3.3 Dispersion of Graphene by in situ Polymerization of Vinyl-

benzyltrimethylammonium Chloride . . . . . . . . . . . . . . . 1084.4 Solvent-Based Liquid-Phase Exfoliation . . . . . . . . . . . . . . . . . 115

4.4.1 Dispersion of Graphite in Different Organic Solvents . . . . . 1164.4.2 Exfoliation Ability of Different Organic Solvents . . . . . . . . 1194.4.3 Graphene Dispersion by a Cosolvent Approach . . . . . . . . . 1254.4.4 Exfoliation of Pretreated Graphite in a Cosolvent Mixture . . 128

5 Application of Liquid-Phase Exfoliated Graphene in Supercapacitors 1335.1 Liquid-Phase Exfoliated Graphene Electrodes - Material Overview . . 134

5.1.1 Evaluation of Flexible Graphene Films . . . . . . . . . . . . . 1345.1.2 Evaluation of the Electrochemical Performance of LPEG-Based

Supercapacitors . . . . . . . . . . . . . . . . . . . . . . . . . . 1375.1.3 Comparison with Commercial Graphene Samples . . . . . . . 143

x

5.1.4 Cycling Stability of LPEG Supercapacitors . . . . . . . . . . . 1455.2 Flexible All-Solid-State Supercapacitors . . . . . . . . . . . . . . . . . 1465.3 Performance in Organic Electrolyte . . . . . . . . . . . . . . . . . . . 1515.4 Performance in Ionic Liquid Electrolyte . . . . . . . . . . . . . . . . . 1545.5 Ragone Plot - Comparison with State-of-the-Art Devices . . . . . . . 158

6 Conclusion and Outlook 161

IV Experimental Details 170

7 Materials and Methods 171

8 Preparative Equipment 177

9 Preparation of Graphene Dispersions 179

10 Preparation of Supercapacitors 183

List of Figures xv

List of Tables xxviii

References xxix

A Appendix lviii

xi

List of Abbreviations

ε . . . . . . . . . . . . . . . Extinction coefficient

γ . . . . . . . . . . . . . . . Surface tension

σ . . . . . . . . . . . . . . . Electrical conductivity

2D . . . . . . . . . . . . . Two-dimensional

3D . . . . . . . . . . . . . Three-dimensional

[BMIM][NTf2] . . 1-Butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide

AC . . . . . . . . . . . . . Activated carbon

ACN . . . . . . . . . . . Acetonitrile

AFM . . . . . . . . . . . Atomic force microscopy

Au . . . . . . . . . . . . . Gold

BA . . . . . . . . . . . . . Butyl acetate

BLG . . . . . . . . . . . Bilayer graphene

BZ . . . . . . . . . . . . . Brillouin zone

c . . . . . . . . . . . . . . . Concentration

CDC . . . . . . . . . . . Charging-discharging curve

CMG . . . . . . . . . . . Chemically modified graphene

CNT . . . . . . . . . . . Carbon nanotube

CV . . . . . . . . . . . . . Cyclic voltammetry

xii

CVD . . . . . . . . . . . Chemical vapor deposition

D . . . . . . . . . . . . . . Raman D peak

DBE . . . . . . . . . . . Dibenzyl ether

DMA . . . . . . . . . . . N,N-Dimethylacetamide

DMEU . . . . . . . . . 1,3-Dimethyl-2-imidazolidione

DMF . . . . . . . . . . . Dimethlyformamide

DMP . . . . . . . . . . . Dimethyl phthalate

DOL . . . . . . . . . . . 1,3-Dioxolane

E . . . . . . . . . . . . . . . Energy

EAC . . . . . . . . . . . Ethyl acetate

EDL . . . . . . . . . . . Electric double-layer

EDLC . . . . . . . . . . Electrochemical double-layer capacitor

EG . . . . . . . . . . . . . Ethylene glycol

ESR . . . . . . . . . . . . Equivalent series resistance

EtOH . . . . . . . . . . Ethanol

EtOH/W . . . . . . . Ethanol/water

FLG . . . . . . . . . . . Few-layer graphene

FWHM . . . . . . . . Full width at half maximum

GLB . . . . . . . . . . . γ-Butyrolactone

GO . . . . . . . . . . . . . Graphene oxide

xiii

I . . . . . . . . . . . . . . . Intensity

I2D . . . . . . . . . . . . . Intensity of the 2D Raman peak

iapp . . . . . . . . . . . . . Charging or discharging current of a supercapacitor

IL . . . . . . . . . . . . . . Ionic liquid

ITO . . . . . . . . . . . . Indium tin oxide

LIB . . . . . . . . . . . . Li-ion battery

LPE . . . . . . . . . . . . Liquid-phase exfoliation

LPEG . . . . . . . . . . Liquid-phase exfoliated graphene

LSG . . . . . . . . . . . . Laser-scribed graphene

MLG . . . . . . . . . . . Multilayer graphene

N . . . . . . . . . . . . . . Number of layers

NaOH . . . . . . . . . . Sodium hydroxide

NMP . . . . . . . . . . . N-methyl-2-pyrrolidone

o-DCB . . . . . . . . . 1,2-Dichorobenzene

OLED . . . . . . . . . . Organic light-emitting diode

P . . . . . . . . . . . . . . . Power density

PET . . . . . . . . . . . Polyethylene terephthalate

PVA . . . . . . . . . . . Polyvinyl alcohol

PVBTA . . . . . . . . Poly(vinylbenzyl)ammonium chloride

R . . . . . . . . . . . . . . Resistance

xiv

RESR . . . . . . . . . . . Equivalent series resistance

rGO . . . . . . . . . . . . Reduced graphene oxide

rpm . . . . . . . . . . . . Revolutions per minute

SDBS . . . . . . . . . . Sodium dodecylbenzenesulfonate

SDS . . . . . . . . . . . . Sodium dodecylsulfate

SLG . . . . . . . . . . . . Single-layer graphene

SSA . . . . . . . . . . . . Specific surface area

SWCNT . . . . . . . . Single-walled carbon nanotube

T . . . . . . . . . . . . . . Temperature

t . . . . . . . . . . . . . . . Time

TEABF4 . . . . . . . Tetraethylammonium tetrafluoroborate

TEM . . . . . . . . . . . Transmission electron microscopy

TP . . . . . . . . . . . . . Terpineol

UV . . . . . . . . . . . . . Ultraviolet

V . . . . . . . . . . . . . . Voltage

Vcell . . . . . . . . . . . . Maximum voltage of the supercapacitor at full charge

VBTA . . . . . . . . . . Vinylbenzyltrimethylammonium chloride

vis . . . . . . . . . . . . . Visible

xv

1 Introduction and Motivation

The world is facing a severe energy challenge driven by an exponential populationgrowth, increasing economic development, and limited availability of fossil fuels. In2014, about 13 trillion tons of oil equivalent were consumed (Figure 1.1 a)) and itis estimated that the world demand for energy will double by 2050.[1,2] These days,the majority (about 86%) of energy consumed is still produced by fossil fuels, suchas oil, gas, and coal, although fossil fuel resources are progressively depleted - seeFigure 1.1 b).[3] While uncertainties in fossil fuel supply often directly or indirectlytrigger regional and global conflicts, the production, transmission, and consumptionof fossil fuels also significantly harms our environment. In particular, the combustionof fossil fuels does not only lead to the release of air pollutants, such as sulfur oxideand heavy metals, but also generates large amounts of anthropogenic CO2, which isbelieved to be the main cause of global warming.[4,5]

a) b)

Figure 1.1: a) World energy consumption growth by fuel type from 1965 to 2014. b)The share of different energy sources for global energy consumption in2014. Data is taken from Ref. [3].

In light of the depletion of oil and gas reserves and the critical rise in CO2 emissions,the development of efficient renewable energy generation and storage technologiesbecomes one of the major challenges of our time. However, due to their intermittentcharacter, the development of a high-performance energy storage and conversionsystem is of paramount importance. Moreover, there is a rising demand for portableenergy storage technologies, not only for portable devices, such as laptops, mobilephones, and tablets but also for next-generation transportation. In fact, suchtechnologies would enable the widespread use of electric vehicles, which would further

1

1 INTRODUCTION AND MOTIVATION

reduce the dependence on fossil fuels.[6] To this end, highly efficient electrochemicalenergy storage devices, such as rechargeable batteries and supercapacitors, have tobe developed. However, current energy storage devices are facing a trade-off betweenenergy density and power density. Supercapacitors, on one hand, feature high powerdensity, but are limited in energy density. Batteries, on the other hand, are able tostore higher amounts of energy, but exhibit lower power densities. In order to bridgethe gap between high power and high energy density, novel electrode materials withexceptional electrochemical properties are strongly needed.[7]

Carbon plays a major role in this research area, because of its low cost, highelectrical conductivity, and high versatility of texture and surface functionality, andis currently used as the state-of-the-art electrode material in batteries, electrolyzers,supercapacitors, and fuel cells.[8] Benefiting from the rapid growth in nanotechnology,tremendous research efforts have been dedicated recently to develop advanced carbonnanomaterials, such as templated carbons, carbon nanotubes, and graphene, whichhave been demonstrated to be a key enabling technology for creating next-generationenergy storage devices.[9,10]

Since its discovery in 2004 by Geim and Novoselov, graphene has emerged as themost promising carbon allotrope, being ideally suited for the implementation inelectrochemical devices.[11,12] An essential criterion for electrode materials, especiallyimportant in energy storage, is a high accessible surface area. Being a true 2Dmaterial, graphene features an exceptionally large surface area of 2630 m2 g−1, sur-passing graphite and even single-walled carbon nanotubes (SWCNT).[13] Additionally,graphene exhibits a theoretical capacitance of 550 F g−1, which is the upper limitfor all carbon-based electrode materials.[14,15] The combination of these strikingfeatures renders it an auspicious candidate for high energy density electrode materials,especially for the application in supercapacitors. Another important factor for energyrelated applications is the electrical conductivity of the electrode material, whichultimately determines the power density achievable of the devices. Resulting fromits extensive conjugated sp2 carbon network, graphene exhibits an extraordinaryin-plane electrical conductivity of 20,000 S cm−1, which is about 60 times higher thanthat of SWCNTs.[16,17] Moreover, the electrical conductivity is stable over a widerange of temperatures, which is an important prerequisite for reliable energy storagesystems.[11,12] Furthermore, the one atom thick graphene sheets offer a unique 2Dplanar geometry, which further facilitates ion and electron transport, especially in

2

1 INTRODUCTION AND MOTIVATION

comparison to SWCNTs, and, hence, are more effective electrode materials.[9]

Considering these one of a kind properties, graphene is one of the few materials thatcan bridge the gap between batteries and supercapacitors, enabling next-generationenergy storage devices with high energy and high power density. However, torealize its true potential for real life applications, graphene has to be producedin large quantities, high quality, and at an affordable cost. Several methods forthe production of graphene have been reported recently, inducing micromechanicalcleavage,[18] chemical vapor deposition,[19] reduction of graphene oxide,[20] and liquid-phase exfoliation of graphite.[21] However, each of the above methods faces its ownlimitations, for example low yields, high process complexity, low material quality,high cost, need for hazardous chemicals, or non-scalability.[22] In order to surmountthese barriers, concerted efforts have to be made to explore, develop, and implementa strategy for the scalable fabrication of high-quality graphene, which is effective,reliable, low in cost, and environmentally friendly. Indeed, this is crucial to empowerthe properties of graphene to create powerful applications in the field of clean andsecure energy generation and storage.

Therefore, this thesis focuses on the development of new production routes towardshigh-quality graphene-based materials for the application in energy storage devices,such as supercapacitors.

3

I

Theoretical Background

4

2 From Graphite to Graphene

2.1 Introduction to Graphene

2.1.1 The Atomic Structure and Electronic Properties ofGraphene

Carbon plays a significant role in today’s science and technology due to its variety ofstructural forms and high availability. In fact, its versatility and usefulness arise fromthe ability to form different types of valence bonds through atomic hybridization.[23]

Carbon provides four valence electrons and has a ground state electron configurationof 1s2 2s2 2p1

x 2p1y. In graphite, the 2s, 2px, and 2py orbitals form three sp2-orbitals.

The sp2-orbitals are oriented perpendicularly to the remaining 2pz-orbital. Thus,every carbon atom forms three covalent in-plane σ bonds with its nearest neighborleading to the planar hexagonal “honeycomb” structure of graphene, which is thefundamental building block of graphite - see Figure 2.1 a). These σ bonds are amongthe strongest bonds (615 kJ mol−1) in nature and give graphene its extraordinarymechanical strength.[24] In addition, the C-C bonding is enhanced by the fourth bond,which is associated with the overlap of the remaining pz-orbital. The π-electrons aredelocalized over graphene and are responsible for its extraordinary electronic andthermal properties.[25]

In bulk graphite, these graphene layers are stacked with a constant interlayer spacingof 0.335 nm in a way that one half of the atoms of each graphene layer lie directlyover the center of a hexagon in the adjacent graphene layer, and the other half of theatoms lie directly over an atom in the lower sheet - see Figure 2.1 b). The resultingAB stacking of the graphene layers is called Bernal stacking.

Nowadays, the term “graphene” is widely used for ultra-thin graphite layers ornanographites, a new class of materials, which recently attracted enormous interestamong both academic and industrial research.[26–28] However, it strictly refers onlyto a quasi-two-dimensional isolated monolayer of carbon atoms, as the electronicproperties strongly alter with the number of layers.[29]

5

2.1 INTRODUCTION TO GRAPHENE

a) b)

Figure 2.1: a) Schematic representation of the bonding structure in graphene featuringthe delocalized π-electron cloud in the out of plane z-axis. b) Schematicrepresentation of AB-Bernal stacked graphite. One set of carbon atomsis over the center of the hexagons in the adjacent layers.

In order to gain a deeper understanding of the extraordinary electronic properties ofgraphene, the electronic structure of isolated single-layer graphene (SLG) needs to beconsidered. Figure 2.2 a) shows the honeycomb lattice of graphene, which consistsof two interpenetrating sublattices noted A and B and depicted in blue and gray.Figure 2.2 b) shows the first Brillouin zone of graphene with the high-symmetrypoints M, K, K’, and Γ. In undoped graphene, the Fermi level lies exactly at theDirac point, which separates the conduction and valence band. Thus, graphene canbe considered as a zero-gap semiconductor or a semi-metal.[30] One significant featureof graphene is the linear energy dispersion at the Dirac point, which is responsiblefor the unique electronic properties of graphene. It significantly differs from the bandstructure of common semiconductors and metals, which exhibit parabolic energydispersion characteristics and free electron behavior.[12,31] As a consequence, electronsin graphene behave as massless Dirac fermions, which results in an extreme intrinsiccarrier mobility, the observation of the anomalous integer quantum Hall effect (QHE)at room temperature, and Klein tunneling.[29,30] In fact, this renders graphene apromising candidate for applications in next-generation molecular electronics, suchas field-effect transistors.[32]

As mentioned earlier, the electronic properties of graphene depend on the number oflayers. Only SLG and bilayer graphene (BLG) are zero-gap semiconductors with onlyone single type of electrons and holes. Turning to few-layered graphene the conductionand valence band start to overlap, and several charge carriers appear.[23,33,34] Thus,for classification four types of graphene are commonly considered: SLG (1 layer),BLG (2 layers), few-layer graphene (FLG, number of layers ≤5) and multilayer

6


graphene (MLG, number of layers >5). Notably, a structure consisting of more thanten graphene layers can be considered as a graphite thin film as it essentially exhibitsthe electronic properties of bulk graphite.[23]

a) b)

c)

Figure 2.2: a) Structure of graphene. �a1 and �a2 represent the unit vectors. b)Reciprocal lattice of graphene. The gray hexagon is the first Brillouinzone (BZ). �b1 and �b2 represent the reciprocal lattice vectors. c) The bandstructure of graphene calculated from tight binding model and a zoom-inof the dispersion relation close to the K point or Dirac point for smallenergies revealing the high degree of electron-hole symmetry.

Moreover, the electronic properties of graphene are influenced by the way thegraphene layers are stacked. In particular, the graphene layers can be stacked indifferent sequences, such as simple hexagonal (AAA), Bernal stacking (ABAB), andrhombohedral stacking (ABC). Disordered graphite, so-called turbostratic graphite,does not exhibit any preferred stacking order, and the parallel ordered graphene sheets

7


are rotated relative to each other without any preferred orientation.[35] Interestingly,the introduction of such rotational stacking faults causes the decoupling of adjacentgraphene layers.[36–38] As a consequence, the dispersion relation close to the Kpoint is altered from a parabolic (AB) to a linear band behavior. Hence, electronicproperties typical for isolated SLG can be observed for turbostratic bilayer andfew-layer graphene structures.[23,39,40] In fact, such rotational stacking faults havebeen reported for few-layer graphene dispersions prepared by liquid-phase exfoliationof graphite.[41]

2.1.2 Properties and Applications of Graphene-BasedMaterials

Arising from its unique structural and electronic properties, graphene features anumber of other exceptional properties ranging from mechanical stiffness, strength,and elasticity,[42,43] electrical and thermal conductivity[30,44,45] to optical transparencyand high surface area.[46,47] However, what distinguishes graphene from other mate-rials are not its impressive properties per se but the intriguing fact that it combinesso many of them. This paves the grounds for a number of technical applicationsthat range from energy storage to composite materials.[6,28,48] Figure 2.3 shows thecorrelation between the properties of graphene and its future potential in differentfields of application.

Graphene, with its extraordinary strength, enhanced electron conductivity, lightness,and flexibility is expected to play a major role in the development of next-generationsolar cells, batteries, fuel cells, and supercapacitors.[11,49] In fact, the widespreadconcern about energy sources has created a surge in effort to explore graphene for solarcells. Graphene has an optical transmission of 97.3% in the visible region and, thus,has the potential to replace indium tin oxide (ITO) as an optical transparent conductor(OTC) in solar cells and organic light-emitting diodes (OLED).[46] Nevertheless,although graphene has a theoretical in-plane conductivity of 20,000 S cm−1, todate the measured electrical resistance of graphene films prepared by chemicalvapor deposition are still much higher than those of ITO (2000-5000 Ω and 50 Ω,respectively).[16] However, graphene’s real advantage over ITO is its high mechanicalflexibility compared to the brittleness of ITO, which may open the way to flexibleelectronics. As such, graphene is expected to play a prominent role in dye-sensitized

8


solar cells, especially in applications where mechanical flexibility is paramount, suchas wearable electronics.[28] Moreover, owing to its various properties, graphene couldserve multiple other functions in photovoltaic cells next to the transparent electrodelayer including antireflective layer, photoactive material, channel for charge transport,and catalyst.[50]

Figure 2.3: Overview of graphene’s properties and different potential fields of appli-cation. Photographs are taken from Ref. [51–53].

In addition, graphene may serve as an ideal conductive additive for hybrid nano-structured electrodes in next-generation lithium-ion batteries. On one hand, itshigh surface area (theoretical ∼2630 m2 g−1)[49] enables improved interfacial contact,which results in an increased electrical conductivity and, thus, enables higher specificpower densities. On the other hand, graphene may also help to significantly improve

9


the energy density of lithium-ion batteries. In this context, graphene has beenrecently employed to form composites with SnO2 in order to increase specific capacityand cyclic stability of lithium-ion battery anodes.[54] In the short-term, graphene mayalso be used as a current collector in several energy storage systems. To this end, free-standing or substrate-bound graphene films could replace traditional activated carbonmaterials in the cathode and serve as current collectors in transparent devices.[48]

Moreover, graphene has attracted great interest during the past few years for super-capacitor applications.[47,55,55–57] Traditional supercapacitor electrodes are made ofactivated carbon (AC), where most of the surface area resides in the micropores and,thus, does not efficiently contribute to the charge storage. AC also possesses lowelectrical conductivity, which further hinders high power densities.[58,59] Graphenewith its two-dimensional sheet-like structure, on the contrary, features a large opensurface area, which is readily accessible to the electrolyte with a small diffusionbarrier. Graphene, thus, offers the potential for designing high power density devices,outrivaling any other form of carbon-based electrode materials. Furthermore, thehigh specific surface area of graphene-based materials enables extremely high energydensity devices.[60] Graphene-based supercapacitors will be discussed in detail inChapter 3.6.

In order to enable the continuation of performance improvements according toMoore’s law, it is predicted that silicon-based transistors in digital logic devices haveto be replaced in the long-term.[61] Indeed, graphene has the potential to producesmaller and faster devices than silicon due to its exceptional in-plane conductivityand high room temperature charge carrier mobility of ∼200,000 cm2 V−1 s−1, whichis more than 100 times higher than that of silicon and over 20 times higher thanthat of gallium arsenide.[7] In fact, the first demonstration of graphene’s exceptionalconductivity in a real application was a field-effect transistor.[32] However, the majorobstacle of graphene-based transistors is the lack of a band gap, which is required indigital electronics in order to achieve ”on” and ”off” states. Several ways to open aband gap are being targeted including the formation of graphene nanoribbons withfinite band gaps,[62,63] introduction of a band gap in BLG by applying an electricfield,[64,65] and selective chemical functionalization.[66,67] However, so far it has notbeen possible to achieve a sizable band gap, while maintaining a high charge carriermobility.[28,68] As such, graphene-based transistors are not expected to be seen inthe next ten years.[48]

10


Not all electronics require extremely high charge carrier mobilities, though. Low-costdevices like ink-jet printed electronics, such as radio-frequency identification (RFID)tags, low-resolution displays and backlights, sensors, and flexible connectors, onlyrequire modest electronic performance.[69] Commonly, the required charge carriermobility for those technologies is less than 1 cm2 V−1 s−1. By using graphene-based conductive inks, mobilities of up to ∼95 cm2 V−1 s−1 and sheet resistances of200 kΩ m−2 at a transmittance of ∼90% can be obtained, which paves the groundsfor transparent flexible electronics.[10,70] Moreover, the successful formulation ofink-jet printable graphene inks opens the path for the inexpensive and scalableintegration of graphene in various applications. In fact, graphene-based inks arethe first commercially available products based on graphene and are commercializedby several manufacturers including Vorbeck Materials, XG Science, and GrapheneSupermarket.[71–73]

Other low-cost applications of graphene dispersions are antistatic, electromagnetic-interference shielding,[74] and gas barrier applications.[28,48] Moreover, graphene’sextremely high Young’s modulus (1 TPa) and intrinsic strength (∼130 GPa) combinedwith its high aspect ratio renders it attractive for the applications in compositematerials.[43] In this context, graphene can be used as a highly efficient nano-filler inpolymers,[75] ceramics,[76] and metals[77] to create composites that feature increasedmechanical stiffness, electrical conductivity, and resistance to heat and pressure. Infact, the incorporation of well-dispersed graphene sheets into polymers has shownto significantly increase the Young’s modulus,[75] the tensile strength,[78] and theelectrical[79] and thermal conductivity,[80] even at very low volume fractions downto <1%.[79,81] Such advanced composites are expected to play a key role in thedesign of new windmill blades, aircraft, and other applications that require ultra-light and high-strength materials. Indeed, first applications of graphene-enhancedpolymer composites are already commercialized in tennis rackets and bicycle racewheels.[82–84]

Nevertheless, several key challenges, including production costs, processability, andchemical functionalization have to be addressed before graphene becomes a largevolume substitute to the current state-of-the-art reinforcement materials, such ashigh-quality carbon fibers and low-quality glass fibers composites.[28] On the basis offuture developments towards low-cost, easily processable bulk graphene materials,functional composites for packaging and consumer goods are expected to enter the

11

2.2 PRODUCTION OF GRAPHENE-BASED MATERIALS

market in the short-term, while in the long-run functional composites for mechanical,photonic, and energy applications might emerge. For further information on thecurrent state of graphene-based applications the reader is referred to Ref. [48] and[28].

2.2 Production of Graphene-Based Materials

For fully exploiting the outstanding properties of graphene in real life applications,it is necessary to produce graphene in large quantities at an affordable cost and,depending on each application, to tailor it to a suitable form and required quality.In order to meet these challenges various methodologies for the production ofgraphene and chemically modified graphene (CMG) have been described duringthe last years, including micromechanical cleavage,[18] chemical vapor deposition,[19]

graphitization of silicon carbide,[85] anionic bonding,[86] unzipping single-walled carbonnanotubes,[87] organic synthesis,[88] reduction of graphene oxide,[89] liquid-phaseexfoliation of graphite,[21] graphite intercalation,[90,91] and ball milling processing ofgraphite.[92–94]

Depending on the production method different graphene types with varying flakesize, flake thickness, grain boundaries, and chemical functionalization sites evolve.These types can be considered as different materials with different properties anddifferent potential application fields - see Figure 2.4. Hence, it is more precise to talkabout a graphene material family rather than a single material.

Among these production techniques, chemical vapor deposition and epitaxial growthproduce the best quality of thin graphene films featuring high electrical and ther-mal conductivity along with excellent optical transparency.[95] Therefore, theseare the preferable methods for graphene production for high-value electronic andoptoelectronic applications. However, the complexity of the process, harsh conditions(such as high temperature and vacuum), high costs, and difficulties encountered intransferring the produced films between various substrates limit their applicability.

Additionally, large quantities of graphene at a reasonable cost are required for thedevelopment of composite materials, conductive inks, and electrode materials forbatteries and supercapacitors. To this end, liquid-phase exfoliation processes startingfrom graphite as an inexpensive and abundant starting material are explored for

12


the mass fabrication of graphene. Solution-based graphene synthesis may start fromoxidation of graphite to graphite oxide or from pure graphite - see Figure 2.5. Themain advantage of these processes is that the resulting flakes are suspended, whichenables chemical modification, purification, and the deposition on various substrateswithout any limitations.[96]

Figure 2.4: Schematic representation of the main production routes of graphene,which allow a wide choice in terms of flake size, quality, and price for anyparticular application. Pictures are taken from Ref. [28].

2.2.1 Graphene Synthesis by Oxidation of Graphite

Nowadays, the majority of studies on graphene-based materials focus on the reductionof graphene oxide as a production process due to the proven scalability and maturityof this technology. In general, graphite oxide can be synthesized via oxidation ofgraphite using concentrated acids and strong oxidants according to either Brodie,[97]

Studenmaier,[98] or Hummers and Offermans[99] method. One of the advantages ofgraphite oxide is that it can be easily dispersed in water up to a concentration of3 mg mL−1.[95] Afterward, graphite oxide can be separated into individual grapheneflakes by rapid annealing at 1050 ◦C or by means of mild ultrasonication. However, theobtained graphene oxide (GO) flakes are insulating due to breakage of sp2-sp2 bondsand due to presence of sp3-sp3 bonds through the introduction of hydroxy, carboxylic

13


acid, and epoxy groups.[100] This renders GO unsuitable for electronic applications,where a high conductivity is needed, such as batteries and supercapacitors. Therefore,several methods for the transformation of insulating GO into fairly conductive reducedgraphene oxide (rGO) have been proposed.[89,96,101–104] Still, the resulting graphenesheets bear a substantial amount of structural defects and residual oxygen from theoxidation step. This constitutes a serious drawback, as many of the unique propertiesof graphene are degraded by the presence of disorder.[105] For this reason, GO willnot be discussed in the present work.

Figure 2.5: Schematic representation of different graphene synthesis approaches start-ing from raw graphite. Oxidation of graphite and subsequent reductionof graphene oxide (GO) to reduced graphene oxide (rGO) (right). Liquid-phase exfoliation of graphene by either solvent-assisted liquid-phaseexfoliation (middle) or surfactant-assisted liquid-phase exfoliation (left).

14


2.2.2 Liquid-Phase Exfoliation of Graphite

In order to minimize oxide defects in the graphene basal plane, several attemptstowards the oxidation-free liquid-phase exfoliation of pure graphite have beenmade.[21,106–112] Unmodified graphite can be successfully exfoliated in liquid me-dia by means of ultrasound to yield individual layers. The liquid-phase exfoliation(LPE) process generally involves three main steps: 1) dispersion of graphite in asolvent, 2) exfoliation by ultrasonication or other mechanical dispersing processes,and 3) purification.[48]

Ultrasonication-Assisted Exfoliation

Ultrasonication is a versatile and simple tool for the exfoliation of graphite in liquid-phase, which has proven to be superior to most other dispersing processes in creatinglaminated materials and stable dispersions even in the case of non-oxidized graphenesheets.[113] The strong shear forces that exfoliate graphite during sonication comesfrom the cavitation process, which involves bubble formation, growth, and collapsein liquids irradiated with high-intensity ultrasound. While collapse on the surfaceitself will cause direct damage by the shock waves produced and will induce defectsin the graphene basal plane, collapse in the liquid close to the surface causes amicrojet of solvent to hit the solid with great force and thereby peeling individualgraphene sheets apart.[113,114] Additionally, localized hot spots with ∼5000 K andpressure of hundreds of bars are formed within the liquid during the bubble collapse.These hot spots generate highly reactive species involving radicals, such as peroxylradicals, from the solvent dissociation.[115] These radical reactions are usually verydestructive, which renders ultrasonic treatment very effective in breaking C-C bonds.In fact, prolonged periods of ultrasonication treatments damage the graphene sheetsresulting in small flake sizes and nanometric graphitic impurities.[13]

After ultrasound-induced exfoliation, the solvent-graphene interaction needs tobalance the inter-sheet attractive forces in order to avoid immediate reagglomerationof the as-produced graphene sheets. This can be achieved by either solvents, whichminimize the interfacial tension between the liquid and the graphene flakes, or bysteric stabilizers, which allow the exfoliated sheets to remain suspended in solution.

15


Graphene Dispersions in Organic Solvents

The exfoliation of unmodified natural graphite via sonication in different solventswas first reported by Hernandez et al.[21] They proposed a gentle method for theliquid-phase exfoliation of graphite with the help of organic solvents, such as N-methyl-2-pyrrolidone (NMP), N,N-dimethylacetamide, γ-butyrolactone (GBL), and1,3-dimethyl-2-imidazolidione. Among the tested solvents, NMP provided the highestpercentage of monolayer graphene, while cyclopentanone yielded the highest absoluteconcentration of FLG with a solubility of 0.0085 ± 0.0012 mg mL−1. In general,solvent-based exfoliation works with solvents, which have an interaction energy withgraphene that is, at least, equal to the graphene-graphene interaction energy. Thisresults in a minimal energy cost to overcome the van der Waals forces betweengraphene sheets. Accordingly, liquids with a surface tension close to that of graphite(∼40 mN m−1) and Hansen solubility parameters of δD∼18 MPa1/2, δP ∼9.3 MPa1/2,and δH∼23 MPa1/2 were found to be most suitable for the exfoliation of graphite.[44]

The main disadvantages of this method are low stability, low concentration, alow single-layer content of the resulting dispersions (0.01 mg mL−1), and highboiling points of the solvents (<450 K).[21] This may not only cause problems whenprocessing graphene into films or composites but also makes it nearly impossible todeposit individual flakes from the dispersion due to aggregation during slow solventevaporation. To address this problem, the successful exfoliation in volatile solvents,such as acetone, ethanol, and isopropanol, with the help of sonication and ballmilling is reported. However, these suspensions contain even lower concentrations ofgraphene than those in the favored high boiling solvents.[94,116]

Graphene Dispersions in Surfactant Solutions

A facile approach towards the production of stable graphene dispersions in processingfriendly solvents, such as water, is the utilization of amphiphilic molecules as stericstabilizers. The noncovalent interactions between graphene and the stabilizer arebased on van der Waals interactions or π-π stacking of amphiphilic molecules ontothe graphene plane. The distinct advantages of this route compared to covalentchemistry are easy accessibility, high scalability, reversibility, and no altering ofgraphene’s electronic structure.[117] Amphiphilic molecules feature both hydrophobicand hydrophilic moieties and are often referred to as surfactants, stabilizers, or

16


additives. The hydrophobic part, which exhibits little attraction to the solvent, buthigh attraction to the graphene surface, is typically an alkyl chain or aromatic moietyand can potentially alter the properties of graphene by doping.[118] In contrast, thehydrophilic part, which has a strong attraction to the solvent, can be an ionizedfunctional group, ethylene oxide group, or a combination of them. In general, thehydrophobic part binds to the hydrophobic surface of graphene via hydrophobiceffects, while the hydrophilic part of the surfactant interacts with the solvent andimpedes sheet aggregation.

Figure 2.6: Structures of the most common surfactants employed for the exfoliationand stabilization of graphene in aqueous medium.

Recently, a variety of surfactants has been investigated for the exfoliation anddispersion of graphite in aqueous media. Lotya et al., for example, proposedthe liquid-phase exfoliation of graphite in water by employing sodium dodecyl-benzenesulfonate (SDBS) as a surfactant.[106] Later on, several other surfactantshave been successfully applied for the stabilization in aqueous medium by eithersteric or electrostatic repulsion - see Figure 2.6. These include sodium dodecylsul-fate (SDS),[92,119–122] sodium cholate (SC),[108,120,122–125] sodium taurodeoxycholate-

17


hydrate,[119,126] sodium deoxycholate,[119,120] sodium poly(4-vinylbenzenesulfonate)(PSS),[119] lithium dodecylsulfate[120] as anionic surfactants, hexadecyltrimethyl-ammonium bromide (CTAB),[119,120,122,127–129] tetradecyltrimethylammonium bro-mide[120,128,129] as cationic surfactants, and Pluronics,[119,122,130–132] Tetronics[122,130,133]

IGEPAL CO-890,[120] Triton X-100,[119,122,134] polyvinylpyrrolidone,[107,119] ethylcellu-lose,[135] Tween 20, Tween 80, and Tween 85[119,120,136] as nonionic surfactants.

In particular, ionic aliphatic surfactants are effective in the exfoliation of graphite inaqueous medium down to FLG with concentrations of 0.002-0.3 mg mL−1. SC hasbeen shown to be superior to other surfactants regarding concentration (0.3 mg mL−1)and exfoliation (1-10 layers and 20% monolayer content).[123] This can be attributedto the geometrical similarity of SC with the flat graphene surface and its enhancedsteric repulsion induced by the bulky set of aliphatic rings.[122] However, prolongedultrasonication times of 430 hours lead to only small graphene flakes ranging from50-200 nm.[123] Moreover, the application of sodium cholate demonstrated adverseeffects on the electrochemical properties of the resulting graphene sheets.[13] Incomparison, nonionic surfactants lead to much higher graphene concentrations ofup to 15 mg mL−1.[132] However, the observed dispersions usually contain few-layerand multilayer graphene flakes with a lower level of exfoliation (1-50 nm).[119,137]

This can be attributed to the fact that most nonionic surfactants, such as PluronicP-123 (P123), possess very long hydrophobic parts compared to ionic aliphaticsurfactants. Indeed, these bulky hydrophobic parts are able to wrap around and,thereby, stabilizing multilayer graphene stacks, which, in turn, leads to a decreasedfraction of single- and bilayer graphene in dispersion. This is a distinct disadvantagein cases, where thin graphene sheets with high specific surface area are desired.However, graphene flakes exfoliated using nonionic surfactants show great promisein biomedical applications.[119]

Surfactants containing aromatic groups, such as polycyclic aromatic hydrocarbons(PAHs), are capable of forming more specific and more directional π-π interactionswith the graphene basal plane and, therefore, facilitate the exfoliation down tomonolayer graphene in aqueous medium by hydrophobic effects.

Pyrene derivatives have been successfully applied to stabilize graphene dispersions inboth water and organic solvents by various groups.[109,137–144] Pyrene derivatives thatare functionalized with water-soluble groups serve a dual purpose by firstly helpingto exfoliate the individual graphene flakes from the parent graphite piece. Afterward,

18


they form stable polar functional groups on the graphene surface via noncovalentπ-π stacking, which keeps the exfoliated graphene flakes suspended in solution asillustrated in Figure 2.5.[109] Different polar and apolar moieties were attached tothe pyrene aromatic core including sulfonic groups,[141] amines,[145] carboxylic,[109]

butyric,[141] or butylphenyl groups.[146] Among them, 1-pyrenesulfonic acid has beenfound to be most effective, yielding final graphene concentrations as high as 0.8-1 mg mL−1.[141] Moreover, several studies revealed that pyrenes are highly efficientin exfoliating graphite towards single- and few-layer graphene (1-4 nm).[109,138–143]

Interestingly, noncovalent functionalization with 1-pyrenecarboxylic acid was foundto improve the graphene’s electrochemical performance in supercapacitor devices.[140]

Although strong π-π interactions render amphiphilic pyrenes as excellent choicesfor the production of thin graphene flakes with high specific surface area, the smallhydrophilic domains are insufficient in preventing graphene sheets from agglomerationand restacking. First attempts to address this issue were made by developing stericallydemanding pyrene derivates, including a pyrene-based hydrophilic dendron and apoly(ethylene oxide) functionalized pyrene.[137,147,148]

Other examples for aromatic stabilizers include amphiphilic perylenes,[149–151] 7,7,8,8-tetracyanoquinodimethane,[152] 1,2,3,4-coronenetetracarboxylic acid,[153] 9-anthracene-carboxylic acid,[154] and 6-amino-4-hydroxynaphthalene-2-sulfonic acid[155] to name afew.

Since ultrasonication is not scalable beyond liter scale, alternative methods for theexfoliation of graphite in liquid-phase, such as ball milling,[92,136,156] high-pressure ho-mogenization,[157] and high-shear mixing,[158] have been studied recently. These meth-ods significantly improve graphene production rates up to ∼100 mg h−1. However,they achieve only moderate concentrations of FLG of 0.1-0.4 mg mL−1. Moreover,only a low percentage of SLG (0.5%) and small flake sizes (<1 μm) were observed.Notably, due to the large batch size all of the experiments were carried out in NMPor aqueous medium using cheap commercial aliphatic surfactants as stabilizers (e.g.Tween 80). From this point of view, it is impossible to distinguish if the low monolayercontent and the poor material quality were determined by the intrinsic properties ofthe used exfoliation methods or by insufficient stabilization. Therefore, an in-depthanalysis of the stabilization mechanism is mandatory to achieve quantitative insightsinto the effectiveness of the proposed exfoliation methods.

19

2.3 EVALUATION OF GRAPHENE DISPERSION

In general, liquid-phase exfoliation of bulk graphite is an extremely mild, versatile,and potentially scalable approach to obtain high-quality FLG for a range of applica-tions, such as inks, composite materials, and electrode materials for batteries andsupercapacitors. The main advantages of surfactant-based liquid-phase exfoliationare high yields and low flake thickness combined with the solubility in user-friendlysolvents like water. However, there seems to be a trade-off between dispersionstability and exfoliation degree. Moreover, aromatic stabilizers are expensive finechemicals, which are insufficient for large-scale processing. In order to fully exploitthe potential of liquid-phase exfoliated graphene towards large-scale applications,future efforts have to be devoted to suitable, commercially available, and cheapsurfactants, which would improve the exfoliation and dispersion stability, featurean enhanced affinity for the basal plane of graphene, and provide sufficient stericstabilization.

2.3 Evaluation of Graphene Dispersion

In addition to the development of facile synthetic approaches towards stable graphenedispersions, scientists have been challenged by the search for reliable characterizationtechniques, which enable a detailed investigation of the dispersed graphene samples.One of the most important characteristics of a given exfoliation method is the liquid-phase exfoliation (LPE) yield, which can be defined in two different ways. One is theyield by weight, Yw, defined as the ratio of the weight of dispersed graphitic particlesand the starting material.[21] However, this ratio can be misleading, since besidessingle- and few-layer graphene, also graphitic nanoparticles with low hydrodynamicdiameters might remain dispersed after centrifugation. This is a post-treatmentthat is typically applied before the analysis of the dispersion in order to remove anymacroscopic graphitic particles. As such, the overall yield by weight may include alarge fraction of graphitic particles, which are expected to impair the performance.A more useful metric to quantify the quality of a graphene dispersion is the analysisof the single-layer content or percentage of a dispersion (YM), which is defined asthe ratio of the number of SLG and the total number of graphitic flakes within adispersion.[21,22,112] Depending on the requirements of the desired application, the few-layer content/percentage may also be of sufficient interest. In order to determine theexfoliation yield within an LPE sample, both qualitative and quantitative information

20


have to be acquired. In the following chapters, a short overview of the most relevantcharacterization techniques, which are absorption spectroscopy, transmission electronmicroscopy, atomic force microscopy, and Raman spectroscopy, will be given.

2.3.1 Determination of the Graphene Concentration

The concentration (c) of dispersed graphitic material, which is of particular interestfor the evaluation of the dispersion efficiency of different surfactants, is commonlydetermined by optical absorption spectroscopy.[21,106,123,159] In order to calculatethe concentration from the UV/vis absorbance according to the Beer-Lambert law(A=εcl), the extinction coefficient (ε) for the material has to be known. ε canbe experimentally determined by filtering a known volume of graphene dispersionand measuring the resulting mass by using a microbalance.[21,106,123,159] In theliterature, different values for the extinction coefficient for both aqueous[106] andnon-aqueous[21] graphene dispersions have been reported. These values range from1390 to 6600 mL mg−1 m−1, mainly depending on post-processing conditions, suchas centrifugation and the production method per se.[21,22,48,112]

2.3.2 Microscopic Techniques

The optical visibility of graphene, which is enhanced by an appropriately chosensubstrate, renders it possible to distinguish between single- and few-layer grapheneflakes by inspection in an optical microscope.[160,161] For a more systematic analysis,the number of graphene layers 〈N〉 and, thus, the thickness of the graphene sheets,can be determined by high-resolution microscopy techniques, such as transmissionelectron microscopy and atomic force microscopy.

Transmission Electron Microscopy

Transmission electron microscopy (TEM) is a frequently used tool for imaging ofnano-sized materials, as it offers atomic scale resolution. In principle, a transmittedelectron beam passes through an ultra-thin sample and is then detected by theimaging lenses and detector.[162] Due to its high sensitivity, TEM is suitable toresolve the atomic features of graphene and can, therefore, be used to investigate

21


the flake thickness (by edge counting),[163] graphene layer stacking faults,[41] anddefects in the basal plane.[164,165] Unfortunately, the utilization of traditional TEMsis limited by their resolution at low operating voltages. Thus, no quantitativestatements on the flake thickness, but rather qualitative information on the flakesize can be gathered with this setup. However, high operation voltages can damagethe graphene sample.[164,166,167] In addition, high-resolution transmission electronmicroscopy (HRTEM) is often assisted by electron diffraction, a TEM techniqueperformed by shifting the image focus from the sample to the back focal plane ofthe microscope, which allows the observation of the lattice structure through theinterference pattern.[168] Hence, the relative intensities of the electron diffractionpattern from the (2-110) and (1-110) planes can be used to determine the number oflayers. For example, if I(1−110)/I(2−110) is >1, it is recognized as SLG, whereas it isreferred to multilayer graphene if the ratio is <1.[164,165,169] Nevertheless, the pitfallsof TEM imaging are the extensive sample preparation and the determination of asmall fraction of the sample only. Within this thesis, TEM imaging is used to gainqualitative insights into the flake size distribution and morphology of the preparedgraphene samples.

Atomic Force Microscopy

Atomic force microscopy (AFM) can be further used to determine the height, and bydividing it by the graphite interlayer distance of 0.34 nm,[160] the layer thickness ofsingle graphene flakes. AFM imaging offers three different operation modes, namelytapping, non-contact, and contact mode. Graphene investigations are commonlyundertaken in tapping mode, as contact mode might drag or tear the graphene sheets.In this mode, the probe is vibrated at a given resonance frequency by a piezo actuator.The change of the cantilever amplitude resulting from the interaction of the tip withthe sample is used as a measure of probe-sample interactions. The change is kept ata pre-set level during the scan. Tapping mode is less destructive, because permanentshear forces are almost eliminated. Another advantage of the tapping mode is thepossibility of probing local mechanical properties such as elasticity.[170,171] AFMimaging gives a topographical contrast of the sample only and does not distinguishbetween graphene, GO, and potential surfactant residuals present on the sample.Nevertheless, phase imaging can be applied in order to gain deeper insights intothe chemical composition of the sample.[20,172,173] The observed thickness for SLG is

22


highly dependent on the substrate and the environmental conditions, such as relativehumidity. The height of SLG was reported to be ∼1 nm on SiO2 and ∼0.4 nm onmica surfaces.[22,29,174]

2.3.3 Raman Spectroscopy

While giving valuable insights into the structural composition of single grapheneflakes, neither TEM nor AFM is suitable for the fast mapping of large sample areas.In order to characterize large probe areas in a high throughput manner, which isa prerequisite for industrial metrology, fast and non-destructive characterizationmethods applicable at both laboratory and mass-production scales, are required.[168]

In this context, statistical Raman mapping analysis has been proven to be the goldstandard for graphene sample characterization.[175]

As illustrated in Figure 2.7 there are two primary types of scattering, which occurwhen a photon impinges a material: Rayleigh and Raman scattering. Wheneverthe laser interacts with the electrons and causes polarization, a virtual energy stateis created. The energy of these states depends on the frequency of the laser used.During the Rayleigh process, no energy transfer takes place, and the molecule returnsto its initial state. Thus, the frequency of the emitted photon has the same energyas the incident photon. Consequently, Rayleigh scattering can be described as anelastic scattering process.[176]

Raman scattering processes, which occur from the vibrational ground state m, canalso lead to an absorption of energy. The molecule is promoted to a higher vibrationalexcited state n and the photon loses energy. This process is called Stokes scattering.Due to thermal energy, some molecules may be present in an excited state such as n.Scattering from these states leads to a transfer of energy to the scattered photon,which results in a loss of energy of the molecule and is called anti-Stokes scattering.The relative intensities of Stokes and anti-Stokes processes depend on the populationof the various states of the molecule.[177] However, the amount of molecules in anexcited vibrational state will be small in comparison to molecules in the groundstate. Thus, anti-Stokes scattering is much weaker compared to Stokes scattering,but increases with increasing temperature.[178]

23


Besides the excitation to virtual states, an excitation to a real electronic excitedstate is also possible. When the incident photons have energies close to a realelectronic-vibrational energy state, excitation to an excited energy state occurs. Thisprocess is called resonance Raman scattering and is consequently much more likelyto occur (intensity enhancement by a factor of 103 with regards to non-resonantRaman scattering).[177,179–181]

Figure 2.7: Schematic diagram of the Raman scattering processes. The lowest energyvibrational state m is shown at the bottom with states of increasingenergy above it. Both, the excitation energy (upward arrows) and thescattered energy (downward arrows) is much larger than the energy of avibration.

Graphene’s electronic structure is captured in its Raman spectrum, which evolveswith the number of graphene layers 〈N〉.[48] Hence, Raman spectroscopy givesvaluable insights into the number and orientation of layers, the quality and types ofedges, and the effects of perturbations, such as electric and magnetic fields, strain,doping, disorder, functional groups, and unwanted by-products.[182]

Figure 2.8 shows a typical Raman spectrum of single-layer graphene. The majorRaman features of graphene and graphite are the so-called G-band (∼1580 cm−1) and2D-band (∼2700 cm−1). The G-band in the first-order Raman spectrum of graphitecorresponds to the optical vibration of two neighboring carbon atoms in a graphene

24


layer and is a double degenerated phonon mode (E2g symmetry) at the Brillouinzone center - see Figure 2.9 a) and b).[175] It occurs at 1582 cm−1 in graphite and isupshifted by approximately 5 cm−1 in SLG. The upshifts follow a 1/n dependenceon the number of layers n. This can be used to characterize the number of layers inFLG.[183,184]

Figure 2.8: Raman spectrum of single-layer graphene prepared by micromechanicalcleavage showing the main Raman features, the D-, G-, and 2D-band.The sample was transferred onto a Si/SiO2 wafer and excited at 532 nm.

The second feature appearing at approximately 2700 cm−1 is the 2D-band. The2D-band arises from a second-order two-phonon process as shown in Figure 2.9 d). Itexhibits a strong frequency dependence on the excitation laser energy Elaser. It is asecond-order process related to a phonon near the K point in graphene activated bydouble resonance processes, which are responsible for its dispersive nature and causea strong dependence on any perturbation of the electronic structure of graphene.[185]

The 2D-band changes in shape, width, and position for increasing N reflecting thechange in electron bands. Hence, it is a sensitive probe for differentiating betweensingle- and bilayer graphene, and graphite. In SLG, the 2D-band can be fit by a singleLorentzian peak while fitting the 2D-band of a bilayer requires four Lorentzian. This

25


results from the splitting of the π electronic structure of graphene when a second layeris present. In BLG, four double resonance scattering processes are possible comparedto one for SLG. With an increasing number of layers, the number of double resonanceprocesses increases and the spectral shape converges to that of graphite, where twopeaks are observed.[184–187] As mentioned earlier, not all multilayer graphenes areBernal stacked, and basically any relative orientation and/or stacking of graphenelayers are possible, which would be reflected in a significant change of the bandstructure.[40] Notably, some stacking orders result in a Dirac-like spectrum and, thus,single-layer like 2D-bands can be observed even in the case of few-layer graphenesamples.[40] It was reported, for example, that turbostratic graphite has a single2D-band, however, with an almost double full width at half maximum (FWHM)compared to SLG. In fact, turbostratic graphite often features a D-band, which leadsto a defect-induced broadening of the 2D-band.[188,189] Indeed, the 2D-band shapegives valuable information on the multilayer interactions and the presence of Diracelectrons in FLG.[182] Thus, care should be taken in affirming that a single Lorentzian2D peak is proof for SLG. In order to fully characterize a graphene sample, the shapeand width of the 2D-band have to be considered. As demonstrated by Graf et al.[187]

and Lee et al.,[190] the FWHM of the 2D-band of SLG is in the range of 25-30 cm−1,whereas FLG features an FWHM between 39-65 cm−1. For multilayer graphene,FWHMs broader than 65 cm−1 can be observed. Furthermore, the intensity of the2D-band, I2D, increases in respect to the G-band (I2D/IG>4) for SLG, which is insharp contrast to the situation in graphite, where the G-band dominates the Ramanspectrum.[183]

Besides these prominent features, which are essential for differentiating betweengraphene and graphite, the D-band is also important for graphene characterization.This band arises from the lattice motion away from the center of the Brillouin zoneand appears between 1270 cm−1 and 1450 cm−1 (Figure 2.9 c)). It depends on theexcitation energy of the laser and is not Raman active for pristine graphene butcan be observed when symmetry is broken at defects or edges. Hence, an intenseD-band signifies the presence of defects within the graphene sample. Quantifyingdisorder and defects in graphene is usually done by analyzing the ID/IG intensityratio between the disorder-induced D-band and the Raman allowed G-band.[182,191,192]

Within this thesis, statistical Raman analysis is applied for the determination of theexfoliation yield over a large sample size and the statistically gathered informationis further confirmed by TEM and/or AFM measurements of representative probe

26


areas.

a) b)

c) d)

Figure 2.9: a) Schematic representation of the physical origin of the G-band. b)Atomic displacements for the E2g modes. Schematic representation ofthe physical origin of c) the D-band and d) the 2D-band.[185]

27

3 Supercapacitors

3.1 Introduction

Supercapacitors, also known as ultracapacitors or electrochemical double-layer ca-pacitors (EDLCs), are a new type of electrochemical energy storage systems usedfor harvesting energy and delivering high power in a short time. Their main energystorage mechanism is based on the formation of an electrostatic double-layer ofelectronic and ionic charge accumulated on each side of the electrode/ electrolyteinterface in response to an applied potential.[193]

Figure 3.1 shows the so-called Ragone plot of typical energy storage and conversionsystems regarding specific power and energy density. Supercapacitors are consideredas an alternative energy storage device to Li-ion batteries (LIB) since they possessan intermediate value of power density and energy density when compared withclassical capacitors, batteries, and fuel cells. In detail, supercapacitors have higherpower densities than batteries and fuel cells and higher energy density than classicalcapacitors.[194] Moreover, batteries undergo chemical changes and phase transitionsupon charge and discharge, which leads to degradation of their performance overseveral hundred to several thousand cycles. In contrast, EDLCs only require thephysical rearrangement of electronic and ionic charges with no chemical change andcan therefore be cycled over a million times without any performance degradation.Additionally, they are not limited by reaction kinetics and solid-state mass transport,as it is typically the case for batteries, enabling EDLCs to be charged and dischargedrapidly with nearly 100% efficiency.[193,195–198]

Therefore, they are widely used in the case of stationary and mobile systems requiringhigh power pulses, such as car acceleration, tramways, cranes, forklifts, and emergencysystems. Moreover, due to their low time constant, they can quickly harvest energy,for example, during deceleration of vehicles.[199,200] However, EDLCs encounter theproblem of low energy density compared to batteries, which limits the dischargetime to less than a minute, whereas many applications clearly need more. To thisend, major efforts are underway to improve the energy density of EDLCs, while tomaintain their advantageous attributes, such as high power density, long cycle life,

28

3.2 THE ELECTRICAL DOUBLE-LAYER

and no short circuits in comparison to batteries. It is known that the energy densityof EDLCs depends on the specific capacitance and the potential window. In turn,the specific capacitance and energy density of the electrode rely on the electrodematerial.[194,196] Hence, the electrode materials play a main role in the developmentof future EDLCs.

Figure 3.1: The Ragone plot for various electrochemical energy storage devices. Ifa supercapacitor is used in an electric vehicle, the specific power showshow fast one can go, while the specific energy shows how far one can goon a single charge. Times shown are the time constant of the devices,obtained by dividing the energy density by the power. Adapted fromRef. [194].

3.2 The Electrical Double-Layer

On the basis of their energy storage mechanism, supercapacitors can be classified intotwo categories. One is the EDLC, in which the electrode material, typically carbonparticles, is not electrochemically active. Thus, there are no electrochemical reactionstaking place at the electrode material during charging and discharging, and purephysical charge accumulation occurs at the electrode/electrolyte interface.[195] Theother category is the faradaic supercapacitor, also called pseudocapacitor, in which

29

3.2 THE ELECTRICAL DOUBLE-LAYER

fast and reversible faradaic processes take place upon the application of a potentialdue to electro-active species (e.g., RuO2, IrO2, MnO2, and Co3O4).[195,201,202] In thefollowing, we will only focus on EDLCs, but the readers should be referred to anumber of recent reviews on pseudocapacitors - see Ref. [203–205].

EDLCs store the charge electrostatically by reversible adsorption of ions of theelectrolyte onto electrochemically stable carbonaceous electrode materials formingan electric double-layer (EDL). The EDL mechanism, which describes the chargestorage mechanism of EDLC electrode materials, was first modeled by Helmholtz -see Figure 3.2.[206]

a) b) c)

Figure 3.2: Schematic representation of the electrical double-layer at a positivelycharged surface: a) the Helmholtz model, b) the Gouy-Chapman model,and c) the Stern model. IHP and OHP represent the inner Helmholtzplane and outer Helmholtz plane, respectively. The OHP is also the planewhere the diffuse layer begins. H is the double-layer distance describedby the Helmholtz model. Ψ0 and Ψ are the potentials at the electrodesurface and the electrode/electrolyte interface, respectively.[207]

However, this model does not take adsorption of water molecules and counter ionsinto account. Therefore, Gouy and Chapman fine-tuned the model by proposingthe presence of a diffuse layer, which arises from the continuous distribution ofelectrolyte ions in the electrolyte solution driven by thermal motion.[208,209] Still, theGouy-Chapman theory leads to an overestimation of the EDL capacitance. Thus,Stern suggested an even further refined model combining the Helmholtz and Gouy-Chapman models, which predicted the existence of two layers of ion distribution.[210]

The first layer is called compact layer, Stern layer, or Helmholtz layer, and arisesfrom specifically adsorbed ions and non-specifically adsorbed counter ions. These two

30

3.3 OPERATING PRINCIPLES OF SUPERCAPACITORS

types of ions are distinguished as inner Helmholtz plane (IHP) and outer Helmholtzplane (OHP). The second layer is called diffusion layer, which is the same as inthe Gouy-Chapman model.[195,199,211] As a result, the capacitance of EDL (Cdl) canbe treated as a combination of the capacitance of the Helmholtz/compact layerdouble-layer capacitance (CH) and the diffusion region capacitance (Cdiff ), and canbe expressed by the following equation:[207]

1Cdl

= 1CH

· 1Cdiff

. (3.1)

In general, the double-layer capacitance is about 10-20 μF cm−2 for a smoothelectrode in concentrated electrolyte solutions. Factors affecting the double-layercapacitance are electrolyte concentration (celectrolyte), solvent (εr), and temperature(T ).[212]

3.3 Operating Principles of Supercapacitors

A supercapacitor is a charge storage device, similar to batteries in design andmanufacturing. As shown in Figure 3.3, it consists of two electrodes, which arefabricated by laminating high specific surface area carbon-based active materialsonto current collectors. Both electrodes are identical and are charged by an externalpower supply to hold opposite charges, that is, negative and positive.

Figure 3.3: Schematic representation of a supercapacitor test cell assembly.[47]

Next, a separator made of a porous electrically insulating material is sandwichedbetween the two electrodes, which is used to prevent contact and short circuitingas well as to provide pathways for the electrolyte ions.[195,213] The cell is thenimmersed into an electrolyte solution, which can be either aqueous (e.g. H2SO4,

31


KOH), organic (e.g. tetraethylammonium tetrafluoroborate in acetonitrile), or anionic liquid (IL).[194,214–216] At each side of the electrode, a double-layer is created,which contains the Helmholtz and diffusion layer along the carbon particle-electrolytesolution interface. The resulting interface can be treated as a capacitor with anelectrical double-layer capacitance. Thus, the stored capacitance on the electrodecan be described by the following equation:

C = εrε0A

d, (3.2)

where εr is the relative dielectric constant of the electrolyte, ε0 is the dielectricconstant of the vacuum, d is the effective thickness of the double-layer (chargeseparation distance; Debye length), and A is the electrode’s surface area. The chargeseparation distance d corresponds to the radius of a solvated ion and depends onthe electrolyte and the size of its ions and is typically on the order of 5-10 A forconcentrated electrolytes.[194,212,217,218]

The capacitance of the electrode/electrolyte interface in an EDLC relates to anelectrode-potential-dependent accumulation of electrostatic charges at the interface.Here, an excess or a deficit of electric charges is accumulated on the electrode surface,and electrolyte ions with counterbalancing charges are built up at the electrolyte sidein order to meet electro neutrality.[217] The electrochemical processes in an EDLCcan be expressed by equation 3.3 - 3.8:[219–221]

Reactions on the positive electrode:

ES1 + A− charging E+S1 ‖A− + e− (3.3)

E+S1 ‖A− + e− discharging ES1 + A− (3.4)

Reactions on the negative electrode:

ES2 + C+ + e− charging E−S2 ‖C+ (3.5)

E−S2 ‖C+ discharging ES2 + C+ + e− (3.6)

32


And the overall reaction:

ES1 + ES2 + A− + C+ charging E+S1 ‖A− + E−

S2 ‖C+ (3.7)

E+S1 ‖A− + E−

S2 ‖C+ discharging ES1 + ES2 + A− + C+, (3.8)

where ES1 and ES2 represent the two electrode surface areas, respectively, ‖ isthe electrode/electrolyte interface, and C+ and A− are cations and anions of theelectrolyte, respectively. During a charging process, the electrons travel from thepositive electrode to the negative electrode through an external load. Within theelectrolyte, anions move towards the positive electrode and cations move towardsthe negative electrode. During the discharging process, the reverse process takesplace and the electrons shift from the negative to the positive electrode through anexternal load and the ions return from the surface into the bulk of the electrolyte - seeFigure 3.4.[221] Noteworthy, there is no charge transfer across the electrode/electrolyteinterface and no net ion exchange between the electrode and the electrolyte takes place.Thus, the electrolyte concentration remains constant during the charge/dischargeprocess and the energy is stored in the double-layer interface.[217]

Figure 3.4: Schematic representation of an electric double-layer supercapacitor ineither charged (left) or discharged (right) state.

Compared to a parallel-plate capacitor, which uses two electrodes to form one ca-pacitor, in the EDCL each electrode/electrolyte interface represents a capacitor.Consequently, the simplified equivalent circuit of the complete device can be repre-sented by two capacitors in series. The total capacitance of a device (Cdevice) can be

33

3.4 THE PERFORMANCE OF SUPERCAPACITORS

written as[217,218,222]

1Cdevice

= 1C1

+ 1C2

= C1

2 , for C1 = C2, (3.9)

where C1 and C2 are the capacitances of the positive and negative electrode, re-spectively. Thus, in a symmetric supercapacitor, with both electrodes having thesame capacitance, the full device capacitance is only half the capacitance of oneelectrode.[214] Typically, the capacitance of a single electrode normalized by themass of a single electrode is reported as the specific or gravimetric capacitance perelectrode Cs(electrode) in F g−1. To estimate the total specific capacitance expectedfrom the device Cs(device), the single electrode capacitance has to be divided by four.A factor of two accounts for the configuration of two capacitors in series and anotherfactor of two accounts for double the mass.[223]

3.4 The Performance of Supercapacitors

The specific energy and power density of supercapacitors are the most importantcriteria for evaluating performances in comparison to other energy storage concepts.The power density describes how fast the energy stored in the device can be deliveredto an external load. The power density is typically expressed as:

P = 14RESR

V 2, (3.10)

where V is the maximum cell voltage, RESR is the equivalent series resistance (ESR),and P is the maximum power.[224]

In supercapacitors, a number of sources contribute to the internal resistance, whichare collectively referred to as ESR. The ESR involves:[214,215,225]

• the resistance of the electrode layer interparticles due to the particulate natureof the electrode matrix,

• the resistance of the external load contact,

• the resistance of the electrolyte,

• the contact resistance between the current collector and the electrode layer,

34

3.4 THE PERFORMANCE OF SUPERCAPACITORS

• the ionic (diffusion) resistance of ions moving in small pores, and

• the ionic resistance of ions moving through the separator.

Because of the electrostatic storage mechanism of EDLCs, the ESR does not containany charge transfer resistance contributions due to redox reactions. Therefore, theESR is much smaller than that of batteries, which explains the high power densitiesof supercapacitors.[8,226] Nevertheless, reducing the ESR will further improve thesupercapacitor’s performance in terms of power density. The ESR can be obtainedby either electrochemical impedance spectroscopy or constant current charging-discharging measurements as described in Chapter 3.5.

The maximum energy density is given by:

E = 12CV 2, (3.11)

where V is the maximum voltage, C is the capacitance, and E is the energy.[218]

In comparison to batteries, where the charge is stored in the bulk material, the chargeis stored on the surface of the active material in a supercapacitor. This renders themlow energy density devices. The energy density of a supercapacitor is proportional toits capacitance. Thus, in order to achieve high energy density supercapacitors, thecapacitance must be increased. The capacitance of an electrode material is referredto either per mass or volume. The specific capacitance or gravimetric capacitanceCs(electrode) of an electrode is most commonly used to evaluate promising electrodematerials for supercapacitors.[223] The theoretical specific capacitance of an electrodematerial can be calculated from the mass of the electrode material m, its differentialcapacitance density Cdl, and the total specific surface area (SBET ) of the electrodematerial measured by Brunauer-Emmett-Teller (BET) technique, as following:

Cs = SBET Cdl

m. (3.12)

The capacitance depends on the characteristics of the electrode material. In detail, itdepends on the specific surface area of the electrode material and the electrode layerstructure, which may hinder or enable access to the surface area, pores, and holesfor ion diffusion. Cs(electrode) is typically reported as the capacitance per mass of theactive electrode material and does not take the mass of the electrolyte into account.However, many nanomaterials, such as carbon nanotubes, and graphene, have very

35

3.5 EVALUATION OF SUPERCAPACITORS BY ELECTROCHEMICALMEASUREMENTS

low packing density. This may lead to an empty space in the electrode layer, whichwill require a large fraction of the electrolyte and will thereby increase the weight ofthe device significantly without adding any capacitance.[223,227] Another importantaspect for rechargeable energy storage devices is the volume of the device. Therefore,the volumetric capacitance Cv has to be considered as well. For this reason, thematerial of choice would be electrode materials with a high ion-accessible surfacearea, which can be packed into dense configurations.

The maximum voltage of the supercapacitor determines both, its power densityand energy density. It is dependent on the material characteristics of the electrodeand electrolyte. The operating voltage of the supercapacitor is mainly limited bythe electrolyte stability. When using aqueous-based electrolytes, such as alkalis(e.g. KOH) and acids (e.g. H2SO4), the maximum cell voltage is limited to 1.0V.[228] However, those electrolytes have the advantage of a high ionic conductivity(up to ∼1 S cm−1), low cost, a wide acceptance, and enable a higher specificcapacitance in comparison to organic and ionic liquid electrolytes.[214,217,229] Incontrast, organic electrolytes enable voltages of up to 3.5 V.[230,231] As the energydensity of a supercapacitor is proportional to the square of the operating voltage,numerous researchers are focusing on the design of highly conductive and stableelectrolytes. While organic electrolytes in acetonitrile present the state-of-the-artfor commercial supercapacitors, tremendous research efforts are made towards theapplication of ionic liquid electrolytes.[232–234] Ionic liquids are room temperatureliquid solvent-free electrolytes. Their unique properties, such as an electrochemicalstability over a wide voltage window (>3 V), a high thermal and chemical stability,and a low vapor pressure, render them promising candidates for supercapacitorelectrolytes.[194,217,224,235,236]

3.5 Evaluation of Supercapacitors by ElectrochemicalMeasurements

The performance of a supercapacitor and the respective electrode material is bestprobed by a two-electrode measurement because its test configuration closely matchesthe performance of a fully packaged supercapacitor cell.[237] Two-electrode test cellsare commercially available, but can also be easily fabricated from two current

36


collector foils packed by Kapton tape. The main performance metrics for a packagedsupercapacitor are the gravimetric and volumetric energy and power densities, thespecific capacitance of the active electrode material, and the life cycle stability.[223]

These metrics can be evaluated by either galvanostatic charging-discharging curvesor cyclic voltammetry.

3.5.1 Cyclic Voltammetry

Cyclic voltammetry (CV) is a commonly used potential-dynamic electrochemicaltechnique, which can be employed to gain insights into the characteristics of the activeelectrode material, such as the potential range of activity, capacitance, cyclability,and kinetics.[8] During the measurement, the potential of the target electrode ismeasured against the reference electrode via linearly scanning back and forth acrossa given potential window at a constant scan rate - see Figure 3.5.[215] The slope ofthe potential-time curve is called the potential scan rate and can be expressed by

ν = dV

dt. (3.13)

The potential plot can be further expressed as

V = Vi + νt, (3.14)

where Vi is the initial potential, t is the time, and ν is the potential scan rate.[215]

Common sweep rates range from 0.001 to 100 mV s−1 and common current densitiesare in the range of 0.01 to 10 mA cm−2.[8] During the potential scan, the currentpassing between the working electrode and the counter electrode is measured bythe potentiostat and is then plotted as a function of the electrode potential. Figure3.5 shows the cyclic voltammetry (CV) curve of an ideal supercapacitor featuring ahighly symmetric rectangular shape. In the case of an ideal capacitative behavior,the capacitance is constant regardless of the scan rate and can be calculated by thefollowing equation:

C = I

ν, (3.15)

where I is the measured current and ν is the applied scan rate.[238,239]

37


Figure 3.5: Cyclic voltammetry (CV) protocol for supercapacitor characterization,showing the potential-time curve (left) and the CV curve of an idealsupercapacitor featuring a highly symmetric rectangular shape (right).[60]

Nonetheless, supercapacitors feature such an ideal behavior at extremely low scanrates only. At higher scan rates, this ideal behavior is changed with a gradual loss incell-specific capacitance - see Figure 3.6 a). The decrease in specific capacitance arisesfrom the limited mass transfer kinetics within the porous electrode. In particular,the electrolyte penetration into the electrode pores is limited at higher scan rates.Therefore, the surface area of the electrode is not fully accessible by the electrolyteat higher scan rates and, thus, less surface area can be utilized for charge storageleading to lower capacitance in comparison to lower scan rates. Nevertheless, havinga high charging and discharging rate without any capacitance loss is highly desirablefor supercapacitors.[215] Other non-idealities arise from electrolyte degradation andpseudocapacitive behavior - see Figure 3.6 b) and c). Electrolyte degradation doesnot contribute to the charge storage, but significantly reduces the performance ofthe device during lifetime cycling. On the contrary, pseudocapacitive current maycontribute to the energy storage capabilities of a device, which can be utilized forpseudocapacitors.[223,240]

38


a) b) c)

Figure 3.6: Schematic representation of CV non-idealities: a) Scan rate limitation,b) electrolyte degradation occurring at a potential window which exceedsthe electrolyte stability window, and c) pseudocapacitive charge storage.

The total capacitance of the device Cdevice can be calculated by using the measuredcharge quantity Q, which is the total charge transferred during the forward orbackward charging direction in the potential range from V1 to Vcell. If V1 is theinitial cell voltage and assuming it equals to 0 and Vcell is the end potential at a fullycharged state, Cdevice can be expressed as

Cdevice = Q

Vcell

. (3.16)

Next, the charge Q can be obtained from the CV through integration, as expressedby:

Q =Vcell∫

V1

I(V ) dt =Vcell∫

V1

I(V ) ν dV, (3.17)

where ν is the scan rate, I(V ) is the current as a function of the cell voltage (V ), anddV and dt are the cell voltages and the time changes within the voltage scanning,respectively.[241]

3.5.2 Galvanostatic Cycling

The characterization via galvanostatic charging-discharging curves (CDC) is one ofthe most reliable techniques for the determination of capacitance energy density,power density, ESR, and cycle life of a supercapacitor. It is, therefore, the mostestablished measurement method in the supercapacitor industry.[215,223] Commonly,CDCs are recorded by applying a constant charging current (iapp) until the maximum

39


operating potential (Vcell) is reached. The current is then reversed in order todischarge the supercapacitor to a minimum voltage (V1). Figure 3.7 shows a typicalcharge-discharge-profile of a symmetric carbon double-layer supercapacitor.

Figure 3.7: Galvanostatic charging-discharging protocol for supercapacitor characteri-zation, showing the current-time-profile (left) and the highly symmetricaltriangular charging-discharging curve of an ideal supercapacitor (right).[60]

In general, the CDC can be divided into two parts, namely a capacitive part, whichrepresents the voltage change due to the change in energy, and a resistive part,representing the voltage change due to the ESR of the supercapacitor - see Figure3.8.[242] An ideal capacitive behavior leads to the same charge and discharge times,and a symmetrical triangular curve evolves. If the curve deviates from linearity, itis an indication of charge transfer across the electrode/electrolyte interface. Suchcharge transfer reactions can arise from both, reversible faradaic reactions within theelectrode, which contribute to pseudocapacitance and irreversible faradaic reactions,such as oxidation, and reduction of the electrolyte.[8]

40


Figure 3.8: Charging-discharging curve (CDC) of a supercapacitor illustrating thecapacitative and resistive parts of the CDC.[60]

In a similar way to CV experiments, the charge Q can be obtained through integration,as expressed by

Q =tend∫

0

I dt = iapp tend, (3.18)

where tend is the discharge time at which the supercapacitor is fully discharged.[241]

Cdevice can further be calculated from the slope of the discharge curve, as expressedby[223]

Cdevice = iapp

dV/dt. (3.19)

Moreover, the series resistance of the device can be deduced from the voltage drop(Vdrop), which occurs when the current changes during switching from charging todischarging - see Figure 3.8. For a full device, the ESR can be calculated from thevoltage drop by[223]

ESR = Vdrop

iapp

. (3.20)

Additionally, the degradation of a supercapacitor can be monitored by repeatingboth capacitance and resistance measurements over galvanostatic cycling. Typically,supercapacitors feature a much longer cycle life than batteries, exceeding over 100,000cycles compared to 1200 cycles. Commonly, prolonged charging-discharging cycling

41


leads to a gradual decrease in the capacitance and increase in the ESR of thesupercapacitor. In turn, this causes a decrease in both, the energy density and powerdensity of the supercapacitor device.[215]

3.5.3 Electrochemical Impedance Spectroscopy

Electrochemical impedance spectroscopy (EIS), also known as AC (alternatingcurrent) impedance spectroscopy, is a powerful technique for determining the ESRand capacitance of a supercapacitor. This technique utilizes very small AC amplitudes.During EIS measurements, a small AC amplitude signal is applied to a supercapacitorcell over a wide range of frequencies. Either current control (galvanostatic) or voltagecontrol (potentiostatic) mode can be applied. In voltage control mode, an alternatingpotential is applied, and the amplitude and phase shift of the resulting current ismeasured to obtain the impedance of the system.[215]

As stated in Ohm’s law (Z=R), the impedance Z corresponds to the ratio of voltageand current as expressed by

Z = V

I, (3.21)

where V is the voltage, I is the current, and Z is the impedance. The impedancecan also be expressed in an exponential form

Z = |Z| eiΦ, (3.22)

where |Z| is the modulus of the impedance expressed as

|Z| =√

Z2re + Z2

im, (3.23)

and Φ is the phase shift between input and output signal expressed as

tanΦ = Zre

Zim

. (3.24)

Here, Zre and Zim are the real and imaginary parts of Z, respectively. The impedanceof an ohmic resistance (ZR) represents a real quantity, whereas the impedance of the

42


capacitance (ZC) is an imaginary quantity.

ZC = −i1

ωC(3.25)

The angular frequency ω is expressed by 2πf, where f is the frequency in Hz. C isthe capacitance of the supercapacitor.[199,243]

Generally, the impedance is plotted in the form of Nyquist plots, where the negativeimaginary part (y-axis) is plotted versus the real part (x-axis).[243] Figure 3.9 a)shows a Nyquist plot of an ideal electrical double-layer capacitor featuring a verticalline parallel to the imaginary axis.

a) b)

Figure 3.9: Schematic representation of the Nyquist impedance plot of a) an idealcapacitor with a series resistance Rs and b) an electrochemical capacitorwith a porous electrode, featuring the equivalent series resistance (ESR)and the equivalent distributed resistance (EDR).[212]

Actually, real capacitor behavior is slightly more complex due to dispersion factors,such as electrode porosity and roughness. The AC behavior of a supercapacitorwith a porous electrode, which is shown in Figure 3.9 b), was first described byDe Levie.[244,245] In general, three parts are discernible. At very high frequencies,only the outer surface of the electrode is accessible to the electrolyte ions. Thus,the impedance of the supercapacitor is purely ohmic in this region, and the valueof the impedance is the resistance of the electrolyte outside the pores Rs. Whenthe frequency is decreased, the influence of the electrode porosity and thicknesson the ion migration rate becomes more prominent. More resistor and capacitorelements arising from the electrode pores become active. Accordingly, the Nyquistplot displays an increase in both the real and imaginary parts of the impedanceuntil a constant value of Zre is reached at the so-called knee frequency. The latter is

43

3.6 CARBON-BASED ELECTRODES FOR SUPERCAPACITORS

reached at low frequencies, where the whole surface is accessible to the electrolyte,and the capacitance and resistance have reached their maximum values.[8,199] As aconsequence, an electrochemical capacitor can be described as an ideal capacitorwith an ESR, which increases by the equivalent distributed resistance (EDR). Inpractice, the ESR is defined as the x-intercept of the real impedance, whereas theEDR can be determined by extrapolating the 45◦ region (Warburg region) back tothe x-intercept - see Figure 3.9 b).[212]

3.6 Carbon-based Electrodes for Supercapacitors

As discussed in Chapter 3.4, the capacitance and charge storage of a supercapacitorhighly depend on the electrode material. Thus, developing new materials with highcapacitance and improved performance compared to existing electrode materials isone of the major challenges in supercapacitor research. Although the capacitanceof the supercapacitor highly depends on the specific surface area of the electrodematerial, no linear increase of the capacitance with increasing specific surface area canbe found. As small pore sizes or restacking of the electrode material particle can lowerthe accessible surface area for the electrolyte, a definition called the electrochemicalactive surface area may be more accurate in describing the electrode capacitancebehavior.[217] Thus, the capacitance of a supercapacitor depends on the surfacearea accessible to the electrolyte. In general, electrode materials for supercapacitorsshould fulfill the following requirements:[214]

• high surface area,

• high electrical conductivity,

• thermodynamic stability for a large potential window of operation,

• high chemical and cycling stability,

• controllable morphology, pore size, particle size, and material distribution, and

• surface wettability.

As carbon materials are excellent conductors, chemically stable, and have a highsurface area, they are the material of choice for supercapacitors. However, carbon

44


comes in many varieties and not all are applicable as electrode materials. Table 3.1compares the properties of different carbon structures.

Table 3.1: Comparison of different carbon electrode materials for supercapacitorsincluding carbon nanotubes (CNTs), graphene, activated carbon (AC),and templated carbon. Gravimetric and volumetric capacitances werereported for aqueous electrolyte, respectively. Data is taken from Ref.[225] and [246]. Pictures are adapted from Ref. [247].

Characteristics CNTs Graphene AC Templatedcarbon

Dimensionality 1-D 2-D 3-D 3-D

Specific surface area [m2/g] 150-500 2630 1000-3500 500-3000

Electrical conductivity [S/cm] 104-105 106 0.1-1 0.3-10

Gravimetric capacitance [F/g] 50-100 100-280 <200 120-350

Volumetric capacitance [F/cm3] <60 >100 <80 <200

Cost High High Low High

Structure

3.6.1 Activated Carbon

Activated Carbon (AC) materials present the current industrial standard for super-capacitor devices due to their high surface area and low cost. Thus, most commerciallyavailable devices employ ACs as their electrode material.[248] ACs are generatedby carbonizing a solid or liquid carbon precursor at high temperatures in an inertatmosphere, followed by a selective oxidation in CO2, water vapor, or KOH to increasethe specific surface area (SSA) and pore volume. ACs can be obtained from severalprecursors, such as coconut shell, sawdust, black ash, wood, pitch, and petroleum coke.The activation process leads to a porous network with a broad distribution of poresizes consisting of micropores (<2 nm size), mesopores (2-50 nm), and macropores(>50 nm).[249,250] As a result, specific surface areas of up to 3000 m2 g−1 can beachieved after activation. Activation with KOH shows by far the best results and adouble-layer capacitance of 100-250 F g−1 in organic electrolytes and 150-400 F g−1 inaqueous electrolytes are reported.[251–254] However, ACs face several challenges, whichhinder their wide application as portable rechargeable energy storage devices with

45


nearly limitless cycle life.[255] Due to their relatively low conductivity, commercialelectrodes are limited in their thickness and usually contain conductive additives,such as carbon black, which enable a rapid electrical charge transfer from the cell.Nevertheless, these conductive fillers are low surface area materials, which do notsignificantly contribute to the capacitance of the electrode and, thus, are referred to“dead mass”. The high electrical conductivity of new carbon materials, such as carbonnanotubes and graphene, eliminate the need for conductive fillers and further allowthe increase of the electrode thickness. In turn, increasing the electrode thicknessand eliminating conductive additives leads to an increased active electrode materialto full device ratio and, thus, enables higher energy densities.[47]

3.6.2 Graphene

Graphene is a new carbon material with unique properties, including a large theo-retical specific surface area (∼2630 m2 g−1), high conductivity, good chemical andthermal stability, a wide potential window, and excellent mechanical flexibility, thatdistinguishes it from other supercapacitor electrode materials.[18,43,256,257] The intrin-sic electrochemical double-layer capacitance of single-layer graphene was measuredto be ∼21 μF cm−1.[15] Thus, a maximum theoretical electrochemical double-layercapacitance of up to ∼550 F g−1 could be achieved, if the entire surface area ofsingle-layer graphene is utilized. This sets the upper limit for all carbon materials.[258]

However, graphene’s characteristics are rather sensitive and primarily determinedby the size of the sheets, the number of layers, and the presence of defects, whichin turn depends on the production method. Moreover, the observed surface areafor various graphene-based materials is significantly lower than that of single-layergraphene, because individual graphene sheets tend to aggregate and restack duringboth, electrode manufacturing and cycling, owing to interplanar π-π interactions andvan der Waals forces between the graphene layers.[259] This agglomeration reduces thesurface area significantly, lowers the accessible surface area, and results in a decreasedelectrochemical performance. Therefore, a number of strategies have been developedto prevent aggregation of graphene sheets in order to increase the accessible surfacearea and promote the transport of the electrolyte ions, including adding surfactantsor spacers, a template-assisted growth, and crumpling the graphene sheets.[260–265]

Generally, graphene fabricated by micromechanical exfoliation or chemical vapor

46


deposition offers the best quality, because of their large crystal domains, single- andbilayer structure, and few defects in the basal plane. However, due to their highcost and low yield, graphene produced by these techniques is unsuitable for themass production of electrode materials.[260] In contrast, graphene-based electrodematerials of lower quality can be easily fabricated at low cost and high scale byreduction of GO. GO can be reduced either by chemical reduction or by physicalmethods.[20,96,101–104]

Figure 3.10: Schematic representation of a graphene-based supercapacitor featuringa stacked geometry.

Ruoff and coworkers have pioneered developing a solution-based process for the pro-duction of chemically modified graphene (CMG) by the reduction of suspended GO inwater using hydrazine hydrate.[20] Later, Stoller et al. processed the powdery CMGinto active electrodes of supercapacitors using polytetrafluoroethylene (PTFE) as apolymeric binder.[47] The resulting electrode material provided specific capacitancesof 135 and 99 F g−1 in aqueous and organic electrolytes, respectively, with a specificsurface area of 705 m2 g−1. Wang et al. further improved hydrazine reduced CMGby using gas-phase hydrazine reduction at room temperature. The CMG producedby this method had a lower degree of agglomeration and, thus, a higher capacitanceof 205 F g−1 in an aqueous electrolyte could be observed.[266] Other reducing agents,including hydroquinone, NaBH4, ethylenediamine, and strong alkalis have been usedto produce CMGs for supercapacitors applications, demonstrating capacities between100-200 F g−1 and, thus, outperform CNT-based supercapacitors.[233,267,268] Interest-ingly, activation of GO with KOH showed similar results compared to AC devices,increasing the surface area of CMG to up to 3100 m2 g−1.[55] The resulting electrodematerial showed an excellent conductivity and a capacitance of up to 170 F g−1 inorganic electrolyte, which could be retained to 97% after 10,000 cycles. However,

47


although CMGs offer promising characteristics as electrode materials for supercapaci-tors, the use of highly toxic reducing agents, such as hydrazine and dimethylhydrazine,remains a serious issue for large-scale production and application.

Physical methods for the reduction of GO include thermal reduction and laser treat-ment. In fact, thermal treatment of GO suspensions has been reported to producerGO with less agglomeration, high specific surface area, and higher electrical conduc-tivity compared to chemically rGO.[173,269] Accordingly, Rao and coworkers producedhigh surface area rGO (925 m2 g−1) through exfoliation of GO at 1050 ◦C.[270]

These samples have shown specific capacitance values reaching up to 117 F g−1.However, the high-temperature annealing process is energy consuming and difficultto control. Therefore, Jian et al. investigated the reduction process at temperaturesranging from 200-900 ◦C. Interestingly, the peak capacitance (315 F g−1 in aqueouselectrolyte) was observed at 200 ◦C, which can be explained by residual functionalgroups facilitating the penetration of the aqueous electrolyte and pseudocapacitiveeffects contributing to the specific capacitance.[271] Apart from improving the re-duction temperature, methods to achieve non-stackable rGO using curved graphenesheets rather than flat ones can further improve the specific capacitance of the devices.Fan and coworkers reported on a scalable approach for the synthesis of corrugatedgraphene sheets by thermal reduction of GO at an elevated temperature followedby rapid cooling using liquid nitrogen.[272] The porous, loose, and highly wrinkledmorphology of the produced rGO hinders restacking of the graphene sheets on theelectrode layer and therefore enables a high specific capacitance of 349 F g−1 inaqueous electrolyte.

Although high electrochemical performance can be achieved by thermally reducedgraphene, the relative harsh preparation environment and the limited amount ofproducts may restrict practical applications of the thermal reduction. In orderto further improve the energy density and the large-scale application of graphene-based supercapacitors, laser reduction of GO has been intensively studied.[57,273–276]

Gao et al. utilized a laser writing technique to convert GO directly to rGO.[273]

As a result of the decomposition of functional groups and water during the lasertreatment, the structure of laser-induced rGO is porous with gaps between thesheets, which leads to an enhanced ionic diffusion of the electrolyte. Kaner andcoworkers improved the laser reduction method by using an inexpensive commerciallyavailable LightScribe CD/DVD optical drive.[57,274] The resulting laser-scribed

48


graphene (LSG)[274] film possessed 3D open networks with a specific surface areaof 1520 m2 g−1. Moreover, they exhibited relatively high electrical conductivity(1738 S m−1) and were mechanically robust, which enabled the fabrication of flexiblesupercapacitors. The specific capacitance was reported to be 276 F g−1 in ionic liquidelectrolytes and maintained 97% of the initial capacitance after charging/dischargingfor 10,000 cycles.[277]

However, rGO electrode materials exhibit significant defects in their basal plane, in-cluding residual oxygen-containing groups, vacancy defects, edges, and deformations,which lower their conductivity and overall performance in supercapacitor devices.Moreover, temperature programmed desorption experiments indicate that the decom-position of GO starts at a relatively low temperature of 70 ◦C. Therefore, GO-basedsupercapacitors may face a temperature restriction, which requires the use of near orbelow room temperature in order to avoid serious stability problems.[278,279] In lightof the limitations of GO-based supercapacitors, first studies have been conducted toutilize liquid-phase exfoliated graphene.

An et al., for example, presented a scalable and facile approach for noncovalentfunctionalization of graphene with 1-pyrenecarboxylic acid, which directly exfoliatesfew-layer graphene flakes into stable aqueous dispersions. Graphene suspensionswere vacuum-filtered through nano-porous membranes, leaving thin graphene filmson the membrane, which were used to fabricate supercapacitors. Specific capacitancevalues as high as 120 F g−1 with a high power density of (∼105 kW kg−1) andenergy density (∼9.2 Wh kg−1) were obtained.[109] Lee and coworkers used thesame approach testing 9-anthracenecarboxylic acid as a stabilizer, reporting slightlyhigher capacitances of 148 F g−1 in aqueous electrolyte.[154] More recently, Li andcoworkers employed imidazole as an exfoliant for graphene in aqueous medium, whichled to significantly higher graphene concentrations (1 mg mL−1).[280] The producedgraphene paper electrode showed good electrical conductivity of 131 S cm−1 andhigh area capacitance of 71.9 mF cm−1, which exceeds that of laser-scribed rGO bya factor of 10. However, only moderate specific capacitances of 20 to 50 F g−1 couldbe observed in aqueous electrolyte.

Although liquid-phase exfoliation of graphite is expected to be the production ofchoice for obtaining defect-free graphene sheets at large scale and low cost, only afew papers have been focusing on liquid-phase exfoliation of graphite for applicationsin supercapacitors. The use of liquid-phase exfoliated graphene in supercapacitors

49


is still hindered by several factors. One key challenge is the production of highlyconcentrated graphene dispersions providing a large amount of graphene electrodematerial, which is required for electrode manufacturing. Moreover, only a smallfraction of single-layer graphene is obtained by most liquid-phase exfoliation methods,leading to low specific surface areas and, thus, insufficient capacitance values. In orderto fully employ liquid-phase exfoliated graphene in supercapacitors, the discovery ofa low-cost, commercially available, and effective exfoliant and a proper exfoliationtechnique is of paramount importance. In fact, this would enable a scalable productionof non-oxidized graphene in high yield and high quality, which can then be easilyused as an active material in future supercapacitor devices.

50

II

Objective

51

OBJECTIVE

Graphene is a promising candidate for next-generation supercapacitor electrodes,owing to its low mass density, excellent electronic conductivity, mechanical flexibility,and high surface area.[225,281–283] However, the major drawback for the techno-logical exploitation of graphene in supercapacitor devices is the lack of scalable,high throughput, eco-friendly, and low-cost production routes towards high-qualitygraphene. Recently, considerable efforts have been made on developing new syntheticroutes towards single- to multilayer graphene.[111,112]

Most of them rely on the reduction of graphite oxide, which is soluble in aqueousmedia.[101,102] Nevertheless, the reduction of graphite oxide is not quantitative andresults in irreversible defects in the basal plane of the graphene sheets produced,which significantly decreases graphene’s electrical conductivity. This, in turn, impairsthe electrochemical performance of the supercapacitor device, especially in terms ofpower density.Recently, the liquid-phase exfoliation of graphite in solvents with appropriate surfaceenergies, such as NMP or DMG, has been demonstrated.[21] However, the hightoxicity and the high boiling points of the solvents hinder their application in super-capacitor electrodes.Other approaches focus on the covalent and noncovalent functionalization of graphenewith different molecules, which minimize interplate adhesion in aqueous media.[109,112]

Although these approaches enable the scalable production of non-oxidized high-qualitygraphene, they still face several challenges. One key challenge is the production ofhighly concentrated graphene dispersions, which is restricted by the limited disper-sion efficiencies of currently used surfactants. Moreover, low yields in single-layergraphene hinder their use in applications where high surface areas are required, suchas supercapacitors. However, large surface area few-layer graphene samples with anincreased fraction of single-layer graphene are strongly needed in order to enable highenergy density devices, which can bridge the gap between batteries and conventionalcapacitors. Conclusively, the development of new efficient stabilization approaches,which involve the use of low-cost, commercially available, and eco-friendly stabiliz-ers or low-toxicity solvents, are of paramount importance to improve exfoliation,especially towards single-layer graphene. In fact, this would pave the way towardsnext-generation graphene-based supercapacitors.

Accordingly, the objective of this study is to develop scalable stabilization andexfoliation approaches for the preparation of high-quality few-layer graphene dis-persions with large amounts of single-layer graphene in liquid media. Main aspects

52

OBJECTIVE

are the noncovalent functionalization with industrial low-cost stabilizers and thesolvent-based liquid-phase exfoliation in low-toxic volatile solvents. Importantly, theprepared graphene dispersions should be suitable for the application in supercapacitorelectrodes.

In order to yield highly stable graphene dispersions from liquid-phase exfoliation ofgraphite two different aspects have to be considered:

Firstly, the exfoliation of the graphene sheets has to be achieved by overcomingthe strong interlayer attractive forces, which can be accomplished by meansof ultrasonication. As there are no standard procedures available for the exfoliationof graphite by ultrasonication, and the reproducibility of ultrasonication is stronglyaffected by several parameters, a reproducible exfoliation procedure is establishedin the first place. Moreover, the procedure is verified by testing its dispersion andexfoliation capability by means of a commonly used graphene stabilizer (CTAB),which further allows a comparison to previous studies. These investigations buildthe first part of the study (Chapter 4.1).

Secondly, another challenging task regarding the production of graphene in liquidmedia is to efficiently reduce the aggregation of exfoliated graphene sheetsby restacking and, thus, preserving the exfoliation state of the sample. This can beaccomplished by the use of either different stabilizers, which prevent reaggregationof graphene sheets by steric or electrostatic repulsion, or by organic solvents, whichmatch the surface tension of graphene.[111]

Hence, the second part (Chapter 4.2) focuses on the investigation of the suitabilityof different surfactants for graphene production in aqueous medium. In particular,amphiphilic pyrene, naphthalene, and perylene derivatives are studied in regard totheir graphene dispersion and exfoliation efficiency. The dispersion, stabilization,and exfoliation ability of the different stabilizers are compared on the foundationof absorbance spectroscopy, emission spectroscopy, statistical Raman analysis, andtransmission electron microscopy. Since the surfactant concentration is expected tohave a strong impact on the graphene dispersibility, the surfactant concentration ofthe best stabilizer is further optimized.

In the third part (Chapter 4.3), the stability of the graphene dispersion is addressedby studying a new exfoliation approach, which involves the use of tetraalkylammo-nium ions as small intercalant additionally to the dispersant. Firstly, the graphene

53

OBJECTIVE

exfoliation and stabilization mechanism of this approach is investigated by conductinga comparative study of several parameters. Secondly, the approach is transferredto several surfactants, which feature different molecular structures, and the effecton the graphene dispersion efficiency is analyzed by means of absorbance spec-troscopy. Thirdly, the exfoliation efficiency of different intercalant-surfactant systemsis compared on the foundation of statistical Raman analysis, transmission electronmicroscopy, and atomic force microscopy measurements. On the basis of thesefindings, an in situ polymerization approach of partially intercalated graphite isstudied towards its applicability for graphite exfoliation and dispersion.

Turning to solvent-based liquid-phase exfoliation in the fourth part (Chapter 4.4),the suitability of different organic solvents for graphene production in liquid-phaseis considered. Next, a cosolvency approach is tested towards its applicability forgraphene dispersion, stabilization, and exfoliation. On the basis of the findings inPart 4.3, the pretreatment approach is finally transferred to the concept of cosolvencyand the dispersion and exfoliation efficiency of this approach is studied by meansof absorbance spectroscopy, statistical Raman analysis, and transmission electronmicroscopy.

In the last part of this work, the prepared liquid-phase exfoliated graphene (LPEG)dispersions are tested towards their applicability for electrode materials in super-capacitor devices. In this regard, supercapacitor devices of LPEG electrodes are builtand their electrochemical performance is studied in different electrolytes by galvano-static charging-discharging experiments, cyclic voltammetry, and electrochemicalimpedance spectroscopy.

54

III

Results and Discussion

55

4 Liquid-Phase Exfoliation ofGraphite in Different Media

In order to yield highly stable graphene dispersions by means of liquid-phase exfolia-tion processes, two different aspects have to be considered. Firstly, the exfoliationof graphene sheets by overcoming the strong interlayer attractive forces has to beachieved. This can be accomplished by means of ultrasonication. Exfoliation ofgraphene via ultrasonication relies on the effect of acoustic cavitation of high-frequencyultrasound, which results in the formation, growth, and collapse of microbubblesin solution. This induces shock waves on the surface of the bulk material, whicheventually causes exfoliation. This technique has been shown to play a crucial rolein the liquid-phase exfoliation of graphite and is widely used for the preparation ofgraphene dispersions in different media.[127,159,284,285]

Another challenging task regarding the production of graphene dispersions is toreduce the aggregation of exfoliated graphene sheets by restacking. This can beaccomplished by either the usage of solvents, which match the surface tension ofgraphene, or by the usage of different surfactants, which prevent the reaggregationof the graphene sheets. Surfactant-assisted liquid-phase exfoliation relies on thestabilization of graphene flakes upon noncovalent methodologies, such as π-π stackingand/or electrostatic interactions.[143,286–288] Selecting or designing proper surfactantsand solvents for the efficient stabilization of graphene in different solvent media is acrucial task in the large-scale production of graphene dispersions.

4.1 Establishment of a Dispersion Procedure

Despite the fact that ultrasonication is universally employed for the exfoliation ofgraphite into single-layer graphene (SLG) and few-layer graphene (FLG), there areno standard procedures available. Instead, different groups apply different sonicationtreatments to their samples as indicated in Table 4.1. This renders the comparisonof different stabilization methodologies a challenging task, as small differences inthe preparation procedure, such as sonication power, temperature, and precipitationconditions, tremendously affect the dispersion efficiency.

56

4.1 ESTABLISHMENT OF A DISPERSION PROCEDURE

Therefore, the first section focuses on the investigation of different ultrasonicationparameters, such as sonication time, temperature, and applied power, searching foroptimum conditions to produce highly exfoliated graphene dispersions. In particular,a protocol should be established, which allows a controlled and reproducible exfo-liation of graphite by means of ultrasonication and, thus, enabling a quantitativecomparison of different stabilization routes.As this thesis targets the evaluation of different stabilization routes towards stablefew-layer graphene dispersions, the main focus of the ultrasonication procedure isreproducibility and material quality rather than the final concentration. It is alreadyknown from the literature that highly concentrated graphene dispersions can beproduced by prolonged ultrasonication of up to 430 h.[123] However, the processlacks reproducibility and the produced flakes are very small and partially damaged.Still, after the quantitative analysis of suitable stabilization methodologies, the bestsurfactant-solvent-system can be transferred to other large-scale dispersion techniquesto yield higher graphene concentrations and larger batch sizes. However, this isbeyond the scope of this thesis.

Table 4.1: Experimental conditions and results reported in previous studies on liquid-phase exfoliation of graphite by either bath sonicator (BS) or tip sonicator(TS). The surfactants are abbreviated as P123 (Pluronic P-123) and SC(sodium cholate). The graphene concentration after centrifugation isabbreviated as cac.

Sonicator type Surfactant Sonication Centrifugation cac / mg mL−1 Ref.

BS P123 6.7 h 5 min at 5000 g 1.0 [119]BS (80 W) SC 430 h 90 min at 1500 rpm 0.3 [123]BS (80 W) SC 430 h 90 min at 5000 rpm 0.05 [123]TS (50 W) SC 1 h 60 min at 21,130 g 0.09 [108]

4.1.1 Optimization of the Dispersion Conditions

In order to optimize the ultrasonication parameters, we employed two differentsurfactant-water-systems. On one hand, hexadecyltrimethylammonium bromide(CTAB) was chosen as a reference surfactant, as it has been shown by several groupsto yield highly stable graphene dispersions and, therefore, enables the comparison toprevious works concerning graphene dispersibility and exfoliation.[119,120,122,127–129]

57


On the other hand, 1-pyrenebutyric acid methoxypolyethylene glycol ester (1) wasused in a graphite-tetraethylammonium hydroxide solution, as this system provedto be promising in terms of graphene stabilization in former experiments - see alsoSection 4.2.

Initially, ultrasonication was applied to exfoliate graphene flakes yielding stablegraphene dispersions with the help of either CTAB (G-CTAB) or 1 (G-1). Asa next step, the dispersions were centrifuged in order to remove any poorly dis-persed graphitic material, and the supernatant after centrifugation containing sta-ble dispersed graphene material was subjected to further analysis. To provide aquantitative comparison of the different parameters in terms of graphene dispersibility,the absorbance of the supernatant was measured at a specific wavelength (660 nm).The corresponding graphene concentration after centrifugation cac was calculatedusing the extinction coefficient for graphene dispersions in surfactant-water solutionsε=1390 mL mg−1 m−1 at 660 nm.[106]

Figure 4.1: Structure of the amphiphilic pyrene derivative 1-pyrenebutyric acidmethoxypolyethylene glycol ester (1). The structure of CTAB is shownin Figure 2.6.

While bath sonication usually has a low level of ultrasonic power, tip sonicators areavailable in numerous ranges of power densities ranging from 50 to 1000 Watt. Inthis study, a low-level tip sonicator with a power output of 50 Watt was applied tominimize defects and/or damage induced during cavitation. Although low energybath sonication is known to yield graphene dispersions of higher concentrations evenin comparison to high power tip sonicators,[122] we decided to use a tip sonicator forfurther sample preparation as it offers better reproducibility as shown in Figure 4.2 a).Additionally, the reproducibility of bath sonication is strongly affected by smallchanges in the sonication probe positioning and the water volume. Using a tipsonicator, the probe to vial ratio can be optimized, and the energy input can becontrolled more easily.Moreover, the sonication temperature can be easily controlled, which is in fact of high

58


importance as the cavitation is strongly influenced by the macroscopic temperatureof the bulk solution. In particular, more microbubbles are generated at a highertemperature, which hinders the propagation of ultrasound waves and, thus, maylower overall exfoliation and dispersion efficiency.[113]

Therefore, the ultrasonication vial was equipped with a cooling jacket, and a coolingunit was installed. Figure 4.2 b) shows the dependence of the graphene concentrationafter centrifugation on the temperature (Tsonic), which peaks at a temperatureof 17 ◦C. This is in accordance with previous reports, which report a decrease ingraphene concentration at Tsonic>25 ◦C.[126] Thus, we assume that at Tsonic=17 ◦C theadsorption of CTAB on the graphene surface is improved compared to Tsonic=5 ◦C,which leads to a better dispersibility. However, at Tsonic=27 ◦C the effect of thetemperature on the cavitation may play a key role as the intensity of ultrasounddramatically decreases with increasing temperature resulting in lower levels of exfolia-tion. Consequently, subsequent graphene dispersions were performed at a constanttemperature of 17 ◦C. Additionally, we used a magnetic stir bar in combination witha magnetic stirrer to stir the solution continuously during ultrasonication. Thisshould prevent the formation of any local high energy spots and, thus, enables ahomogeneous exfoliation process.

a) b)

Figure 4.2: Calculated graphene concentration remaining in the supernatant of G-CTAB after centrifugation cac a) for a tip sonicated and bath sonicatedsample and b) as a function of the temperature Tsonic. The error barscorrespond to the standard deviation of the average values obtained byperforming three independent experiments.

Next, we analyzed the effect of the power output of the tip sonicator on the dis-persibility of graphene and the defects present in the sample. In general, increasingultrasound power intensity leads to a larger acoustic amplitude and collapse pres-sure and, thus, to faster and more violent bubble collapse.[289] In order to prevent

59


overheating and the formation of local high energy spots within the sample, whichmay reduce the overall efficiency of the dispersion process, the cycle was set to 0.5.Figure 4.3 a) shows the concentration after centrifugation of G-CTAB and G-1 as afunction of power output for a fixed ultrasonication time of 2 h. The concentrationof G-CTAB and G-1 reach a maximum for a power output of 60%. Indeed, thisvalue is surprising as an increase in the graphene concentration with increasingenergy was expected. One explanation for this phenomenon may be that a too highultrasonication power may destroy aggregate structures of the surfactant, whichare adsorbed on the graphene basal plane and, thus, leads to degradation of theexfoliation efficiency.

a) b)

Figure 4.3: a) Calculated graphene concentration after centrifugation cac of G-CTABand G-1 dispersions as a function of sonicator power output. b) ID/IG

ratio of G-CTAB and G-1 dispersions as a function of sonicator poweroutput.

Moreover, ultrasonication not only affects the exfoliation of graphite to graphene, butit also induces defects on the graphene skeleton and promotes scission of the sheetsresulting in smaller flake sizes.[290] In this respect, Raman spectroscopy was employedto gain further insights in any defects present in the G-CTAB and G-1 hybrids.In general, three peaks have to be considered: the G-band at about 1582 cm−1,which arises from the in-plane vibration of sp2 carbons and which is typical for allgraphitic materials; the disorder-related D-band, which comes from the activation inthe first-order scattering process of sp3 carbons at about 1351 cm−1; as well as the D‘peak, which appears as a high-frequency shoulder to the G-band.[291] The extent ofdefects within a sample is commonly monitored via the intensity ratio of the D- andG-bands, ID/IG, which expresses the sp3/sp2 carbon ratio.[191] Moreover, the ID/ID‘

ratio can be used to determine the nature of defects in graphene. One can divide

60


such defects into three main types: sp3 defects (ID/ID‘∼13), vacancy-like defects(ID/ID‘∼7), and defects arising from the boundaries in graphite (ID/ID‘∼3.5).[192]

The introduction of edge defects is unavoidable, as ultrasonication cuts the initiallylarge graphite crystallites up into smaller flakes. These small flakes have more edgesper unit mass, resulting in an increase in edge defect population and, thus, an increaseof the ID/IG ratio.[191]

Figure 4.3 b) shows the ID/IG ratio of G-CTAB and G-1 dispersions dip coatedon Si/SiO2 wafers as a function of power output, respectively. It is observed, thatthe ID/IG ratio strongly increases with increasing power output for both surfactant-graphene systems. This indicates a substantial increase in the defect density. Notably,the mean ID/ID‘ ratio was estimated to be 3.98 for G-1. Conclusively, the resultingdisorder can be predominately assigned to edge related defects induced by the scissionof graphene flakes due to ultrasonic treatment. Thus, an increase in power outputseems to result in a nearly linear decrease of the average graphene lateral sizes, whichis in accordance with previous findings from Khan et al.[159] Hence, we chose toprepare all subsequent surfactant tests with a power output of 60% and a cycle of0.5, as it presents a fair compromise between the scission of the graphene flakes andoverall dispersibility in terms of graphene concentration.

Figure 4.4: Calculated concentration after centrifugation cac for CTAB and G-1 asa function of ultrasonication time, respectively.

Moreover, the concentrations of dispersed graphene after centrifugation of G-CTAB

61


and G-1 dispersions were plotted as a function of ultrasonication time (tsonic) - seeFigure 4.4. In both cases, the concentration increases steadily with tsonic and reachesafter 6 h values of 0.045 and 0.036mg mL−1 for G-CTAB and G-1, respectively.Hence, we decided to perform subsequent surfactant tests with tsonic=6 h, as a certainamount of material is required for a quantitative analysis of the material quality andsubsequent application tests in supercapacitors.

In a next step, the effect of the centrifugation rate on the graphene concentrationand defect density was studied. To this end, the as-prepared graphene dispersionswere centrifuged at different conditions. The concentration after centrifugation atvarious rates (ω) is summarized in Table 4.2. As expected, a steady decrease, from0.027 mg mL−1 at 500 rpm to about 0.013 mg mL−1 at 15,000 rpm, is observed.This is due to the fact that milder centrifugation usually does not completely removelarge flakes and aggregates.

Moreover, it was shown that high rpm rates lead to lower standard deviations and,therefore, reproducibility of the samples is strongly improved. Notably, no cleartrend could be observed from the ID/IG ratio. As reproducibility is our primary goalin order to carry out a quantitative and reliable analysis of different stabilization andexfoliation approaches, we decided to centrifuge subsequent samples at 15,000 rpmfor 10 min.

Table 4.2: Concentration after centrifugation cac, standard deviation (SD), and ID/IG

ratio for different centrifugation conditions applied to a G-1 dispersion.

ω / rpm Time / min cac / mg mL−1 SD ID/IG

500 90 0.027 0.009 1.822500 30 0.025 0.003 2.035000 20 0.023 0.003 1.5015,000 10 0.013 0.0005 1.62

In summary, we established a highly reproducible protocol for the preparationof graphene dispersions by means of ultrasonication with optimized sonicationpower output, sonication time, temperature, and centrifugation conditions - seeTable 4.3. All further tests will be performed according to this protocol, as notstated otherwise.

62


Table 4.3: Optimized parameters for the ultrasonication-assisted liquid-phase exfolia-tion of graphite.

Paramater Value

Sonication power 60% (cycle 0.5)Sonication time 6 hTemperature 17 ◦CCentrifugation 10 min at 15,000 rpm

4.1.2 Characterization of Graphene Dispersions inHexadecyltrimethylammonium Bromide as an InternalReference

In order to allow a more detailed comparison of our preparation procedure with theliterature and to set-up an internal benchmark for graphene dispersion, 1.25 mg mL−1

of CTAB was mixed with 5 mg mL−1 of graphite in water followed by sonicationand centrifugation according to the aforementioned conditions (see Chapter 4.1.1).Afterward, the dispersion was analyzed in terms of concentration, exfoliation degree,flake size, and defect content. Please note that the CTAB concentration is aboveits critical micelle concentration. However, former experiments revealed maximumgraphene dispersibility at this concentration. In order to prevent micelle formation,CTAB was added gradually to the graphite-water suspension, allowing the CTABmolecules to bind subsequently to the graphite surface.

To quantify the dispersion efficiency, the optical absorbance of G-CTAB was mea-sured in the ultraviolet and visible region of the electromagnetic spectrum - seeFigure 4.5. The G-CTAB dispersion displays a sharp peak at ∼266 nm, arisingfrom the π→π∗ transition of aromatic C-C bonds, which are commonly observedin graphitic materials.[292] For the visible region, the spectrum is featureless witha monotonic decrease in intensity at increasing wavelength - see Figure 4.5. Thegraphene concentration after centrifugation was calculated from three samples and av-eraged obtaining a final concentration of 0.047 mg mL−1. This is in good accordancewith previous studies reporting concentrations of ∼0.02 mg mL−1.[119,120] Moreover,the stability of the dispersion was measured by recording the absorption spectra (andso the concentration remaining dispersed) 4 weeks after sample preparation. Note-worthy, the prepared dispersions possessed superior stability compared to those from

63


literature.[120] In detail, only negligible amounts of 0.53% graphene have sedimentedfrom the dispersion over one month, indicating that the dispersions are highly stableover time.

Figure 4.5: Optical absorbance spectrum of G-CTAB diluted by a factor of 10. Inset:Photograph of the as-prepared and centrifuged G-CTAB dispersion.

Initial insights into the graphite exfoliation were gained from Raman spectroscopyfollowing laser excitation at 532 nm after dip coating the dispersion onto a Si/SiO2

wafer. Raman spectroscopy is a valuable tool to characterize graphitic materialsand to distinguish between single-, bi-, and few-layered graphene, turbostraticgraphite, and bulk graphite. Figure 4.6 shows two representative Raman spectra ofG-CTAB with D-, G-, D’, and 2D-bands located at 1346, 1587, 1628, and 2732 cm−1,respectively. The ID/IG ratio is clearly increased compared to the starting material,which does not have any D-band - see Figure A.2 in the Appendix. The ID/ID′

in G-CTAB is ∼3.1, which testifies that the resulting disorder is mainly edgerelated defects, resulting from flake size reduction due to ultrasonication treatment.Additionally, the 2D-band gives valuable insights into the exfoliation state of thegraphene sample. In detail, the full width at half maximum (FWHM) of the 2D-band is used to distinguish between graphite and multilayer graphene (>66 cm−1),few-layered/turbostratic graphite (65-40 cm−1), and SLG (<40 cm−1).[190,291,293,294]

Further, the I2D/IG ratio is used to determine the degree of exfoliation. For bulkgraphite, the I2D/IG ratio is ∼0.4 (Figure A.2 in the Appendix). Moreover, a ratio

64


of 0.7 or less is attributed to few- to multilayered graphite.[295] For the given spectraof G-CTAB, the 2D-bands converges to a single and highly symmetric Lorentzianpeak with an FWHM of 52 and 43 cm−1, respectively. Moreover, the 2D-band isenlarged compared to the graphite starting material featuring an I2D/IG ratio of 0.89and 2.04, respectively. These findings clearly reveal the enhanced exfoliation state ofthe G-CTAB sample down to FLG.

Figure 4.6: Representative solid-state Raman spectra of G-CTAB, indicating theexfoliation state of the sample. The samples were dip coated fromdispersion onto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting ofthe 2D-band. Black dots - experimental data; blue curve - fitting.

In order to gain statistical insights, a Raman map with around 2800 spectra was

65


evaluated - Figure 4.7. The data further reveal the high exfoliation degree of thesample, displaying a distribution of 2.5% multilayered/bulk graphite and about 97.5%few-layered/turbostratic graphene. Furthermore, the FWHM of the 2D-band wasexamined. Noteworthy, none of these spectra features a 2D-FWHM <40 cm−1, whichleads to the conclusion that no SLG was probed. However, the samples were preparedby dip coating the G-CTAB sample onto the corresponding Si/SiO2 wafers, whichmay lead to reaggregation of previously exfoliated graphene sheets during depositionand, thus, SLG is unlikely to be found. Moreover, the 2D-FWHM band is enlargedby defects, edges, and dopants, which may further lead to an underestimation ofthe single-layer content.[156] Nevertheless, the high amount of turbostratic grapheneis indicative for a successful exfoliation. Moreover, only negligible amounts ofmultilayer/bulk graphite were found, which, in fact, reveals the overall efficiency ofthe exfoliation procedure towards quantitative amounts of FLG.

Figure 4.7: Histogram resulting from the Raman mapping of G-CTAB, showingrelative counts versus I2D/IG ratio and the corresponding log-normaldistribution. The sample was dip coated from dispersion onto Si/SiO2wafer and excited at 532 nm.

Along the same lines, atomic force microscopy (AFM) images reveal the presenceof highly exfoliated graphene flakes with heights as low as 2-3 nm, which tend tostack on the Si/SiO2 wafer in close contact with each other - see Figure 4.8. Pleasenote that the laser spot of the Raman instrument is ∼1 μm and, as such, it becomesapparent that several flakes are probed for one spectral acquisition, and the presenceof SLG is unlikely to be determined.

66


Figure 4.8: Atomic force microscopy image (left) of G-CTAB hybrids from a dipcoated and washed Si/SiO2 wafer and the corresponding height profiles(right).

Furthermore, transmission electron microscopy (TEM) images of G-CTAB weretaken in order to estimate the graphene flake size. Figure 4.9 shows representativeTEM images of G-CTAB. The observed flakes have lateral dimensions in the orderof 500 nm up to 4 μm. The smaller flakes are flat while the larger ones tend tobe folded and restacked. Until now it is not clear if the restacking takes placein dispersion or is favored during the solvent evaporation throughout the samplepreparation. Notably, these findings differ from AFM investigations, which revealflake sizes <500 nm. However, flakes with very small lateral dimensions can get lostduring sample preparation, as holey carbon grids are used as a sample support forTEM measurements.

Nevertheless, we proved that our exfoliation procedure results in highly concen-trated graphene dispersions with quantitative amounts of FLG. Regarding grapheneconcentration, our process is comparable with the ones reported in the literatureusing CTAB as surfactant and even outperforms them in terms of long-term stabil-ity.[119,120] However, the comparison in terms of exfoliation degree is more challengingas most reports do not provide any statistical Raman analysis. Nevertheless, theyreport FLG dispersions with roughly 60-85% FLG (AFM analysis of less than 50flakes) and flake sizes varying from 30-750 μm (AFM and TEM analysis).[119,120,127]

Conclusively, our procedure proved to be superior in terms of exfoliation, flake size,and overall graphene dispersibility. Taking this into account, we are able to use thisprocedure to gain deeper insights into the dispersibility and the exfoliation of FLGby means of different stabilization and exfoliation approaches.

67

4.2 GRAPHENE DISPERSION BY AROMATIC AMPHIPHILES

Figure 4.9: Representative TEM micrographs of the G-CTAB sample with differentmagnifications. The sample was drop coated on a lacey carbon grid.

4.2 Graphene Dispersion by Aromatic Amphiphiles

Recent publications have established that amphiphilic aromatic molecules, such aspyrene and perylene derivatives, can noncovalently anchor onto the hydrophobicgraphitic surface of carbon nanotubes (CNTs) and graphene, stabilizing them inaqueous medium.[109,147,149,151,296,297] Such a supramolecular approach is based onthe strong and specific π-π interactions of aromatic surfactants with the graphenebasal plane, which are favorable over non-specific hydrophobic interactions utilizedby most classical surfactants. Moreover, hydrophilic head groups, which are attachedto the aromatic backbone of the stabilizer, provide water solubility and, thus, enablethe dispersion and stabilization of graphene in aqueous medium.[117] However, thetedious synthesis procedures of the molecules used for fundamental research renderthem as expensive fine chemicals that are only available in small amounts. Moreover,the small hydrophilic domains of these derivatives demonstrated to be insufficientin preventing the graphene sheets from reagglomeration and restacking through,for example, van der Waals interactions.[137] Hence, overall graphene concentration,dispersion stability, and yield of single-layer graphene are relatively low. Thus, thesestabilizers are not suitable for the large-scale production of graphene by liquid-phaseexfoliation.

In this context, this section investigates the suitability of different industriallyavailable aromatic amphiphiles with respect to graphene dispersion and exfoliation

68


in aqueous medium. In addition, aromatic amphiphiles with bulky hydrophilicparts were in the focus of research, as those molecules should enable more stablegraphene dispersions - see Figure 4.10. Notably, most of these stabilizers are availablein large amounts and at low cost. However, they are expected to be not free ofcontaminations, as elaborated purification is not commonly applied in industriallysurfactant manufacturing. As such, pure samples of the respective molecules may evenenhance dispersibility and exfoliation of graphene in aqueous medium. Nevertheless,we were aiming for an explicit statement about those materials, which will beemployed in industrial scale later on.

In particular, the dispersion of graphene by the amphiphilic pyrene 1-pyrenebutyricacid methoxypolyethylene glycol ester (1) was investigated featuring two key points -see Figure 4.1. Firstly, the aromatic motif of the new surfactant consists of 1-pyrene-butyric acid, which is known to interact strongly with the graphene surface.[141]

Secondly, this moiety is covalently linked to a hydrophilic poly(ethylene glycol)methyl ether group, which is expected to protrude away from the graphene surfaceand, thus, increases hydrophilicity in order to disperse the graphene sheets in aqueousmedia.Moreover, several naphthalene derivatives (2-7) were studied as theoretical calcu-lations predict strong interactions with the graphene surface via π-π interactions -see Figure 4.10.[298–300] Based on such interactions, naphthalene derivatives showedpromising results in the stabilization of CNTs.[301–304] Additionally, naphthalenederivatives are commonly used in industrial surfactant design as naphthalene precur-sors are cheap and easily available in ton scale. Interestingly, naphthalene derivativesare only rarely applied for the dispersion of graphene in water in comparison to otheraromatic building blocks.[155,305]

In detail, 6-amino-4-hydroxy-2-naphthalenesulfonic acid (4) has been employed forgrinding-assisted liquid-phase exfoliation of graphite recently.[155] In order to furtherinvestigate the dispersion and exfoliation efficiency of 4 in ultrasound-assisted liquid-phase exfoliation we included this molecule in our study. Another study reports on theliquid-phase exfoliation of graphite by means of 1,4,5,8-naphthalenetetracarboxylicdianhydride.[305] To this end, we investigated 6 featuring a sterically more demandinghydrophilic moiety, which consists of a poly(propylene glycol)-block-poly(ethyleneglycol) block copolymer.

69


Figure 4.10: Structures of aromatic amphiphiles, which are employed for the nonco-valent stabilization of graphene in aqueous medium. The structure of 1is shown in Figure 4.1.

Moreover, 2 and 3 present representative naphthalene derivatives of small molecularweight with different bulky side chains. In comparison, 5 and 7 were chosen as repre-

70


sentatives of high molecular weight naphthalene derivatives (>1000 Da) featuring twodifferent molecular designs. The molecular structure of 5 is similar to 1, featuringa single naphthalene core as aromatic anchor attached to the hydrophilic stericallydemanding poly(ethylene glycol) side chains, which enable the interaction with theaqueous medium. Contrary, 7 can interact with the graphene surface through variousnaphthalene units enabling strong π-π stacking, while hydrophilic interaction withthe solvent is achieved by sulfonic acid groups, which further promote electrostaticrepulsion.Finally, 8 was tested as a representative for a sterically demanding perylene deriva-tive, as perylenes showed promising results in the dispersion of graphene andCNTs.[117,149,151]

4.2.1 Overview of Graphene Dispersability

In a typical graphene dispersion experiment, defined quantities of the respectivesurfactant (1-8) were mixed with graphite in water and stirred overnight. Thisprocess should ensure that the aromatic derivatives either anchor onto the graphitesurface or partially insert into small gaps between the edges of the graphene sheets,which may open up during sonication. Afterward, ultrasonication was applied in orderto yield highly stable graphene dispersions followed by centrifugation to remove anypoorly dispersed graphitic material. The supernatant after centrifugation, containingthe stably dispersed graphene material, was then subjected to further analysis. Theinitial graphite concentration (5 mg mL−1) was chosen to be rather low in orderto prevent reagglomeration and to achieve high yields in relation to the initialgraphite concentration. For direct comparison of the dispersion efficiency of thedifferent aromatic stabilizers, all dispersions were prepared with an identical surfactantconcentration of 1.25 mg mL−1. As such, this also enables direct comparison with thereference surfactant CTAB - see Section 4.1.2. Another advantage of using such alow surfactant concentration is that there will be less surfactant to remove when thedispersion is further processed e.g. in films for supercapacitor electrodes. However,it is suggested that the dispersed graphene concentration depends critically on thesurfactant concentration. In order to yield higher final graphene concentrations, thesurfactant concentration of the best graphene-surfactant system was optimized lateron.

71


Notably, graphite/graphene was not dispersible in aqueous solutions of either methoxy-poly(ethylene glycol), poly(ethylene glycol), or poly(propylene glycol)-block-poly-(ethylene glycol) block copolymer, revealing the importance of an aromatic anchorunit to interact with the graphene surface. Furthermore, first dispersion tests inaqueous medium indicated that 6 and 8 are also not capable of dispersing graphene.Although perylene and naphthalene diimides are known to bind to the curved CNTsurface via strong π-π interactions,[117,303] we assume that binding to the planargraphene surface is less favored. In fact, the polar oxygen atoms of the imide groupswould be in close contact with the hydrophobic graphene surface upon adsorption ofthe aromatic anchor unit, which may weaken the hydrophobic interactions and may,thus, affect the overall exfoliation efficiency. Additionally, 3 was not effective in thedispersion of graphite towards few-layer graphene dispersions. Moreover, 5 did notlead to stable graphene dispersions in aqueous medium by ultrasonication-assistedexfoliation of pure graphite. However, in combination with an auxiliary step forthe pretreatment of graphite, we succeeded to prepare highly stable and concen-trated dispersions of graphene using 5 - as discussed in Chapter 4.3. Consequently,we excluded 3, 5, 6, and 8 from the study on graphene stabilization by aromaticamphiphiles.

To quantify the dispersion efficiency of the different aromatic derivatives, the op-tical absorbance of the graphene suspensions was measured at 660 nm and theconcentration after centrifugation was calculated. Additionally, long-term stabilitywas determined by measuring the absorption (and, thus, the concentration) of thedispersions after storing them for 4 weeks at room temperature in a sealed tube.Figure 4.11 shows the calculated graphene concentration after centrifugation of theas-prepared dispersion and 4 weeks after preparation for G-1, G-2, G-4, and G-7,respectively.

Obviously, 7 is the most effective dispersing agent, which yields the highest grapheneconcentration of 0.058 mg mL−1, followed by 1 (0.024 mg mL−1). 4 is the least effec-tive surfactant for graphene dispersion, revealing concentrations of about0.019 mg mL−1, which are in good accordance with graphene concentrations re-ported for grinding-assisted liquid-phase exfoliation of graphite with the help of4.[155] We attribute this to the fact that 4 possesses relatively small side chainslinked to the naphthalene aromatic anchor, which might be unsuitable for effectivestabilization of exfoliated graphene sheets through steric repulsion. However, the

72


overall dispersibility is comparable with well-established pyrene derivatives, suchas trimethyl-(2-oxo-2-pyrene-1-yl-ethyl)-ammonium bromide and 1-aminomethylpyrene.[155]

Figure 4.11: Calculated graphene concentration cac directly after preparation and 4weeks after preparation for G-1, G-2, G-4, and G-7, respectively.

The efficient dispersion of graphene in aqueous solution of 7 implies strong interactionsof 7 with the basal plane of graphene through π-π interactions, which keep thenaphthalene adsorbed onto the graphitic surface. In order to study π-π interactions,fluorescence measurements were carried out. Fluorescence quenching experimentsreveal a nearly quantitative fluorescence quenching (80%) of the fluorescence of 7mixed with graphite upon ultrasonication-assisted exfoliation - see Figure 4.12. Thisprovided direct evidence of strong π-π interactions between 7 and graphene. Notably,quenching experiments of G-1 revealed an even higher quenching efficiency in the G-1sample (90%), due to charge or energy transfer, which is consistent with the presenceof π-π interactions. Thus, enhanced π-π interactions between graphene and 1, arisingfrom the larger π-π network, in comparison to graphene and 7 are corroborated.Nevertheless, the enhanced dispersion efficiency of 7 may arise from both improvedelectrostatic and steric repulsion. In contrast to 1, which binds to the graphenesurface via one single pyrene group, 7 can interact with the graphene basal plane viavarious naphthalene units. Moreover, every naphthalene unit is equipped with one

73


sulfonic acid group, thus, coating the graphene surface with a bulky layer of negativecharges. This can provide strong repulsive electrostatic and steric interactions and,thus, hinders the graphene layers from restacking and reagglomerating.

a)

b)

Figure 4.12: a) Fluorescence spectra of 7 mixed with graphite in water before ul-trasonication (Graphite + 7) and after ultrasonication (G-7) withλ=260 nm. b) Fluorescence spectra of 1 mixed with graphite in waterbefore ultrasonication (Graphite + 1) and after ultrasonication (G-1)with λ=340 nm (bottom). Insets: Photographs of the G-7 and G-1dispersion, respectively.

The surface potential (zeta potential, ζ) is a useful parameter to further characterizeour samples, as it is a measure of electrostatic repulsion between surfactant-coated

74


graphene sheets and, thus, the stability of the dispersion. Accordingly, Smith etal. found that the graphene concentration increases with increasing |ζ| for ionicsurfactants.[120] As such, zeta potential measurements were carried out in order togain further insights into the stabilization mechanism of 7 and 1. For the as-preparedG-7 dispersion, the peak zeta potential of -65 mV is well beyond the accepted value forcolloidal stability of -25 mV and previously reported zeta potentials of graphene flakesstabilized by ionic surfactants, which indicates efficient stabilization of graphene by7 through electrostatic repulsion (Figure A.3 in the Appendix).[106,120] Interestingly,rather than being neutral, the graphene sheets coated by 1 display a zeta potential ofabout -17 mV (Figure A.3 in the Appendix). Such negative zeta potentials of nonionicsurfactant-stabilized particles were previously observed for surfactant-stabilized CNTand graphene samples and may arise from the adsorption of charged impuritieson the graphene surface and edge functionalities.[120,155,306] In light of this, evengraphene sheets coated by 1 may be partially stabilized by repulsive electrostaticinteractions.

Conclusively, absorbance, emission, and zeta potential measurements proved theeffective functionalization of graphene by 1 and 7 and their ability to dispersegraphene in aqueous media through efficient electrostatic and sterical repulsiveinteractions. In terms of graphene dispersibility and stabilization, 7 has proven tobe superior to other aromatic amphiphiles analyzed in this study, yielding grapheneconcentrations as high as 0.058 mg mL−1. However, since all dispersions wereprepared with a fixed surfactant concentration, it is expected that the amount ofgraphene stabilized by 7 can be further increased upon optimization of the surfactantconcentration of 7.

In this context, two different strategies were considered. Firstly, the initial concen-tration of the surfactant solution, which was added to the graphite feedstock priorto stirring, was varied in order to find the optimal surfactant concentration (strategy1). Secondly, starting with a low initial surfactant concentration of 1.25 mg mL−1,additional surfactant was added throughout the ultrasonication process, yieldingfinal surfactant concentrations compared to strategy 1 (strategy 2). The continuousaddition of surfactant during the ultrasonication process was expected to be beneficialfor the exfoliation and stabilization of graphene. The initial surfactant is expected tobind to the graphite flake surface during stirring. Thus, an equilibrium between thestrongly bond surfactant molecules and the free surfactant in the solution is formed.

75


However, during the ultrasonication induced exfoliation free surfactant starts toassemble around as-exfoliated graphene sheets, while surfactant molecules start todetach from other graphitic particles in order to maintain equilibrium and, thus,limit the concentration of stabilized particles. In this context, continuous additionof free surfactant during the exfoliation process should further facilitate graphenedispersion and stabilization by shifting the equilibrium towards the adsorption of 7onto the exfoliated graphene sheets.

Figure 4.13 shows the calculated graphene concentration observed by differentsurfactant concentrations, which are achieved by either strategy 1 or strategy 2.

Figure 4.13: Calculated graphene concentration (cac) for the G-7 sample as a functionof surfactant concentration (c7) achieved by either an initial surfactantaddition prior to ultrasonication or a continuous surfactant additionduring ultrasonication treatment, respectively. The inset shows pho-tographs of the dispersion, featuring highest (right) and lowest (left)graphene concentration, respectively.

The maximum graphene concentration appears at an initial concentration of2.5 mg mL−1, rather than increasing monotonically with the surfactant concentration.This is in good accordance with previous reports on other graphene stabilizers, whichshow maximum graphene dispersibility at a peak surfactant concentration.[126,307,308]

The decrease in the graphene concentration at higher concentrations of 7 can beattributed to attractive depletion interactions.[309] Interestingly, a continuous addi-

76


tion of the surfactant leads to lower graphene concentrations compared to an initialsurfactant addition in the case of each final surfactant concentration.

This is in contrast to previous findings by Notely et al., who report on a substantialincrease of graphene concentration in dispersion by means of continuous surfactantaddition.[132] However, in their case, the improvement of a continuous surfactantaddition is expected to relate to a readjustment of the surface tension to the optimumvalue for graphene exfoliation at about 40 mN m−1.

In general, surfactants are considered to influence the exfoliation of graphite tographene in two different ways. Firstly, they lower the liquid-vapor interfacial energyof the solution to an optimal range, which corresponds to the required energy forseparating graphene sheets beyond the range of van der Waals interaction and,thus, enabling the exfoliation. Secondly, surfactants adsorb onto the exfoliatedgraphene sheets, creating a repulsive layer and, thus, prevent the reaggregation of thesheets.[120,128,310] However, as 7 is only weakly surface active, the surface tension for allstabilizer concentrations is well beyond the optimal range for the graphene exfoliationand, thus, surface tension is expected to play a negligible role for graphene exfoliationby 7. Nevertheless, the significant increase in final graphene concentration in thecase of an initial addition of 7 compared to a continuous addition indicated a strongimpact of 7 on the graphite feedstock, resulting in enhanced graphene exfoliation anddispersion and, thus, higher final graphene concentrations. In this context, we assumethat 7 partially intercalates between the edges of graphene sheets of the graphitefeedstock during stirring and, thus, triggers graphene exfoliation - see Figure 4.14.Such contributions have also been reported for pyrene derivatives recently.[109] Insummary, we demonstrated a further improvement of graphene dispersibility byoptimizing the concentration of 7, yielding final graphene concentrations of up to0.15 mg mL−1.

77


Figure 4.14: A schematic representation of the proposed dispersion and exfoliationmechanism of 7. Firstly, 7 partially intercalates into the grapheneedges during the stirring process. Secondly, exfoliation of graphene byultrasonication and subsequent adsorption of 7 onto the graphene basalplane, yielding stabilized graphene flakes in aqueous medium.

4.2.2 Exfoliation Ability of Aromatic Amphiphiles

Although absorption measurements give valuable insights into the dispersibility ofgraphitic particles by different surfactants, it has to be considered that besides single-layer and few-layer graphene, also graphitic nanoparticles with small hydrodynamicdiameter may remain dispersed after centrifugation and, thus, contribute to theabsorbance. To this end, the exfoliation state of the samples was further studied togain deeper insights into overall material quality. Graphene characterization methodsinclude Raman spectroscopy, TEM, and AFM. However, TEM microscopy allows thedetermination of a small fraction of the sample only. Additionally, flakes with verysmall lateral sizes may get lost through the holey carbon grid, thus, leading to anoverestimation of the flake size. Moreover, AFM imaging encounters the challenge ofremoving residual surfactant molecules, which may disturb the measurements.

Therefore, we mainly focus on statistical Raman analysis, which is less influenced bysurfactant residuals and allows the investigation of a large sample area. Hence, itgives quantitative information on the exfoliation state of the sample and, as such, onthe exfoliation ability of the different aromatic surfactants. To this end, Raman maps

78


of G-1, G-2, G-4, and G-7 dip coated on SiO2 wafers were carried out, respectively,and more than 200 Raman spectra were evaluated for each sample.

A selected Raman spectrum of G-1 is shown in Figure 4.15, displaying the D-, G-,and 2D-band located at 1342, 1582, and 2684 cm−1, respectively. The spectrum ofthe exfoliated material clearly differs from those of the starting material - see FigureA.2. In particular, the D-band arises upon exfoliation, indicating the presence ofdefects and, thus, an increased contribution of double resonant Raman scatteringprocesses. Moreover, the 2D-band transforms into a highly symmetric peak that isred-shifted to 2684 cm−1 upon excitation at 532 nm with a I2D/IG ratio of 2.3 andan FWHM of 28 cm−1, indicating SLG.

Figure 4.15: Selected solid-state Raman spectrum of G-1, indicating the exfoliationstate of the sample. The sample was dip coated from dispersion ontoa Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the 2D-band.Black dots - experimental data; blue curve - fitting.

To gather sufficient statistics, a Raman map with around 700 different spectra wasevaluated - Figure 4.16. Taking into account that the I2D/IG ratio of few- to multilayergraphene is 0.7 or less,[295] the data reveals a distribution of 22% multilayered/bulkgraphite and about 78% single-layered/few-layered/turbostratic graphite within thesample - see Figure 4.16 a). Notably, the spectral shape of the 2D peak feature allows amore precise determination of the number of deposited graphene layers. In particular,peak shape or symmetry analysis and examination of the FWHM of the 2D-band

79


can be used to determine whether single-layered, few-layered/turbostratic graphite,or few- to multilayered graphite is probed. As demonstrated by Graf et al., the2D-FWHM for single-layer graphene is <40 cm−1, whereas few-layered/turbostraticgraphite exhibits a FWHM between 40 and 65 cm−1.[187] Multilayered/bulk graphiteshow an FWHM that is broader than 65 cm−1. With this information at hand,the sample can be quantified to consist of 20% multilayered/bulk graphite and ofabout 80% few-layered/turbostratic graphite, with small amounts of single-layergraphene (<1%) - see Figure 4.16 b). However, as already discussed in Chapter 4.1.2,sample preparation and the presence of defects may lead to an underestimation of thesingle-layer graphene content. Nevertheless, the high amount of turbostratic graphiterevealed the successful exfoliation and, thus, proved the capability of 1 to exfoliateand stabilize FLG in aqueous medium. While the analysis of the I2D/IG ratio and2D-FWHM of the Raman spectra gave valuable insights into the exfoliation state,it is crucial to consider the defects present in the exfoliated graphene flakes, whichcould alter the electronic structure of the few-layer graphene sheets. Defects in thegraphene structure break symmetries and thereby allow forbidden inter/intravalleyprocesses, which give rise to D- and D’-bands. These bands are not observed in theRaman spectrum of the pristine starting material - see Figure A.2 in the Appendix.The quantity of defects strongly relates to the ID/IG ratio. In particular, the largerthe ratio, the larger the defect density (and, thus, the larger the typical distancebetween defects).[191] These defects can arise from either the sample edges or bulkdefects. Recent studies have been shown that the intensity ratio of the D- andD’-band is sensitive to the type of defect, being around 3.5 for edge related defects,7 for vacancies, and 13 for sp3 defects.[192] The fact that the spectra show a roughlyconstant ID/ID′ ratio of 3.1 rules out vacancies, substitutions, and sp3 defects and,thus, testifies that the resulting disorder is mainly edge related defects that resultfrom the reduction in flake size due to ultrasonication treatment. Conclusively, thehigh mean ID/IG ratio of 1.75 indicates the presence of relative small flake sizes -Figure 4.16 c).

80


a)

b)

c)

Figure 4.16: Histograms resulting from the Raman mapping of G-1, showing a)counts versus I2D/IG ratio, b) counts versus 2D-FWHM, and c) countsversus ID/IG and the corresponding log-normal distribution, respectively.The sample was dip coated from dispersion onto a Si/SiO2 wafer andexcited at 532 nm.

81


Notably, statistical Raman analysis of G-7 gives rise to even better exfoliation results.In Figure 4.17 and 4.18 a representative Raman spectrum and statistical distributionsof the I2D/IG ratio and 2D-FWHM of about 330 measured spectra are displayed.The D-, G-, and 2D-bands of G-7 dip coated onto a silicon wafer are centered ataround 1350, 1580, and 2697 cm−1, respectively.

Figure 4.17: Selected solid-state Raman spectra of G-7, indicating the exfoliationstate of the sample. The sample was dip coated from dispersion ontoa Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the 2D band.Black dots - experimental data; blue curve - fitting.

Statistical evaluation of the I2D/IG ratio indicates that 15% of the collected spectrareveal multilayered/bulk graphite character while 85% of the collected spectracorrespond to turbostratic exfoliated graphite and monolayered graphene - seeFigure 4.18 a). In order to gain deeper insights into the exfoliation state of thesample, the FWHM of the 2D-band was taken into account. To this end, a shapeanalysis of the 2D-band was performed. In all cases, the spectra could be fitwith a single Lorentzian function, but differ in their FWHM. In particular, 76%of the spectra exhibit an FWHM between 40 and 65 cm−1 revealing their few-layer/turbostratic character, while 3% show FWHMs above 65 cm−1, thus, beingreferred as multilayered/bulk graphite. Notably, 21% of the spectra display anFWHM <40 cm−1, representing SLG - see Figure 4.18 b). Conclusively, G-7 is bestdescribed as FLG with a large amount of SLG. Moreover, a relative weak D-band(mean ID/IG = 0.49) implies a low defect density that originates from edge defects

82


(mean ID/ID′ = 1.8) and, thus, larger flake sizes compared to G-1 are corroborated -Figure 4.18 c).

a)

b)

c)

Figure 4.18: Histograms resulting from the Raman mapping of G-7, showing a)counts versus I2D/IG ratio, b) counts versus 2D-FWHM, and c) countsversus ID/IG the corresponding log-normal distribution, respectively.The sample was dip coated from the dispersion onto Si/SiO2 wafers andexcited at 532 nm.

In contrast, Raman maps of G-2 and G-4, which were prepared under the same

83


conditions, led to lower I2D/IG ratios and higher 2D-FWHMs. In particular, G-2is composed of 31% multilayered/bulk graphite particles and of about 69% few-layered/turbostratic graphite sheets with very low single-layer content (<0.01%) -see Figure 4.19 a). Moreover, the intense ID/IG ratio of 2.4 indicates a stabilizationof only very small graphene sheets - see Figure 4.19 b). This might be related to thefact that the relatively small molecule does not provide effective coating and, thus,the stabilization of large graphene sheets is hindered.

a)

b)

Figure 4.19: Histograms resulting from the Raman mapping of G-2, showing a) countsversus I2D/IG ratio and b) counts versus ID/IG and the correspondinglog-normal distribution, respectively. The samples were dip coated fromthe dispersion onto Si/SiO2 wafers and excited at 532 nm, respectively.

For G-4, Raman analysis reveals a multilayered/bulk graphite content of 61% and afew-layered/turbostratic graphite content of about 39%, indicating that 4 is more

84


likely to stabilize multilayered graphite sheets than few-layer graphene sheets - seeFigure 4.20 a). Interestingly, a relatively low mean ID/IG ratio of 0.55 further impliesthe stabilization of larger graphite stacks - see Figure 4.20 b). Both findings lead tothe conclusion that 4 is insufficient in preventing graphene aggregation by restacking.This may be attributed to the small molecular structure of 4, which might notprovide an effective barrier for reaggregation of exfoliated graphene flakes. Moreover,4 exhibits three hydrophilic substituents arranged around the aromatic core unit,which may hinder 4 from intercalation between the hydrophobic graphene sheetsand, thus, hamper exfoliation.

a)

b)

Figure 4.20: Histograms resulting from the Raman mapping of G-4, showing a) countsversus I2D/IG ratio and b) counts versus ID/IG and the correspondinglog-normal distribution, respectively. The samples were dip coated fromthe dispersion onto Si/SiO2 wafers and excited at 532 nm, respectively.

85


In conclusion, Raman analysis revealed the successful exfoliation of graphite to yieldfew-layered/turbostratic graphite by using 1 and 7 as stabilizers. However, only 7featured high yields of SLG. Moreover, statistical analysis of the ID/IG ratio indicateda lower defect content of G-7 and, thus, larger flake sizes are corroborated.

In order to gain deeper insights into structural characteristics of the prepared graph-ene flakes, TEM analysis was performed. Representative TEM images of G-1 areshown in Figure 4.21. The TEM observations confirm that the graphene obtainedfrom G-1 dispersions is likely to be in the forms of FLG. Presumably, thin exfoliatedgraphene sheets, which consist of individual mono- and few-layer graphene, randomlyrestack and fold, forming turbostratic graphene stacks, as Raman studies alreadyimplied. Notably, neither very large nor thick aggregates were observed and, thus, theoverall exfoliation degree seems to be high. The thin flakes exhibit lateral dimensionsin the order of 250 nm up to 2 μm together with flake thicknesses in the rangeof single- to few-layer graphene. Here, the high transparency of the flakes to theelectron beam is used as evidence.

Figure 4.21: Representative TEM micrographs of G-1 with different magnifications.The sample was drop coated on a lacey carbon grid.

Next to G-1 also G-7 was examined by TEM imaging in order to compare theexfoliation degree and flake size. Figure 4.22 shows representative TEM imagesof G-7. Conclusions from the TEM investigations are qualitatively such to thosederived from G-1. In most cases, aggregates of a number of rather disordered thingraphene sheets are discernible. However, the amount of flakes per aggregate isincreased in comparison to the G-1 hybrid. This observation arises from the fact that

86


the concentration of the G-7 sample is much higher than the concentration of G-1,which results in a general increase in the number of graphene sheets. It is expectedthat these graphene sheets reaggregate on the TEM grid during the slow solventevaporation while preparing the sample to form turbostratic graphene aggregates.Again, the overall exfoliation degree of the sample seems to be high, which is in goodaccordance with Raman analysis. The flakes exhibit lateral dimensions in the rangeof 250 nm to a few micrometers. Notably, no clear difference in the lateral size ofG-1 and G-7 could be observed. However, flakes with very small lateral dimensionsare expected to get lost during TEM sample preparation and, thus, are not probedby TEM imaging.

Figure 4.22: Representative TEM micrographs of G-7 with different magnifications.The sample was drop coated on a lacey carbon grid.

To summarize the results so far, graphene was dispersed by four aromatic amphiphiles,namely 1, 2, 4, and 7. The graphene concentration varies significantly by a factorof 2-3 from 0.058 mg mL−1 for G-7 to 0.019 mg mL−1 for G-4, rendering 7 asthe best aromatic surfactant for graphene stabilization investigated in this study.Furthermore, the optimization of the stabilizer concentration demonstrated thatthe concentration of dispersed graphene can be increased by a factor of 2-3, yield-ing graphene concentrations of about 0.15 mg mL−1 according to our preparationprocedure. We stress that our production procedure focused on the quantitativeevaluation of different stabilization methods and, thus, is not improved in terms ofoverall graphene dispersibility. Notably, prior studies have shown that prolongedultrasonication treatment along with high initial graphite concentrations can further

87


increase the final concentration of graphene dispersions.[123] However, this may affectthe structural quality of the resulting graphene sheets. Nonetheless, it is expectedthat the graphene concentration can be further increased by optimizing the exfolia-tion procedure and adjacent scale-up, which is beyond the scope of the present work.Moreover, it was found that the surfactant strongly influences the flake size anddegree of exfoliation. As revealed by Raman analysis of the flake thickness distribu-tion (I2D/IG ratio and 2D-FWHMs), the exfoliation degree of the samples increases inthe order 4 < 2 < 1 < 7. Additional evaluation of the defect-related D-band indicatesthat the defect content of the samples increases in the order of 7 < 4 < 1 < 2, whichcould be assigned to the reduction of the flake size by analyzing the ID/ID′ ratio. Inparticular, highly concentrated graphene dispersions with good structural quality andlow defect content were prepared by means of 7. These dispersions consist of mainlyfew-layered/turbostratic graphite with high amounts of single-layer graphene (21%)and lateral sizes ranging from sub- to micrometer size. However, a more generalconclusion is that aromatic amphiphiles with a large molecular structure tend toperform better than aromatic amphiphiles with small hydrophilic parts in regardto their dispersion and exfoliation ability towards few-layer graphene dispersions.Such results suggest that steric repulsion by either sterically demanding hydrophobicparts (1) or high molecular weight surfactants (7) is essential in the stabilization ofgraphene sheets in water.

Comparing the exfoliation and dispersion ability of 7 to the reference surfactantCTAB - see Chapter 4.1.2 - 7 proved to be superior in terms of dispersion and exfo-liation efficiency, particularly towards the exfoliation of SLG. Nonetheless, G-CTABdispersions featured enhanced long-term stability. However, further developmentregarding the long-term stability of graphene dispersions prepared by aromaticamphiphiles are discussed in Chapter 4.3.

Conclusively, this work demonstrates a straightforward route towards aqueous few-to single-layer graphene dispersions with very low defect content and high grapheneconcentrations. Such dispersions are made possible via the use of the specific aromaticsurfactant 7, which features several naphthalene units, enabling strong interactionwith the graphene basal plane. Moreover, it is also equipped with sulfonic acidgroups, which further promote strong electrostatic repulsion and, thus, prevent theexfoliated graphene sheets from restacking. To exfoliate graphite and subsequentlystabilize the produced high-quality few- and monolayer graphene in water by an

88

4.3 PRETREATMENT OF GRAPHITE AS A DISPERSION ENHANCER

industrially available low-cost surfactant, such as 7 ($2 per kg), will further facilitatethe practical use of this novel material in applications, such as supercapacitors - seealso Chapter 5.

4.3 Pretreatment of Graphite as a DispersionEnhancer

As discussed in Chapter 4.2, amphiphilic aromatic stabilizers with a large molecularstructure tend to be superior in terms of graphene dispersion and exfoliation com-pared to small molecules. However, since the favored aromatic stabilizers carry ratherbulky hydrophilic and hydrophobic parts to ensure efficient stabilization throughsteric interactions, it is reasonable to suggest that these large molecules cannot easilyslip between adjacent graphene layers during the sonication induced exfoliation.Moreover, the graphene dispersions prepared by amphiphilic stabilizers turned out topossess limited long-term stability with about 40% of the sample precipitating withinthe first month (G-7). As a matter of fact, such limited long-term stability is notsufficient for commercial graphene production. In an attempt to further increase thedispersion and stabilization efficiency of surfactants, a dispersion concept was applied,which was first introduced by Geng and coworkers,[311] and involves the use of small in-tercalant molecules additionally to the dispersant. Accordingly, the additional use ofsmall intercalant molecules, such as quaternary alkylammonium ions, which possess ahigh affinity to the graphene edges[312,313] and, thus, may open up these edges, are ex-pected to increase graphene exfoliation and dispersibility even though these moleculescannot stabilize graphene sheets by themselves. In fact, recent studies have shownan improvement of the liquid-phase exfoliation of graphite in N-methyl-2-pyrrolidone(NMP) by adding tetraalkylammonium salts, such as tetrabutylammonium hydrox-ide and tetraethylammonium tetrafluoroborate.[311,314] In addition, several groupsrecently reported on the successful intercalation of quaternary alkylammonium saltsinto the layered structure of graphite oxide, in order to produce quaternary alky-lammonium ion-graphene oxide intercalation compounds.[261,312,313,315] Moreover,first screening experiments within my master thesis indicated that the addition oftetraalkylammonium ions, such as tetraethylammonium hydroxide, is also beneficialfor the liquid-phase exfoliation of pure graphite in aqueous medium.[137,316]

89


4.3.1 Influence of Tetraalkylammonium Salts on GrapheneDispersibility

In order to evaluate this concept, three approaches were tested by preparing differentdispersions. Firstly, graphite was dispersed in a solution of 1 by stirring and subse-quent ultrasonication for 6 h, yielding stable graphene dispersions (G-1; approach 1)- see also Chapter 4.2. Secondly, graphite was dispersed in solutions of 1 containingtetraethylammonium hydroxide (I1, 0.082 mol L−1) in water. Then, these sampleswere treated under the same conditions as the ones with 1 only, resulting in stablegraphene dispersions (G-I1-1; approach 2). In the third approach, graphitic materialwas pretreated with ultrasonication in an aqueous solution of tetraethylammoniumhydroxide (I1*, 0.082 mol L−1) for one hour. Afterward, the dispersion was leftto settle overnight and the supernatant was removed the next day. The resultinggraphitic precipitate was then redispersed in an aqueous solution of 1 with equalconcentration as used in approach 1 and 2. Then, the dispersion was stirred overnightand subsequently sonicated for 5 h, yielding stable graphene dispersions (G-I1*-1;pretreatment of graphite in the presence of I1 is indicated by *). Please note thatthe sonication time after redispersion was adjusted in order to enable comparisonwith previous experiments in terms of total energy input. However, following thisapproach, the time in which stabilization and exfoliation of the graphene particlestake place is reduced by 17% in approach 3.

Nevertheless, we observed that pretreating graphite in a solution of I1 yielded thehighest graphene concentration of 0.073 mg mL−1 and, thus, improved the graphenedispersibility of 1 by a factor of 3 - see Figure 4.23. Notably, direct addition ofI1 and pretreatment of graphite in I1 led both to an enhanced dispersibility ofgraphene. However, we note that the direct addition of I1 led to lower grapheneconcentrations (0.047 mg mL−1) compared to the pretreatment of graphite in I1,even though the exfoliation process is reduced by 1 h in the case of pretreatment.This can be attributed to the fact that in the case of direct addition of I1 to thegraphite-surfactant mixture, 1 and I1 may compete with each other in the interactionwith graphene. This may partially hinder the interaction of I1 with the graphiteedges. Moreover, I1 may partially adsorb on the graphene surface, thus, hinderingthe adsorption of 1 and, therefore, further reducing graphene dispersibility. Inthe case of the pretreatment procedure, I1 can interact first with the raw graphitematerial by partially intercalating into the graphene edges. Afterward, any free I1 is

90


removed by discarding the supernatant and 1 can freely adsorb on the graphene basalplane and subsequently stabilize the graphene flakes, which are exfoliated during theultrasonication process.

Figure 4.23: Calculated graphene concentration cac directly after preparation and4 weeks after preparation for G-1, G-I1-1, and G-I1*-1, respectively.G-1: Graphite + 1 / 6 h ultrasonication; G-I1-1: Graphite + I1 + 1 /6 h ultrasonication; G-I1*-1: Graphite + I1 / 1 h ultrasonication →precipitate, precipitate + 1 / 5 h ultrasonication

Further, the addition of I1 remarkably improved the stability of the resulting graph-ene dispersions. To this end, the stability of the dispersions was measured byrecording the absorption spectra 4 weeks after sample preparation. Accordingly,we found that only small amounts of graphene (∼8%) had sedimented over thisperiod within dispersions of G-I1-1. In the case of G-I1*-1, ∼19% of the graphenehad precipitated over 4 weeks, while G-1 was the least stable sample, displaying agraphene precipitation of about 32%. These findings can be explained by an increasedstability of graphene sheets in alkaline media due to the deprotonation of the edgefunctionalities, which further contributes to repulsive electrostatic interactions.[317]

This was further proven by zeta potential measurements, which reveal a zeta potentialof -53 mV and -54 mV for G-I1*-1 and G-I1-1, respectively. This is in line withprevious reports, which measured the zeta potential of multilayered graphene sheets

91


in alkaline media to be about -47 mV (pH=11) (Figure A.4 in the Appendix).[317]

In order to further shed light on the nature of the exfoliation and stabilizationmechanism of the intercalant approach, four control experiments were performed.Firstly, graphite was pretreated in an aqueous solution of 1 in order to investigatethe effect of pretreatment itself. Secondly, graphite was pretreated in an aqueoussolution of sodium hydroxide (NaOH) in order to understand the effect of the pH ongraphene dispersibility. In addition, graphite was pretreated in an aqueous solution oftetraethylammonium chloride (I2) to gain further insights into the importance of thecounter ion and the pH. Last but not least, the pretreatment concept was transferredto tetra-n-butylammonium hydroxide (I3) in order to prove the transferability of theapproach towards other quaternary alkylammonium salts. In this context, the rawgraphitic material was pretreated with 1 hour of ultrasonication in solutions of 1,NaOH, I2, and I3. Afterward, the samples were treated under the same conditionsas G-I1*-1. Then, the resulting dispersions were centrifuged to remove any graphiticmaterial, and the supernatant was subjected to further analysis.

Figure 4.24 shows the calculated graphene concentrations of the dispersions afterpreparation and 4 weeks after preparation, respectively. Interestingly, the pretreat-ment of graphite with 1 led to small improvements only. Moreover, these dispersionswere highly unstable, precipitating to a large extent (∼60%) within the first 4 weeksafter preparation. Accordingly, these dispersions possessed lower zeta potentialvalues of ∼-3 mV in comparison with G-1 (∼-17 mV), which further indicates a lowerstability - see Figure A.4 a) in the Appendix. Additionally, pretreatment of graphitewith NaOH led to quite similar graphene concentrations of about 0.028 mg mL−1.These results suggest, that the ultrasound-assisted pretreatment of graphite has asmall impact on the dispersibility of graphene in general, which is in accordance withBarwich et al.[318] In fact, they found that sonication of graphite in non-stabilizingsolvents leads to partial exfoliation of graphite flakes, resulting in aggregated graphenestacks, which can be dispersed in stabilizing solvents afterward. Again, the enhancedstability of G-NaOH*-1 may be attributed to electrostatic repulsive interactions,arising from the deprotonation of edge functionalities. This was further confirmedby zeta potential measurements revealing a zeta potential of about -46 mV, which isindicative for an efficient stabilization via repulsive electrostatic interactions - seeFigure A.4 in the Appendix. Nevertheless, these findings also reveal the importanceof the presence of quaternary alkylammonium ions during the pretreatment process,

92


which in fact led to much higher graphene concentrations.

Figure 4.24: Calculated graphene concentrations cac for graphene dispersions pre-pared by pretreatment of graphite, namely G-1*-1, G-NaOH*-1, G-I2-1, and G-I3*-1.

However, control experiments employing I2 as an intercalant molecule led to grapheneconcentrations comparable to G-1, indicating a negligible effect on the exfoliationand dispersion of graphene by tetraethylammonium chloride and, thus, stressingthe importance of the counter ion and the pH. In fact, these dispersions displayedlow zeta potential values of about -5 mV, as edge functionalities are not dissociatedand, thus, carry no charge (Figure A.4 e)). Moreover, these findings imply that thetetraethylammonium ions are not adsorbed onto the basal plane of the stabilizedgraphene sheets and, thus, do not contribute significantly to the stabilization of thegraphene flakes. Consequently, it is expected that both, the presence of quaternaryalkylammonium ions and the alkaline pH of the solution, are of crucial importancefor the improved dispersion of graphene according to the pretreatment process. Inorder to further evaluate these findings, tetrabutyl-n-ammonium hydroxide was usedas an intercalant during the pretreatment. Again, we found that the graphenedispersibility of 1 is significantly improved upon pretreatment of graphite yieldinggraphene concentrations of 0.066 mg mL−1. Please note that direct addition of I3 ledto even lower graphene concentrations than those observed in the case of G-1. This

93


might be attributed to the large molecular size of I3, which might further hinder theadsorption of 1 to the graphite/graphene basal plane due to steric effects. Indeed,the successful utilization of I3 in the pretreatment approach further confirmed theconcept of pretreating raw graphite in alkaline solutions of tetraalkylammoniumsalts in order to enhance graphene dispersibility and the stability of the prepareddispersions.

According to these findings, we propose a mechanism for the pretreatment of graphitein solutions of I1 as illustrated in Figure 4.25. In general, it can be assumedthat the interaction of I1 with graphite is ascribed to the edge oxygen-containinggroups, which are introduced by means of ultrasonication during the pretreatmentprocess (see also Chapter 4.1.2).[312,319] We suggest that in a first step, the edgefunctional groups (C-OH and COOH) are dissociated into C-O− and COO− inthe alkaline solution of I1.[320] The positively charged tetraethylammonium cationcan then be adsorbed and partially intercalated into the edge planes of graphitevia cation-anion interactions. This results in an expansion of the edge surface ofgraphite and may also switch the edges wettability from hydrophilic to hydrophobic.This further facilitates co-intercalation of 1 into the interlayer space of graphite viahydrophobic interaction between the alkyl chains of I1 and the aromatic pyrene unitof 1 and via π-π interactions between the graphene basal plane and the aromaticpyrene unit, during the stirring process. These partially expanded and intercalatedgraphene layers may then easily exfoliate during the ultrasonication process and aresubsequently stabilized by 1. Moreover, zeta potential measurements indicated thatthese layers are additionally stabilized by electrostatic repulsive interactions arisingfrom deprotonated edge functionalities. Please note that we assume the interactionto take place mainly at the graphene edges because XRD measurements did notreveal full intercalation of I1 into the graphite layer structure - see Figure A.6 inthe Appendix.

94


Figure 4.25: A schematic representation of the dispersion and exfoliation mechanism.Firstly, selective interaction of I1 with hydroxyl and carboxyl groups atthe graphite edges. Secondly, 1 partially intercalates into the grapheneedges. Thirdly, exfoliated graphene sheets with 1 coating on the surfaceand, thus, stabilizing the flakes in aqueous medium.

In an attempt to further increase the graphene dispersibility of 1, the interactionof 1 with the I1 decorated graphene edges was enhanced by an increase in theoverall interaction time before ultrasonication. In this context, we left the dispersionsof pretreated graphene in solutions of 1 for 1, 2, 3, and 4 weeks undisturbedand sonicated them afterward for 5 h to yield stable graphene dispersions. Inanother experiment, the dispersions were stirred for 1, 2, 3, and 4 weeks followedby ultrasonication. Figure 4.26 shows the calculated graphene concentrations aftercentrifugation subsequent to different stirring and stand times, respectively. Ingeneral, the graphene concentration increases as a function of time in both casesyielding graphene concentrations of 0.26 mg mL−1 for 4 weeks of stirring and0.16 mg mL−1 for 4 weeks stand time without stirring, respectively. Notably, inthe case of all time variations stirring yielded higher graphene concentrations. Thissuggests stirring the sample further facilitates the intercalation of 1 into the grapheneedges and the adsorption of 1 onto the basal plane of the graphene stacks. Althoughlong stirring times are not favorable for large-scale production processes, theseresults further demonstrated the potential of the pretreatment process towards the

95


production of highly concentrated few-layer graphene dispersions. In order to employthis concept in an industrial manner, more efficient dispersion techniques, such as aspeedmixer or an ultraturrax, can be utilized.

Figure 4.26: Calculated graphene concentration cac for the long-term intercalation of1 into the I1 decorated graphene sheets by either stirring or no stirring.

Conclusively, we showed that the pretreatment of graphite in alkaline solutions oftetraalkylammonium hydroxide significantly improves the graphene dispersibilityof 1. In order to further evaluate the impact of the pretreatment approach ongraphene dispersibility of different surfactants, we transferred the process to severalsurfactants, featuring different molecular structures. In detail, we prepared differentgraphene dispersions by employing either a pretreatment step with tetraethylammo-nium hydroxide as an intercalant molecule or employing no additional intercalant,respectively. In this context, we tested the graphene dispersibility of CTAB as acationic surfactant, sodium dodecylsulfate (SDS) as an anionic surfactant, PluronicP-123 (P123) as a nonionic aliphatic surfactant, 5 as a nonionic aromatic amphiphile,and 7 as an anionic aromatic amphiphile.

Figure 4.27 shows the calculated graphene concentrations of the dispersions preparedwith and without the utilization of an auxiliary pretreatment step for each surfactant,respectively. Notably, in all cases except for CTAB, pretreatment of graphite inI1 leads to a significant increase in the graphene concentration after centrifugation.However, in the case of CTAB, it is expected that repulsive electrostatic interactions,

96


arising from CTAB and I1, hinder the intercalation of CTAB into the I1 decoratedgraphene edges and, thus, limit the effect of pretreatment. Particularly, evensurfactants, which are not able to disperse graphene on their own, showed very goodgraphene dispersibility in the case of pretreatment. Interestingly, 5 led to the highestgraphene concentration after centrifugation, showing superior long-term stabilitywith no signs of precipitation after 4 weeks and only small amounts of grapheneprecipitating after 3 months (5% precipitation). Again, pretreatment of graphitein a solution of NaOH did not give rise to the same results. As 5 is not capableof exfoliating and dispersing graphene on its own, pretreatment of raw graphite insolutions of tetraethylammonium hydroxide is expected to be the crucial step in thedispersion of graphene by means of 5.

Figure 4.27: Calculated graphene concentration cac after preparation and 4 weeksafter preparation, showing the effect of pretreatment of raw graphite insolutions of I1 on graphene dispersibility of different surfactants andthe stability of the resulting dispersions, respectively.

Additionally, the utilization of an auxiliary pretreatment step remarkably improvedthe stability of the resulting graphene dispersions for all surfactants except CTAB -see Table 4.4. Pretreating graphite in I1 only had a small impact on the stabilityof the dispersions prepared with the help of CTAB. In fact, we assume that,besides its interaction with the graphene basal plane, CTAB also interacts with thecharged hydroxy and carboxy groups on the graphene edges through anion-cationinteractions. This results in charge neutrality at the graphene edges and, thus,

97


no additional repulsive electrostatic interaction can be introduced by switching toan alkaline medium. However, in all other cases, the stability of the dispersionsprepared by the pretreatment approach is remarkably enhanced compared to the onesprepared without any auxiliary pretreatment step. Again, we assume the improvedstability to arise from deprotonated edge functionalities, providing additional repulsiveelectrostatic interactions.

Table 4.4: Comparison of the precipitation (%, after 4 weeks) of the graphene dis-persions prepared by either pretreatment with I1 or no pretreatment fordifferent surfactants, respectively.

Surfactant No pretreatment / % Pretreatment with I1 / %

CTAB 5.0 4.4P123 78.3 10.6SDS 7.9 6.81 31.1 19.05 90.2 07 39.6 11.7

In summary, we presented a facile way to effectively improve graphene dispersibilityof several types of surfactants, namely an anionic aliphatic surfactant, a nonionicaliphatic surfactant, a nonionic aromatic surfactant, and an anionic aromatic sur-factant, by employing an auxiliary pretreatment step, involving the use of tetra-alkylammonium hydroxides.[321] The dispersion efficiencies of graphene could besignificantly increased by the highly efficient pretreatment step, resulting in anincrease in graphene concentration by up to 21 times for 5, 20 times for P123, andaround 1-3 times for all other surfactants, respectively. Moreover, the stability of theresulting dispersions is remarkably enhanced, which is affected by the deprotonationof edge functionalities, providing additional electrostatic stabilization.

4.3.2 Influence of an Auxiliary Pretreatment Step on theExfoliation Degree

In order to gain further insights into the effect of the auxiliary pretreatment stepon the exfoliation degree of the graphene samples, statistical Raman measurements

98


were performed. To this end, Raman maps of G-I1*-1, G-I1*-5, and G-I1*-7 dipcoated on SiO2 wafers were carried out.

Figure 4.28 shows the Raman spectrum of G-I1*-1 with the D-, G-, D’-, and 2D-bandlocated at 1344, 1582, 1623, and 2680 cm−1, respectively. Upon exfoliation, the2D-band converges to a single highly symmetrical Lorentzian peak with an FWHMof 35 cm−1, which is indicative for SLG. Additionally, an increase of the D- andD’-band intensity parallels the exfoliation and is attributed to smaller flake sizes(ID/ID′=3.1) and, thus, to larger edge contribution, which provides crystal defectsfor the activation of the underlying double resonant Raman process.[288]

Figure 4.28: Selected solid-state Raman spectrum of G-I1*-1, indicating the exfolia-tion state of the sample. The sample was dip coated from dispersiononto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black curve - experimental data; blue curve - fitting.

Statistical evaluation of the I2D/IG ratio further confirms the high exfoliation state ofthe sample, indicating a distribution of 6% multilayered/bulk graphite and of about94% turbostratic exfoliated graphite and SLG within the sample - see Figure 4.29 a).Notably, shape analysis of the 2D-band reveal a single-layer content of about 25% -see Figure 4.29 b). This is in sharp contrast to G-1, where only very small amountsof SLG (<1%) were observed - see Chapter 4.2.2.

99


a)

b)

c)

Figure 4.29: Histograms resulting from the Raman mapping of G-I1*-1, showing a)counts versus I2D/ID, b) counts versus 2D-FWHM, and c) counts versusID/IG and the corresponding log-normal distribution, respectively. Thesample was dip coated from dispersion onto a Si/SiO2 wafer and excitedat 532 nm.

These findings indicate that the pretreatment process does not only affect the

100


graphene dispersibility of 1 but also the graphene exfoliation ability of 1 down toSLG. In particular, we assume that the intercalation of I1 and 1 into the grapheneedges facilitates the delamination of the graphene sheets from the bulk graphite byenlarging the interlayer space and, thus, disturbing the van der Waals interactionsbetween the graphene layers. As a consequence, higher amounts of SLG can beobserved. In summary, pretreating graphite with I1 followed by exfoliation andstabilization of graphene flakes in solutions of 1 led to higher amounts of few-and single-layer graphene (FLG+SLG: 94%, SLG: 25%) compared to exfoliationand stabilization of graphene in solutions of 1 without any pretreatment step(FLG+SLG: 78%, SLG: <1%). Additionally, the ID/IG ratio decreases for G-I1*-1(mean ID/IG=1.13) compared to G-1, indicating that larger graphene flakes arestabilized - see Figure 4.29 c).

Next, G-I1*-7 was investigated by means of Raman spectroscopy upon 532 nm laserexcitation. In Figure 4.30 a representative Raman spectrum is shown. The D-, G-,and 2D-band of G-I1*-7 dip coated on silicon wafers are centered at around 1346,1577, and 2694 cm−1, respectively.

Figure 4.30: Selected solid-state Raman spectrum of G-I1-1, indicating the exfolia-tion state of the sample. The sample was dip coated from dispersiononto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black dots - experimental data; blue curve - fitting.

Statistical Raman evaluation of the I2D/IG ratio reveals that the sample consists of

101


about 18% multilayered/bulk graphite and of about 82% few-layered/turbostraticgraphite and single-layer graphene - see Figure 4.31 a). Again, a large amount ofSLG is found (23%) - see Figure 4.31 b). Moreover, the relatively weak D-bandintensity (mean ID/IG=0.54) implies a low defect content and, thus, large flake sizes- see Figure 4.31 c).

a)

b)

c)


102


These findings are comparable with G-7. Hence, we assume that the pretreatmentof graphite in I1 has a negligible effect on the exfoliation of graphite in solutionsof 7 only. However, taking into account that the overall graphene concentrationincreases in the case of pretreatment, the concentration of SLG of 0.012 mg mL−1 inG-7 increases to 0.016 mg mL−1 in G-I1*-7, respectively.

Additionally, the exfoliation state of G-I1*-5 was of particular interest, as 5 onlyyielded stable graphene dispersions in the case of pretreatment. Figure 4.32 shows aselected Raman spectrum of G-I1*-5. Notably, a highly intense D-band arises uponexfoliation. We stress that this D-band is both narrower and less intense than the D-band reported in literature for graphene oxide and for reduced graphene oxide.[20,322]

However, the ID/ID‘ ratio in G-I1*-5 is 5.4, which indicates that besides edge relateddefects also some defects associated with sp3 hybridization may be present in thesample. Nevertheless, the 2D-band exhibits a highly symmetric Lorentzian shapewith an FWHM of 42 cm−1, indicating non-AB stacked FLG.

Figure 4.32: Selected solid-state Raman spectrum of G-I1*-5, indicating the exfolia-tion state of the sample. The sample was dip coated from dispersiononto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black line - experimental data; blue curve - fitting.

Along the same lines, statistical Raman analysis reveals a distribution of 51%multilayered/bulk graphite and about 49% turbostratic graphite and single-layergraphene within the sample - see Figure 4.33 a). Interestingly, 2D shape analysisreveals a single-layer content of about 18% - see Figure 4.33 b). This is consistent

103


with G-I1*-1 and G-I1*-7, where we also observed high amounts of SLG. However,the high mean ID/IG ratio of 1.9 indicates that 5 mainly stabilizes small grapheneflakes, which may also exhibit some sp3 defects in their basal plane - see Figure 4.33c).

a)

b)

c)


104


Taking these results into account, 5 is not suitable for the stabilization of high-qualityFLG, although it yielded the highest graphene concentrations and large amountsof monolayer graphene. However, especially the high amount of unexfoliated multi-layered/bulk graphite flakes present in the sample would require intense purificationof the sample in order to make use of the single- and few-layer graphene sheets.

In conclusion, Raman analysis revealed the successful exfoliation of pretreated graph-ite to yield few-layered/turbostratic graphite with high yields of SLG by employing1, 5, and 7 as stabilizers. However, in the case of 5 the high defect density and thefairly high yield of multilayered/bulk graphite sheets renders it unfavorable for thestabilization of high-quality few-layer graphene sheets.

In order to gain deeper insights into the structural characteristics of G-I1*-1 andG-I1*-7, TEM analysis was performed. Representative TEM images of G-I1*-1 areshown in Figure 4.34. In the upper part, a lacey carbon grid was used, which iscommonly used for TEM imaging of graphene flakes. The images prompt to thinpartially folded flakes with mean side lengths of up to 500 nm. In the lower row, weemployed a carbon coated lacey carbon grid, wherein the thin carbon coating servesas a substrate for thin graphene sheets and, thus, prevents them from falling throughthe grid. As such, very small flakes are observed next to large flakes, ranging frommean side lengths of 100 nm to 1 μm.

Figure 4.34: Representative TEM micrographs of G-I1*-1 with different magnifica-tions. The sample was drop coated either on a lacey carbon grid (upperpart) or on a carbon coated lacey carbon grid (lower part).

105


TEM analysis of G-I1*-7, however, reveals larger flake sizes even in the case of usinga carbon coated lacey carbon grid. In particular, thin few-layer graphene flakes areobserved, which exhibit lateral dimensions in the range of 250 nm up to 1 μm. Pleasenote that these measurements are rather qualitative than quantitative, however, theyfurther confirm the results gained from statistical Raman analysis.

Figure 4.35: Representative TEM micrographs of G-I1*-7 with different magnifica-tions. The sample was drop coated either on a lacey carbon grid (upperpart) or on a carbon coated grid (lower part).

Additionally, AFM measurements were carried out in order to evaluate the mor-phology of the graphene sheets. AFM characterization of surfactant-stabilizedgraphene suspensions is a difficult task due to the presence of surfactant residueson the substrate. Thus, the samples were washed with water and absolute ethanolimmediately after dip coating on a SiO2 wafer to remove any residual surfactantfrom the wafer. Figure A.12 in the Appendix shows the AFM image of a SiO2 wafer,which was dip coated in a solution of 1 and washed afterward. Any spots, which arenot discernible in this control image, will be assigned to graphene sheets. However,the presence of small amounts of residual surfactant cannot be excluded.

In Figure 4.36 the AFM image of G-I1*-1 hybrids is shown. The observed graphenesheets exhibit lateral dimensions ranging from 50 to 250 nm. These findings are ingood accordance with graphene sheets observed on a carbon coated lacey carbon gridin TEM imaging. Cross section analysis reveals that the graphene sheets mainly havethicknesses ranging from 1 to 3 nm. However, the theoretical thickness of 0.34 nmfor single-layer graphene is rarely found in experiments due to the fact that at such

106


high resolutions the apparent thickness obtained by AFM includes the chemicaland the van der Waals contrast. Moreover, there might be a layer of adsorbedwater or solvent between the graphene sheet and substrate. Furthermore, theinstrumental offset of approximately 0.5 nm always exists, which exceeds the thicknessof a single-layer graphene sheet. These three factors lead to an overestimation ofthe actual thickness of graphene sheets. Therefore, single-layer graphene sheetswith a thickness of approximately 1 nm have been commonly reported on Si/SiO2

substrates.[18,21,96,311,323] Taking these factors into account, the AFM measurementsclearly feature that the G-I1*-1 dispersion consists of FLG with high amounts ofSLG.

Figure 4.36: Atomic force microscopy image (left) of G-I1*-1 hybrids from a dipcoated and washed Si/SiO2 wafer and the corresponding height profiles(right).

Along the same lines, AFM analysis of G-I1*-7 confirms a high exfoliation state ofthe sample featuring thin graphene sheets with a thickness of 1-4 nm next to someaggregated graphene stacks - see Figure 4.37. However, the observed flakes havelarger dimensions ranging from 50 nm to 1 μm, which again is in good accordancewith the Raman and TEM analyses.

In summary, we developed an efficient and simple method for the preparation ofhighly stable and concentrated few-layer graphene dispersions. Pretreatment ofgraphite in solutions of tetraethylammonium hydroxide enabled the preparationof single- and few-layer graphene sheets upon sonication in different surfactantsolutions by the efficient interaction of deprotonated graphite edge functionalitiesand tetraethylammonium ions in alkaline solution. In particular, the graphenedispersibility of several surfactants could be improved by the pretreatment step,

107


leading to higher final graphene concentrations in every case. In fact, 5 yieldedthe highest graphene concentration of 0.095 mg mL−1, featuring superior stabilitywith only small amounts precipitating within the first three months (∼5%). Ingeneral, the stability of all surfactant-based dispersions was improved by utilizationof the pretreatment step, leading to highly stable graphene dispersions. Additionally,Raman, TEM, and AFM investigations confirmed that these dispersions consist ofhighly exfoliated graphene sheets, featuring large amounts of SLG. In particular,dispersions of G-I1*-1 revealed about 94% few-layered/turbostratic graphite andmonolayer graphene and about 25% SLG within the sample. This points to anoverall concentration of 0.068 mg mL−1 few- and single-layer graphene and about0.018 mg mL−1 SLG within the sample. It is expected that this highly efficientpreparation of graphene dispersions will further promote the production of highlyexfoliated graphene by liquid-phase processes and thereby enabling the applicationof graphene in important technologies, such as energy storage - see also Chapter 5.

Figure 4.37: Atomic force microscopy image (left) of G-I1*-7 hybrids from a dipcoated and washed Si/SiO2 wafer and the corresponding height profiles(right).

4.3.3 Dispersion of Graphene by in situ Polymerization ofVinylbenzyltrimethylammonium Chloride

As discussed in Chapter 4.3 the pretreatment of graphite in alkaline solutions oftetraalkylammonium hydroxide leads to an enhancement in the graphene dispersibilityof cac=0.072 mg mL−1 for G-I1*-1 through the interaction of tetraalkylammoniumcations with the deprotonated edge functionalities of graphite. Accordingly, the

108


halide salts of the respective tetraalkylammonium cations did not give rise to thesame results (cac=0.017 mg mL−1 for G-I2*-1). However, we found that pretreatinggraphite in the aromatic ammonium salt vinylbenzyltrimethylammonium chloride(VBTA) leads, in fact, to a remarkable enhancement of the graphene dispersibilityof 1, indicating a beneficial interaction of VBTA with the graphitic raw material(cac=0.043 mg mL−1 for G-VBTA*-1). Thus, we assumed that the interaction ofgraphite and VBTA takes place through π-π interactions between the aromaticbenzyl unit of the intercalant molecule, which may partially slip between the graphenesheets and the aromatic graphene basal plane. Such interactions have already beenreported for small aromatic molecules, such as pyrenes and anthracenes.[109,140,154]

Accordingly, a positive zeta potential of 23 mV for G-VBTA-1 was found, whichfurther indicates that positively charged VBTA are bound to the graphene surface(Figure A.5 in the Appendix). Please note that we did not observe positive zeta po-tentials in the case of pretreatment with several tetralkylammonium halide salts, suchas tetraethylammonium chloride, tetraheptylammonium bromide, and tetrapropyl-ammonium bromide. Although VBTA seemed to be beneficial for the exfoliation anddispersion of graphite and may partially engage in the stabilization of the dispersionsthrough π-π interactions, it is not able to disperse and stabilize graphite/grapheneby itself. This can be attributed to its small molecular structure, which does notprovide sufficient stabilization through repulsive steric interactions. Notably, it hasbeen reported that in situ polymerization of VBTA in the presence of SWCNTs[324]

and noncovalent stabilization of SWNTs by poly(vinylbenzyl)ammonium chloride(PVBTA)[325–327] lead to stable SWCNT dispersions.

Having this information in hand, the question arises if PVBTA may also be suitablefor the stabilization of graphene. And more interestingly, if the strong interactionof VBTA with graphene after the pretreatment of graphite in solutions of VBTAcan further be exploited for the exfoliation and dispersion of graphene throughan in situ polymerization process. Assuming that VBTA partially inserts intothe edges of graphene, it can be expected that these edges may further open upduring the polymerization of VBTA, which may, thus, present an additional drivingforce for the exfoliation of graphene besides the well-known mechanical exfoliationtechniques, such as ultrasonication. Notably, such processes have been recentlyemployed for the dispersion of graphene oxide to yield stable graphene oxide-polymercomposites.[96,328–332] Moreover, expanded graphite obtained by microwave treatment

109


of graphite intercalation compounds has been intensely studied for the productionof graphene/graphite polymer composites through in situ polymerization processesaccompanied by mechanical dispersion techniques.[333–335] Nevertheless, completeexfoliation of graphite to the level of individual graphene sheets with high structuraland chemical quality is yet to be accessed. As such, the in situ polymerizationof VBTA in pretreated graphitic material was employed in order to yield stabledispersions of non-oxidized graphene without the need of intense ultrasonication.

First, raw graphite was mixed with an aqueous solution of VBTA. Afterward, thedispersion was pretreated with ultrasonication for one hour and left to settle overnight.The next day the supernatant was discarded to leave the pretreated graphiticmaterial (G-VBTA*) without any free VBTA, which would also polymerize andadd redundant amounts of stabilizer to the sample. Next, water was added tothe pretreated graphite and a free-radical polymerization of VBTA,[336] which isexpected to interact strongly with the graphene basal plane and to be partiallyintercalated into the graphene edges, was carried out.

In order to gain insights into the effect of in situ polymerization on the grapheneinterlayer space, XRD patterns of the raw graphitic material, the pretreated graphite,and the in situ polymerized sample were recorded - see Figure 4.38.

Figure 4.38: XRD spectra of raw graphite, pretreated graphite in VBTA, and in situpolymerized pretreated graphite, respectively.

The XRD patterns show strong and sharp [002] peaks at 26.5◦ for all samples,corresponding to an ideal layered-structure with a d-spacing of 0.335 nm.[337] Ac-

110


cordingly, these findings indicated that VBTA does not fully intercalate into thegraphene layers during the pretreatment step, but might rather interact with thegraphene edges and partially open them as it is already reported for other smallaromatic molecules.[109,142] Upon polymerization two new peaks at 20.5◦ and 18.7◦

arise, corresponding to an interlayer spacing of 0.437 and 0.473 nm, respectively.This can be assigned to the attachment of the bulkier PVBTA along the grapheneedges, which disturbs the van der Waals interaction and enlarges the d-spacing of thegraphene layers. This is in good accordance with the interlayer spacing reported foredge-intercalated polystyrene graphene composites prepared by in situ polymerization(d=0.41 nm).[338]

In order to fully exfoliate and disperse the partially expanded graphite structure, thedispersion was mildly ultrasonicated for 1 hour, yielding stable graphene dispersionsof G-iPVBTA. The dispersions were subsequently centrifuged, and the absorptionwas measured at 660 nm. The concentration after centrifugation was calculatedas 0.026 mg mL−1. Notably, these concentrations are comparable with graphenedispersions prepared by 6 h of ultrasonication, employing 1 as a surfactant and, thus,demonstrate the efficient exfoliation and dispersion of graphene by utilization of thismild in situ polymerization process - see Figure 4.39.

In a control experiment, graphite was stirred overnight in solutions of PVBTA withdifferent molecular weight and was subsequently sonicated for 1 h. Afterward, thedispersions were treated under the same conditions as the ones from the in situpolymerization process, yielding transparent dispersions with very low grapheneconcentrations of 0.003 to 0.009 mg mL−1. These findings further revealed that thein situ polymerization of VBTA after the pretreatment of graphite in solutions ofVBTA is much more effective in the dispersion of graphene by means of PVBTAthan pure surfactant-based exfoliation.

111


Figure 4.39: Schematic representation of the exfoliation mechanism according tothe in situ polymerization approach. Firstly, selective interaction ofVBTA with the graphite layers during the pretreatment process. Sec-ondly, in situ polymerisation of VBTA. Thirdly, exfoliation of PVBTAintercalated graphite to yield PVBTA stabilized graphene flakes.

In order to gain further insights into the exfoliation state of the sample, Ramananalysis was performed. Figure 4.40 shows the Raman spectrum of G-iPVBTAdip coated on SiO2 wafers. Upon laser excitation of 532 nm, the typical Ramansignals for exfoliated graphene, namely the D-band at 1348 cm−1, the G-band at1584 cm−1, and the 2D-band at around 2702 cm−1 emerge. In particular, the2D-band converges into a highly symmetric shape with an FWHM of 59 cm−1,indicating few-layered/turbostratic graphite. Although it has been reported thatfree-radical polymerization of VBTA in the presence of SWCNT leads to covalentattachment of PVBTA to the SWCNT core, no signs for covalent functionalizationof the graphene sheets by PVBTA were observed by Raman spectroscopy. Forhigh covalent functionalization densities, high D-band intensities accompanied by adownshift and disappearance of the 2D-band and an overall broadening of the linewidth would be expected.[91,324] However, low densities of covalent functionalizationbesides noncovalent functionalization of graphene by PVBTA cannot be excluded.

Raman mapping further provided statistical insights into the sample, revealing about

112


41% multilayered/bulk graphite and about 59% few-layered/turbostratic graphiteand monolayer graphene within the sample - see Figure 4.41 a). Moreover, 2D-bandshape analysis indicates that about 12% of the graphene sheets exhibit single-layergraphene character - see Figure A.7 in the Appendix. Notably, relatively low D-bandintensities (mean ID/IG=0.63) indicate high structural quality of the sample resultingfrom the mild exfoliation procedure - see Figure 4.41 b). In summary, Ramananalysis proved the successful exfoliation of graphite towards FLG by the in situpolymerization approach.

Figure 4.40: Selected solid-state Raman spectrum of G-iPVBTA, indicating the ex-foliation state of the sample. The sample was dip coated from dispersionsonto a Si/SiO2 wafer and excited at 532 nm.

Unfortunately, due to high polymer coating on the graphene sheets, which could notbe removed by several washing steps and treatment with ionic exchangers, no TEM orAFM analysis could be performed. In this regard, excessive coating of the graphenesheets by the polyelectrolyte might also be challenging to remove during large-scaleproduction processes and, thus, renders this production procedure unfavorable forapplications where pure graphene sheets are required.

Since in situ polymerized graphene composites have not been the focus of this thesisand the polymer coating on the graphene sheets is unfavorable for supercapacitorapplications where a large accessible surface area is required, which would be blockedby the polyelectrolyte coating, no further experiments concerning this approach wereperformed. Nevertheless, we believe that these results further indicate the potential ofthe pretreatment of graphite with different ammonium salts and subsequent dispersion

113


of these pretreated graphite stacks by means of either sonication-assisted liquid-phase exfoliation processes, in situ polymerization processes, or other dispersiontechniques. As such, we presented a facile way to in situ polymerize partiallyintercalated graphite stacks yielding highly stable graphene dispersions in the formof few-layered/turbostratic graphite with moderate amounts of SLG. It is expectedthat by further improvement of this experimental concept new polymer-graphenecomposites can be developed, which may find a prominent position in electro- andphotoactive nanocomposites.[339]

a)

b)

Figure 4.41: Histograms resulting from the Raman mapping of G-iPVBTA, showinga) counts versus I2D/IG and b) counts versus ID/IG and the correspondinglog-normal distribution, respectively. The sample was dip coated fromdispersion onto a Si/SiO2 wafer and excited at 532 nm.

114

4.4 SOLVENT-BASED LIQUID-PHASE EXFOLIATION

4.4 Solvent-Based Liquid-Phase Exfoliation

One of the primary obstacles in achieving individual graphene layers is their insolubil-ity in aqueous medium and common organic solvents due to strong dispersion forcesbetween them provoking aggregation. Such aggregation can be overcome throughnoncovalent functionalization of graphene with water-soluble aromatic amphiphilesand surfactants, as already detailed in the previous chapters. In particular, bulkyamphiphilic naphthalene and pyrene derivatives have been demonstrated to be ef-ficient exfoliation agents in the liquid-phase exfoliation of graphite - see Chapters4.2 and 4.3. However, when aiming for applications that require pristine graphene,residual surfactant molecules may impair the device performance.

In order to address such requirements, the surfactant-free liquid-phase exfoliation ofgraphite in organic solvents seems to be straight forward. The success of exfoliationand subsequent stabilization of graphene in liquid media strongly depends on the smallnet energetic cost of the whole process.[21] This energy balance can be expressed as theenthalpy of mixing per unit volume, which becomes smaller the closer the grapheneand solvent surface energies are and, thus, the more favored the exfoliation andstabilization of graphene in the liquid media will be. As such, matching the surfaceenergies of graphene and a solvent is an important criterion for the successful solvent-based liquid-phase exfoliation of graphene. Coleman et al. pioneered the field ofsolvent-based liquid-phase exfoliation by testing a broad range of solvents with varyingsurface energies, further confirming that solvents with a surface tension close to40 mN m−1 are most suitable for the exfoliation and stabilization of graphene,[21] whichmatched the reported values for the surface tension of graphite.[340,341] These findingshave implied a surface energy criterion for graphene exfoliation, which has furtherbeen validated by several posterior studies.[342,343] However, these solvents sufferfrom drawbacks originating from their high boiling point that urges for additionalprocessing to remove solvent residues from electrode layers upon deposition.[21]

Another crucial aspect to consider is the high toxicity of the solvents, such asN-methyl-2-pyrrolidone and dimethylformamide (DMF).

As the main aim of this study is to find suitable stabilization for few-layer grapheneflakes in liquid media, which could further be scaled-up in an industrial manner,we believe that it is of crucial importance to exfoliate graphite into graphene usingnon-toxic or low-toxic solvents. Moreover, the use of harmful organic solvents should

115


be restricted to protect the environment and guarantee industrial safety, even thoughlower graphene concentrations may be achieved.

This chapter focuses on the surfactant-free solvent-based liquid-phase exfoliation ofgraphite in order to gain sufficient insights into the exfoliation and dispersion ability ofthose approaches compared to surfactant-based liquid-phase exfoliation. Concerningscalability, solvents with low toxicity are of particular interest. In terms of theapplication of the prepared graphene sheets as electrode materials in supercapacitors,low boiling points are required.

4.4.1 Dispersion of Graphite in Different Organic Solvents

In a first attempt, we examined the literature for low-toxic solvents, which aresuitable for the exfoliation and dispersion of graphene in liquid-phase processes and,thus, would meet the requirements for large-scale production. A summary of commonsolvents utilized in the solvent-based liquid-phase exfoliation of graphite is displayedin Table A. In close coordination with the technical team responsible for the scale-up ethyl acetate (EAC), dibenzyl ether (DBE), 1,3-dioxolane (DOL), dimethylphthalate (DMP), ethylene gylcol (EG), butyl acetate (BA), γ-butyrolactone(GBL), and terpineol (TP) were selected. Please note that exfoliation in polarsolvents, such as ethanol, isopropyl alcohol, acetone, tert-butanol, and water didnot yield stable graphene dispersions after centrifugation. In a control experiment,N-methyl-2-pyrrolidone (NMP) was tested as it is the most widely used organicexfoliation agent due to its well-matched surface tension and Hansen solubilityparameters and, thus, serves as a benchmark.[343,344]

To study the effectiveness of the selected solvents towards the exfoliation and dis-persion of graphene, graphite was dispersed in the respective solvent by stirringthe dispersion overnight followed by sonication. Afterward, the suspensions werecentrifuged to remove any large unexfoliated graphite. The same procedure wasapplied to all solvents, and no further processing was carried out at this stage tomaintain comparability. The post-centrifugation concentration of dispersed mate-rial was used as an indicator for the graphene dispersibility of a solvent. To thisend, the absorption at 660 nm was measured, and the corresponding concentrationafter centrifugation cac was calculated using the extinction coefficient for graphenedispersions in solvents reported as ε=2490 mL mg−1 m−1 at 660 nm.[21]

116


The measured concentrations vary from 10−4 mg mL−1 in GBL to 0.038 mg mL−1

in DMP - see Figure 4.42. Interestingly, DMP yielded the highest grapheneconcentration of 0.038 mg mL−1 in our study, which was even higher than thegraphene concentration observed in NMP. This is in sharp contrast to earlier findingsby Hernandez et al., which reported on higher graphene concentrations in the caseof NMP and even in the case of GBL in comparison to DMP, respectively.[21]

Figure 4.42: Calculated graphene concentration after centrifugation cac of graphenedispersions prepared by different solvents, namely ethyl acetate (EAC),dibenzyl ether (DBE), 1,3-dioxolane (DOL), dimethyl phthalate(DMP), ethylene gylcol (EG), butyl acetate (BA), γ-butyrolactoneGBL, and terpineol (TP), respectively.

Nevertheless, DMP features a surface tension of ∼40 mN m−1, which is expected tobe highly beneficial for the exfoliation and dispersion of graphene. The nonconformityof the observed trends can be assigned to the fact that Hernandez et al. used verymild centrifugation conditions (500 rpm for 90 min). As already discussed in Chapter4.1, such centrifugation conditions leave large flakes and aggregates in solution and,thus, larger multilayer graphene stacks are contributing to the final concentration nextto thin graphene sheets. In contrast, we applied very harsh centrifugation conditions(15,000 rpm for 10 min), which are expected to leave smaller and thinner grapheneflakes in the supernatant. As such, we assume that the graphitic material aftercentrifugation is not fully comparable, and as the stabilization ability of the solvents

117


towards certain flake thicknesses and flake sizes are expected to vary, different resultsare observed. Moreover, we stress that we used solvents of technical grade rather thananalytical grade in order to gain insights into the dispersion and exfoliation abilityof those materials. However, these solvents are expected to contain contaminations,such as additives, by-products, and varying amounts of residual water, which mayfurther impact the stabilization and exfoliation of graphene. This might also explainthe relatively high graphene dispersibility of 0.016 mg mL−1 in EG, which has asurface tension (49 mN m−1) that is expected to be non-ideal for graphene dispersion.In fact, control experiments with EG of analytical grade did not lead to such highconcentrations but were rather unstable after centrifugation. In this context, westress that not only the molecular structure and the chemical properties of a givensurfactant or solvent are important for the efficient exfoliation and dispersion ofgraphene but also material quality and purity.

118


4.4.2 Exfoliation Ability of Different Organic Solvents

In order to gain further insights into the structural quality and exfoliation stateof the samples, Raman spectroscopy analysis was performed. A selected Ramanspectrum of G-DMP is shown in Figure 4.43, displaying the D-, G-, and 2D-bandlocated at 1351, 1582, and 2694 cm−1, respectively.

Figure 4.43: Selected solid-state Raman spectrum of G-DMP, indicating the exfoli-ation state of the sample. The sample was dip coated from dispersionsonto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black dots - experimental date; blue curve - fitting.

Upon ultrasonication-assisted exfoliation, the 2D-band converges to a symmetricalLorentzian peak with an FWHM of 58 cm−1, which is indicative for turbostraticallyexfoliated graphene. Additionally, a slight increase of the D and D’-band intensitiesis observed, which can be attributed to ultrasonication induced scission as alreadydiscussed for previous samples (ID/ID′=2.5).

Further, statistical evaluation of the I2D/IG ratio reveals a rather low exfoliation stateof the sample, indicating that it consists of about 55% multilayered/bulk graphiteand of about 45% few-layered/turbostratic graphite with single-layer graphene - seeFigure 4.44 a). Moreover, shape analysis of the 2D-band reveals a single-layer contentof about 3%, which is in good accordance with previous reports on solvent-basedexfoliated graphene samples - see Figure A.8 in the Appendix.[21,285] Notably, relative

119


weak D-band intensities (mean ID/IG=0.60) imply a low defect density within thesample and, thus, revealing the integrity of the sp2 network - see Figure 4.44 b).

a)

b)

Figure 4.44: Histograms resulting from the Raman mapping of G-DMP, showing a)counts versus I2D/IG and b) counts versus ID/IG and the correspondinglog-normal distribution, respectively. The sample was dip coated fromdispersion onto a Si/SiO2 wafer and excited at 532 nm.

Next to G-DMP also G-EG was examined by Raman spectroscopy. In Figure 4.45a representative Raman spectrum is shown. The D-, G-, and 2D-bands of G-EGdip coated on a silicon wafer are centered at around 1344, 1583, and 2683 cm−1,respectively.

120


Figure 4.45: Selected solid-state Raman spectrum of G-EG, indicating the exfoliationstate of the sample. The sample was dip coated from dispersions ontoa Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the 2D-band.Black dots - experimental date; blue curve - fitting.

Statistical Raman evaluation of the I2D/IG ratio reveals that the sample consists of36% multilayered/bulk graphite and about 64% few-layered/turbostratic graphiteand, thus, features a higher exfoliation degree than G-DMP - see Figure 4.46a). Noteworthy, none of the spectra features a 2D-FWHM <40 cm−1 indicatingthat no SLG was probed within the sample - see Figure A.9 in the Appendix.However, it remains unclear, if this arises from the sample preparation process,which allows exfoliated graphene to restack during the slow solvent evaporation,or if it is an intrinsic feature of the sample. Interestingly, rather high D-bandintensities are observed (mean ID/IG=1.6), which arise from the scission of graphenesheets during the ultrasonication process and, thus, a larger contribution of edgedefects (ID/ID′=3.0) indicate smaller flake sizes in comparison to G-DMP - seeFigure 4.46 b).

121


a)

b)

Figure 4.46: Histograms resulting from the Raman mapping of G-EG, showing a)counts versus I2D/IG and b) counts versus ID/IG and the correspondinglog-normal distribution, respectively. The sample was dip coated fromdispersion onto a Si/SiO2 wafer and excited at 532 nm.

To allow a comparison with previous studies, Raman analysis of G-NMP was carriedout. Figure 4.45 shows a selected Raman spectrum of G-NMP dip coated on asilicon wafer and excited at 532 nm. The typical features of exfoliated graphite,namely the D-, G-, and 2D-bands, arise at 1344, 1580, and 2689 cm−1, respectively.

122


Figure 4.47: Selected solid-state Raman spectrum of G-NMP, indicating the exfoli-ation state of the sample. The sample was dip coated from dispersionsonto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black dots - experimental date; blue curve - fitting.

Conclusions of the statistical evaluation of the I2D/IG are comparable to those derivedfrom G-EG. In particular, 34% of the spectra reveal multilayered/bulk graphitecharacter and about 66% featured few-layered/turbostratic graphite character withsingle-layer graphene - see Figure 4.48 a). Again, 2D-band shape analysis revealsonly very small amounts of single-layer graphene (<0.1%) - see Figure A.10 in theAppendix. Moreover, relative low D-band intensities confirm the high structuralquality of the sample (mean ID/IG=0.58) - see Figure 4.48 b).

To summarize the results so far, graphene was dispersed in several low-toxicitysolvents, namely EAC, DBE, DOL, DMP, EG, BA, GBL, and TP and thedispersion efficiencies of the solvents were compared to those of NMP. The grapheneconcentrations varied significantly by a factor of 10, ranging from 10−4 mg mL−1 forGBL to 0.038 mg mL−1 for G-DMP, thus, rendering DMP as the best solvent forgraphene dispersion. Interestingly, nonconformities in the correlation between surfacetension and graphene concentration could be found, which are in contradiction toprevious reports. However, this can be correlated with the technical grade of thesolvents used. These are expected to be contaminated with additives, byproductsand/or varying amounts of residual water, which affect the stabilization of graphenein solution.

123


a)

b)

Figure 4.48: Histograms resulting from the Raman mapping of G-NMP, showing a)counts versus I2D/IG and b) counts versus ID/IG and the correspondinglog-normal distribution, respectively. The sample was dip coated fromdispersion onto a Si/SiO2 wafer and excited at 532 nm.

Moreover, it was found that the solvent only weakly influences the degree of exfoliation.As revealed by Raman analysis of the flake thickness distribution, all samples exhibitfew-layer character with large amounts of multilayered/bulk graphite sheets and onlyvery low yields of SLG. Comparing these results to those derived from surfactant-assisted liquid-phase exfoliation, we conclude that the solvent-based exfoliationprocess is less effective in the exfoliation of graphite down to few- and single-layergraphene. In fact, it is expected that the surfactants strongly impact the exfoliation of

124


graphite by slipping between the edges of the graphene layers and, thus, contributingto the delamination of graphite next to the ultrasonication induced exfoliation.However, solvents are expected to have a weaker effect on the exfoliation, whichis enabled by the low energetic cost when the solvent’s surface tension approachesaround 40 mN m−1, but mainly driven by the energy input by ultrasonication.As such, this approach is less effective in the exfoliation of graphite. Additionalevaluation of the defect-related D-band indicated that defect content of the samplesincreases in the order of NMP < DMP < EG, which could be assigned to thereduction of flake sizes during the ultrasonication process by analysis of the ID/ID′

ratio. As such, NMP and DMP are expected to stabilize larger graphene flakes,whereas EG stabilizes mainly small graphene flakes.

Unfortunately, the best solvents for graphene dispersion within this study are non-volatile. This can be attributed to the fact, that the energetic cost of exfoliationincreases as the solvent’s Hildebrand solubility parameter approaches 23 MPa1/2 orin other words the surface tension approaches about 40 mN m−1.[343,345] Accordingto Trouton’s rule, the solubility parameter of a solvent is directly linked to its boilingpoint through the enthalpy of vaporization and, thus, the best solvents for grapheneexfoliation usually have high boiling points.[116] However, this causes problems whenprocessing them in electrode layers and renders them unfavorable for the liquid-phaseproduction of graphene for supercapacitor applications.

4.4.3 Graphene Dispersion by a Cosolvent Approach

As already discussed in Chapter 4.4, the favored solvents for solvent-based liquid-phase exfoliation of graphite have high boiling points, which renders it difficultto process the graphene sheets derived. As such, the dispersion of graphene inlow boiling point solvents, such as water and alcohols is of paramount importance.However, on the one hand, the surface tension of water (72.75 mN m−1) is too highto fulfill the surface tension criterion for graphene exfoliation and, thus, surfactantshave to be employed to tune the water surface energy to a proper level for aqueousliquid-phase exfoliation of graphite. On the other hand, common non-toxic lowboiling solvents, such as ethanol, isopropyl alcohol, and acetone possess surfacetensions (22-25 mN m−1), which are too low to yield stable graphene dispersions -

125


see Chapter 4.4. As such, the liquid-phase exfoliation of graphene in volatile solventshas been met with very limited success.[116,346]

In order to disperse graphene in non-toxic low boiling solvents, such as water oralcohols, according to the surface tension criterion, a cosolvency approach was tested.Cosolvency is a well-studied phenomenon in the field of polymer and drug science. Ithas been demonstrated that the solubility of a material can be greatly improved byusing a specific mixture of solvents while showing no or low solubility in either of theindividual solvents.[347–350] As such, we assumed that the right solvent mixture, whichfulfills the surface tension criterion, could effectively disperse graphene sheets in liquidmedia. Indeed, it has been shown that inorganic graphene analogues, such as MoS2

and WS2, can be exfoliated using a solvent mixture of water and ethanol.[351,352]

To this end, we measured the surface tension of different ethanol-, isopropyl alcohol-, and acetone/water mixtures in order to determine the solvent/water ratios forideal graphene dispersion at around 40 mN m−1 - see Figure A.11 in the Appendix.In order to testify the dispersion and exfoliation ability of the solvent mixtures,graphite was dispersed in mixtures of ethanol/water (3:20), acetone/water (7:25), andisopropyl alcohol/water (1:10). The mixtures were stirred overnight and subsequentlyultrasonicated. Afterward, the dispersions were centrifuged in order to removeany poorly exfoliated material. Although TEM analysis reveals that thin few-layergraphene sheets could be exfoliated by this approach - see Figure 4.49, the dispersionswere highly unstable and precipitated within one day. These findings indicate thatadjusting the surface tension of a solvent/water mixture to the ideal surface tensionfor graphene exfoliation of about 40 mN m−1 benefits the exfoliation of graphitetowards few-layer graphene sheets. However, fast agglomeration of the exfoliatedgraphene sheets in the polar media takes place due to hydrophobic effect. Hence, nostable graphene dispersions could be prepared in the solvent/water mixtures.

126


Figure 4.49: Representative TEM micrographs of few-layer graphene sheets stabilizedin an acetone/water mixture (7:25). The sample was drop coated on alacey carbon grid.

As already discussed in Chapters 4.4 and 4.2.1, graphene exfoliation agents, suchas surfactants or solvents, influence the exfoliation of graphite to graphene in twodifferent ways. Firstly, they adjust the liquid-vapor interfacial energy of the solutionto an optimum range corresponding to the energy required to separate graphenesheets beyond the range of van der Waals interactions and, thus, enabling theexfoliation. Secondly, another key requirement for the long-term stability of theas-prepared graphene dispersions is the exfoliation agent’s ability to colloidal stabilizegraphene. For this purpose, the graphene exfoliation agent interactions should besufficiently strong to compensate for the enormous van der Waals attractive forcespresent between the graphene sheets. Additionally, the exfoliation agent should createa repulsive layer on the graphene surface and, thereby, prevent the reaggregation ofthe sheets. In fact, it has been shown by Shih et al. that the favored solvents forgraphene exfoliation, such as NMP, DMF, and GBL, have a very strong affinity for thegraphene basal plane, and the interaction force between two graphene sheets changefrom attractive to repulsive upon adsorption of those solvent molecules. However,among the tested solvents, water clearly showed the lowest energy barrier for thedesorption from the graphene sheets leading to the strongest interlayer van der Waalsinteractions, thus, rendering it the least efficient solvent for stabilizing graphene.[310]

127


In fact, these findings explain our observation that after successful exfoliation ofgraphene by ultrasonication in solvent/water mixtures with ideal surface tension, thegraphene sheets lose their dispersibility, aggregate and finally precipitate during avery short time. Thus, we assume that solvent/water mixtures are not capable ofdispersing graphene in a stable manner unless suitable stabilizers are added.

4.4.4 Exfoliation of Pretreated Graphite in a CosolventMixture

In a further attempt to disperse graphene in non-toxic low boiling solvents, thecosolvency approach was applied to pretreated graphite. As already discussed inChapter 4.3 pretreatment of graphite with tetraethylammonium hydroxide can im-prove the graphene dispersibility of several surfactants by opening the graphene edgesand, thus, favoring the exfoliation of graphite down to SLG. Moreover, pretreatinggraphite with tetraethylammonium hydroxide led to an enhanced dispersion stabilityarising from the deprotonation of graphene edge functionalities.

Accordingly, graphite was pretreated in a solution of tetraethylammonium hydroxideby 1 hour of ultrasonication. Afterward, the dispersion was left to settle, and thesupernatant was discarded the next day. In order to remove any residual water,the pretreated graphite was dried. Then, the pretreated graphite was dispersed inan ethanol/water solvent mixture (3:20) and stirred overnight. Next, ultrasonica-tion was applied, and the resulting graphene dispersion was centrifuged to removeany unexfoliated graphitic material leaving a stable opaque dispersion of G-I1*-EtOH/W. The graphene concentration after centrifugation was calculated fromthe adsorption at 660 nm using the extinction coefficient for water-based graphenedispersions revealing a final graphene concentration of 0.039 mg mL−1. Notably,these dispersions were highly stable with only very small amounts precipitatingwithin one month after preparation (∼3% precipitation). Thus, these dispersions arecomparable with those derived from surfactant-assisted liquid-phase exfoliation withamphiphilic aromatic stabilizers in terms of graphene dispersibility - see Chapter 4.2.Notably, they even outperform them in terms of long-term stability. Nevertheless,concentrations achieved with both, that is, the pretreatment step and surfactant-assisted liquid-phase exfoliation by amphiphilic aromatic stabilizers featured highergraphene concentrations by a factor of 2. This can be attributed to the fact, that the

128


stabilizer molecules can actively take part in the exfoliation of graphene by slippingbetween adjacent graphene layers and, thus, peeling them apart, which has beenreported previously.[109] Moreover, the stabilizer molecules build a repulsive layeron the graphene surface, which efficiently prevents as-exfoliated graphene sheetsfrom restacking and reagglomeration and, thus, enable the production of higherconcentrated graphene dispersions.

In order to gain quantitative insights into the exfoliation state of the sample, Ramanmapping analysis was performed. Figure 4.50 shows a selected Raman spectrumof G-I1*-EtOH/W with the D-, G-, and 2D-bands located at 1347, 1582, and2693 cm−1, respectively.

Figure 4.50: Selected solid-state Raman spectrum of G-I1*-EtOH/W, indicatingthe exfoliation state of the sample. The sample was dip coated fromdispersions onto a Si/SiO2 wafer and excited at 532 nm. Inset: Fittingof the 2D-band. Black dots - experimental date; blue curve - fitting.

Statistical evaluation of the I2D/IG ratio reveals that the sample consists of about40% multilayered/bulk graphite and about 60% few-layered/turbostratic graphitewith single-layer graphene - see Figure 4.51 a). Notably, shape analysis of the2D-band gives rise to a single-layer content of about 15% - see Figure 4.51 b).These findings are in line with the ones derived from surfactant-assisted liquid-phaseexfoliation of pretreated graphite, which also featured high yields of SLG. Hence,we assume that the pretreatment of graphite with tetraethylammonium hydroxidefacilitates the exfoliation of graphene down to SLG. This can, in fact, be explained

129


by the mechanism of partial intercalation of tetraethylammonium hydroxide into thegraphene edges, which we proposed previously - see Chapter 4.3. Moreover, relativelylow D-band intensities (ID/IG=0.70) imply a low defect content and, thus, large flakesizes (ID/ID′=2.8) - see Figure 4.51 c).

a)

b)

c)

Figure 4.51: Histograms resulting from the Raman mapping of G-I1*-EtOH/W,showing a) counts versus I2D/IG, b) counts versus 2D-FWHM, andc) counts versus ID/IG and the corresponding log-normal distribution,respectively. The sample was dip coated from dispersion onto a Si/SiO2wafer and excited at 532 nm.

130


In summary, Raman analysis revealed a fairly good exfoliation state of the sample,which predominantly consists of turbostratic graphite with moderate amounts offew- to multilayered and bulk graphite and considerable high amounts of SLG beingdispersed in the sample.

To gain further insights into the structural characteristics of G-I1*-EtOH/W, TEManalysis was carried out. Representative TEM images of G-I1*-EtOH/W areshown in Figure 4.52. In the upper part, a lacey carbon grid was used as a substrate,whereas the images of the lower part were carried out on a carbon coated substrate inorder to analyze smaller graphene flakes. In particular, small thin flakes in the rangeof 50-250 nm, as well as larger few-layer graphene flakes with lateral dimensions onthe order of 250 nm up to 2 μm appear folded and partially restacked to minimizesurface energy.

Figure 4.52: Representative TEM micrographs of G-I1*-EtOH/W with differentmagnifications. The sample was drop coated on a lacey carbon grid.

In summary, we presented the successful exfoliation and dispersion of pretreatedgraphite in an ethanol/water mixture of a surface tension close to 40 mN m−1

towards FLG. In detail, highly stable graphene dispersions with concentrations ofup to 0.039 mg mL−1, which show very small amounts of graphene precipitationof 3% within the first month of controlled storage were achieved. Further, Ramanand TEM analysis confirmed the successful exfoliation down to turbostratic graphiteand SLG with some amounts of unexfoliated multilayer graphene stacks being stillpresent. The superior exfoliation and dispersibility characteristics of the cosolvent

131


exfoliation of pretreated graphite can be explained by the co-operative effects ofideal surface tension, partial intercalation of tetraethylammonium ions and openingof graphene edges, and the stabilization of the prepared graphene sheets by repulsiveinteractions of the deprotonated graphene edges. Nevertheless, this approach stillfalls behind the surfactant-based liquid-phase exfoliation presented in Chapters 4.2and 4.3. In fact, this can be attributed to enhanced repulsive steric interactionsbetween surfactant-coated graphene flakes.

132

5 Application of Liquid-PhaseExfoliated Graphene inSupercapacitors

Supercapacitors represent a promising technology for energy storage and mobilepower supply. On the one hand, they feature superior power density and cyclelife, on the other hand, however, they are limited in their energy density, which isat least one order of magnitude lower than those of conventional batteries.[194,353]

The supercapacitor’s electrode material is the primary determining factor of itsperformance and in particular its energy storage capacity. Recently, great efforts havebeen dedicated to the development of new electrode materials, which can significantlyincrease the device’s energy density without any sacrifice in the power density or cyclelife.[194,217,224,354] The key requirements for such high-performance supercapacitorelectrodes are high electrical conductivity, high ion-accessible surface area, highionic transport rate, and high electrochemical stability.[355] As carbon materials areexcellent conductors, chemically stable, and have a large surface area, they are thematerial of choice for supercapacitors. Today, commercially available supercapacitorsmainly utilize porous activated carbons as electrode material featuring a typicalgravimetric capacitance of 80-120 F g−1 and a stack energy density of 4-5 Wh kg−1,which is inferior to lead acid batteries (25-35 Wh kg−1).[355,356]

Recently, graphene has attracted great interest as an electrode material for super-capacitors due to its excellent and unique properties, such as high chemical and ther-mal stability, superior electrical conductivity, high mechanical flexibility and strength,broad electrochemical window, an exceptionally large surface area of 2630 m2 g−1,and a theoretical capacitance of 550 F g−1.[11,225,282,357,358] Many attempts have beenmade to utilize graphene as an electrode material for supercapacitors recently. Theymainly have focused on the use of reduced graphene oxide and functionalized graphenematerials instead of pristine graphene as those materials can already be manufacturedat high scale and low cost.[47,57,274,359,360] However, in contrast to pristine graphene,functionalized and reduced graphene oxide sheets produced by thermal, chemical,or laser-induced reduction of graphite oxide contain residual functional groups and

133

5.1 LIQUID-PHASE EXFOLIATED GRAPHENE ELECTRODES - MATERIALOVERVIEW

lattice defects. This strongly affects the electrical conductivity of the electrodematerial and, thus, limit the achievable power density.[20,47] Therefore, a low-costscalable production of non-oxidized pristine graphene is of paramount importance tofully exploit the potential of graphene as an electrode material in supercapacitorstowards high power and energy density devices. In regard to the production ofnon-oxidized pristine graphene, liquid-phase exfoliation of graphite has proven tobe superior in terms of scalability, production cost, and defect density.[111] However,the low yield in single-layer graphene still limits its utilization in applications thatrequire a high surface area, such as supercapacitors.[361] In order to simultaneouslyachieve high specific capacitances, which are comparable to those reported for highlyexfoliated reduced graphene oxide, and a high power density of the device, largesurface area pristine graphene with an increased fraction of single-layer graphene isstrongly needed.[362]

In the previous chapters, we discussed the liquid-phase exfoliation of graphene towardshighly exfoliated graphene dispersion with relatively large amounts of single-layergraphene. This was accomplished by employing different stabilization approaches,such as the stabilization with aromatic amphiphiles (e.g. 7) and utilizing an auxiliarypretreatment step, which further facilitated the exfoliation of graphene down tosingle-layer graphene. In order to study the potential of these liquid-phase exfoliatedgraphene (LPEG) materials as electrode materials for high power and energy densitysupercapacitors, this chapter focuses on their application in supercapacitors.

5.1 Liquid-Phase Exfoliated Graphene Electrodes -Material Overview

5.1.1 Evaluation of Flexible Graphene Films

In the first step, the electrical properties of the exfoliated graphene sheets weretested in order to evaluate their applicability as electrode materials. To this end, theas-prepared and centrifuged graphene dispersions were concentrated, and desaltedusing a centrifugal dialysis device - see Figure 5.1. By repeatedly using this approach,the concentration of the graphene dispersions could be increased by up to 80 timesyielding graphene concentrations of 2-5 mg mL−1. Moreover, free surfactants and

134


intercalants are removed from the dispersion, which would not only disturb theelectrical measurements but also extensively coat the graphene sheets, thus, blockingthe ion-accessible surface area in the supercapacitor device. Please note that this isa very mild approach for the concentration of graphene dispersions as the graphenesheets remain dispersed and do not restack and agglomerate as it would occur in thecase of vacuum filtering or solvent evaporation.

Figure 5.1: Schematic representation of the preparation procedure for LPEG super-capacitors. Firstly, the graphene dispersions are concentrated using acentrifugal dialysis device. Secondly, the graphene dispersion is dropcoated onto a PET film or Au-coated polyimide film, respectively. After-ward, the samples were dried yielding a thin graphene film. Thirdly, thecoated graphene films are assembled into supercapacitor cells.

In particular, the dispersions of G-7, G-I1*-1, G-I1*-5, and G-I1*-7 were concen-trated in order to evaluate their applicability as electrode materials in supercapacitors,as those graphene dispersions gave rise to the highest single-layer graphene content- see Chapters 4.2 and 4.3. Unfortunately, G-I1*-EtOH/W could not be concen-trated as the graphene sheets tended to reagglomerate in the dispersion throughoutthe concentration process. This can be attributed to the fact that the graphenesheets in G-I1*-EtOH/W are not stabilized by any surfactants, which would in-troduce strong repulsive interactions. This would limit the reagglomeration of thegraphene sheets when they approach each other during the reduction of the solventvolume. As a matter of fact, these non-stabilized graphene sheets tend to restack and-agglomerate at higher concentrations. Additionally, G-CTAB was concentratedand tested towards its applicability as an electrode material in supercapacitors inorder to allow the comparison to previous studies on liquid-phase exfoliated graphenesheets in terms of material quality and suitability.

Accordingly, thin films of the concentrated graphene dispersion of G-7, G-I1*-1, G-I1*-5, G-I1*-7, and G-CTAB were drop cast onto a flexible polyethyleneterephthalate (PET) substrate - see Figure 5.1. Figure 5.2 a) shows a typical

135


current-voltage (I-V) curve of a graphene film prepared by G-I1*-7 at room temper-ature. The resulting graphene films showed excellent conductivities ranging from679 S m−1 for G-CTAB to 2989 S m−1 for G-I1*-7, respectively, which are in goodaccordance with the values reported in the literature for graphene films fabricatedfrom liquid-phase exfoliated graphene[119,139,305,363,364] and superior to those fromlaser-scribed[274] and flash converted[359] graphene oxide, as well as activated carbons(10 to 100 S m−1).[365] The latter is the state-of-the-art material used in commercialsupercapacitor devices.

Moreover, the graphene films were tested towards their mechanical flexibility. In thiscontext, the graphene films were bend using a PET ring - see inset Figure 5.2 b).Notably, the graphene films show excellent mechanical flexibility and structuralintegrity during bending - see Figure 5.2 b). In particular, the electrodes show adecrease in the electrical resistance of ∼2.5% upon bending to a 5 mm radius, whichcan be attributed to the fact that the layers slide alongside each other upon bending,which leads to an enhanced overlap of the graphene sheets as was previously reportedfor reduced graphene oxide (rGO) and graphene films.[274,280] Given the high yield ofsingle-layer graphene and the excellent electrical properties, LPEG can be directlyused as a supercapacitor electrode without the need for any additional binder orconductive additive.

a) b)

Figure 5.2: a) Electrical characterization of a LPEG electrode. b ) The electricalresistance change of a LPEG electrode as a function of the bending radius.The inset shows the electrode upon bending. The electrode was preparedby drop coating G-I1*-7 on a polyethylene terephthalate (PET) flexiblesupport.

136


5.1.2 Evaluation of the Electrochemical Performance ofLPEG-Based Supercapacitors

In order to demonstrate the superior performance of LPEG electrodes in super-capacitor devices, symmetrically LPEG supercapacitors were assembled by sandwich-ing an ion porous separator between two identically LPEG electrodes drop castedonto an ultra-thin gold current collector and using an aqueous liquid gel electrolyte(H2SO4/PVA) as electrolyte. The devices are super-thin with a total thickness of<100 μm and, thus, are referred to as micro-supercapacitors. The supercapacitorperformance of graphene electrodes made from dispersions of G-7, G-I1*-1, G-I1*-5, G-I1*-7, and G-CTAB was analyzed by cyclic voltammetry and galvanostaticcharging-discharging experiments, respectively.

Figure 5.3 a) shows the cyclic voltammetry (CV) curves of the supercapacitors madefrom G-7, G-I1*-1, G-I1*-5, G-I1*-7, and G-CTAB. In general, the shape of the CVcurve should be rectangular for an ideal electrical double-layer supercapacitor.[195,366]

In comparison to the pure and the surfactant-coated Au current collector (FigureA.13 in the Appendix), the LPEG supercapacitors show enhanced electrochemicalperformance revealing a nearly rectangular shape at a scan rate of 100 mV s−1,which is indicative of a nearly ideal and reversible capacitive behavior and a lowcontact resistance. In particular, G-I1*-7 shows the largest curve area and, thus,the highest specific capacitance, which was calculated as 170 F g−1 at a scan rate of100 mV s−1 and is superior to those reported previously for supercapacitors madefrom liquid-phase exfoliated graphene.[109,154,280] In comparison, G-CTAB gave riseto the lowest specific capacitance of 81 F g−1 at 100 mV s−1. In addition, theLPEG supercapacitors can be charged and discharged over a wide range of scanrates (100 to 10,000 mV s−1), revealing their remarkably fast scan rate capability- see Figure 5.3 b) and c). Notably, the CV shape and, thus, the capacitance ofactivated carbon-based electrodes rapidly degrade as the voltage scan rate increasesbeyond 50 mV s−1.[367] In comparison, the CV shape of the LPEG devices remainsrectangular with a capacitance retention of 80% even at an increased scan rateof 1000 mV s−1 - see Figure 5.3 c). Moreover, the linear increase in current withincreasing scan rate indicates that the charge is primarily of non-faradaic nature,and no contributions of pseudocapacitance arising from functional groups of thegraphene sheets or surface moieties are observable - see Figure 5.3 d).[368]

137


a) b)

c) d)

Figure 5.3: Evaluation of LPEG supercapacitors in aqueous H2SO4/PVA electrolyte.a) Cyclic voltammetry (CV) of G-7, G-I1*-1, G-I1*-5, G-I1*-7, andG-CTAB supercapacitors at a scan rate of 100 mV s−1, respectively.The rectangular CV shape indicates an efficient double-layer formation.b) and c) CV profiles of a G-I1*-7 electrochemical capacitor in aqueousH2SO4/PVA electrolyte at different scan rates. d) The dependency ofthe capacitative current (extracted from the CV profiles at 0.5 V forthe charge and discharge curves) on the applied scan rate. A linearrelationship is observed with R2= 0.997 and 0.994 for the charge anddischarge curves, respectively.

Figure 5.4 a) shows the galvanostatic charging-discharging curves (CDC) of theLPEG supercapacitor devices at a constant current of 5 mA (corresponding toa charge/discharge rate of ∼8 A/g). All samples show rapid potential changesduring the charging-discharging process, which indicates the rapid polarization of theelectrodes. Moreover, their nearly triangular shape further confirms the formationof a highly efficient electrochemical double-layer and fast ion transport within thesample. Furthermore, the CDC reveal only a small voltage drop of ∼0.001 V at theinitial point of discharge, which is indicative for devices with very low equivalent series

138


resistance (ESR) (0.5-2 Ω). In order to compare the different electrode materials, thespecific capacitances were calculated over a wide range of charge/discharge currentdensities - see Figure 5.4 b) and c).

a) b)

c)

Figure 5.4: Evaluation of LPEG supercapacitors in aqueous H2SO4/PVA electrolyte.a) Galvanostatic charging-discharging curves (CDC) of G-7, G-I1*-1, G-I1*-5, G-I1*-7, and G-CTAB supercapacitors measured at 5 mA, corre-sponding to a high current density of ∼8 A g−1, respectively. b) The gravi-metric capacitance of G-7, G-I1*-1, G-I1*-5, G-I1*-7, and G-CTABsupercapacitors calculated from CDC at different charge/discharge cur-rent densities. c) The volumetric stack capacitances of the LPEG su-percapacitors obtained from G-7, G-I1*-1, G-I1*-5, G-I1*-7, and G-CTAB calculated from the galvanostatic curves as a function of thecharge/discharge current density, respectively. The volumetric capac-itances are presented on the basis of the entire device stack volumewithout packaging, including the volume of the graphene electrodes, theinterspace between the electrodes, the Au current collectors, and theseparators.

139


The LPEG electrodes give rise to gravimetric capacitances ranging from ∼202 F g−1

for G-I1*-7 to ∼145 F g−1 for G-I1*-5 at a current density of ∼1 A g−1, respectively.Notably, the electrodes made from G-7, G-I1*-1, G-I1*-5, and G-I1*-7 exhibitedsignificantly higher gravimetric capacitances in comparison to the electrode madefrom G-CTAB. These findings can be attributed to the fact that G-7, G-I1*-1,G-I1*-5, and G-I1*-7 featured a high yield of SLG (∼20%), whereas G-CTABmostly consisted of few-layer graphene sheets, which in fact may lead to a lowerion-accessible surface area of the electrode material - see Chapter 4.1.2 and 4.3. Inthe case of G-7, G-I1*-1, G-I1*-5, and G-I1*-7 it is expected that the high yieldof SLG results in a high ion-accessible surface area that can be utilized for chargestorage in the supercapacitor device. In addition, the devices showed very good ratecapabilities when the current density was increased from ∼1 to ∼20 A g−1 retaining>60% of their specific capacitance. Interestingly, the devices further revealed highareal capacitances ranging from ∼102 mF cm−2 for G-I1*-7 to ∼85 mF cm−2 forG-I1*-5 at ∼1 A g−1, which are superior to those reported for supercapacitors madefrom reduced graphene oxide,[362] electrochemically exfoliated graphene,[369] carbonnanotubes,[370] laser-scribed graphene,[274] and other liquid-phase exfoliated graphenematerials[280] but lower than those reported for graphene hydrogel films.[366] Inaddition, the volumetric stack capacitances were calculated ranging from ∼9 F cm−3

for G-I1*-7 to ∼5 F cm−3 for G-I1*-5, which are well beyond the values reportedfor supercapacitors made from laser-scribed,[274] flash converted graphene,[359] andreduced graphene oxide.[362]

Such high gravimetric and volumetric capacitances are made possible due to theunique structural characteristics of the exfoliated graphene sheets. In particular,the efficient exfoliation of graphite down to single-layer graphene sheets provideshigh specific surface area materials (around ∼1080 m2 g−1 for G-I1*-7 according tomethylene blue adsorption). Moreover, the stabilization of these sheets by stericallydemanding surfactants is expected to prevent the graphene sheets from restackingto form AB stacked graphite-like films. Instead, turbostratic graphite stacks withsurfactant molecules being present between individual layers may be formed, aspreviously discussed in Chapter 4.3. This structure is expected to allow the electrolyteto penetrate between small gaps present between the graphene sheets - see Figure5.5. In fact, the sterically demanding structure of 7, which already proved to besuperior in terms of graphene stabilization, might be most efficient in the preventionof restacking of the graphene sheets and, thus, leads to the higher capacitances

140


compared to 1 and 5, respectively. Additionally, scanning microscopy (SEM) imagesrevealed the presence of small gaps between the turbostratically aggregated graphenelayers, which further should allow the electrolyte to access the whole volume ofthe electrode - see Figure 10.1 in the Experimental Part IV. This ensures that theelectrolyte can thoroughly wet the entire surface, which, thus, is accessible to theelectrolyte ions and is electrochemically active.

These structural characteristics are of particular importance, especially as parallelaligned AB-restacked graphene sheets lead to a decrease in the accessible surface areaof the graphene sheets and further block the electrolyte from accessing the wholeelectrode volume. This results in an unsatisfactory gravimetric capacitance, whichwas previously observed for reduced graphene oxide electrode materials.[47,264,371]

To this end, porous graphene structures have been investigated recently, such aslaser-scribed[274] and flash converted[359] graphene oxide, which provide high accessiblesurface area and, thus, allow gravimetric capacitances of up to 276 F g−1. Nevertheless,those materials exhibit a very low packaging density resulting in low volumetriccapacitance on the order of 10−1 F cm−3. In fact, there seems to be a trade-offbetween volumetric and gravimetric capacitance. However, next to the gravimetriccapacitance the volumetric capacitance becomes more and more important as manypractical applications, such as portable electronics, are limited in space.

Figure 5.5: Schematic representation of noncovalently functionalized graphene inte-grated into LPEG supercapacitors. The adsorbed surfactant molecules(7) prevent the graphene sheets from restacking leading to a large ion-accessible surface area. The graphene sheets are randomly oriented in aturbostratically stacked geometry.

Taking these considerations into account, the studied LPEG electrode materialsrepresent promising electrode materials in order to simultaneously achieve highgravimetric (up to 202 F g−2) and high volumetric (up to 9 F cm−3) capacitances

141


while still retaining excellent rate capability. Therefore, they may be of specialinterest for applications requiring both low weight and low volume. On the basisof the gravimetric and volumetric capacitance and a voltage window of 1 V, energydensities of up to 7 Wh kg−1 and up to 1.2×10−3 Wh cm−3 were achieved.

Even more important, the devices featured very low ESR. The Nyquist plots revealnearly vertical lines in the low-frequency region, which indicates a nearly ideal capac-itive behavior of the electrode materials - see Figure 5.6. With decreasing frequency,the Nyquist plots transition into a Warburg-like behavior. This is characteristic fordouble-layer charging of porous electrodes consisting of a high-frequency region witha slope of about 45◦, which is a result of the frequency dependence of the electrolytediffusion in the electrode interface, and a low-frequency region in which capacitivebehavior is approached.[195,212,226,372] The frequency at which the impedance responsefirst approaches that of an ideal capacitor is in the MHz region for all samples,e.g. 4141 Hz for G-I1*-7, revealing the fast ion transport within the electrode.[373]

Moreover, the Warburg curve is shorter than that of other supercapacitor electrodesreported, indicating a short ion diffusion path within the samples.[195,266] Additionally,no distinct semi-circles were observed, which confirms that no electrical charge trans-fer arising from Faradaic processes takes place, and the charge storage of the deviceis mainly non-faradaic.[372–374] Moreover, this indicates a very small charge transferresistance (Rct) at the LPEG electrode/electrolyte interface, which can be attributedto the low thickness (<10 μm) of the LPEG films favoring the ion transfer kineticsof the LPEG films.[372,375] The ESRs were estimated to be <2 Ω for all samples,which is in consistence with galvanostatic charging-discharging studies and furtherconfirms the good ionic conductivity of the electrolyte and low internal resistance ofthe LPEG electrodes - see Figure 5.6. This translated in a high volumetric powerdensity of up to ∼100 W cm−3 and a gravimetric power density of 5×105 W kg−1 inaqueous electrolyte employing a voltage window of 1 V.

Notably, such a high volumetric and gravimetric power density is far superior tothose reported for hydrazine reduced graphene,[376] carbon nanotubes,[370] laser-scribed graphene,[274] holey graphene frameworks,[355] and other liquid-phase ex-foliated graphene materials[109] in aqueous electrolyte. Moreover, it even exceedsactivated microwave reduced graphene oxide,[55] sponge-like graphene nanoarchitec-tures,[377] pillared graphene paper,[378] a thermally reduced melamine resin function-alized graphene oxide and hydrazine reduced graphene oxide,[376] and laser-scribed

142


graphene[274] in organic electrolyte.

Figure 5.6: Nyquist plot for G-7, G-I1*-1, G-I1*-5, G-I1*-7, and G-CTAB super-capacitors. The inset expands the high-frequency region.

Moreover, the gravimetric power density of LPEG devices is one order of magnitudehigher than that of several common commercial devices.[359] In fact, these findingscan be explained by the improved electrical conductivity of pristine graphene incomparison to reduced graphene oxide, which enables faster ion charge transfer acrossthe electrode/electrolyte interface. Moreover, no binder or other non-conductive addi-tives are added, which would increase the resistance of the electrode and additionallyadd “dead mass”, which cannot be utilized for the charge storage.

5.1.3 Comparison with Commercial Graphene Samples

The energy and power densities of the G-I1*-7 supercapacitor measured over a widerange of current densities are summarized in a Ragone plot in Figure 5.7. To furtherput these results in perspective with current graphene technology, we characterizeda number of commercially available graphene materials in supercapacitor devices.These devices were built and tested under the same dynamic conditions as the LPEGdevices. For all devices, the calculations for volumetric energy and power densitywere made based on the volume of the cell without packaging, that includes thecurrent collector, active material, separator, and electrolyte. The calculations for thegravimetric energy and power density were made based on the mass of the electrode

143


material. All devices were prepared with an aqueous H2SO4/PVA electrolyte. Forcomparison with the literature, laser-scribed graphene (LSG)[274] was included - seeFigure 5.7.

a)

b)

Figure 5.7: Ragone plot comparing a) the gravimetric and b) the volumetric energyand power density of LPEG supercapacitors with supercapacitors builtfrom a number of commercially available graphene materials, respectively.For comparison laser-scribed graphene (LSG) was included - see Ref.[274].

Notably, LPEG outperforms all commercial samples in terms of gravimetric energydensity and power density. These findings can be attributed to the significantly lowerspecific surface area of the commercial graphene samples compared to LPEG.[379–382]

However, LSG enables slightly higher energy densities compared to LPEG, whicharises from its higher capacitance of up to 276 F g−1. LPEG, on the contrary, gives

144


rise to higher gravimetric power densities resulting from its excellent conductivity asa non-oxidized graphene material. In fact, most of the commercial graphene samplesare synthesized via an oxidative route, which impairs their electrical conductiv-ity.[379–382] Turning to volumetric energy and power density, LPEG supercapacitorsare superior to both, supercapacitors from commercially available graphene andLSG. These findings can be attributed to the significantly lower specific surface areaof the commercial graphene samples and the low packing density of laser-scribedgraphene.[274] Nevertheless, we stress that LSG and LPEG are produced in lab scaleonly while the other graphene materials are produced in gram to ton scales.[69] Inorder to fully exploit the potential of LPEG in supercapacitor devices, the scale-upfrom lab to mass production has to be accomplished first.

5.1.4 Cycling Stability of LPEG Supercapacitors

Finally, the cycle stability of LPEG devices was tested. A device of G-I1*-7 wastaken as an example and charged and discharged over 10,000 times at a constantcurrent of 5 mA without any rest time between each cycle - see Figure 5.8. TheLPEG device retained 90% of its initial capacitance after 10,000 cycles. However,in practice devices are usually not immediately discharged after charging. Hence,another experiment was performed, in which the devices was allowed to rest for30 min after every 999 cycles. In this experiment, the device retained even 95% of itscapacitance after 10,000 cycles demonstrating the excellent electrochemical stabilityof the LPEG supercapacitor device.

Conclusively, LPEG electrodes proved to be promising materials for electrochemicalcapacitors. In particular, they can provide gravimetric capacitances of up to 202 F g−1

and volumetric capacitances of up to 9 F cm−3 in aqueous electrolytes. Moreover, thedevices can deliver gravimetric and volumetric energy densities of up to 7 Wh kg−1

and up to 1.2 mWh cm−3, respectively. Additionally, they feature high gravimetricand volumetric power densities of up to 5×105 W kg−1 and up to ∼100 W cm−3,respectively, and, thus, are expected to bridge the gap between conventional capacitorsand batteries. Moreover, they exhibit an excellent cycle stability and therefore holdpromise for commercial applications. In order to further explore their potential insupercapacitor applications for mobile power supply, such as electrical vehicles andmobile electronics, devices of G-I1*-7 were further investigated.

145

5.2 FLEXIBLE ALL-SOLID-STATE SUPERCAPACITORS

Figure 5.8: Constant current charging-discharging cycling at a constant current of5 mA (∼10 A g−1) for a G-I1*-7-supercapacitor device. 10,000 cycleswere performed without any rest time and with 30 min rest time afterevery 999 cycles, respectively.

5.2 Flexible All-Solid-State Supercapacitors

Multifunctional electronics such as wearable electronics, electronic newspapers, paper-like mobile phones, and smart sensors are emerging technologies, which requirebendable and flexible energy storage devices.[383–385] In this regards, supercapacitorsare considered as one of the most promising technologies due to their high powerdensity, long cycle lifetime, low environmental impact, and safety.[193] However,the rigid microstructure of activated carbons hinders their application in flexibilitydevices, as the electrode material would easily crack and peel-off the current collectorduring frequent bending.[260] To this end, electrode materials with robust mechanicalflexibility, high energy density, power density, and excellent cycling stability, such asgraphene, are of particular interest.[260,277,386]

As already discussed in Chapter 5.1, the current-voltage curves of the LPEG electrodesare nearly unchanged at different bending states, which reveals that the conductanceof the graphene films is hardly affected by bending stress and, moreover, indicatesthat the LPEG electrodes are suitable for the application in flexible all-solid-statesupercapacitor devices. In order to test the durability of LPEG electrode materials

146


for flexible energy storage, a device of G-I1*-7 was tested. To this end, the liquidelectrolyte was replaced by a solid-state H2SO4/PVA polymer gelled electrolyte,which was drop cast on the LPEG electrode and solidified overnight. Remarkably,this does not only avoid the harmful leakage of the liquid electrolyte, but alsosignificantly reduces the device thickness and weight as the electrolyte also actsas a separator, and no additional packaging materials are required. Again, theelectrochemical performance of the flexible all-solid-state LPEG supercapacitors wasevaluated by CV and galvanostatic charging-discharging tests. Figure 5.9 a) showsthe CV curves of the device at various scan rates. The observed CV curves are nearlyrectangular at a high scan rate of 100 mV s−1, indicating their ideal capacitativebehavior and low contact resistance. The specific capacitance of the G-I1*-7 LPEGelectrode film estimated from the CV curves is ∼162 F g−1 at 100 mV s−1 and126 F g−1 at 1000 mV s−1, revealing the good rate capability of the device (∼78%).Along these lines, the linear profile of the galvanostatic CDC and their symmetricaltriangular shape indicate nearly ideal capacitative behavior - see Figure 5.9 b).

a) b)

Figure 5.9: a) Cyclic voltammetry and b) galvanostatic charging-discharging curvesof the flexible all-solid-state device, respectively. The device was madeby assembling two G-I1*-7 electrodes with a small amount of gelledelectrolyte in between, which also acted as a separator.

As shown in Figures 5.10 a) and b), the gravimetric and volumetric capacitancevalues for the all-solid-state device were comparable with those obtained in liquidaqueous H2SO4/PVA electrolyte. In particular, the LPEG film in the all-solid-state device exhibits a gravimetric capacitance of 188 F g−1 at a current density of∼1 A g−1, which is slightly lower than those measured in liquid aqueous H2SO4/PVAelectrolyte (202 F g−1). Moreover, the volumetric stack capacitance increases slightlyto ∼10.3 F cm−3 for the all-solid-state device in comparison to ∼8.8 F cm−3 for

147


the liquid electrolyte at a current density of ∼1 A g−1, respectively, which can beattributed to the smaller volume of the all-solid-state supercapacitor. However, theall-solid-state device features lower gravimetric capacitance and lower rate capability,when the current density was increased from 1 to 20 A g−1 of about 111 F g−1 and,thus, reveals about 60% capacitance retention. These findings can be explainedby the higher resistance and slower electrolyte diffusion within the all-solid-statedevice. This was further confirmed by the Nyquist plot obtained from electrochemicalimpedance spectroscopy - see Figure 5.10 c).

Nevertheless, the gravimetric capacitances of the LPEG-based all-solid-state super-capacitor are substantially higher than those of most previously reported all-solid-state supercapacitor devices made from carbon nanotubes[370,387,387,388] and graphenefilms,[389–391] and are comparable to those reported for laser-scribed graphene[57,274]

and graphene hydrogel films.[366]

In a next step, the potential of these LPEG all-solid-state supercapacitors for flexibleenergy storage was evaluated under real conditions by placing them under constantmechanical stress and analyzing their performance. The CV curve of the devicebent at a large bending angle of 150◦ shows nearly the same capacitive behavior asthe flat device - see Figure 5.11 b). Moreover, the device was wrapped around aPET tube, and the electrochemical performance at different bend radii was tested- see Figure 5.11 c). The LPEG all-solid-state supercapacitor shows exceptionalelectrochemical stability regardless the degree of bending, demonstrating its excellentmechanical stability and robustness. In order to further characterize the stability ofthe all-solid-state device, galvanostatic charging-discharging tests were carried out,and the device was tested for more than 1000 cycles under a 150◦ bending angle. Theall-solid-state device proved to be very robust as it maintains over 97% of its originalcapacitance after 1000 cycles. Such exceptional performance durability and flexibilitycan be attributed to the excellent mechanical and electrical robustness of the LPEGelectrodes along with a well-interconnected network between the electrodes and thepolymer gel electrolyte.

148


a) b)

c)

Figure 5.10: Comparison of a) the gravimetric and b) the volumetric capacitance ofa G-I1*-7 electrode in liquid and solid-state H2SO4/PVA electrolyte,respectively. c) Nyquist plots of the G-I1*-7 electrode in liquid andsolid-state H2SO4/PVA electrolyte. The inset shows the magnifiedhigh-frequency regions of the Nyquist curves.

In order to further demonstrate the practical usage of the flexible LPEG all-solid-state supercapacitors, four supercapacitor units were connected in series to create atandem device. Each supercapacitor unit had the approximately same mass loading ofLPEG. The tandem device was characterized by cyclic voltammetry and galvanostaticcharging-discharging measurements. Figure 5.12 b) shows that the potential windowcan be extended from 1 to 4 V for the tandem device. Furthermore, the tandemdevice display almost unchanged charge/discharge times and product current (definedas the area of the CV curves) compared with the individual devices at the samecurrent density, revealing that the performance of each supercapacitor unit is wellretained in the tandem device - see Figure 5.12 a). Moreover, the tandem exhibits

149


nearly ideal triangular charging-discharging curves with a very small voltage drop,as observed for the individual devices, revealing its excellent capacitive behavior andminimal internal resistance. Notably, this outstanding performance was achievedwithout the need for any voltage balance, which is usually used in a series connectionin order to prevent the cells from going into over-voltage.[274] After being chargedat 4.0 V, the tandem device can light up a red LED that operates at a minimumvoltage of 1.9 V (inset Figure 5.11 a)), demonstrating the practical potential of thefabricated flexible all-solid-state supercapacitors.

a) b)

c) d)

Figure 5.11: a) A schematic presentation of the all-solid-state LPEG supercapacitor,which illustrates that the polymer gel electrolyte can serve as boththe electrolyte and the separator. A photograph demonstrating theflexibility of the device. b) Cyclic voltammetry curves of the flexibleall-solid-state supercapacitor for the flat device (0◦) and for a largebending angle of 150◦ at 100 mV s−1. c) Capacitance change of theLPEG electrode as a function of the bending radius. d) Cycling stabilityof the all-solid-state device under bending state (150◦) at a constantcurrent of 5 mA.

150

5.3 PERFORMANCE IN ORGANIC ELECTROLYTE

a) b)

Figure 5.12: a) Galvanostatic charging-discharging curves at 5 mA and b) cyclicvoltammetry curves at 100 mV s−1 of a single all-solid-state supercapac-itor (black) and four supercapacitors in series (red), respectively. Inset:Photograph of a red LED powered by four supercapacitors in series.

5.3 Performance in Organic Electrolyte

One of the key issues in the development of high-performance supercapacitors isthe utilization of suitable non-aqueous electrolytes in order to overcome the limitedelectrochemical window of 1 V in aqueous electrolytes. Hence, we further examinedan organic electrolyte, which allows the operation of the device at higher voltages and,thus, enables higher energy and power densities. To this end, tetraethylammoniumtetrafluoroborate in acetonitrile (TEABF4/ACN) was employed as an electrolyte, asit is commonly used for commercial devices.[194] Moreover, the devices were fabricatedthe same way as it is done in industry using a coin cell assembly - see Figure 5.13.

Figure 5.13: Schematic representation of a LPEG-based supercapacitor device usedfor tests with organic electrolyte (left). Optical image of an industry-grade coin cell LPEG supercapacitor device used in this study (right).

151


As shown in Figure 5.14 a), the CV curves are close to being rectangular from 0 to 2 Veven at a high scan rate of 10,000 mV s−1, indicating an excellent capacitive behaviorand low contact resistance of the supercapacitor devices. Furthermore, galvanostaticcharging-discharging measurements reveal a nearly symmetrical triangular shapewith small voltage drops at the initial point of discharge, which further indicates anexcellent electrical double-layer capacitative behavior and a very low ESR (∼1.5 Ω)of the LPEG devices - see Figure 5.14 b). This was further confirmed by the Nyquistplot obtained from electrochemical impedance spectroscopy - see Figure 5.14 c).

a) b)

c) d)

Figure 5.14: Evaluation of LPEG supercapacitors using an organic electrolyte of 1.0 Mtetraethylammonium tetrafluoroborate in acetonitrile (TEABF4/ACN).a) Cyclic voltammetry and b) galvanostatic charging-discharging curvesof the device. c) The Nyquist plot of a G-I1*-7 electrode inTEABF4/ACN organic electrolyte. The inset shows the magnifiedhigh-frequency regions of the Nyquist plot. d) Cycling stability of thedevice at a constant current of 5 mA.

The LPEG electrodes display a lower gravimetric capacitance of 120 F g−1 incomparison to that observed in aqueous electrolyte (202 F g−1) at a current density

152


of ∼1 A g−1, respectively. In fact, higher specific capacitances of high surfacearea carbons in aqueous electrolyte compared to organic electrolyte have beenreported previously and can be explained by the higher dielectric constant of aqueoussystems.[214,229] Moreover, the aqueous electrolyte provides higher concentration,lower resistance, and a smaller ionic radius, which further propagates their diffusionwithin the graphene pore structures.[217] Increasing the current density to 20 A g−1,the LPEG electrodes retained a high gravimetric capacitance of 76 F g−1 (63%).Such high gravimetric capacitances in an organic electrolyte led to a gravimetricenergy density of 33.3 Wh kg−1.

On the basis of the device volume without packaging, including the aluminum currentcollectors, LPEG electrodes, and separator, the volumetric capacitance and energydensity of the LPEG electrodes were calculated to be 1.5 F cm−3 and 1.1 mWh cm−3,respectively. Moreover, the gravimetric and volumetric power density of the devicewere calculated from the CDC data. Accordingly, a gravimetric power density of∼29×105 kW kg−1 and a volumetric power density of 151 W cm−3 are obtained.Again, such high power density values are superior to those reported for activatedmicrowave reduced graphene oxide,[55] sponge-like graphene nanoarchitectures,[377] pil-lared graphene paper,[378] a thermally reduced melamine resin functionalized grapheneoxide and hydrazine reduced graphene oxide,[376] flash converted graphene oxide,[359]

and laser-scribed graphene[274] in organic electrolyte. In fact, such extraordinarypower densities can only be achieved by a highly conductive active material, whichdoes not significantly contribute to the internal resistance of the system. In thiscontext, LPEG graphene outperforms reduced graphene oxide materials due to itsstructural integrity and, thus, retained electrical conductivity enabling fast chargetransfer across the electrode/electrolyte interface.

Another important factor for practical applications of LPEG supercapacitors is along cycle life. As such, we measured the cycle life of the LPEG device by chargingand discharging it over 10,000 times at a constant current of 5 mA without any resttime between the cycles - see Figure 5.14 d). Importantly, the LPEG electrodes showexcellent cycling stability with 74% of its initial capacitance retained after 10,000charging-discharging cycles. This indicates that the performance is not limited by anyparasitic chemical reactions and further demonstrates the excellent electrochemicalstability of the LPEG electrodes.

In fact, we demonstrated that LPEG can be easily employed as an active electrode

153

5.4 PERFORMANCE IN IONIC LIQUID ELECTROLYTE

material in state-of-the-art commercial coin cell supercapacitors. The prepareddevices featured high gravimetric and volumetric capacitance of up to 120 F g−1

and 1.5 F cm−3, respectively. This translated in gravimetric and volumetric en-ergy densities of 33 Wh kg−1 and 1.1 mWh cm−3, respectively. Importantly, thedevices further featured very high gravimetric and volumetric power densities of∼29×105 kW kg−1 and 151 W cm−3, respectively, which are higher than that ofmany state-of-the-art graphene-based coin cell supercapacitors.[55,266,274,359,376–378]

Nevertheless, organic electrolytes suffer from various drawbacks including electrolytedepletion upon charge, narrow operational temperature range, and low safety.[283]

To this end, the devices can be further improved by using ionic liquid electrolytes,which enable a wider operational potential window.

5.4 Performance in Ionic Liquid Electrolyte

As depicted by Equation 3.11 and discussed in Chapter 3.4, two key factors limitthe energy density of a supercapacitor. One is the capacitance of the electrode,which depends on the ion-accessible surface area of the electrode active material.The second is the maximum operating voltage Vcell of the device. As a matterof fact, Vcell is mainly limited by the electrolyte stability. Recently, new kinds ofelectrolytes based on ionic liquids (ILs) have gained tremendous interest as theyfeature significant higher potential windows in comparison to conventional organicelectrolytes enabling operating voltages >3 V.[212,234,354] Thus, they are expectedto open the path towards higher energy and power density devices.[234] Moreover,ionic liquids enable the operation at high temperature, which is highly desirable forelectric vehicle power sources.[392]

To this end, we further tested the electrochemical performance of LPEG supercapaci-tors in 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([BMIM][NTf2])ionic liquid electrolyte. Accordingly, a device of G-I1*-7 was utilized.

The CV testing shows nearly rectangular curves from 0 to 3.5 V even at very highscan rates of 2000 mV s−1, indicating a nearly ideal capacitive behavior and a lowcontact resistance - see Figure 5.15 a). Although the electrochemical window of[BMIM][NTf2] with a value of 4.6 V is much higher than the applied potential, a slightcurrent increase in the highest voltage range occurring from electrolyte degradation

154


is noticed. In fact, the ionic liquid is not free of contaminations and impurities, suchas small amounts of water (≤0.5% water), which are electrochemical decomposed atsuch high potentials.[393] As such, few ppm of water in the ionic liquid electrolyte cansignificantly depress the potential window and may further lead to capacity fading. Inthis context, pure and water-free ionic liquids are strongly needed. Moreover, smallamounts of water adsorbed on the graphene film surface and surface groups presentat the graphene edges may be decomposed at a voltage higher than 1.23 V.[234,394]

a) b)

c)

Figure 5.15: Electrochemical characterization of an LPEG supercapacitor in 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([BMIM][NTf2])ionic liquid electrolyte. a) Cyclic voltammetry curves at different scanrates. b) Galvanostatic charging-discharging curves at different currentdensities. c) Nyquist plot of the LPEG supercapacitor. The inset showsthe close-up view of the high-frequency regime.

Meanwhile, galvanostatic charging-discharging measurements reveal a nearly sym-

155


metric triangular shape with small voltage drops at the initial point of the dischargecurve (0.076 at a current density of 50 A g−1) - see Figure 5.15 b). Both resultssuggest an excellent electrical double-layer capacitive behavior and a very low ESR(∼3.8 Ω) in ionic liquid electrolyte. The LPEG electrodes showed a gravimetric ca-pacitance of ∼143 F g−1 at ∼1 A g−1, which is about 30% lower than that in aqueouselectrolyte. This can be explained by the higher resistance (∼5 mS cm−1), higherviscosity (38 cP at 20 ◦C), and higher ionic radius ([BMIM] ∼0.79 nm and [NTf2]∼0.42 nm, respectively) of the ionic liquid electrolyte in comparison to the aqueouselectrolyte, respectively.[377,395–397] In fact, the gravimetric capacitance values arelower than those observed for laser-scribed graphene (LSG, ∼276 F g−1 in ionic liquidelectrolyte).[274] Here, we assume that the open porous structure of LSG enablesbetter electrolyte diffusion in comparison with our graphene flake sample, where onlysmall gaps between the turbostratically aggregated graphene flakes exist, which mightlimit the efficient diffusion of the ionic liquid electrolyte into the graphene pore struc-ture. Nevertheless, the observed capacitances are still superior to those derived fromreduced graphene oxide[270,397,398] and sponge-like graphene nanoarchitectures.[377] Todate, no comparison with other liquid-phase exfoliated graphene samples is possible asthose materials were not tested in ionic liquid electrolytes yet. To our best knowledge,this represents the first study on liquid-phase exfoliated graphene supercapacitorsoperating at a very high voltage window of 3.5 V. Increasing the current densityto ∼50 A g−1, the LPEG electrodes retained 70% of the capacitance at 1 A g−1.These excellent capacitance values translate into a gravimetric energy density ofthe device of ∼121 Wh kg−1, which is nearly comparable to the theoretical valueof lead acid battery electrodes (165 Wh kg−1).[355] On the basis of the packagingdensity of 0.07 g cm−3, the volumetric stack capacitance of the LPEG supercapacitorwas calculated to be ∼2.5 F cm−3. Notably, these values are superior to those oflaser-scribed graphene electrodes[274] and flash-converted graphene electrodes,[359]

but still lower than those of activated graphene.[55] In fact, such high volumetriccapacitance values arise from the high electrode density of LPEG graphene electrodescompared to most porous state-of-the-art graphene-based supercapacitor electrodes,respectively. Consequently, the devices feature high volumetric energy densities ofup to ∼4.3 mWh cm−3.

To further evaluate the device performance, the frequency response of the LPEG super-capacitor was analyzed using electrical impedance spectroscopy (EIS). Figure 5.15 c)shows the Nyquist plot featuring a nearly vertical curve in the low-frequency regime,

156


which confirms an excellent capacitative behavior. A knee frequency at 180 Hz isobserved, which is the maximum frequency at which capacitive behavior is dominant.Moreover, the absence of a semi-circle and the presence of a rather linear 45◦ War-burg area further reveals that the charge is stored through a mainly non-faradaicprocess.[195] The ESR is estimated as ∼4.5 Ω, which is slightly higher than the ESRobserved in aqueous and organic electrolyte, respectively. As discussed earlier, thisis likely due to lower electrical conductivity and higher viscosity of [BMIM][NTf2]in comparison to aqueous and organic electrolytes, respectively. Nevertheless, thenegative impact from the lower ionic conductivity of the ionic liquid electrolyteis offset by the increase in the operating voltage and, thereby, improvement inenergy and power density. Accordingly, the ESR translated in a gravimetric andvolumetric power density of ∼48×105 W kg−1 and 346 W cm−3, respectively. Infact, these values set an upper benchmark for graphene-based supercapacitors beingsuperior to laser-scribed graphene,[274] reduced graphene oxide,[397,398] graphene hy-drogel films,[366] activated graphene,[55] graphene oxide, mesoporous carbon plateletscomposites,[399] and graphene and carbon nanotubes composites[400] in ionic liquidelectrolyte, respectively.

In summary, we demonstrated that the general concept of a supercapacitor designbased on liquid-phase exfoliated graphene electrodes, and a compatible ionic liquidelectrolyte holds potential as an electrical energy storage device enabling higherenergy and power densities, respectively. In fact, the use of [BMIM][NTf2] takesadvantage of its larger electrochemical stability window, allowing for operation at upto 3.5 V, which in turn increased both the energy density and power density of thedevice. Cyclic voltammetry and galvanostatic charging-discharging results indicatestable electrochemical performance. In particular, a specific capacitance as highas ∼142 F g−1 and a volumetric stack capacitance of ∼2.5 F cm−3 was measured.This translated in gravimetric and volumetric energy densities of ∼121 Wh kg−1 and∼4.3 mWh cm−3, respectively. Moreover, the devices featured excellent gravimetricand volumetric power densities of ∼48×105 W kg−1 and ∼346 W cm−3, respectively,which are far superior to many state-of-the-art graphene-based supercapacitorsemploying ionic liquid electrolyte.[55,274,366,397–400] Nevertheless, the electrochemicaldecomposition of impurities and residual water within the ionic liquid electrolytedepresses the operational voltage window and will lead to capacity fading in thelong-term. As such, pure and water-free ionic liquids are strongly needed in order toemploy this concept for commercial supercapacitors.

157

5.5 RAGONE PLOT - COMPARISON WITH STATE-OF-THE-ART DEVICES

5.5 Ragone Plot - Comparison with State-of-the-ArtDevices

In the previous chapters, we presented the excellent performance of LPEG electrodesin supercapacitors in various electrolytes. The energy and power densities of theLPEG devices in different electrolytes are summarized in a Ragone plot - see Figure 6.2.Next to the data of LPEG, the metrics of commercially available energy storagedevices including a high-energy lithium thin-film battery (4 V/500 μAh), a high-power aluminum electrolytic capacitor (3 V/300 μF), and an activated carbonelectrochemical capacitor are included for comparison.[274] Please note that thesecommercially available energy devices were disassembled and measured from El-Kadyet al.[274] under very similar dynamic conditions as the LPEG samples. Remarkably,LPEG supercapacitors delivered a volumetric energy density of up to 4 mWh cm−3

in ionic electrolyte, which is almost one order of magnitude higher that that ofa typical activated carbon supercapacitor (<1 mWh cm−3)[354] and comparable tothose of lithium thin-film batteries (1-10 mWh cm−3).[401] Notably, the increase inenergy density of supercapacitors is usually accompanied by a decrease in powerdensity and/or cyclability. In fact, these are the most important characteristicsof a supercapacitor cell and without them it becomes a mediocre battery.[227,355]

Nevertheless, featuring fast electron transport and efficient ion transport throughoutthe whole LPEG film electrode, the LPEG supercapacitors were also able to deliverhigh power densities. In detail, the power densities of the LPEG supercapacitorsare comparable to those of the aluminum electrolytic capacitor, about two ordersof magnitude higher than those of commercial activated carbon supercapacitors,more than three orders of magnitude higher than those of conventional lead acidbatteries, and more than four orders of magnitude higher than that of the lithiumthin-film battery.[355,401] This superior energy and power density should enable themto compete with micro batteries and electrolytic capacitors in a variety of applications,such as electrical vehicles and mobile electronics. To the best of our knowledge, thisis the first report on LPEG supercapacitors with such excellent performance in termsof ultrahigh energy and power density.

158


Figure 5.16: Ragone plot comparing the energy and power density of LPEG superca-pacitors in different electrolytes with a number of commercially availableenergy storage devices. Data for the activated carbon supercapacitor(AC-EC), the lithium thin-film battery, and the aluminum electrolyticcapacitor are taken from Ref. [274].

The superior performance of LPEG supercapacitors can be explained by the co-operative effects of the high conductivity of the pristine few-layer graphene materialand the structural characteristics of the graphene electrode films. In fact, mosttraditional porous carbon materials, such as activated carbon, activated graphene, 3Dgraphene foams, laser-scribed, and flash converted graphene, exhibit lower volumetriccapacitance compared to LPEG despite their high specific surface area (∼1000-3000 m2 g−1), respectively.[55,194,274,359,402,403] This arises from their highly porousstructure featuring abundant wormlike pore channels and gaps between the carbonsurfaces, which may not be accessible to the electrolyte ions and lead to very lowelectrode densities. On the one hand, this enables very high gravimetric capacitanceof up to 300 F g−1, but, on the other hand, fairly low volumetric capacitance(10−2-10−1 F cm−3). In contrast, LPEG features a nearly ideal two-dimensionalflat graphene surface. Moreover, adsorbed surfactant molecules on the graphenesurface and edges may enable small gaps between the turbostratically aggregatedgraphene flakes leading to a porous network structure within the electrode film. Assuch, the graphene surface may be readily accessible to the electrolyte ions, while theelectrode itself remains very dense. However, such small pores may not be accessibleto large electrolyte ions and, thus, smaller capacitance values in organic and ionicliquid electrolyte were observable in comparison to aqueous electrolyte, respectively.

159


Nevertheless, these lower gravimetric and volumetric capacitance values were offsetby the increase in the operating voltage and, thereby, improved energy and powerdensity of the devices, especially in the case of ionic liquid electrolyte.

In summary, we have developed a simple and scalable approach for the fabrication ofLPEG-based supercapacitors that are applicable in different aqueous, organic, andionic liquid electrolytes enabling operational voltages of up to 3.5 V. Moreover, theycombine the energy density of micro batteries with the power density of electrolyticcapacitors, thus, bridging the gap between batteries and capacitors. Furthermore,LPEG supercapacitors showed excellent cycling stability, which is a prerequisitefor their application in permanent structures, such as RFID tags, embedded micro-sensors, and biomedical implants.[57] In fact, such a long cycle life cannot be achievedwith conventional micro batteries, which have a finite lifetime of about <500 cycles.Taking advantage of their mechanical flexibility and superior performance underbending, LPEG supercapacitors may also find their way into flexible electronics, suchas flexible displays, roll-up portable displays, and wearable electronics.[57,404,405]

160

6 Conclusion and Outlook

The work presented aimed to systematically develop noncovalent stabilization meth-odologies, which enable the scalable production of high-quality few-layer graphene forthe application as high energy density electrodes in supercapacitors. The focus wasdirected towards establishing stabilization approaches for the liquid-phase exfoliationof graphite. On the one hand, they should enable high yields of single-layer grapheneand, on the other hand, should be suitable for large-scale production of graphene interms of non-toxicity and easy processability. Furthermore, the use of low boilingsolvents was in the center of our research, as this enables direct processing of thegraphene sheets into electrode films towards supercapacitor applications.

Despite the extraordinary progress in the development of graphene synthesis, theefficiency of graphene dispersion has been limited by the exfoliation efficiencies oforganic surfactants used for fundamental research.[109,111,112] This is exactly the pointwhere this work tied in and enabled to pave the way towards improved graphenedispersion and exfoliation down to high amounts of single-layer graphene.

Establishment of a Dispersion Procedure

Although ultrasonication serves as the state-of-the-art exfoliation process for theexfoliation of graphite in liquid media, no standard protocol had been available forthe sonication process. As such, the examination of the exfoliation mechanism in theliquid-phase, especially the influence of different parameters, were at the forefront ofthe investigations (Chapter 4.1.1). By systematically adjusting the ultrasonicationpower, time, temperature, and centrifugation conditions, a highly reproducibleprotocol for the liquid-phase exfoliation of graphite by means of ultrasonication couldbe established.

As a next step, cetyltrimethylammonium bromide (CTAB) was tested as an in-ternal standard, enabling the comparison to previous works concerning graphenedispersibility and exfoliation (Chapter 4.1.2). Absorbance measurements revealedthe high efficiency of the dispersing procedure, confirming graphene concentrationsof 0.047 mg mL−1. These were comparable to the ones reported in the literatureemploying CTAB as a stabilizer and even outperformed them in terms of long-term

161

6 CONCLUSION AND OUTLOOK

stability.[119,120] Furthermore, statistical Raman spectroscopy, transmission electronmicroscopy (TEM), and atomic force microscopy (AFM) measurements confirmedthe few-layer character of the exfoliated graphene sheets consisting of about 2.5%multilayered/bulk graphite and of about 97.5% few-layered/turbostratic graphenewith flake sizes varying from sub- to micrometer. The comparison with the lit-erature in terms of exfoliation degree turned out to be a challenging task, as nostandard characterization protocols are established so far. While most studies focuson non-statistical Raman, TEM, and AFM measurements of selected graphene flakes,we decided to choose statistical Raman analysis as our primary characterizationtool. This allowed us to gain insights into a large sample size and, thus, derivingquantitative information on the exfoliation degree of the samples. Nevertheless,comparing the herein obtained results with those reported in the literature, ourexfoliation procedure was superior in terms of exfoliation degree, flake size, andoverall graphene dispersibility.

Based on these results, we were able to use this standard procedure to gain quantita-tive insights into the dispersibility and exfoliation of few-layer graphene by means ofdifferent stabilization and exfoliation approaches, namely noncovalent functionaliza-tion in aqueous medium, pretreatment of graphite by small intercalant molecules,solvent-based graphene exfoliation, and cosolvency.

Graphene Dispersion by Aromatic Amphiphiles

In the noncovalent approach, different aromatic amphiphiles were tested with respectto graphene dispersion and exfoliation in aqueous media (Chapter 4.2). In particular,graphene was dispersed by four aromatic amphiphiles, 1, 2, 4, and 7 - see Figure 6.1.The graphene concentration determined by absorbance spectroscopy increased in theorder of 4 < 2 < 1 < 7, rendering 7 as the best aromatic surfactant for graphenestabilization investigated in this study.

The superior graphene dispersibility of 7 could be attributed to highly effective π-πinteractions between 7 and graphene, which were confirmed by fluorescence measure-ments. Moreover, zeta potential measurements revealed the efficient stabilization ofgraphene by 7 through electrostatic repulsion arising from the negative charge of thedeprotonated sulfonic acid groups of 7.

162


Figure 6.1: Structures of aromatic amphiphiles, which were employed for the nonco-valent stabilization of graphene in aqueous medium within this study.

Next, the optimization of the stabilizer concentration enabled an increase in theconcentration of dispersed graphene by a factor of 3, yielding graphene concentrationsof about 0.15 mg mL−1. We stress that our production procedure focused on thequantitative evaluation of different stabilization methods and, thus, is not improved interms of overall graphene dispersibility. Nonetheless, it is expected that the grapheneconcentration can be further increased by the optimization of the exfoliation procedureand subsequent scale-up, which should be in the focus of future work.

Moreover, the influence of the surfactant on the degree of exfoliation was investigated.Statistical Raman analysis of the flake thickness distribution revealed that theexfoliation degree of the samples increases in the order 4 < 2 < 1 < 7. In particular,the dispersion of G-7 consisted of mainly few-layered/turbostratic graphite with highamounts of single-layer graphene (21%). Moreover, TEM measurements indicatedthat the lateral flake size ranges from sub- to micrometer size. Additional evaluationof the defect-related D-band indicated that the defect content of the samples increasesin the order of 7 < 4 < 1 < 2, which was mainly assigned to the reduction of the

163


flake size during ultrasonication.

In general, aromatic amphiphiles with a large molecular structure tended to showa better graphene dispersibility and graphene exfoliation ability than aromaticamphiphiles with small hydrophilic parts. These results suggested that steric repulsionby either sterically demanding hydrophobic parts or high molecular weight surfactantsis essential to the stabilization of graphene sheets in aqueous media.

Pretreatment of Graphite as a Dispersion Enhancer

In order to address the stability of the graphene dispersions, a dispersion conceptinvolving the use of tetraalkylammonium cations as small intercalant in additionto the dispersant was applied (Chapter 4.3). Despite the fact that tetraethyl-ammonium hydroxide cannot disperse graphene on its own, a pronounced effect couldbe observed upon the pretreatment of graphite in solutions of tetraethylammoniumhydroxide. To shed light on the interaction mechanism of tetraethylammoniumcations with graphite/graphene, several parameters of the intercalant approach,such as addition mode, pH, counter ion, and the presence of tetraalkylammoniumsalts, were investigated by using 1 as a stabilizer. The findings indicated that boththe presence of quaternary ammonium ions and an alkaline pH of the solution areessential for a significant enhancement of graphene dispersibility.

Based on these findings, we propose a mechanism for the pretreatment approach,which involves the deprotonation of graphene edges in alkaline solutions and subse-quent expansion of the graphene edges by partial intercalation of tetraalkylammoniumcations into the graphite layers.

Having this information in hand, we successfully transferred the pretreatment ap-proach to several surfactants, namely an anionic aliphatic surfactant (sodium do-decylsulfate, SDS), a cationic aliphatic surfactant (CTAB), a nonionic aliphaticsurfactant (Pluronic P-123, P123), nonionic aromatic surfactants (1 and 5), and ananionic aromatic surfactant (7), significantly improving the graphene dispersibility -see Figure 6.1.

In summary, we developed an efficient and simple method for the preparation ofhighly stable and concentrated few-layer graphene dispersions. In particular, the

164


dispersion efficiency could be significantly increased by the highly effective pretreat-ment step, resulting in an increase in the graphene concentration for all testedsurfactants, respectively. Interestingly, 5 yielded the highest graphene concentrationof 0.095 mg mL−1 featuring superior stability with only negligible amounts precipi-tating within the first three months of controlled storage. Moreover, the stability ofall surfactant-based dispersions increased upon the pretreatment approach. Addi-tionally, statistical Raman analysis, TEM, and AFM investigations confirmed thatthese dispersions consist of highly exfoliated graphene sheets featuring large amountsof monolayer graphene. In particular, dispersions of G-I1*-1 revealed about 94%of few-layered/turbostratic graphite and single-layer graphene and about 25% ofsingle-layer graphene within the sample. This pointed to an overall concentration of0.068 mg mL−1 few- and single-layer graphene and about 0.018 mg mL−1 single-layergraphene within the sample.

Based on these results, we further investigated the in situ polymerization of vinyl-benzyltrimethylammonium chloride (VBTA) pretreated graphite to yield stablegraphene dispersions (Chapter 4.3.3). In fact, the in situ polymerization of VBTA-pretreated graphite followed by subsequent mild sonication yielded stable graphenedispersions in the form of few-layered/turbostratic graphite (59%) with moderateamounts of single-layer graphene (12%). These dispersions were comparable tothose derived by extended ultrasonication employing 1 as a stabilizer and, therefore,demonstrated the efficient exfoliation and dispersion of graphene by employing thismild exfoliation procedure. Moreover, the concentration of the prepared samplesexceeded the concentration of pure surfactant-based graphene exfoliation by meansof poly(vinylbenzyl)ammonium chloride (PVBTA) by a factor of 8. However, ex-cessive polymer coating rendered those samples unsuitable for the application insupercapacitor devices.

Solvent-Based Liquid-Phase Exfoliation

In search for non-toxic eco-friendly solvents applicable for the liquid-phase exfoliationof graphite, we tested several organic solvents, namely ethyl acetate (EAC), dibenzylether (DBE), 1,3-dioxolane (DOL), dimethyl phthalate (DMP), ethylene gylcol(EG), butyl acetate (BA), γ-butyrolactone (GBL), and terpineol (TP), in regardto their graphene dispersibility. Further, the dispersion efficiencies were compared to

165


those of N-methyl-2-pyrrolidone (NMP) (Chapter 4.4.1). The graphene concentra-tions varied significantly by a factor of 10 and increased in the order GBL < DOL< BA < EAC < TP < DBE < EG < NMP < DMP, rendering DMP as thebest solvent for graphene dispersion investigated in this study. Moreover, it wasfound that the solvent influences the degree of exfoliation only weakly. StatisticalRaman analysis of the flake thickness distribution revealed that all samples exhibitfew-layer character with large amounts of multilayered/bulk graphite sheets and onlyvery low yields of single-layer graphene. Comparing these results to those derivedfrom surfactant-assisted liquid-phase exfoliation, we concluded that the solvent-basedexfoliation process is less efficient in the exfoliation of graphite down to few- andsingle-layer graphene. Additional evaluation of the defect-related D-band indicatedthat the defect content of the samples increases in the order of NMP < DMP< EG, which was assigned to the reduction of flake size. Unfortunately, the bestsolvents for graphene dispersion were non-volatile, which renders it difficult to processthese dispersions into electrode layers for supercapacitor applications.

In order to disperse graphene in low boiling point solvents, we tested a cosolvencyapproach (Chapter 4.4.3). Unfortunately, we were not able to produce stablegraphene dispersions in different alcohol-water mixtures with a surface tensionclose to 40 mN m−1. However, we succeeded in the exfoliation and dispersionof pretreated graphite in an ethanol/water mixture of a surface tension close to40 mN m−1 (Chapter 4.4.4). In detail, highly stable graphene dispersions withconcentrations of up to 0.039 mg mL−1 were produced. Further, Raman, and TEManalysis confirmed the successful exfoliation towards few-layer graphene/turbostraticgraphite (60%) and single-layer graphene (15%) with some amounts of unexfoliatedmultilayer graphene stacks (40%) being present. However, this approach still fellbehind the surfactant-based liquid-phase exfoliation presented in Chapters 4.2 and4.3. In fact, this can be attributed to enhanced repulsive steric interactions betweensurfactant-coated graphene flakes.

Application of Liquid-Phase Exfoliated Graphene in Supercapacitors

Finally, the prepared liquid-phase exfoliated graphene (LPEG) was employed aselectrode material for supercapacitors in order to test its applicability for energystorage devices (Chapter 5). Indeed, LPEG electrodes proved to be promisingmaterials for electrochemical capacitors. The Ragone plot in Figure 6.2 summarizes

166


the energy and power densities of the LPEG supercapacitor devices in differentelectrolytes, namely liquid H2SO4/PVA aqueous electrolyte, solid-state H2SO4/PVAgel electrolyte, tetraethylammonium fluoroborate/acetonitrile organic electrolyte,and 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([BMIM][NTf2])ionic liquid electrolyte. Moreover, the results are compared to current state-of-the-artenergy storage devices.

Figure 6.2: Ragone plot comparing the energy and power density of LPEG super-capacitors, made from G-I1*-7 dispersions, with a number of commer-cially available energy storage devices. Data for the activated carbonsupercapacitor (AC-EC), the lithium thin-film battery, and the aluminumelectrolytic capacitor are taken from Ref. [274].

Featuring a large ion-accessible surface area, efficient electron and ion transportpathways, as well as a high packaging density, the LPEG electrodes deliveredgravimetric capacitances of up to 202 F g−1 and volumetric capacitance of 9 F cm−3

for G-I1*-7 in aqueous electrolyte. This translated into gravimetric and volumetricenergy densities of 7 Wh kg−1 and 1.2 mWh cm−3, respectively. Even more important,the devices featured exceptional gravimetric and volumetric power densities ofup to 5×105 W kg−1 and up to 100 W cm−3, respectively. Most strikingly, ourLPEG supercapacitors represented a milestone in terms of energy and power density,when compared with previously investigated liquid-phase exfoliated graphene-basedsupercapacitors and supercapacitors built from commercially available graphenesamples.

Taking advantage of higher operational voltage windows in organic and ionic liq-

167


uid electrolytes, ultra high energy and power density of up to 4 mWh cm−3 and346 W cm−3 in [BMIM][NTf2] ionic liquid electrolyte could be attained, respectively.In fact, such extraordinary energy densities are one order of magnitude higher thanthose of typical activated carbon supercapacitors, approaching the upper limit oflithium thin-film batteries. In terms of power density, the devices are comparable tothose of aluminum electrolyte capacitors, about two orders of magnitude higher thanthose of commercial activated carbon supercapacitors, and more than four orders ofmagnitude higher than that of the lithium thin-film battery. Hence, combining theenergy density of micro batteries with the power density of electrolytic capacitorsour LPEG supercapacitors are expected to bridge the gap between batteries andcapacitors.

Furthermore, galvanostatic cycling tests revealed the excellent cycling stability of theLPEG supercapacitor devices, which further renders them promising candidates forcommercial applications. In order to explore the potential of LPEG electrodes forflexible energy storage devices, flexible all-solid-state supercapacitors were built fromG-I1*-7 dispersions, and their electrochemical performance under mechanical stresswas evaluated. Indeed, those devices showed exceptional electrochemical stabilityregardless the bending degree, demonstrating their excellent mechanical stability,which further proved their exciting potential for high-performance flexible devices,such as flexible displays, roll-up portable displays, and wearable electronics.

Outlook

In summary, we have developed a simple and scalable approach for the liquid-phaseexfoliation of graphite towards high-quality few-layer graphene. Moreover, we suc-cessfully integrated these few-layer graphene dispersions into LPEG supercapacitorsthat are compact, reliable, energy and power dense, and possess a long cycle life.Given that graphite is an abundant material, which is already used in a variety ofenergy storage systems, in addition to the scalability of the liquid-phase exfoliationprocess, we believe that LPEG electrodes offer promise for industrial applications.

However, the quantity and graphene concentration of the dispersions still need to beimproved until our material may graduate from lab to marketplace. As such, thescale-up of the stabilization approaches and the transfer towards scalable dispersiontechniques should be in the focus of future work. Additionally, the development of

168


scalable and efficient methods for the removal of free surfactant from the dispersionprior to processing them into working devices is of paramount importance. Moreover,graphene has to find its value proposition in comparison to other emerging nano-materials, such as two-dimensional graphene analogues and graphene foams, andexisting technologies. In light of this, it will be critical to find the balance betweencost and performance.

169

IV

Experimental Details

170

7 Materials and Methods

Materials

Lamellar graphite with a d50 particle size of 590 μm, a density of about 2.25 g cm−3,and a specific surface area of 23 m2 g−1 was purchased from SGL Carbon under thebrand name GFG 350. The graphite was sieved to a particle size of 200-400 μm inorder to remove very large particles. Raman, SEM, and XRD analysis of the rawgraphite starting material can be found in the Appendix A.

6-amino-4-hydroxy-2-naphthalenesulfonic acid (4), tetra-n-butylammonium hydrox-ide, tetraethylammonium chloride, tetraethylammonium hydroxide, Pluronic P-123,vinylbenzyltrimethylammonium chloride, poly(vinylbenzyl)ammonium chloride, 4,4-azobis-4-cyanovaleric acid, and 1-butyl-3-methylimidazolium bis(trifluoromethyl-sulfonyl)imide were purchased from Sigma-Aldrich. Cetyltrimethylammonium bro-mide was purchased from AppliChem GmbH. The dispersant 2 was purchased fromCarl Roth under the brand name Nekal BX, 3 was purchased from Cytec IndustriesInc. under the brand name Aerosol OS, 5 was purchased from Lubrizol under thebrand name Solsperse 27000, 7 was purchased from BASF under the brand nameTamol NN4501, and 1, 6, and 8 were synthesized in the group of Dr. Bernd Gobeltat BYK-Chemie GmbH, respectively. Commercial graphene samples were receivedfrom Graphene Supermarket, Io-li-tec, Angstron Materials, and Vorbeck under thebrand names UHC-NPD, CP-0081, N002-PDR, and F101, respectively. All chemicalswere used as received without further purification. All solvents were purchasedfrom VWR, Sigma-Aldrich, or Merck and applied without further treatment. AllTEM grids were received from Plano. All separators used for the fabrication ofsupercapacitors were purchased from Celgard. Graphene films were prepared ongold Kapton foils purchased from Astral Technologies, aluminum current collectorsreceived from Maxwell Technologies, and PET foils purchased from 3M. All sampleswere prepared in Millipore water received from Sigma Aldrich.

171

7 MATERIALS AND METHODS

Methods

Absorption Spectroscopy

Steady-state UV/vis absorption measurements were performed with a Nicolet Evolu-tion 100 from Thermo Electron Corporation at room temperature. The spectra havebeen measured with a data interval of 1 nm and a detection speed of 480 nm/min.Before each measurement, a baseline correction was carried out. Precision quartzcuvettes with a light path of 1 cm were used for all measurements. The absorbanceat 660 nm was evaluated, and the graphene concentration was calculated usingthe extinction coefficient of α=1390 mL mg−1 m−1 determined by Loyta et al. forfew-layer graphene in surfactant solutions.[106]

Steady-State Emission

Steady-state fluorescence measurements were performed with a Fluoromax-3P spectro-meter from Horiba Jobin Yvon. In order to correct the results for the instrumentresponse, the sample signal was divided by the reference signal of the lamp. Theintegration time was 0.5 s with a data interval of 1 nm. All measurements wereperformed at room temperature with a 1 cm×1 cm cuvette.

Raman Spectroscopy

Raman spectroscopy measurements were carried out with a Renishaw inVia Reflexconfocal Raman microscope using laser excitations of 532 nm. The samples wereprepared on silicon wafers coated with a 90 nm SiO2 layer.

Statistical Raman analysis was performed by creating a square map with spacings of1 μm between two adjacent data points in each x- and y-direction. At each point,a Raman spectrum was taken. For evaluation of the exfoliation degree the D-, G-,and 2D-Raman peaks were fitted by a single Lorentzian function using OriginLab toanalyze the height and full width at half maximum (FWHM) of the Raman peaks.

172


Wafer Coating

The supernatants of the different graphene dispersions after centrifugation weredeposited onto silicon wafers coated with a 90 nm SiO2 layer. To this end, the SiO2/Sisubstrates were first surface modified with N-[3-(trimethoxysilyl)propyl]ethylene-diamine, then immersed in the graphene suspension for 15 min and subsequentlydried under a stream of pure synthetic air. Residuals of the surfactants, which woulddisturb further investigations, were removed by rinsing the wafers with absoluteethanol and deionized water.

Optical Light Microscopy

Optical microscopy images were taken with a Zeiss Axio Scope.A1 with white lightillumination using bright field image mode.

Transmission Electron Microscopy

Bright-field TEM images were recorded with an 80 kV EM 900 from Carl Zeiss AG.The samples were prepared by drying one drop (5 μL) of the suspension on coppergrids covered with a lacey carbon film.

Atomic Force Microscopy

AFM images were obtained with a Nanoscope IIIa Multimode, Veeco in tappingmode. AFM was performed in the tapping mode of operation, using a NSC15/AIBSprobe with a spring constant of ∼40 N/m and a resonance frequency of ∼307-341 kHz.All obtained AFM images were analyzed by contrast imaging in Gwyddion 2.36.

Zeta Potential

Zeta potential measurements were carried out on a Delsa Nano C Particle Analyzerfrom Beckman Coulter with irradiation from a 633 nm He-Ne laser. The samples wereinjected in folded capillary cells, and the electrophoretic mobility (μ) was measured

173


using a combination of electrophoresis and laser Doppler velocimetry techniques. Theelectrophoretic mobility relates to the drift velocity of a colloid (υ) to the appliedelectric field (E), υ=μE. All measurements were conducted at 20 ◦C and at thenatural pH of the surfactant solution unless stated otherwise. The zeta potential(ζ) can be calculated from the electrophoretic mobility using the Smoluchowskiexpression for plate-like particles: ζ=ημ/ε, where η is the solution viscosity, and ε isthe solution permittivity, ε = εrε0. This expression applies for plates with uniformsurface charge, which are large enough for edge charge to be neglected and whoseradius is much larger than the double-layer thickness.[106]

Surface Tension

The surface tensions of the solvent mixtures were measured using the ring methodwith a KRUSS Processor Tensiometer K100.

Conductivity Measurements

Conductivity measurements of bendable graphene electrodes were performed bytwo-point probe measurements using a voltameter. Several equidistant squares werepainted onto the surface of the graphene film using silver paint to provide improvedcontact with lower resistance to the voltmeter probes. The length, width, andheight (thickness) between the two squares were measured, and the conductivity wasdetermined using equation: σ=d/RsA, where Rs is the sheet resistance, d is the filmthickness, and A is the area.

Film Tickness Measurements

The thickness of graphene films was measured with a Mitutoyo IP65 micrometer anda surface profiler, DektakxT from Bruker. Cross section measurements were carriedout with a SEM, Zeiss - Supra 35.

174


Electrochemical Measurements

Electrochemical characterization was performed using a Biologic VMP3 potentiostatand a VMP3b-10 Amp booster. The devices were connected to the potentiostat byeither using alligator clips or a four-point holder for coin cells. Cyclic voltammetrycurves and galvanostatic charging-discharging curves were obtained from 0 to 1 V forH2SO2/PVA electrolytes, from 0 to 2 V for organic electrolyte, and from 0 to 3.5 V forionic liquid electrolyte, respectively. The stable voltage window for each electrolytewas determined through CV at 100 mV s−1 at different voltages. Electrochemicalimpedance measurements were also taken on the Biologic VMP3 over a frequencyrange from 1 MHz to 10 mHz at open circuit potential. For bending tests, thefabricated device was conformally wrapped around PET cylinders of known radiiand tested in the flexed configuration.

Calculations[223,274]

The capacitance can be calculated from the galvanostatic charging-discharging curvesat different current densities using following formula:

Cdevice = iapp

(−dV/dt) , (7.1)

wherein iapp is the applied current (in A), and dV/dt is the slope of the dischargecurve (in V s−1). The specific capacitances were calculated based on the area (1 cm2),volume (including current collector, active material, and separator with electrolyte)or mass (including the electrode mass) of the device stack according to the followingequations:

Areal capacitance = Cdevice

A, (7.2)

V olumetric stack capacitance = Cdevice

V, (7.3)

175


Gravimetric capacitance = Cdevice

m, (7.4)

wherein A is the area of the device (in cm2), V is the volume of the device (cm3)and m is the mass of the active material (in g).

Moreover, the equivalent series resistance RESR was calculated from the constantcurrent data by using the voltage drop Vdropp at the beginning of the discharge curveaccording to the following formula:

RESR = Vdrop

2iapp

. (7.5)

Next, the specific and volumetric power density, Pgrav and Pvol, were calculated usingthe following equations:

Pvol = (ΔE)2

4RESRVand (7.6)

Pgrav = (ΔE)2

4RESRm, (7.7)

wherein P is the power (in W cm−3 and W kg−1, respectively), ΔE is the operatingvoltage window (in V), V is the volume stack (in cm3), and RESR is the internalresistance of the device (in Ω).

Moreover, the energy density of the devices was obtained from constant currentcharging-discharging experiments by following formulas:

Evol = Cv · (ΔE)2

2 · 36000 and (7.8)

Egrav = Cs · (ΔE)2

2 · 36000 , (7.9)

wherein E is the energy density (in Wh cm−3 and Wh kg−1, respectively), ΔE is theoperating voltage window (in V), and Cv and Cs are the volumetric stack capacitanceand gravimetric capacitance obtained from Equation 7.3 and 7.4, respectively.

176

8 Preparative Equipment

Centrifugation

The sonicated graphene/graphite samples were centrifuged to remove any poorlydispersed graphitic material. To this end, the dispersions were transferred to 1.5 mLcentrifugation tubes and spun in a Mikro 200 (Hettich GmbH) centrifuge with amaximum speed of 15,000 rpm. The top 1 mL of graphene suspension, correspondingto a maximum sedimentation distance of approximately 1 cm, was carefully extractedfrom the centrifuge tubes after centrifugation. Different centrifugation conditions wereemployed in order to find optimal centrifugation conditions. Dispersions preparedby low centrifugation speed produced excess levels of poorly dispersed graphiticmaterials while stronger centrifugation conditions lead to well-dispersed graphene. Tothis end, dispersions obtained using 10 minutes of centrifugation at 15,000 rpm wereused to study different stabilization routes presented in Chapter 4. The centrifugalacceleration was 15,091 g.

Ultrasonic Bath

Bath ultrasonication was performed with a Bandelin SONOREX RK 100 ultrasoni-cation bath cleaner with a power output of 80 W. In order to guarantee a constantwater temperature fresh cold water was refilled every 30 min.

Tip Sonicator

Tip ultrasonication was performed with an UP50H tip sonicator from Dr. HielscherGmbH. In order to guarantee constant temperature, a cooling unit from Huber wasused. Moreover, the dispersion was stirred homogeneously during ultrasonicationwith a magnetic stirrer - see Figure 8.1.

177

8 PREPARATIVE EQUIPMENT

Figure 8.1: Schematic representation of the experimental setup used for the liquid-phase exfoliation of graphite by ultrasonication.

The specific energy input delivered to the graphene suspension was estimated bythe calorimetric method based on the temperature increase in the liquid over timeaccording to E = m · CP , where E is the energy input (in J), T is the temperature(in K), Cp is the specific heat capacity of the liquid (in J g−1 K−1), and m isthe mass of the liquid.[406] The energy input of 6 h ultrasonication with a poweroutput of 60% and a cycle of 0.5 using 40 mL of probe volume was calculated to be63.86 kJ. Moreover, the delivered acoustic power was 2.95 W, the energy per volumewas 1.59×10−6 kJ m−3, and the energy per mass of graphite starting material was3.20×10−5 kJ kg−1.

178

9 Preparation of GrapheneDispersions

Surfactant Solutions for the Optimization of theSonication Procedure

For the dispersion tests, 200 mg of graphite was added to 40 mL of aqueous solutionsof either CTAB or 1 and tetraethylammonium hydroxide (0.082 mol L−1) to givean initial graphite concentration of 5 mg mL−1. The surfactant concentrationwas 1.25 mg mL−1. Afterward, the resulting dispersions were stirred overnight(>15 h). Next, the dispersions were sonicated by either tip or bath sonication anddifferent sonication power outputs, sonication times, and temperatures were explored.Afterward, the samples were centrifuged for 10 min at 15,000 rpm, as not statedotherwise. Then, the top two-thirds of the dispersions were gently extracted andsubsequently used for further analysis.

Graphene Dispersions with the Aid of AromaticAmphiphiles

For each surfactant type, 200 mg of graphite was added to 40 mL of aqueoussurfactant solution (1.25 mg mL−1) to give an initial graphite concentration of5 mg mL−1. Afterward, the resulting suspensions were stirred overnight (>15 h).Next, the mixtures were continuously stirred with a magnetic stirrer, while sonicationwas applied for 6 h at 17 ◦C. The top 20 mL was decanted into 1.5 mL vials andcentrifuged for 10 min at 15,000 rpm. The top 1 mL were then decanted into a4 mL vial. UV/vis absorption spectroscopy and wafer preparation were performedimmediately. The dispersions were then left to stand for 4 weeks undisturbed, and theabsorption spectrum was measured again. For each surfactant type, this procedurewas performed at least three times. Samples of the reference surfactant CTAB wereprepared under the same procedure.

179

9 PREPARATION OF GRAPHENE DISPERSIONS

Graphene Dispersions from Pretreated Graphite inSurfactant Solutions

200 mg of graphite was dispersed in an aqueous solution of tetraethylammoniumhydroxide, tetra-n-butylammonium hydroxide, tetraethylammonium chloride, sodiumhydroxide, or 1 (0.082 mol L−1), respectively, to give an initial graphite concentrationof 5 mg mL−1. Afterward, the resulting suspensions were stirred overnight (>15 h).Next, the mixtures were continuously stirred with a magnetic stirrer while sonicationwas applied for 1 h at 17 ◦C. The sample was then allowed to settle overnight.The next day, the supernatant was eliminated to ensure complete removal of freetetraalkylammonium ions/NaOH/1 and the precipitate was redispersed in 40 mL ofa surfactant solution (1.25 mg mL−1). The resulting dispersion was stirred overnightand afterwards the sample was sonicated for 5 h at 17 ◦C during continuous stirring.Next, the top 20 mL was decanted into 1.5 mL vials and centrifuged for 10 minat 15,000 rpm. The top 1 mL were then decanted into a 4 mL vial. UV/visabsorption spectroscopy and wafer preparation were immediately performed. Thedispersions were then left to stand for 4 weeks undisturbed and after that time theabsorption spectrum was measured again. For each surfactant type, this procedurewas performed at least three times.

Graphene Dispersions in Organic Solvents

Graphite was dispersed in the respective solvent at a concentration of 5 mg mL−1.Afterward, the resulting suspensions were stirred overnight (>15 h). Due to safetyreasons, the samples were sonicated in a bath sonicator for 4 h at 17 ◦C, which iscomparable to tip sonication of 6 h. The top 20 mL was decanted into 1.5 mL vialsand centrifuged for 10 min at 15,000 rpm. The top 1 mL were then decanted into a4 mL vial. UV/vis absorption spectroscopy and wafer preparation were immediatelyperformed. The dispersions were then left to stand for 4 weeks undisturbed and afterthat time the absorption spectrum was measured again.

180


Graphene Dispersions from Pretreated Graphite in anEthanol/Water Cosolvent Mixture

200 mg of graphite was dispersed in an aqueous solution of tetraethylammoniumhydroxide (0.082 mol L−1) to give an initial graphite concentration of 5 mg mL−1.Afterward, the resulting suspensions were stirred overnight (>15 h). Next, the mix-tures were continuously stirred with a magnetic stirrer, while sonication was appliedfor 1 h at 17 ◦C. The sample was then allowed to settle overnight. The next day thesupernatant was eliminated to ensure complete removal of free tetraethylammoniumions and the sample was dried. The precipitate was redispersed in 40 mL of anethanol/water (3:20 w/w) mixture. The resulting solution was stirred overnight andafterward the sample was sonicated for 5 h at 17 ◦C during continuous stirring. Thetop 20 mL was decanted into 1.5 mL vials and centrifuged for 10 min at 15,000 rpm.The top 1 mL were then decanted into a 4 mL vial. UV/vis absorption spectroscopyand wafer preparation were performed immediately. The dispersions were then leftto stand for 4 weeks undisturbed and after that time the absorption spectrum wasmeasured again.

Graphene Dispersions from in situ Polymerization ofVinybenzyltriethylammonium Chloride

A mixture of VBTACl monomer (1 g), graphite (400 mg), and deionized water (40 mL)in a 50 mL vial was sonicated for 1 h at 17 ◦C under continuous stirring. The samplewas then allowed to settle overnight. The next day the supernatant was eliminated toensure the removal of any free VBTA and the precipitate was redispersed in 250 mLwater in a Schlenk flask. Then, 0.01275 g 4,4-azobis-4-cyanovaleric acid (ACVA) wasadded to this solution, which was stirred with a stirring bar while the solution waspurged with nitrogen for 2 h. The synthesis was carried out in a thermostated oilbath at 60 ◦C for 6 h while the solution was continuously stirred. The mixture wascooled to 25 ◦C, opened to the air, and sonicated for 1 h.[336] Afterward, the top20 mL was decanted into 1.5 mL vials and centrifuged for 10 min at 15,000 rpm.The top 1 mL were then decanted into a 4 mL vial. UV/vis absorption spectroscopyand wafer preparation were immediately performed.

181


Figure 9.1: Reaction mechanism of the free-radical polymerization of vinylbenzyltri-ethylammonium chloride.

182

10 Preparation of Supercapacitors

Preparation of Graphene Electrode Films

A distinct amount of graphene dispersion was prepared according to Chapter 9. Thedispersions were then concentrated and desalted in centrifugal dialysis devices at acentrifugation rate of 3500 rpm for 10 min (Amicon Ultra - 15, Millipore). Afterward,the dispersions were used to make graphene films on various substrates, includingpolyethylene terephthalate (PET), gold coated Kapton foil, and aluminum foil. Thegraphene films were made by drop casting the graphene dispersions onto substrates,which were previously cut to a size of 4×4 cm. The films were then allowed to dryfor 24 hours under ambient conditions.

Figure 10.1: A cross-sectional SEM image of an LPEG film drop cast on PET sub-strate featuring small gaps between the turbostratically aggregatedgraphene sheets. In order to adjust for the film roughness and inhomo-geneity a film thickness of 2 μm was taken for volumetric calculations.

The thickness of the graphene layer was measured from micrometer analysis, cross-sectional SEM, and profilometry. In an actual device, the area made accessible tothe electrolyte was 1 cm2.

183

10 PREPARATION OF SUPERCAPACITORS

Measurement of the Specific Surface Area ofGraphene

The surface area of the graphene sheets was determined by methylene blue adsorptionmethod. Methylene blue is commonly used to determine the surface area of graphiticmaterials, with each molecule of adsorbed methylene blue representing 1.35 nm2 ofgraphene surface area.[269,274,407–410] The surface area was calculated by adding aknown volume of a methylene blue solution to a graphene dispersion. Afterward, themixture was stirred continuously for 24 hours in order to reach maximum adsorption.The mixture was then allowed to settle overnight and was further centrifuged toremove any suspended material. The methylene blue concentration was determinedby measuring the adsorption of the supernatant at a wavelength of 665 nm andcomparing it to the initial standard concentration of methylene blue.

Fabrication of Electrochemical Capacitors

Gold-Kapton-Based Electrochemical Capacitor Cells

In a typical device, an ion-porous separator (Celgard 3501) was sandwiched betweentwo identical pieces of graphene-coated gold current collector in a layered structure -see Figure 10.2. This layered assembly was wrapped with Kapton tape and dipped inthe electrolyte solution (H2SO4/PVA electrolyte or ionic liquid electrolyte). In orderto evaluate the electrochemical performance of the devices, they were connected toan electrochemical working station using alligator clips. Electrode edges were lightlypainted with conducting silver paint, and a strip of copper tape was attached toensure good electrical contact between the graphene film and the alligator clips. Thedimensions of a typical device are shown in Figure 10.2. The active area was 1 cm2.In an all-solid-state device, the organic separator was replaced with a H2SO4/PVAgelled electrolyte, which has a thickness of 11 μm and, thus, reduces the devicethickness by 14 μm.

184


Figure 10.2: Schematic representation of the supercapacitor device showing the pouchcell assembly (right) and the dimensions of the separator, current col-lector, and LPEG films (left).

Fabrication of All-Solid-State Flexible GrapheneSupercapacitors

All-solid-state devices were made by pouring the H2SO4/PVA gelled electrolyte(100 μL electrolyte at 1 cm2 of the electrode) carefully onto the graphene coatedelectrodes. This assembly was left for 5 hours under ambient conditions to ensurethat the electrolyte completely wets the electrode and to allow evaporation of anyexcess water. Afterward, the two electrodes were assembled face-to-face and leftovernight until the electrolyte solidified. This resulted in mechanically robust andflexible devices in which the polymer electrolyte served as both, the electrolyte andthe ion porous separator.

Preparation of the Polymer Gelled Electrolyte

The gelled electrolyte was prepared according to Ref. [274]. In particular, 1 gpolyvinyl alcohol (PVA) powder was mixed with 10 g water. Afterward, the mixturewas heated at ∼90 ◦C under constant stirring until the solution turned clear. Aftercooling under ambient conditions, 0.8 g of concentrated sulfonic acid was added tothe solution followed by stirring.

185


Assembly of Graphene Electrodes into Coin Cell SupercapacitorDevices

Coin cell supercapacitor devices were assembled from LPEG coated aluminum currentcollectors. In particular, 15 mm diameter disks were punched from the LPEG coatedaluminum foils using a 15 mm diameter round hole arch punch (McMaster-Carr).These disks served as the electrodes in the coin cell assembly. Each electrode wasweighed using a microbalance and the thicknesses were measured using a micrometer.The assembly of the coin cell (CR2032) device was made by starting with the bottomcap and gasket, then adding one electrode, the polypropylene separator (Celgard2501), the electrolyte (1 M tetraethylammonium tetrafluoroborate in acetonitrile),and the other electrode. The spacer, spring, and the top cap were then added toyield the full cell assembly. Finally, the stack was crimped shut. Devices were madein air, as former studies of Wang et al. revealed no noticeable difference in the resultof the cells made in air versus those made under nitrogen.[359] The time it takes tobuild a device, from inserting the electrolyte to sealing the cell is only about 20 s.

Figure 10.3: Schematic representation showing the coin cell assembly (right) andthe dimensions of the separator, current collector, and LPEG film,respectively (left). The picture of the coin cell assembly (CR2032) wasreprinted from Ref. [411].

186

List of Figures

1.1 a) World energy consumption growth by fuel type from 1965 to 2014.b) The share of different energy sources for global energy consumptionin 2014. Data is taken from Ref. [3]. . . . . . . . . . . . . . . . . . . 1

2.1 a) Schematic representation of the bonding structure in graphenefeaturing the delocalized π-electron cloud in the out of plane z-axis.b) Schematic representation of AB-Bernal stacked graphite. One setof carbon atoms is over the center of the hexagons in the adjacent layers. 6

2.2 a) Structure of graphene. �a1 and �a2 represent the unit vectors. b)Reciprocal lattice of graphene. The gray hexagon is the first Brillouinzone (BZ). �b1 and �b2 represent the reciprocal lattice vectors. c) Theband structure of graphene calculated from tight binding model and azoom-in of the dispersion relation close to the K point or Dirac pointfor small energies revealing the high degree of electron-hole symmetry. 7

2.3 Overview of graphene’s properties and different potential fields ofapplication. Photographs are taken from Ref. [51–53]. . . . . . . . . . 9

2.4 Schematic representation of the main production routes of graphene,which allow a wide choice in terms of flake size, quality, and price forany particular application. Pictures are taken from Ref. [28]. . . . . . 13

2.5 Schematic representation of different graphene synthesis approachesstarting from raw graphite. Oxidation of graphite and subsequentreduction of graphene oxide (GO) to reduced graphene oxide (rGO)(right). Liquid-phase exfoliation of graphene by either solvent-assistedliquid-phase exfoliation (middle) or surfactant-assisted liquid-phaseexfoliation (left). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

2.6 Structures of the most common surfactants employed for the exfoliationand stabilization of graphene in aqueous medium. . . . . . . . . . . . 17

2.7 Schematic diagram of the Raman scattering processes. The lowestenergy vibrational state m is shown at the bottom with states ofincreasing energy above it. Both, the excitation energy (upwardarrows) and the scattered energy (downward arrows) is much largerthan the energy of a vibration. . . . . . . . . . . . . . . . . . . . . . . 24

xv

LIST OF FIGURES

2.8 Raman spectrum of single-layer graphene prepared by micromechanicalcleavage showing the main Raman features, the D-, G-, and 2D-band.The sample was transferred onto a Si/SiO2 wafer and excited at 532 nm. 25

2.9 a) Schematic representation of the physical origin of the G-band. b)Atomic displacements for the E2g modes. Schematic representation ofthe physical origin of c) the D-band and d) the 2D-band.[185] . . . . . 27

3.1 The Ragone plot for various electrochemical energy storage devices.If a supercapacitor is used in an electric vehicle, the specific powershows how fast one can go, while the specific energy shows how farone can go on a single charge. Times shown are the time constant ofthe devices, obtained by dividing the energy density by the power.Adapted from Ref. [194]. . . . . . . . . . . . . . . . . . . . . . . . . . 29

3.2 Schematic representation of the electrical double-layer at a positivelycharged surface: a) the Helmholtz model, b) the Gouy-Chapmanmodel, and c) the Stern model. IHP and OHP represent the innerHelmholtz plane and outer Helmholtz plane, respectively. The OHPis also the plane where the diffuse layer begins. H is the double-layer distance described by the Helmholtz model. Ψ0 and Ψ arethe potentials at the electrode surface and the electrode/electrolyteinterface, respectively.[207] . . . . . . . . . . . . . . . . . . . . . . . . . 30

3.3 Schematic representation of a supercapacitor test cell assembly.[47] . . 313.4 Schematic representation of an electric double-layer supercapacitor in

either charged (left) or discharged (right) state. . . . . . . . . . . . . 333.5 Cyclic voltammetry (CV) protocol for supercapacitor characterization,

showing the potential-time curve (left) and the CV curve of an idealsupercapacitor featuring a highly symmetric rectangular shape (right).[60] 38

3.6 Schematic representation of CV non-idealities: a) Scan rate limitation,b) electrolyte degradation occurring at a potential window whichexceeds the electrolyte stability window, and c) pseudocapacitivecharge storage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.7 Galvanostatic charging-discharging protocol for supercapacitor char-acterization, showing the current-time-profile (left) and the highlysymmetrical triangular charging-discharging curve of an ideal super-capacitor (right).[60] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

xvi

LIST OF FIGURES

3.8 Charging-discharging curve (CDC) of a supercapacitor illustrating thecapacitative and resistive parts of the CDC.[60] . . . . . . . . . . . . . 41

3.9 Schematic representation of the Nyquist impedance plot of a) anideal capacitor with a series resistance Rs and b) an electrochemicalcapacitor with a porous electrode, featuring the equivalent seriesresistance (ESR) and the equivalent distributed resistance (EDR).[212] 43

3.10 Schematic representation of a graphene-based supercapacitor featuringa stacked geometry. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

4.1 Structure of the amphiphilic pyrene derivative 1-pyrenebutyric acidmethoxypolyethylene glycol ester (1). The structure of CTAB isshown in Figure 2.6. . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

4.2 Calculated graphene concentration remaining in the supernatant ofG-CTAB after centrifugation cac a) for a tip sonicated and bathsonicated sample and b) as a function of the temperature Tsonic. Theerror bars correspond to the standard deviation of the average valuesobtained by performing three independent experiments. . . . . . . . . 59

4.3 a) Calculated graphene concentration after centrifugation cac of G-CTAB and G-1 dispersions as a function of sonicator power output.b) ID/IG ratio of G-CTAB and G-1 dispersions as a function ofsonicator power output. . . . . . . . . . . . . . . . . . . . . . . . . . 60

4.4 Calculated concentration after centrifugation cac for CTAB and G-1as a function of ultrasonication time, respectively. . . . . . . . . . . . 61

4.5 Optical absorbance spectrum of G-CTAB diluted by a factor of10. Inset: Photograph of the as-prepared and centrifuged G-CTABdispersion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

4.6 Representative solid-state Raman spectra of G-CTAB, indicating theexfoliation state of the sample. The samples were dip coated fromdispersion onto a Si/SiO2 wafer and excited at 532 nm. Inset: Fittingof the 2D-band. Black dots - experimental data; blue curve - fitting. . 65

4.7 Histogram resulting from the Raman mapping of G-CTAB, showingrelative counts versus I2D/IG ratio and the corresponding log-normaldistribution. The sample was dip coated from dispersion onto Si/SiO2

wafer and excited at 532 nm. . . . . . . . . . . . . . . . . . . . . . . . 66

xvii

LIST OF FIGURES

4.8 Atomic force microscopy image (left) of G-CTAB hybrids from adip coated and washed Si/SiO2 wafer and the corresponding heightprofiles (right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

4.9 Representative TEM micrographs of the G-CTAB sample with dif-ferent magnifications. The sample was drop coated on a lacey carbongrid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

4.10 Structures of aromatic amphiphiles, which are employed for the non-covalent stabilization of graphene in aqueous medium. The structureof 1 is shown in Figure 4.1. . . . . . . . . . . . . . . . . . . . . . . . 70

4.11 Calculated graphene concentration cac directly after preparation and4 weeks after preparation for G-1, G-2, G-4, and G-7, respectively. . 73

4.12 a) Fluorescence spectra of 7 mixed with graphite in water beforeultrasonication (Graphite + 7) and after ultrasonication (G-7) withλ=260 nm. b) Fluorescence spectra of 1 mixed with graphite in waterbefore ultrasonication (Graphite + 1) and after ultrasonication (G-1)with λ=340 nm (bottom). Insets: Photographs of the G-7 and G-1dispersion, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . 74

4.13 Calculated graphene concentration (cac) for the G-7 sample as afunction of surfactant concentration (c7) achieved by either an initialsurfactant addition prior to ultrasonication or a continuous surfactantaddition during ultrasonication treatment, respectively. The insetshows photographs of the dispersion, featuring highest (right) andlowest (left) graphene concentration, respectively. . . . . . . . . . . . 76

4.14 A schematic representation of the proposed dispersion and exfoliationmechanism of 7. Firstly, 7 partially intercalates into the grapheneedges during the stirring process. Secondly, exfoliation of grapheneby ultrasonication and subsequent adsorption of 7 onto the graphenebasal plane, yielding stabilized graphene flakes in aqueous medium. . 78

4.15 Selected solid-state Raman spectrum of G-1, indicating the exfoliationstate of the sample. The sample was dip coated from dispersion ontoa Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the 2D-band.Black dots - experimental data; blue curve - fitting. . . . . . . . . . 79

xviii

LIST OF FIGURES

4.16 Histograms resulting from the Raman mapping of G-1, showing a)counts versus I2D/IG ratio, b) counts versus 2D-FWHM, and c) countsversus ID/IG and the corresponding log-normal distribution, respec-tively. The sample was dip coated from dispersion onto a Si/SiO2

wafer and excited at 532 nm. . . . . . . . . . . . . . . . . . . . . . . . 814.17 Selected solid-state Raman spectra of G-7, indicating the exfoliation

state of the sample. The sample was dip coated from dispersion ontoa Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the 2D band.Black dots - experimental data; blue curve - fitting. . . . . . . . . . . 82

4.18 Histograms resulting from the Raman mapping of G-7, showing a)counts versus I2D/IG ratio, b) counts versus 2D-FWHM, and c) countsversus ID/IG the corresponding log-normal distribution, respectively.The sample was dip coated from the dispersion onto Si/SiO2 wafersand excited at 532 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . 83

4.19 Histograms resulting from the Raman mapping of G-2, showing a)counts versus I2D/IG ratio and b) counts versus ID/IG and the cor-responding log-normal distribution, respectively. The samples weredip coated from the dispersion onto Si/SiO2 wafers and excited at 532nm, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

4.20 Histograms resulting from the Raman mapping of G-4, showing a)counts versus I2D/IG ratio and b) counts versus ID/IG and the cor-responding log-normal distribution, respectively. The samples weredip coated from the dispersion onto Si/SiO2 wafers and excited at 532nm, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

4.21 Representative TEM micrographs of G-1 with different magnifications.The sample was drop coated on a lacey carbon grid. . . . . . . . . . . 86

4.22 Representative TEM micrographs of G-7 with different magnifications.The sample was drop coated on a lacey carbon grid. . . . . . . . . . . 87

4.23 Calculated graphene concentration cac directly after preparation and4 weeks after preparation for G-1, G-I1-1, and G-I1*-1, respectively.G-1: Graphite + 1 / 6 h ultrasonication; G-I1-1: Graphite + I1 + 1/ 6 h ultrasonication; G-I1*-1: Graphite + I1 / 1 h ultrasonication→ precipitate, precipitate + 1 / 5 h ultrasonication . . . . . . . . . . 91

xix

LIST OF FIGURES

4.24 Calculated graphene concentrations cac for graphene dispersions pre-pared by pretreatment of graphite, namely G-1*-1, G-NaOH*-1,G-I2-1, and G-I3*-1. . . . . . . . . . . . . . . . . . . . . . . . . . . . 93

4.25 A schematic representation of the dispersion and exfoliation mecha-nism. Firstly, selective interaction of I1 with hydroxyl and carboxylgroups at the graphite edges. Secondly, 1 partially intercalates intothe graphene edges. Thirdly, exfoliated graphene sheets with 1 coatingon the surface and, thus, stabilizing the flakes in aqueous medium. . . 95

4.26 Calculated graphene concentration cac for the long-term intercalationof 1 into the I1 decorated graphene sheets by either stirring or nostirring. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

4.27 Calculated graphene concentration cac after preparation and 4 weeksafter preparation, showing the effect of pretreatment of raw graphitein solutions of I1 on graphene dispersibility of different surfactantsand the stability of the resulting dispersions, respectively. . . . . . . . 97

4.28 Selected solid-state Raman spectrum of G-I1*-1, indicating the exfoli-ation state of the sample. The sample was dip coated from dispersiononto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black curve - experimental data; blue curve - fitting. . . . . 99

4.29 Histograms resulting from the Raman mapping of G-I1*-1, showing a)counts versus I2D/ID, b) counts versus 2D-FWHM, and c) counts versusID/IG and the corresponding log-normal distribution, respectively.The sample was dip coated from dispersion onto a Si/SiO2 wafer andexcited at 532 nm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

4.30 Selected solid-state Raman spectrum of G-I1-1, indicating the exfoli-ation state of the sample. The sample was dip coated from dispersiononto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black dots - experimental data; blue curve - fitting. . . . . 101


xx

LIST OF FIGURES

4.32 Selected solid-state Raman spectrum of G-I1*-5, indicating the exfoli-ation state of the sample. The sample was dip coated from dispersiononto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black line - experimental data; blue curve - fitting. . . . . . 103


4.34 Representative TEM micrographs of G-I1*-1 with different magnifi-cations. The sample was drop coated either on a lacey carbon grid(upper part) or on a carbon coated lacey carbon grid (lower part). . . 105

4.35 Representative TEM micrographs of G-I1*-7 with different magnifi-cations. The sample was drop coated either on a lacey carbon grid(upper part) or on a carbon coated grid (lower part). . . . . . . . . . 106

4.36 Atomic force microscopy image (left) of G-I1*-1 hybrids from a dipcoated and washed Si/SiO2 wafer and the corresponding height profiles(right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

4.37 Atomic force microscopy image (left) of G-I1*-7 hybrids from a dipcoated and washed Si/SiO2 wafer and the corresponding height profiles(right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

4.38 XRD spectra of raw graphite, pretreated graphite in VBTA, and insitu polymerized pretreated graphite, respectively. . . . . . . . . . . . 110

4.39 Schematic representation of the exfoliation mechanism according tothe in situ polymerization approach. Firstly, selective interaction ofVBTA with the graphite layers during the pretreatment process.Secondly, in situ polymerisation of VBTA. Thirdly, exfoliation ofPVBTA intercalated graphite to yield PVBTA stabilized grapheneflakes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

4.40 Selected solid-state Raman spectrum of G-iPVBTA, indicating theexfoliation state of the sample. The sample was dip coated fromdispersions onto a Si/SiO2 wafer and excited at 532 nm. . . . . . . . 113

xxi

LIST OF FIGURES

4.41 Histograms resulting from the Raman mapping of G-iPVBTA, show-ing a) counts versus I2D/IG and b) counts versus ID/IG and thecorresponding log-normal distribution, respectively. The sample wasdip coated from dispersion onto a Si/SiO2 wafer and excited at 532 nm.114

4.42 Calculated graphene concentration after centrifugation cac of graphenedispersions prepared by different solvents, namely ethyl acetate (EAC),dibenzyl ether (DBE), 1,3-dioxolane (DOL), dimethyl phthalate(DMP), ethylene gylcol (EG), butyl acetate (BA), γ-butyrolactoneGBL, and terpineol (TP), respectively. . . . . . . . . . . . . . . . . . 117

4.43 Selected solid-state Raman spectrum of G-DMP, indicating the exfoli-ation state of the sample. The sample was dip coated from dispersionsonto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black dots - experimental date; blue curve - fitting. . . . . 119

4.44 Histograms resulting from the Raman mapping of G-DMP, showing a)counts versus I2D/IG and b) counts versus ID/IG and the correspondinglog-normal distribution, respectively. The sample was dip coated fromdispersion onto a Si/SiO2 wafer and excited at 532 nm. . . . . . . . . 120

4.45 Selected solid-state Raman spectrum of G-EG, indicating the exfolia-tion state of the sample. The sample was dip coated from dispersionsonto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black dots - experimental date; blue curve - fitting. . . . . 121

4.46 Histograms resulting from the Raman mapping of G-EG, showing a)counts versus I2D/IG and b) counts versus ID/IG and the correspondinglog-normal distribution, respectively. The sample was dip coated fromdispersion onto a Si/SiO2 wafer and excited at 532 nm. . . . . . . . . 122

4.47 Selected solid-state Raman spectrum of G-NMP, indicating the exfoli-ation state of the sample. The sample was dip coated from dispersionsonto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting of the2D-band. Black dots - experimental date; blue curve - fitting. . . . . 123

4.48 Histograms resulting from the Raman mapping of G-NMP, showing a)counts versus I2D/IG and b) counts versus ID/IG and the correspondinglog-normal distribution, respectively. The sample was dip coated fromdispersion onto a Si/SiO2 wafer and excited at 532 nm. . . . . . . . . 124

xxii

LIST OF FIGURES

4.49 Representative TEM micrographs of few-layer graphene sheets stabi-lized in an acetone/water mixture (7:25). The sample was drop coatedon a lacey carbon grid. . . . . . . . . . . . . . . . . . . . . . . . . . . 127

4.50 Selected solid-state Raman spectrum of G-I1*-EtOH/W, indicatingthe exfoliation state of the sample. The sample was dip coated fromdispersions onto a Si/SiO2 wafer and excited at 532 nm. Inset: Fittingof the 2D-band. Black dots - experimental date; blue curve - fitting. . 129

4.51 Histograms resulting from the Raman mapping of G-I1*-EtOH/W,showing a) counts versus I2D/IG, b) counts versus 2D-FWHM, and c)counts versus ID/IG and the corresponding log-normal distribution,respectively. The sample was dip coated from dispersion onto a Si/SiO2

wafer and excited at 532 nm. . . . . . . . . . . . . . . . . . . . . . . . 1304.52 Representative TEM micrographs of G-I1*-EtOH/W with different

magnifications. The sample was drop coated on a lacey carbon grid. . 131

5.1 Schematic representation of the preparation procedure for LPEGsupercapacitors. Firstly, the graphene dispersions are concentratedusing a centrifugal dialysis device. Secondly, the graphene dispersion isdrop coated onto a PET film or Au-coated polyimide film, respectively.Afterward, the samples were dried yielding a thin graphene film.Thirdly, the coated graphene films are assembled into supercapacitorcells. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

5.2 a) Electrical characterization of a LPEG electrode. b ) The electricalresistance change of a LPEG electrode as a function of the bendingradius. The inset shows the electrode upon bending. The electrodewas prepared by drop coating G-I1*-7 on a polyethylene terephthalate(PET) flexible support. . . . . . . . . . . . . . . . . . . . . . . . . . . 136

xxiii

LIST OF FIGURES

5.3 Evaluation of LPEG supercapacitors in aqueous H2SO4/PVA elec-trolyte. a) Cyclic voltammetry (CV) of G-7, G-I1*-1, G-I1*-5,G-I1*-7, and G-CTAB supercapacitors at a scan rate of 100 mV s−1,respectively. The rectangular CV shape indicates an efficient double-layer formation. b) and c) CV profiles of a G-I1*-7 electrochemicalcapacitor in aqueous H2SO4/PVA electrolyte at different scan rates.d) The dependency of the capacitative current (extracted from the CVprofiles at 0.5 V for the charge and discharge curves) on the appliedscan rate. A linear relationship is observed with R2= 0.997 and 0.994for the charge and discharge curves, respectively. . . . . . . . . . . . . 138

5.4 Evaluation of LPEG supercapacitors in aqueous H2SO4/PVA elec-trolyte. a) Galvanostatic charging-discharging curves (CDC) of G-7,G-I1*-1, G-I1*-5, G-I1*-7, and G-CTAB supercapacitors measuredat 5 mA, corresponding to a high current density of ∼8 A g−1, re-spectively. b) The gravimetric capacitance of G-7, G-I1*-1, G-I1*-5,G-I1*-7, and G-CTAB supercapacitors calculated from CDC at dif-ferent charge/discharge current densities. c) The volumetric stackcapacitances of the LPEG supercapacitors obtained from G-7, G-I1*-1, G-I1*-5, G-I1*-7, and G-CTAB calculated from the galvanostaticcurves as a function of the charge/discharge current density, respec-tively. The volumetric capacitances are presented on the basis of theentire device stack volume without packaging, including the volumeof the graphene electrodes, the interspace between the electrodes, theAu current collectors, and the separators. . . . . . . . . . . . . . . . . 139

5.5 Schematic representation of noncovalently functionalized grapheneintegrated into LPEG supercapacitors. The adsorbed surfactantmolecules (7) prevent the graphene sheets from restacking leading toa large ion-accessible surface area. The graphene sheets are randomlyoriented in a turbostratically stacked geometry. . . . . . . . . . . . . 141

5.6 Nyquist plot for G-7, G-I1*-1, G-I1*-5, G-I1*-7, and G-CTABsupercapacitors. The inset expands the high-frequency region. . . . . 143

xxiv

LIST OF FIGURES

5.7 Ragone plot comparing a) the gravimetric and b) the volumetric en-ergy and power density of LPEG supercapacitors with supercapacitorsbuilt from a number of commercially available graphene materials, re-spectively. For comparison laser-scribed graphene (LSG) was included- see Ref. [274]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144

5.8 Constant current charging-discharging cycling at a constant current of5 mA (∼10 A g−1) for a G-I1*-7-supercapacitor device. 10,000 cycleswere performed without any rest time and with 30 min rest time afterevery 999 cycles, respectively. . . . . . . . . . . . . . . . . . . . . . . 146

5.9 a) Cyclic voltammetry and b) galvanostatic charging-dischargingcurves of the flexible all-solid-state device, respectively. The de-vice was made by assembling two G-I1*-7 electrodes with a smallamount of gelled electrolyte in between, which also acted as a separator.147

5.10 Comparison of a) the gravimetric and b) the volumetric capacitance ofa G-I1*-7 electrode in liquid and solid-state H2SO4/PVA electrolyte,respectively. c) Nyquist plots of the G-I1*-7 electrode in liquid andsolid-state H2SO4/PVA electrolyte. The inset shows the magnifiedhigh-frequency regions of the Nyquist curves. . . . . . . . . . . . . . . 149

5.11 a) A schematic presentation of the all-solid-state LPEG supercapacitor,which illustrates that the polymer gel electrolyte can serve as boththe electrolyte and the separator. A photograph demonstrating theflexibility of the device. b) Cyclic voltammetry curves of the flexibleall-solid-state supercapacitor for the flat device (0◦) and for a largebending angle of 150◦ at 100 mV s−1. c) Capacitance change of theLPEG electrode as a function of the bending radius. d) Cyclingstability of the all-solid-state device under bending state (150◦) at aconstant current of 5 mA. . . . . . . . . . . . . . . . . . . . . . . . . 150

5.12 a) Galvanostatic charging-discharging curves at 5 mA and b) cyclicvoltammetry curves at 100 mV s−1 of a single all-solid-state superca-pacitor (black) and four supercapacitors in series (red), respectively.Inset: Photograph of a red LED powered by four supercapacitors inseries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151

5.13 Schematic representation of a LPEG-based supercapacitor device usedfor tests with organic electrolyte (left). Optical image of an industry-grade coin cell LPEG supercapacitor device used in this study (right). 151

xxv

LIST OF FIGURES

5.14 Evaluation of LPEG supercapacitors using an organic electrolyte of 1.0M tetraethylammonium tetrafluoroborate in acetonitrile (TEABF4/ACN).a) Cyclic voltammetry and b) galvanostatic charging-dischargingcurves of the device. c) The Nyquist plot of a G-I1*-7 electrodein TEABF4/ACN organic electrolyte. The inset shows the magnifiedhigh-frequency regions of the Nyquist plot. d) Cycling stability of thedevice at a constant current of 5 mA. . . . . . . . . . . . . . . . . . . 152

5.15 Electrochemical characterization of an LPEG supercapacitor in 1-butyl-3-methylimidazolium bis(trifluoromethylsulfonyl)imide ([BMIM][NTf2])ionic liquid electrolyte. a) Cyclic voltammetry curves at different scanrates. b) Galvanostatic charging-discharging curves at different cur-rent densities. c) Nyquist plot of the LPEG supercapacitor. The insetshows the close-up view of the high-frequency regime. . . . . . . . . . 155

5.16 Ragone plot comparing the energy and power density of LPEG su-percapacitors in different electrolytes with a number of commerciallyavailable energy storage devices. Data for the activated carbon super-capacitor (AC-EC), the lithium thin-film battery, and the aluminumelectrolytic capacitor are taken from Ref. [274]. . . . . . . . . . . . . 159

6.1 Structures of aromatic amphiphiles, which were employed for thenoncovalent stabilization of graphene in aqueous medium within thisstudy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163

6.2 Ragone plot comparing the energy and power density of LPEG super-capacitors, made from G-I1*-7 dispersions, with a number of com-mercially available energy storage devices. Data for the activatedcarbon supercapacitor (AC-EC), the lithium thin-film battery, andthe aluminum electrolytic capacitor are taken from Ref. [274]. . . . . 167

8.1 Schematic representation of the experimental setup used for the liquid-phase exfoliation of graphite by ultrasonication. . . . . . . . . . . . . 178

9.1 Reaction mechanism of the free-radical polymerization of vinylbenzyl-triethylammonium chloride. . . . . . . . . . . . . . . . . . . . . . . . 182

xxvi

LIST OF FIGURES

10.1 A cross-sectional SEM image of an LPEG film drop cast on PETsubstrate featuring small gaps between the turbostratically aggre-gated graphene sheets. In order to adjust for the film roughness andinhomogeneity a film thickness of 2 μm was taken for volumetriccalculations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

10.2 Schematic representation of the supercapacitor device showing thepouch cell assembly (right) and the dimensions of the separator,current collector, and LPEG films (left). . . . . . . . . . . . . . . . . 185

10.3 Schematic representation showing the coin cell assembly (right) andthe dimensions of the separator, current collector, and LPEG film,respectively (left). The picture of the coin cell assembly (CR2032)was reprinted from Ref. [411]. . . . . . . . . . . . . . . . . . . . . . . 186

A.1 Scanning electron image of the graphite starting material GFG 350. . lviiiA.2 Raman spectrum of the graphitic starting material GFG 350. The

sample was put onto a Si/SiO2 wafer and excited at 532 nm. Inset:Fitting of the 2D-band. Black dots - experimental data; blue curve -fitting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . lix

A.3 Electrophoretic mobility distribution of a) G-1 and b) G-7 dispersionsafter centrifugation, respectively. The zeta potential was calculatedaccording to Chapter 7. The electrical field was 16.2 V cm−1. . . . . . lix

A.4 Electrophoretic mobility distribution of a) G-1*-1, b) G-NaOH*-1,c) G-I1-1, d) G-I1*-1, and e) G-I2*-1 dispersions after centrifugation,respectively. The zeta potential was calculated according to Chapter7. The electrical field was 16.2 V cm−1. . . . . . . . . . . . . . . . . . lx

A.5 Electrophoretic mobility distribution of G-VBTA*-1 dispersions af-ter centrifugation. The zeta potential was calculated according toChapter 7. The electrical field was 16.2 V cm−1. . . . . . . . . . . . . lxi

A.6 XRD spectra of raw graphite and pretreated graphite in tetraethy-lammonium hydroxide (I1). . . . . . . . . . . . . . . . . . . . . . . . lxi

A.7 Histogram resulting from the Raman mapping of G-iPVBTA, showingcounts versus 2D-FWHM and the corresponding log-normal distribu-tion, respectively. The sample was dip coated from dispersion onto aSi/SiO2 wafer and excited at 532 nm. . . . . . . . . . . . . . . . . . . lxii

xxvii

LIST OF TABLES

A.8 Histogram resulting from the Raman mapping of G-DMP, showingcounts versus 2D-FWHM and the corresponding log-normal distribu-tion, respectively. The sample was dip coated from dispersion onto aSi/SiO2 wafer and excited at 532 nm. . . . . . . . . . . . . . . . . . . lxii

A.9 Histogram resulting from the Raman mapping of G-EG, showingcounts versus 2D-FWHM and the corresponding log-normal distribu-tion, respectively. The sample was dip coated from dispersion onto aSi/SiO2 wafer and excited at 532 nm. . . . . . . . . . . . . . . . . . . lxiii

A.10 Histogram resulting from the Raman mapping of G-NMP, showingcounts versus 2D-FWHM and the corresponding log-normal distribu-tion, respectively. The sample was dip coated from dispersion onto aSi/SiO2 wafer and excited at 532 nm. . . . . . . . . . . . . . . . . . . lxiii

A.11 Surface tension of water-alcohol mixture at various weight fraction at25 ◦C. Tested solvents: a) acetone/water, b) isopropyl alcohol/water,and c) ethanol/water, respectively. . . . . . . . . . . . . . . . . . . . lxiv

A.12 AFM image of a SiO2 which was dip coated in a solution of 1 followedby rinsing with millipore water and absolute ethanol. . . . . . . . . . lxvi

A.13 Cyclic voltammetry profile of an G-I1*-7 supercapacitor, a 7 super-capacitor without any graphitic active material, and a symmetricallyassembled supercapacitor from uncoated Au current collector foils at100 mV s−1, respectively. . . . . . . . . . . . . . . . . . . . . . . . . . lxvi

xxviii

List of Tables

3.1 Comparison of different carbon electrode materials for supercapacitorsincluding carbon nanotubes (CNTs), graphene, activated carbon (AC),and templated carbon. Gravimetric and volumetric capacitances werereported for aqueous electrolyte, respectively. Data is taken from Ref.[225] and [246]. Pictures are adapted from Ref. [247]. . . . . . . . . . 45

4.1 Experimental conditions and results reported in previous studies onliquid-phase exfoliation of graphite by either bath sonicator (BS) ortip sonicator (TS). The surfactants are abbreviated as P123 (PluronicP-123) and SC (sodium cholate). The graphene concentration aftercentrifugation is abbreviated as cac. . . . . . . . . . . . . . . . . . . . 57

4.2 Concentration after centrifugation cac, standard deviation (SD), andID/IG ratio for different centrifugation conditions applied to a G-1dispersion. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

4.3 Optimized parameters for the ultrasonication-assisted liquid-phaseexfoliation of graphite. . . . . . . . . . . . . . . . . . . . . . . . . . . 63

4.4 Comparison of the precipitation (%, after 4 weeks) of the graphenedispersions prepared by either pretreatment with I1 or no pretreatmentfor different surfactants, respectively. . . . . . . . . . . . . . . . . . . 98

xxix

References

[1] T. Duong, 2002 Annual Progress Report for Energy Storage Research andDevelopment, FreedomCar & Vehicle Technologies Program, 2003.

[2] International Energy Agency, Key World Energy Statistics 2015,http://http://www.iea.org/publications/freepublications/publication/KeyWorld_Statistics_2015.pdf, accessed: 2016-05-15.

[3] BP Statistical Review of World Energy June 2015, http://bp.com/statisticalreview, accessed: 2016-05-15.

[4] T. M. Lechter, Climate Change: Observed Impacts on Planet Earth, Elsevier,Amsterdam, NL, 2016.

[5] E. J. Moniz, J. Garcıa-Martınez, Nanotechnology for the Energy Challenge,Wiley-VCH Verlag GmbH, Weihnheim, D, 2013.

[6] R. Ciriminna, N. Zhang, M.-Q. Yang, F. Meneguzzo, Y.-J. Xu, M. Pagliaro,Chemical Communications 2015, 51, 7090–7095.

[7] D. Wei, J. Kivioja, Nanoscale 2013, 5, 10108–10126.

[8] F. Beguin, E. Frackowiak, Carbons for Electrochemical Energy Storage andConversion Systems, CRC Press Taylor & Francis Group, Boca Raton, USA,2010.

[9] L. Dai, D. W. Chang, J.-B. Baek, W. Lu, Small 2012, 8, 1130–1166.

[10] X. Huang, Z. Yin, S. Wu, X. Qi, Q. He, Q. Zhang, Q. Yan, F. Boey, H. Zhang,Small 2011, 7, 1876–1902.

[11] D. A. Brownson, D. K. Kampouris, C. E. Banks, Journal of Power Sources2011, 196, 4873–4885.

[12] A. Geim, K. Novoselov, Nature Materials 2007, 6, 183–191.

[13] A. Ambrosi, C. K. Chua, A. Bonanni, M. Pumera, Chemical Reviews 2014,114, 7150–7188.

xxx

REFERENCES

[14] K. Chen, S. Song, F. Liu, D. Xue, Chemical Society Reviews 2015, 44, 6230–6257.

[15] J. Xia, F. Chen, J. Li, N. Tao, Nature Nanotechnology 2009, 4, 505–509.

[16] S. Bae, H. Kim, Y. Lee, X. Xu, J.-S. Park, Y. Zheng, J. Balakrishnan, T. Lei,H. R. Kim, Y. I. Song, K. S. Kim, Young-Jin ad Kim, B. Ozyilmaz, J.-H. Hn,H. B. Hong, S. Iijima, Nature Nanotechnology 2010, 5, 574–578.

[17] X. Wang, L. Zhi, K. Mullen, Nano Letters 2008, 8, 323–327.

[18] K. S. Novoselov, A. K. Geim, S. Morozov, D. Jiang, Y. Zhang, S. Dubonos, ,I. Grigorieva, A. Firsov, Science 2004, 306, 666–669.

[19] A. Reina, X. Jia, J. Ho, D. Nezich, H. Son, V. Bulovic, M. S. Dresselhaus,J. Kong, Nano Letters 2008, 9, 30–35.

[20] S. Stankovich, D. A. Dikin, R. D. Piner, K. A. Kohlhaas, A. Kleinhammes,Y. Jia, Y. Wu, S. T. Nguyen, R. S. Ruoff, Carbon 2007, 45, 1558–1565.

[21] Y. Hernandez, V. Nicolosi, M. Lotya, F. Blighe, Z. Sun, S. De, I. McGovern,B. Holland, M. Byrne, Y. Gunko, J. Boland, P. Niraj, G. Duesberg, S. Krishna-murti, R. Goodhue, J. Hutchison, V. Scardaci, A. Ferrari, J. Coleman, NatureNanotechnology 2008, 3, 563–568.

[22] F. Bonaccorso, A. Lombardo, T. Hasan, Z. Sun, L. Colombo, A. C. Ferrari,Materials Today 2012, 15, 564–589.

[23] J. H. Warner, F. Schaffel, M. H. Rummeli, A. Bachmatiuk, Graphene: funda-mentals and emergent applications, Elsevier, Waltham, USA, 2013.

[24] W. de Heer, C. Berger, X. Wu, X. Wu, P. First, E. Conrad, H. Li, T. Li,M. Sprinkle, J. Hass, M. Sadowski, M. Potemski, G. Martinez, Solid StateCommunications 2007, 143, 92–100.

[25] K. Novoselov, A. Geim, S. Morozov, D. Jiang, Y. Zhang, S. Dubonos, I. Grig-orieva, A. Firsov, Science 2004, 306, 666.

[26] H. Alcalde, J. de la Fuente, B. Kamp, A. Zurutuza, Proceedings of the IEEE2013, 101, 1793–1800.

xxxi

REFERENCES

[27] P. Shapira, J. Youtie, S. Arora, Journal of Nanoparticle Research 2012, 14,1–15.

[28] K. S. Novoselov, V. Fal’ko, L. Colombo, P. Gellert, M. Schwab, K. Kim, Nature2012, 490, 192–200.

[29] K. Novoselov, D. Jiang, F. Schedin, T. Booth, V. Khotkevich, S. Morozov,A. Geim, Proceedings of the National Academy of Sciences of the United Statesof America 2005, 102, 10451–10453.

[30] S. Morozov, K. Novoselov, M. Katsnelson, F. Schedin, D. Elias, J. Jaszczak,A. Geim, Physical Review Letters 2008, 100, 016602.

[31] A. Castro Neto, F. Guinea, N. Peres, K. Novoselov, A. Geim, Reviews ofModern Physics 2009, 81, 109–162.

[32] Y. Zhang, Y.-W. Tan, H. L. Stormer, P. Kim, Nature 2005, 438, 201–204.

[33] S. Morozov, K. Novoselov, F. Schedin, D. Jiang, A. Firsov, A. Geim, PhysicalReview B 2005, 72, 201401.

[34] B. Partoens, F. Peeters, Physical Review B 2006, 74, 075404.

[35] B. Warren, The Journal of Chemical Physics 1934, 2, 551–555.

[36] J. K. Wassei, M. Mecklenburg, J. A. Torres, J. D. Fowler, B. Regan, R. B.Kaner, B. H. Weiller, Small 2012, 8, 1415–1422.

[37] W. Bao, L. Jing, J. Velasco, Y. Lee, G. Liu, D. Tran, B. Standley, M. Aykol,S. Cronin, D. Smirnov, Nature Physics 2011, 7, 948–952.

[38] K. F. Mak, J. Shan, T. F. Heinz, Physical Review Letters 2010, 104, 176404.

[39] J. Hass, F. Varchon, J.-E. Millan-Otoya, M. Sprinkle, N. Sharma, W. A.de Heer, C. Berger, P. N. First, L. Magaud, E. H. Conrad, Physical ReviewLetters 2008, 100, 125504.

[40] S. Latil, V. Meunier, L. Henrard, Physical Review B 2007, 76, 201402.

[41] J. H. Warner, M. H. Rummeli, T. Gemming, B. Buchner, G. A. D. Briggs,Nano Letters 2008, 9, 102–106.

xxxii

REFERENCES

[42] G.-H. Lee, R. C. Cooper, S. J. An, S. Lee, A. van der Zande, N. Petrone, A. G.Hammerberg, C. Lee, B. Crawford, W. Oliver, J. W. Kysar, J. Hone, Science2013, 340, 1073–1076.

[43] C. Lee, X. Wei, J. W. Kysar, J. Hone, Science 2008, 321, 385–388.

[44] R. S. Edwards, K. S. Coleman, Nanoscale 2013, 5, 38–51.

[45] A. A. Balandin, S. Ghosh, W. Bao, I. Calizo, D. Teweldebrhan, F. Miao, C. N.Lau, Nano Letters 2008, 8, 902–907.

[46] R. Nair, P. Blake, A. Grigorenko, K. Novoselov, T. Booth, T. Stauber, N. Peres,A. Geim, Science 2008, 320, 1308.

[47] M. Stoller, S. Park, Y. Zhu, J. An, R. Ruoff, Nano Letters 2008, 8, 3498–3502.

[48] A. C. Ferrari, F. Bonaccorso, V. Fal’Ko, K. S. Novoselov, S. Roche, P. Bøggild,S. Borini, F. H. Koppens, V. Palermo, N. Pugno, J. A. Garrido, R. Sordan,A. Bianco, L. Ballerini, M. Prato, E. Lidorikis, Nanoscale 2015, 7, 4598–4810.

[49] M. Pumera, Energy & Environmental Science 2011, 4, 668–674.

[50] F. Bonaccorso, L. Colombo, G. Yu, M. Stoller, V. Tozzini, A. C. Ferrari, R. S.Ruoff, V. Pellegrini, Science 2015, 347, 1246501.

[51] Graphene-based Optical Electronics, http://grapheneus.com/wp-content/uploads/2013/09/Graphene-electronics.jpg, accessed: 2016-03-20.

[52] RFID Tag, https://www.sciencedaily.com/releases/2010/03/100318113300.htm, accessed: 2016-03-20.

[53] Tennis Racket, http://www.tennis-warehouse.com/Reviews/HGRMP/HGRMPReview.html, accessed: 2016-03-20.

[54] S.-M. Paek, E. Yoo, I. Honma, Nano Letters 2008, 9, 72–75.

[55] Y. Zhu, S. Murali, M. D. Stoller, K. Ganesh, W. Cai, P. J. Ferreira, A. Pirkle,R. M. Wallace, K. A. Cychosz, M. Thommes, D. Su, E. A. Stach, R. S. Ruoff,Science 2011, 332, 1537–1541.

[56] Y. Zhu, M. D. Stoller, W. Cai, A. Velamakanni, R. D. Piner, D. Chen, R. S.Ruoff, ACS Nano 2010, 4, 1227–1233.

xxxiii

REFERENCES

[57] M. F. El-Kady, R. B. Kaner, Nature Communications 2013, 4, 1475.

[58] C. Merlet, B. Rotenberg, P. A. Madden, P.-L. Taberna, P. Simon, Y. Gogotsi,M. Salanne, Nature Materials 2012, 11, 306–310.

[59] C. Du, N. Pan, Journal of Power Sources 2006, 160, 1487–1494.

[60] M. El-Kady, Graphene Supercapacitors: Charging Up the Future, Dissertation,2013.

[61] G. E. Moore, Electron Devices Meeting 1975, 21, 11–13.

[62] J. Cai, P. Ruffieux, R. Jaafar, M. Bieri, T. Braun, S. Blankenburg, M. Muoth,A. P. Seitsonen, M. Saleh, X. Feng, K. Mullen, R. Fasel, Nature 2010, 466,470–473.

[63] M. Y. Han, B. Ozyilmaz, Y. Zhang, P. Kim, Physical Review Letters 2007, 98,206805.

[64] J. Bai, X. Zhong, S. Jiang, Y. Huang, X. Duan, Nature Nanotechnology 2010,5, 190–194.

[65] T. Ohta, A. Bostwick, T. Seyller, K. Horn, E. Rotenberg, Science 2006, 313,951–954.

[66] D. Elias, R. Nair, T. Mohiuddin, S. Morozov, P. Blake, M. Halsall, A. Ferrari,D. Boukhvalov, M. Katsnelson, A. Geim, K. S. Novoselov, Science 2009, 323,610–613.

[67] R. R. Nair, W. Ren, R. Jalil, I. Riaz, V. G. Kravets, L. Britnell, P. Blake,F. Schedin, A. S. Mayorov, S. Yuan, M. I. Katsnelson, C. Hui-Ming, W. Srupin-ski, L. G. Bulusheva, A. V. Okotrub, I. V. Grigorieva, A. N. Grigorenko, K. S.Novoselov, A. K. Geim, Small 2010, 6, 2877–2884.

[68] F. Schwierz, Nature Nanotechnology 2010, 5, 487–496.

[69] M. Segal, Nature Nanotechnology 2009, 4, 612–614.

[70] J. Li, F. Ye, S. Vaziri, M. Muhammed, M. C. Lemme, M. Ostling, AdvancedMaterials 2013, 25, 3985–3992.

xxxiv

REFERENCES

[71] Vorbeck Materials, Conductive Inks, http://vorbeck.com/pages/products-conductiveinks, accessed: 2016-03-20.

[72] XG Science, Graphene-based Inks and Coatings, http://xgsciences.com/products/inks-and-coatings/, accessed: 2016-03-20.

[73] XG Science, Conductive Dispersion, https://graphene-supermarket.com/Conductive-Graphene-Dispersion-100.html, accessed: 2016-03-05.

[74] A. P. Singh, M. Mishra, A. Chandra, S. Dhawan, Nanotechnology 2011, 22,465701.

[75] R. J. Young, I. A. Kinloch, L. Gong, K. S. Novoselov, Composites Science andTechnology 2012, 72, 1459–1476.

[76] L. S. Walker, V. R. Marotto, M. A. Rafiee, N. Koratkar, E. L. Corral, ACSNano 2011, 5, 3182–3190.

[77] C. Xu, X. Wang, J. Zhu, The Journal of Physical Chemistry C 2008, 112,19841–19845.

[78] S. Vadukumpully, J. Paul, N. Mahanta, S. Valiyaveettil, Carbon 2011, 49,198–205.

[79] F. He, S. Lau, H. L. Chan, J. Fan, Advanced Materials 2009, 21, 710–715.

[80] C.-C. Teng, C.-C. M. Ma, C.-H. Lu, S.-Y. Yang, S.-H. Lee, M.-C. Hsiao, M.-Y.Yen, K.-C. Chiou, T.-M. Lee, Carbon 2011, 49, 5107–5116.

[81] R. Verdejo, M. M. Bernal, L. J. Romasanta, M. A. Lopez-Manchado, Journalof Materials Chemistry 2011, 21, 3301–3310.

[82] Graphene Composites: Introduction and Market Status, http://www.graphene-info.com/graphene-composites, accessed: 2016-03-05.

[83] Head, Graphenext, http://www.head.com/de/sports/tennis/technology/graphene-xt/, accessed: 2016-03-20.

[84] Vittoria, G+, https://www.vittoria.com/wheels-tech/, accessed: 2016-03-20.

xxxv

REFERENCES

[85] K. V. Emtsev, A. Bostwick, K. Horn, J. Jobst, G. L. Kellogg, L. Ley, J. L.McChesney, T. Ohta, S. A. Reshanov, J. Rohrl, Nature Materials 2009, 8,203–207.

[86] A. Balan, R. Kumar, M. Boukhicha, O. Beyssac, J.-C. Bouillard, D. Taverna,W. Sacks, M. Marangolo, E. Lacaze, R. Gohler, W. Escoffier, J.-M. Poumirol,A. Shukla, Journal of Physics D: Applied Physics 2010, 43, 374013.

[87] D. V. Kosynkin, A. L. Higginbotham, A. Sinitskii, J. R. Lomeda, A. Dimiev,B. K. Price, J. M. Tour, Nature 2009, 458, 872–876.

[88] X. Yang, X. Dou, A. Rouhanipour, L. Zhi, H. J. Rader, K. Mullen, Journal ofthe American Chemical Society 2008, 130, 4216–4217.

[89] S. Stankovich, D. Dikin, R. Piner, K. Kohlhaas, A. Kleinhammes, Y. Jia,Y. Wu, S. Nguyen, R. Ruoff, Carbon 2007, 45, 1558–1565.

[90] C. Valles, C. Drummond, H. Saadaoui, C. Furtado, M. He, O. Roubeau,L. Ortolani, M. Monthioux, A. Penicaud, Journal of the American ChemicalSociety 2008, 130, 15802–15804.

[91] J. M. Englert, C. Dotzer, G. Yang, M. Schmid, C. Papp, J. M. Gottfried,H.-P. Steinruck, E. Spiecker, F. Hauke, A. Hirsch, Nature Chemistry 2011, 3,279–286.

[92] C. Knieke, A. Berger, M. Voigt, R. N. K. Taylor, J. Rohrl, W. Peukert, Carbon2010, 48, 3196–3204.

[93] V. Leon, M. Quintana, M. A. Herrero, J. L. Fierro, A. de la Hoz, M. Prato,E. Vazquez, Chemical Communications 2011, 47, 10936–10938.

[94] W. Zhao, F. Wu, H. Wu, G. Chen, Journal of Nanomaterials 2010, 2010, 6.

[95] S. Tkachev, E. Buslaeva, S. Gubin, Inorganic Materials 2011, 47, 1–10.

[96] S. Stankovich, D. Dikin, G. Dommett, K. Kohlhaas, E. Zimney, E. Stach,R. Piner, S. Nguyen, R. Ruoff, Nature 2006, 442, 282–286.

[97] B. Brodie, Annales des Chimie et des Physique 1860, 59, 466–472.

[98] L. Staudenmaier, Berichte der deutschen chemischen Gesellschaft 1898, 31,1481–1487.

xxxvi

REFERENCES

[99] W. Hummers Jr, R. Offeman, Journal of the American Chemical Society 1958,80, 1339–1339.

[100] Y.-J. Park, S. Y. Park, I. In, Journal of Industrial and Engineering Chemistry2011, 17, 298–303.

[101] J. Paredes, S. Villar-Rodil, A. Martinez-Alonso, J. Tascon, Langmuir 2008,24, 10560–10564.

[102] J. Lomeda, C. Doyle, D. Kosynkin, W. Hwang, J. Tour, Journal of the AmericanChemical Society 2008, 130, 16201–16206.

[103] V. Tung, M. Allen, Y. Yang, R. Kaner, Nature Nanotechnology 2008, 4, 25–29.

[104] G. Wang, J. Yang, J. Park, X. Gou, B. Wang, H. Liu, J. Yao, The Journal ofPhysical Chemistry C 2008, 112, 8192–8195.

[105] L. Guardia, M. Fernandez-Merino, J. Paredes, P. Solıs-Fernandez, S. Villar-Rodil, A. Martinez-Alonso, J. Tascon, Carbon 2010, 49, 1653–1662.

[106] M. Lotya, Y. Hernandez, P. King, R. Smith, V. Nicolosi, L. Karlsson, F. Blighe,S. De, Z. Wang, I. McGovern, G. Duesberg, J. Coleman, Journal of theAmerican Chemical Society 2009, 131, 3611–3620.

[107] A. Bourlinos, V. Georgakilas, R. Zboril, T. Steriotis, A. Stubos, C. Trapalis,Solid State Communications 2009, 149, 2172–2176.

[108] A. A. Green, M. C. Hersam, Nano Letters 2009, 9, 4031–4036.

[109] X. An, T. Simmons, R. Shah, C. Wolfe, K. Lewis, M. Washington, S. Nayak,S. Talapatra, S. Kar, Nano Letters 2010, 10 .

[110] S. Park, R. S. Ruoff, Nature Nanotechnology 2009, 4, 217–224.

[111] X. Cui, C. Zhang, R. Hao, Y. Hou, Nanoscale 2011, 3, 2118–2126.

[112] A. Ciesielski, P. Samorı, Chemical Society Reviews 2014, 43, 381–398.

[113] G. Cravotto, P. Cintas, Chemistry - A European Journal 2010, 16, 5246–5259.

[114] S. E. Skrabalak, Physical Chemistry Chemical Physics 2009, 11, 4930–4942.

xxxvii

REFERENCES

[115] H. Xu, K. S. Suslick, Journal of the American Chemical Society 2011, 133,9148–9151.

[116] A. O’Neill, U. Khan, P. Nirmalraj, J. Boland, J. Coleman, The Journal ofPhysical Chemistry C 2011, 115, 5422–5428.

[117] C. Backes, F. Hauke, A. Hirsch, Advanced Materials 2011, 23, 2588–2601.

[118] C.-J. Shih, G. L. Paulus, Q. H. Wang, Z. Jin, D. Blankschtein, M. S. Strano,Langmuir 2012, 28, 8579–8586.

[119] L. Guardia, M. Fernandez-Merino, J. Paredes, P. Solis-Fernandez, S. Villar-Rodil, A. Martinez-Alonso, J. Tascon, Carbon 2011, 49, 1653–1662.

[120] R. Smith, M. Lotya, J. Coleman, New Journal of Physics 2010, 12, 125008.

[121] J. H. Lee, D. W. Shin, V. G. Makotchenko, A. S. Nazarov, V. E. Fedorov, J. H.Yoo, S. M. Yu, J.-Y. Choi, J. M. Kim, J.-B. Yoo, Small 2010, 6, 58–62.

[122] M. Buzaglo, M. Shtein, S. Kober, R. Lovrincic, A. Vilan, O. Regev, PhysicalChemistry Chemical Physics 2013, 15, 4428–4435.

[123] M. Lotya, P. King, U. Khan, S. De, J. Coleman, ACS Nano 2010, 4, 3155–3162.

[124] S. De, P. J. King, M. Lotya, A. O’Neill, E. M. Doherty, Y. Hernandez, G. S.Duesberg, J. N. Coleman, Small 2010, 6, 458–464.

[125] R. J. Smith, P. J. King, C. Wirtz, G. S. Duesberg, J. N. Coleman, ChemicalPhysics Letters 2012, 531, 169–172.

[126] Z. Sun, J. Masa, Z. Liu, W. Schuhmann, M. Muhler, Chemistry-A EuropeanJournal 2012, 18, 6972–6978.

[127] S. Vadukumpully, J. Paul, S. Valiyaveettil, Carbon 2009, 47, 3288–3294.

[128] A. Griffith, S. M. Notley, Journal of Colloid and Interface Science 2012, 369,210–215.

[129] S. M. Notley, Journal of Colloid and Interface Science 2012, 375, 35–40.

[130] J.-W. T. Seo, A. A. Green, A. L. Antaris, M. C. Hersam, The Journal ofPhysical Chemistry Letters 2011, 2, 1004–1008.

xxxviii

REFERENCES

[131] S.-Z. Zu, B.-H. Han, The Journal of Physical Chemistry C 2009, 113, 13651–13657.

[132] S. M. Notley, Langmuir 2012, 28, 14110–14113.

[133] L. Xu, J.-W. McGraw, F. Gao, M. Grundy, Z. Ye, Z. Gu, J. L. Shepherd, TheJournal of Physical Chemistry C 2013, 117, 10730–10742.

[134] R. J. Smith, M. Lotya, J. N. Coleman, New Journal of Physics 2010, 12,125008.

[135] Y. T. Liang, M. C. Hersam, Journal of the American Chemical Society 2010,132, 17661–17663.

[136] T. J. Nacken, C. Damm, H. Xing, A. Ruger, W. Peukert, Nano Research 2015,8, 1865–1881.

[137] A. Schneider, Liquid phase production and characterization of few-layergraphene, Master Thesis, 2011.

[138] M. Zhang, R. Parajuli, D. Mastrogiovanni, B. Dai, P. Lo, W. Cheung, R. Brukh,P. Chiu, T. Zhou, Z. Liu, E. Garfunkel, H. He, Small 2010, 6, 1100–1107.

[139] Y. Xu, H. Bai, G. Lu, C. Li, G. Shi, Journal of the American Chemical Society2008, 130, 5856–5857.

[140] S. Ghosh, X. An, R. Shah, D. Rawat, B. Dave, S. Kar, S. Talapatra, TheJournal of Physical Chemistry C 2012, 116, 20688–20693.

[141] D. Parviz, S. Das, H. T. Ahmed, F. Irin, S. Bhattacharia, M. J. Green, ACSNano 2012, 6, 8857–8867.

[142] H. Yang, Y. Hernandez, A. Schlierf, A. Felten, A. Eckmann, S. Johal, P. Louette,J.-J. Pireaux, X. Feng, K. Mullen, V. Palermo, C. Casiraghi, Carbon 2013, 53,357–365.

[143] G. Katsukis, J. Malig, C. Schulz-Drost, S. Leubner, N. Jux, D. M. Guldi, ACSNano 2012, 6, 1915–1924.

[144] A. Schlierf, H. Yang, E. Gebremedhn, E. Treossi, L. Ortolani, L. Chen, A. Mi-noia, V. Morandi, P. Samori, C. Casiraghi, Nanoscale 2013, 5, 4205–4216.

xxxix

REFERENCES

[145] G. Katsukis, J. Malig, C. Schulz-Drost, S. Leubner, N. Jux, D. M. Guldi, ACSNano 2012, 6, 1915–1924.

[146] M. Cardinali, L. Valentini, J. M. Kenny, The Journal of Physical Chemistry C2011, 115, 16652–16656.

[147] D.-W. Lee, T. Kim, M. Lee, Chemical Communications 2011, 47, 8259–8261.

[148] X. Zheng, Q. Xu, J. Li, L. Li, J. Wei, RSC Advances 2012, 2, 10632–10638.

[149] N. Kozhemyakina, J. Englert, G. Yang, E. Spiecker, C. Schmidt, F. Hauke,A. Hirsch, Advanced Materials 2010, 22, 5483–5487.

[150] H. J. Salavagione, G. Ellis, J. L. Segura, R. Gomez, G. M. Morales, G. Martınez,Journal of Materials Chemistry 2011, 21, 16129–16135.

[151] J. Englert, J. Rohrl, C. Schmidt, R. Graupner, M. Hundhausen, F. Hauke,A. Hirsch, Advanced Materials 2009, 21, 4265–4269.

[152] R. Hao, W. Qian, L. Zhang, Y. Hou, Chemical Communications 2008, 6576–6578.

[153] A. Ghosh, K. V. Rao, S. J. George, C. Rao, Chemistry - A European Journal2010, 16, 2700–2704.

[154] S. Bose, T. Kuila, A. K. Mishra, N. H. Kim, J. H. Lee, Nanotechnology 2011,22, 405603.

[155] A. Viinikanoja, J. Kauppila, P. Damlin, E. Makila, J. Leiro, T. Aaritalo,J. Lukkari, Carbon 2014, 68, 195–209.

[156] C. Damm, T. J. Nacken, W. Peukert, Carbon 2015, 81, 284–294.

[157] T. J. Nacken, C. Damm, J. Walter, a. Ruger, W. Peukert, RSC Advances 2015,5, 57328–57338.

[158] K. R. Paton, E. Varrla, C. Backes, R. J. Smith, U. Khan, A. OŠNeill, C. Boland,M. Lotya, O. M. Istrate, P. King, P. Higgins, S. Barwich, P. May, P. Puczkarski,O. Ahmed, M. Moebius, H. Pettersson, E. Long, C. J. ap, S. O’Brien, E. K.McGuire, B. M. Sanchez, G. S. Duesberg, N. McEvoy, T. J. Pennycook, NatureMaterials 2014, 13, 624–630.

xl

REFERENCES

[159] U. Khan, A. O’Neill, M. Lotya, S. De, J. Coleman, Small 2010, 6, 864–871.

[160] P. Blake, E. Hill, A. C. Neto, K. Novoselov, D. Jiang, R. Yang, T. Booth,A. Geim, Applied Physics Letters 2007, 91, 063124.

[161] C. Casiraghi, A. Hartschuh, E. Lidorikis, H. Qian, H. Harutyunyan, T. Gokus,K. Novoselov, A. Ferrari, Nano Letters 2007, 7, 2711–2717.

[162] V. Singh, D. Joung, L. Zhai, S. Das, S. Khondaker, S. Seal, Progress inMaterials Science 2011, 56, 1178–1271.

[163] C. O. Girit, J. C. Meyer, R. Erni, M. D. Rossell, C. Kisielowski, L. Yang, C.-H.Park, M. Crommie, M. L. Cohen, S. G. Louie, A. Zellel, Science 2009, 323,1705–1708.

[164] J. C. Meyer, C. Kisielowski, R. Erni, M. D. Rossell, M. Crommie, A. Zettl,Nano Letters 2008, 8, 3582–3586.

[165] J. C. Meyer, C. O. Girit, M. Crommie, A. Zettl, Nature 2008, 454, 319–322.

[166] U. Dahmen, R. Erni, V. Radmilovic, C. Ksielowski, M.-D. Rossell, P. Denes,Philosophical Transactions of the Royal Society of London A: Mathematical,Physical and Engineering Sciences 2009, 367, 3795–3808.

[167] C. Kisielowski, B. Freitag, M. Bischoff, H. Van Lin, S. Lazar, G. Knippels,P. Tiemeijer, M. van der Stam, S. von Harrach, M. Stekelenburg, M. Haider,S. Uhlemann, H. Muller, P. Hartel, B. Kabius, D. Miller, I. Petrov, E. A.Olson, T. Donchev, E. A. Kenik, A. R. Lupini, J. Bentley, S. J. Pennycook,I. M. Anderson, A. M. Minor, A. K. Schmid, T. Duden, V. Radmilovic, Q. M.Ramasse, M. Watanabe, R. Erni, E. Stach, P. Denes, U. Dahmen, Microscopyand Microanalysis 2008, 14, 469–477.

[168] J. R. Kyle, C. S. Ozkan, M. Ozkan, Nanoscale 2012, 4, 3807–3819.

[169] P. Sutter, J. T. Sadowski, E. Sutter, Physical Review B 2009, 80, 245411.

[170] N. Yao, Z. Wang, Handbook of Microscopy for Nanotechnology, Kluwer Aca-demic Publishers, Boston, USA, 2005.

[171] P. Braga, D. Ricci, Atomic Force Microscopy: Biomedical Methods and Appli-cations, Bd. 242, Humana Press, Totowa, USA, 2004.

xli

REFERENCES

[172] J. Paredes, S. Villar-Rodil, P. Solis-Fernandez, A. Martinez-Alonso, J. Tascon,Langmuir 2009, 25, 5957–5968.

[173] H. C. Schniepp, J.-L. Li, M. J. McAllister, H. Sai, M. Herrera-Alonso, D. H.Adamson, R. K. Prud’homme, R. Car, D. A. Saville, I. A. Aksay, The Journalof Physical Chemistry B 2006, 110, 8535–8539.

[174] C. Valles, C. Drummond, H. Saadaoui, C. A. Furtado, M. He, O. Roubeau,L. Ortolani, M. Monthioux, A. Penicaud, Journal of the American ChemicalSociety 2008, 130, 15802–15804.

[175] Z. Ni, Y. Wang, T. Yu, Z. Shen, Nano Research 2008, 1, 273–291.

[176] C. F. Bohren, D. R. Huffman, Absorption and scattering of light by smallparticles, John Wiley & Sons, Inc., New York, USA, 1983.

[177] A. Jorio, M. S. Dresselhaus, R. Saito, G. Dresselhaus, Raman spectroscopy ingraphene related systems, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim,D, 2011.

[178] P. Yu, M. Cardona, Fundamentals of Semiconductors. Physics and MaterialsProperties, Springer, Heidelberg, D, 2005.

[179] E. Smith, G. Dent, Modern Raman Spectroscopy: a Practical Approach, JohnWiley & Sons Ltd, West Sussex, UK, 2005.

[180] M. S. Amer, Raman Spectroscopy, Fullerenes and Nanotechnology, The RoyalSociety of Chemistry, Cambridge, UK, 2010.

[181] A. Ferrari, J. Robertson, Physical Review B 2001, 64, 075414.

[182] A. C. Ferrari, D. M. Basko, Nature Nanotechnology 2013, 8, 235–246.

[183] M. Dresselhaus, G. Dresselhaus, M. Hofmann, Philosophical Transactions ofthe Royal Society A: Mathematical, Physical and Engineering Sciences 2008,366, 231.

[184] A. Ferrari, Solid State Communications 2007, 143, 47–57.

[185] A. Ferrari, J. Meyer, V. Scardaci, C. Casiraghi, M. Lazzeri, F. Mauri, S. Pis-canec, D. Jiang, K. Novoselov, S. Roth, A. Geim, Physical Review Letters2006, 97, 187401.

xlii

REFERENCES

[186] M. Dresselhaus, A. Jorio, M. Hofmann, G. Dresselhaus, R. Saito, Nano Letters2010, 10, 751–758.

[187] D. Graf, F. Molitor, K. Ensslin, C. Stampfer, A. Jungen, C. Hierold, L. Wirtz,Nano Letters 2007, 7, 238–242.

[188] P. Lespade, A. Marchand, M. Couzi, F. Cruege, Carbon 1984, 22, 375–385.

[189] C. H. Lui, Z. Li, Z. Chen, P. V. Klimov, L. E. Brus, T. F. Heinz, Nano Letters2010, 11, 164–169.

[190] D. S. Lee, C. Riedl, B. Krauss, K. von Klitzing, U. Starke, J. H. Smet, NanoLetters 2008, 8, 4320–4325.

[191] K. N. Kudin, B. Ozbas, H. C. Schniepp, R. K. Prud’Homme, I. A. Aksay,R. Car, Nano Letters 2008, 8, 36–41.

[192] A. Eckmann, A. Felten, A. Mishchenko, L. Britnell, R. Krupke, K. S. Novoselov,C. Casiraghi, Nano Letters 2012, 12, 3925–3930.

[193] M. Winter, R. J. Brodd, Chemical Reviews 2004, 104, 4245–4270.

[194] P. Simon, Y. Gogotsi, Nature Materials 2008, 7, 845–854.

[195] B. Conway, Electrochemical Supercapacitors: Scientific Fundamentals andTechnological Applications, Bd. 8, Kluwer Academic/ Plenum Publisher, NewYork, USA, 1999, S. 3498–3502.

[196] A. Shukla, A. Banerjee, M. Ravikumar, A. Jalajakshi, Electrochimica Acta2012, 84, 165–173.

[197] P. Simon, Y. Gogotsi, B. Dunn, Science 2014, 343, 1210–1211.

[198] F. Beguin, E. Raymundo-Pinero, Nanocarbons for supercapacitors in Batteriesfor Sustainability, Springer Science + Business Media, New York, USA, 2013,S. 393–421.

[199] M. Lu, F. Beguin, E. Frackowiak, Supercapacitors Materials, Systems andApplications, Wiley-VCH Verlag GmbH & Co. KGaA, Weihnheim, D, 2013.

[200] A. Burke, Journal of Power Sources 2000, 91, 37–50.

xliii

REFERENCES

[201] Y. Zhang, H. Feng, X. Wu, L. Wang, A. Zhang, T. Xia, H. Dong, X. Li,L. Zhang, International Journal of Hydrogen Energy 2009, 34, 4889–4899.

[202] X. Dong, W. Shen, J. Gu, L. Xiong, Y. Zhu, H. Li, J. Shi, The Journal ofPhysical Chemistry B 2006, 110, 6015–6019.

[203] M. Vangari, T. Pryor, L. Jiang, Journal of Energy Engineering 2012, 139,72–79.

[204] M. G. Bakker, R. M. Frazier, S. Burkett, J. E. Bara, N. Chopra, S. Spear,S. Pan, C. Xu, Nanomaterials and Energy 2012, 1, 136–158.

[205] G. Shuyan, F. Hao, Chemistry 2013, 3, 003.

[206] H. Helmholtz, Annals of Physics 1879, 29, 337.

[207] S. Senthilkumar, K. V. Sankar, J. Melo, A. Gedanken, R. K. Selvan, AdvancedFunctional Materials 2015, 399–431.

[208] G. Gouy, Journal of Physics 1910, 9, 457–467.

[209] D. L. Chapman, Philosophical Magazine 1913, 25, 475–481.

[210] O. Stern, Zeitschrift fur Elektrochemie 1924, 30, 1014–1020.

[211] X. Zhao, B. M. Sanchez, P. J. Dobson, P. S. Grant, Nanoscale 2011, 3, 839–855.

[212] R. Kotz, M. Carlen, Electrochimica Acta 2000, 45, 2483–2498.

[213] S. Senthilkumar, K. V. Sankar, J. Melo, A. Gedanken, R. K. Selvan, AdvancedFunctional Materials 2015, 399.

[214] A. Pandolfo, A. Hollenkamp, Journal of Power Sources 2006, 157, 11–27.

[215] A. Yu, V. Chabot, J. Zhang, Electrochemical supercapacitors for energy storageand delivery: fundamentals and applications, CRC Press Taylor & FrancisGroups, Boca Raton, USA, 2013.

[216] Y. Maletin, N. Strizhakova, V. Izotov, A. Mironova, S. Kozachkov, V. Danilin,S. Podmogilny, Supercapacitors: old problems and new trends in New Car-bon Based Materials for Electrochemical Energy Storage Systems: Batteries,Supercapacitors and Fuel Cells, Springer, Dordrecht, NL, 2006, S. 51–62.

xliv

REFERENCES

[217] G. Wang, L. Zhang, J. Zhang, Chemical Society Reviews 2012, 41, 797–828.

[218] E. Frackowiak, Physical Chemistry Chemical Physics 2007, 9, 1774–1785.

[219] J. Zheng, J. Huang, T. Jow, Journal of the Electrochemical Society 1997, 144,2026–2031.

[220] J. Zheng, T. Jow, Journal of The Electrochemical Society 1995, 142, L6–L8.

[221] Y. M. Vol’fkovich, T. Serdyuk, Russian Journal of Electrochemistry 2002, 38,935–959.

[222] J. Zhang, X. Zhao, ChemSusChem 2012, 5, 818–841.

[223] M. D. Stoller, R. S. Ruoff, Energy & Environmental Science 2010, 3, 1294–1301.

[224] L. L. Zhang, X. Zhao, Chemical Society Reviews 2009, 38, 2520–2531.

[225] L. L. Zhang, R. Zhou, X. Zhao, Journal of Materials Chemistry 2010, 20,5983–5992.

[226] C. Portet, P. L. Taberna, P. Simon, E. Flahaut, C. Laberty-Robert, Elec-trochimica Acta 2005, 50, 4174–4181.

[227] Y. Gogotsi, P. Simon, Science 2011, 334, 917–918.

[228] M. Toupin, D. Belanger, I. R. Hill, D. Quinn, Journal of Power Sources 2005,140, 203–210.

[229] B. Conway, W. Pell, Journal of Solid State Electrochemistry 2003, 7, 637–644.

[230] P. Azaıs, L. Duclaux, P. Florian, D. Massiot, M.-A. Lillo-Rodenas, A. Linares-Solano, J.-P. Peres, C. Jehoulet, F. Beguin, Journal of Power Sources 2007,171, 1046–1053.

[231] M. Ue, K. Ida, S. Mori, Journal of the Electrochemical Society 1994, 141,2989–2996.

[232] A. Balducci, U. Bardi, S. Caporali, M. Mastragostino, F. Soavi, ElectrochemistryCommunications 2004, 6, 566–570.

[233] Y. Chen, X. Zhang, D. Zhang, P. Yu, Y. Ma, Carbon 2011, 49, 573–580.

xlv

REFERENCES

[234] M. Galinski, A. Lewandowski, I. Stepniak, Electrochimica Acta 2006, 51,5567–5580.

[235] H. Ohno, K. Fukumoto, Electrochemistry 2008, 76, 16–23.

[236] M. Armand, F. Endres, D. R. MacFarlane, H. Ohno, B. Scrosati, NatureMaterials 2009, 8, 621–629.

[237] M. Inagaki, H. Konno, O. Tanaike, Journal of Power Sources 2010, 195,7880–7903.

[238] W. G. Pell, B. E. Conway, Journal of Power Sources 2001, 96, 57–67.

[239] W. Pell, B. Conway, Journal of Electroanalytical Chemistry 2001, 500, 121–133.

[240] E. Kalu, T. Nwoga, V. Srinivasan, J. Weidner, Journal of Power Sources 2001,92, 163–167.

[241] K.-C. Tsay, L. Zhang, J. Zhang, Electrochimica Acta 2012, 60, 428–436.

[242] N. Devillers, S. Jemei, M.-C. Pera, D. Bienaime, F. Gustin, Journal of PowerSources 2014, 246, 596–608.

[243] W. Plieth, Electrochemistry for Materials Science, Elsevier, Oxford, UK, 2008.

[244] R. De Levie, Electrochimica Acta 1963, 8, 751–780.

[245] R. De Levie, Electrochimica Acta 1964, 9, 1231–1245.

[246] U. K. Sur, International Journal of Electrochemistry 2012, 2012 .

[247] P. Simon, Y. Gogotsi, Accounts of Chemical Research 2012, 46, 1094–1103.

[248] S. Nomoto, H. Nakata, K. Yoshioka, A. Yoshida, H. Yoneda, Journal of PowerSources 2001, 97, 807–811.

[249] H. Marsh, F. R. Reinoso, Activated Carbon, Elsevier, Oxford, UK, 2006.

[250] H. Shi, Electrochimica Acta 1996, 41, 1633–1639.

[251] S. Zhao, C.-Y. Wang, M.-M. Chen, J. Wang, Z.-Q. Shi, Journal of Physics andChemistry of Solids 2009, 70, 1256–1260.

xlvi

REFERENCES

[252] Y.-J. Kim, B.-J. Lee, H. Suezaki, T. Chino, Y. Abe, T. Yanagiura, K. C. Park,M. Endo, Carbon 2006, 44, 1592–1595.

[253] J. Mi, X.-R. Wang, R.-J. Fan, W.-H. Qu, W.-C. Li, Energy & Fuels 2012, 26,5321–5329.

[254] M. Biswal, A. Banerjee, M. Deo, S. Ogale, Energy & Environmental Science2013, 6, 1249–1259.

[255] K. Kierzek, E. Frackowiak, G. Lota, G. Gryglewicz, J. Machnikowski, Elec-trochimica Acta 2004, 49, 515–523.

[256] M. J. Allen, V. C. Tung, R. B. Kaner, Chemical Reviews 2009, 110, 132–145.

[257] C. Rao, A. Sood, K. Subrahmanyam, A. Govindaraj, Angewandte ChemieInternational Edition 2009, 48, 7752–7777.

[258] Y. Wang, Y. Wu, Y. Huang, F. Zhang, X. Yang, Y. Ma, Y. Chen, The Journalof Physical Chemistry C 2011, 115, 23192–23197.

[259] X. Yang, J. Zhu, L. Qiu, D. Li, Advanced materials 2011, 23, 2833–2838.

[260] Y. Shao, M. F. El-Kady, L. J. Wang, Q. Zhang, Y. Li, H. Wang, M. F. Mousavi,R. B. Kaner, Chemical Society Reviews 2015, 44, 3639–3665.

[261] K. Zhang, L. Mao, L. L. Zhang, H. S. O. Chan, X. S. Zhao, J. Wu, Journal ofMaterials Chemistry 2011, 21, 7302–7307.

[262] L. Mao, K. Zhang, H. S. O. Chan, J. Wu, Journal of Materials Chemistry2012, 22, 80–85.

[263] J. Luo, H. D. Jang, J. Huang, ACS Nano 2013, 7, 1464–1471.

[264] C. Liu, Z. Yu, D. Neff, A. Zhamu, B. Z. Jang, Nano Letters 2010, 10, 4863–4868.

[265] J. L. Vickery, A. J. Patil, S. Mann, Advanced Materials 2009, 21, 2180–2184.

[266] Y. Wang, Z. Shi, Y. Huang, Y. Ma, C. Wang, M. Chen, Y. Chen, The Journalof Physical Chemistry C 2009, 113, 13103–13107.

[267] Y. Yoon, K. Lee, C. Baik, H. Yoo, M. Min, Y. Park, S. M. Lee, H. Lee,Advanced Materials 2013, 25, 4437–4444.

xlvii

REFERENCES

[268] A. Yu, I. Roes, A. Davies, Z. Chen, Applied Physics Letters 2010, 96, 253105.

[269] M. J. McAllister, J.-L. Li, D. H. Adamson, H. C. Schniepp, A. A. Abdala,J. Liu, M. Herrera-Alonso, D. L. Milius, R. Car, R. K. Prud’Homme, I. A.Aksay, Chemistry of Materials 2007, 19, 4396–4404.

[270] S. Vivekchand, C. S. Rout, K. Subrahmanyam, A. Govindaraj, C. Rao, Journalof Chemical Sciences 2008, 120, 9–13.

[271] J. Ye, H. Zhang, Y. Chen, Z. Cheng, L. Hu, Q. Ran, Journal of Power Sources2012, 212, 105–110.

[272] J. Yan, J. Liu, Z. Fan, T. Wei, L. Zhang, Carbon 2012, 50, 2179–2188.

[273] W. Gao, N. Singh, L. Song, Z. Liu, A. L. M. Reddy, L. Ci, R. Vajtai, Q. Zhang,B. Wei, P. M. Ajayan, Nature Nanotechnology 2011, 6, 496–500.

[274] M. F. El-Kady, V. Strong, S. Dubin, R. B. Kaner, Science 2012, 335, 1326–1330.

[275] F. Wen, C. Hao, J. Xiang, L. Wang, H. Hou, Z. Su, W. Hu, Z. Liu, Carbon2014, 75, 236–243.

[276] D. Wang, Y. Min, Y. Yu, B. Peng, Electrochimica Acta 2014, 141, 271–278.

[277] J. Chen, C. Li, G. Shi, The Journal of Physical Chemistry Letters 2013, 4,1244–1253.

[278] G. Xiong, C. Meng, R. G. Reifenberger, P. P. Irazoqui, T. S. Fisher, Electro-analysis 2014, 26, 30–51.

[279] I. Jung, D. A. Field, N. J. Clark, Y. Zhu, D. Yang, R. D. Piner, S. Stankovich,D. A. Dikin, H. Geisler, C. A. Ventrice Jr, R. S. Ruoff, The Journal of PhysicalChemistry C 2009, 113, 18480–18486.

[280] I.-W. P. Chen, Y.-S. Chen, N.-J. Kao, C.-W. Wu, Y.-W. Zhang, H.-T. Li,Carbon 2015, 90, 16–24.

[281] D. A. Brownson, D. K. Kampouris, C. E. Banks, Journal of Power Sources2011, 196, 4873–4885.

xlviii

REFERENCES

[282] L. Zhang, G. Shi, The Journal of Physical Chemistry C 2011, 115, 17206–17212.

[283] Y. Huang, J. Liang, Y. Chen, Small 2012, 8, 1805–1834.

[284] K. S. Suslick, G. J. Price, Annual Review of Materials Science 1999, 29,295–326.

[285] U. Khan, H. Porwal, A. O’Neill, K. Nawaz, P. May, J. N. Coleman, Langmuir2011, 27, 9077–9082.

[286] A. Hirsch, J. M. Englert, F. Hauke, Accounts of Chemical Research 2012, 46,87–96.

[287] J. Malig, A. W. Stephenson, P. Wagner, G. G. Wallace, D. L. Officer, D. M.Guldi, Chemical Communications 2012, 48, 8745–8747.

[288] J. Malig, N. Jux, D. Kiessling, J.-J. Cid, P. Vazquez, T. Torres, D. M. Guldi,Angewandte Chemie International Edition 2011, 50, 3561–3565.

[289] Q. Cheng, S. Debnath, E. Gregan, H. J. Byrne, The Journal of PhysicalChemistry C 2010, 114, 8821–8827.

[290] T. Skaltsas, X. Ke, C. Bittencourt, N. Tagmatarchis, The Journal of PhysicalChemistry C 2013, 117, 23272–23278.

[291] L. Malard, M. Pimenta, G. Dresselhaus, M. Dresselhaus, Physics Reports 2009,473, 51–87.

[292] T. Eberlein, U. Bangert, R. Nair, R. Jones, M. Gass, A. Bleloch, K. Novoselov,A. Geim, P. Briddon, Physical Review B 2008, 77, 233406.

[293] C. Casiraghi, A. Hartschuh, H. Qian, S. Piscanec, C. Georgi, A. Fasoli,K. Novoselov, D. Basko, A. Ferrari, Nano Letters 2009, 9, 1433–1441.

[294] J. M. Englert, P. Vecera, K. C. Knirsch, R. A. Schafer, F. Hauke, A. Hirsch,ACS Nano 2013, 7, 5472–5482.

[295] M.-E. Ragoussi, G. Katsukis, A. Roth, J. Malig, G. de la Torre, D. M. Guldi,T. Torres, Journal of the American Chemical Society 2014, 136, 4593–4598.

xlix

REFERENCES

[296] D. Guldi, G. Rahman, F. Zerbetto, M. Prato, Accounts of Chemical Research2005, 38, 871–878.

[297] C. Ehli, G. Rahman, N. Jux, D. Balbinot, D. Guldi, F. Paolucci, M. Marcaccio,D. Paolucci, M. Melle-Franco, F. Zerbetto, S. Campidelli, M. Prato, Journalof the American Chemical Society 2006, 128, 11222–11231.

[298] S. D. Chakarova-Kack, E. Schroder, B. I. Lundqvist, D. C. Langreth, PhysicalReview Letters 2006, 96, 146107.

[299] A. AlZahrani, Applied Surface Science 2010, 257, 807–810.

[300] S. D. Chakarova-Kack, A. Vojvodic, J. Kleis, P. Hyldgaard, E. Schroder, NewJournal of Physics 2010, 12, 013017.

[301] M. Shtein, I. Pri-Bar, O. Regev, Analyst 2013, 138, 1490–1496.

[302] Y. Tan, D. E. Resasco, The Journal of Physical Chemistry B 2005, 109,14454–14460.

[303] K. T. Kim, W. H. Jo, Journal of Polymer Science Part A: Polymer Chemistry2010, 48, 4184–4191.

[304] Y. Tomonari, H. Murakami, N. Nakashima, Chemistry - A European Journal2006, 12, 4027–4034.

[305] L. Zhang, Z. Zhang, C. He, L. Dai, J. Liu, L. Wang, ACS Nano 2014, 8,6663–6670.

[306] B. White, S. Banerjee, S. O’Brien, N. J. Turro, I. P. Herman, The Journal ofPhysical Chemistry C 2007, 111, 13684–13690.

[307] W. Liu, R. Zhou, D. Zhou, G. Ding, J. M. Soah, C. Y. Yue, X. Lu, Carbon2015, 83, 188–197.

[308] C. Yeon, S. J. Yun, K.-S. Lee, J. W. Lim, Carbon 2015, 83, 136–143.

[309] B. Vigolo, A. Penicaud, C. Coulon, C. Sauder, R. Pailler, C. Journet, P. Bernier,P. Poulin, Science 2000, 290, 1331–1334.

[310] S. Lin, C.-J. Shih, M. S. Strano, D. Blankschtein, Journal of the AmericanChemical Society 2011, 133, 12810–12823.

l

REFERENCES

[311] J. Geng, B. Kong, S. Yang, H. Jung, Chemical Communications 2010, 46,5091–5093.

[312] L. Wei, F. Wu, D. Shi, C. Hu, X. Li, W. Yuan, J. Wang, J. Zhao, H. Geng,H. Wei, Y. Wang, N. Hu, Y. Zhang, Scientific Reports 2013, 3 .

[313] Z.-h. Liu, Z.-M. Wang, X. Yang, K. Ooi, Langmuir 2002, 18, 4926–4932.

[314] M. Zhou, T. Tian, X. Li, X. Sun, J. Zhang, P. Cui, J. Tang, L.-C. Qin,International Journal of Electrochemical Science 2014, 9, 810–820.

[315] Y. Matsuo, T. Niwa, Y. Sugie, Carbon 1999, 37, 897–901.

[316] C. Wolfrum, A. Berger, A. Schneider, B. Gobelt, Suspension ContainingGraphene, Method for the Production thereof, Graphene Flakes and Use,WO002014053510 A1, 2014.

[317] K. B. Ricardo, A. Sendecki, H. Liu, Chemical Communications 2014, 50,2751–2754.

[318] S. Barwich, U. Khan, J. N. Coleman, The Journal of Physical Chemistry C2013, 117, 19212–19218.

[319] G. Chen, W. Weng, D. Wu, C. Wu, J. Lu, P. Wang, X. Chen, Carbon 2004,42, 753–759.

[320] M. Nakahara, Y. Sanada, Journal of Materials Science 1995, 30, 4363–4368.

[321] A. Schneider, S. Hildebrand, C. Wolfrum, Suspension Containing Graphene,Method for the Production thereof, Graphene Flakes and Use, WO2015004283A1, 2015.

[322] G. Eda, G. Fanchini, M. Chhowalla, Nature Nanotechnology 2008, 3, 270–274.

[323] A. Gupta, G. Chen, P. Joshi, S. Tadigadapa, P. Eklund, Nano Letters 2006, 6,2667–2673.

[324] G. Rahman, A. Troeger, V. Sgobba, D. M. Guldi, N. Jux, D. Balbino, M. N.Tchoul, W. T. Ford, A. Mateo-Alonso, M. Prato, Chemistry-A EuropeanJournal 2008, 14, 8837–8846.

[325] S. Leubner, G. Katsukis, D. M. Guldi, Faraday Discussions 2012, 155, 253–265.

li

REFERENCES

[326] L. Wang, H. Liu, R. M. Konik, J. A. Misewich, S. S. Wong, Chemical SocietyReviews 2013, 42, 8134–8156.

[327] P. Bilalis, D. Katsigiannopoulos, A. Avgeropoulos, G. Sakellariou, RSC Ad-vances 2014, 4, 2911–2934.

[328] G. Wang, Z. Yang, X. Li, C. Li, Carbon 2005, 43, 2564–2570.

[329] T. Kyotani, H. Moriyama, A. Tomita, Carbon 1997, 35, 1185–1187.

[330] Z. Gu, L. Zhang, C. Li, Journal of Macromolecular Science, Part B 2009, 48,1093–1102.

[331] N. D. Luong, U. Hippi, J. T. Korhonen, A. J. Soininen, J. Ruokolainen, L.-S.Johansson, J.-D. Nam, J. Seppala, Polymer 2011, 52, 5237–5242.

[332] C. Bao, Y. Guo, L. Song, Y. Kan, X. Qian, Y. Hu, Journal of MaterialsChemistry 2011, 21, 13290–13298.

[333] A. Yasmin, J.-J. Luo, I. M. Daniel, Composites Science and Technology 2006,66, 1182–1189.

[334] G. Zheng, J. Wu, W. Wang, C. Pan, Carbon 2004, 42, 2839–2847.

[335] X. Zhang, L. Shen, X. Xia, H. Wang, Q. Du, Materials Chemistry and Physics2008, 111, 368–374.

[336] P. Zarras, O. Vogl, Journal of Macromolecular Science - Pure and appliedChemistry 2000, 37, 817–840.

[337] H. Kamimura, Physics Today 2008, 40, 64–71.

[338] H. Hu, X. Wang, J. Wang, L. Wan, F. Liu, H. Zheng, R. Chen, C. Xu, ChemicalPhysics Letters 2010, 484, 247–253.

[339] F. Bonaccorso, Z. Sun, T. Hasan, A. Ferrari, Nature Photonics 2010, 4,611–622.

[340] L. X. Benedict, N. G. Chopra, M. L. Cohen, A. Zettl, S. G. Louie, V. H. Crespi,Chemical Physics Letters 1998, 286, 490–496.

[341] R. Zacharia, H. Ulbricht, T. Hertel, Physical Review B 2004, 69, 155406.

lii

REFERENCES

[342] J. Shen, Y. He, J. Wu, C. Gao, K. Keyshar, X. Zhang, Y. Yang, M. Ye,R. Vajtai, J. Lou, et al., Nano letters 2015, 15, 5449–5454.

[343] Y. Hernandez, M. Lotya, D. Rickard, S. D. Bergin, J. N. Coleman, Langmuir2009, 26, 3208–3213.

[344] C.-J. Shih, S. Lin, M. S. Strano, D. Blankschtein, Journal of the AmericanChemical Society 2010, 132, 14638–14648.

[345] J. N. Coleman, Advanced Functional Materials 2009, 19, 3680–3695.

[346] E.-Y. Choi, W. San Choi, Y. B. Lee, Y.-Y. Noh, Nanotechnology 2011, 22,365601.

[347] J. M. Cowie, M. A. Mohsin, I. J. McEwen, Polymer 1987, 28, 1569–1572.

[348] S. Edmondson, N. T. Nguyen, A. L. Lewis, S. P. Armes, Langmuir 2010, 26,7216–7226.

[349] R. Sanghvi, R. Narazaki, S. G. Machatha, S. H. Yalkowsky, AAPS Pharm-SciTech 2008, 9, 366–376.

[350] J. Cowie, I. McEwen, Polymer 1983, 24, 1449–1452.

[351] K.-G. Zhou, N.-N. Mao, H.-X. Wang, Y. Peng, H.-L. Zhang, AngewandteChemie International Edition 2011, 50, 10839–10842.

[352] U. Halim, C. R. Zheng, Y. Chen, Z. Lin, S. Jiang, R. Cheng, Y. Huang,X. Duan, Nature Communications 2013, 4 .

[353] J. R. Miller, P. Simon, Science 2008, 321, 651–652.

[354] A. Burke, Electrochimica Acta 2007, 53, 1083–1091.

[355] Y. Xu, Z. Lin, X. Zhong, X. Huang, N. O. Weiss, Y. Huang, X. Duan, NatureCommunications 2014, 5 .

[356] R. Hamlen, T. Atwater, D. Linden, T. Reddy, Handbook of Batteries, McGraw-Hill, New York, USA, 2002, S. 38–100.

[357] Y. Zhu, S. Murali, W. Cai, X. Li, J. W. Suk, J. R. Potts, R. S. Ruoff, AdvancedMaterials 2010, 22, 3906–3924.

liii

REFERENCES

[358] Y. Sun, Q. Wu, G. Shi, Energy & Environmental Science 2011, 4, 1113–1132.

[359] L. J. Wang, M. F. El-Kady, S. Dubin, J. Y. Hwang, Y. Shao, K. Marsh,B. McVerry, M. D. Kowal, M. F. Mousavi, R. B. Kaner, Advanced EnergyMaterials 2015, 5 .

[360] A. K. Mishra, S. Ramaprabhu, The Journal of Physical Chemistry C 2011,115, 14006–14013.

[361] E. P. Randviir, D. A. Brownson, C. E. Banks, Materials Today 2014, 17,426–432.

[362] J. J. Yoo, K. Balakrishnan, J. Huang, V. Meunier, B. G. Sumpter, A. Srivastava,M. Conway, A. L. Mohana Reddy, J. Yu, R. Vajtai, et al., Nano Letters 2011,11, 1423–1427.

[363] C. E. Hamilton, J. R. Lomeda, Z. Sun, J. M. Tour, A. R. Barron, Nano Letters2009, 9, 3460–3462.

[364] S. Sampath, A. N. Basuray, K. J. Hartlieb, T. Aytun, S. I. Stupp, J. F. Stoddart,Advanced Materials 2013, 25, 2740–2745.

[365] R. Chandrasekaran, Y. Soneda, J. Yamashita, M. Kodama, H. Hatori, Journalof Solid State Electrochemistry 2008, 12, 1349–1355.

[366] Y. Xu, Z. Lin, X. Huang, Y. Liu, Y. Huang, X. Duan, ACS Nano 2013, 7,4042–4049.

[367] G. Lota, T. A. Centeno, E. Frackowiak, F. Stoeckli, Electrochimica Acta 2008,53, 2210–2216.

[368] S. Stankovich, R. D. Piner, S. T. Nguyen, R. S. Ruoff, Carbon 2006, 44,3342–3347.

[369] K. Parvez, Z.-S. Wu, R. Li, X. Liu, R. Graf, X. Feng, K. M”ullen, Journal ofthe American Chemical Society 2014, 136, 6083–6091.

[370] M. Kaempgen, C. K. Chan, J. Ma, Y. Cui, G. Gruner, Nano Letters 2009, 9,1872–1876.

[371] J. R. Miller, R. Outlaw, B. Holloway, Science 2010, 329, 1637–1639.

liv

REFERENCES

[372] Z. Niu, W. Zhou, J. Chen, G. Feng, H. Li, W. Ma, J. Li, H. Dong, Y. Ren,D. Zhao, et al., Energy & Environmental Science 2011, 4, 1440–1446.

[373] P. Ruch, R. Kotz, A. Wokaun, Electrochimica Acta 2009, 54, 4451–4458.

[374] R. Shah, X. Zhang, S. Talapatra, Nanotechnology 2009, 20, 395202.

[375] X. Zhao, B. T. Chu, B. Ballesteros, W. Wang, C. Johnston, J. M. Sykes, P. S.Grant, Nanotechnology 2009, 20, 065605.

[376] J. H. Lee, N. Park, B. G. Kim, D. S. Jung, K. Im, J. Hur, J. W. Choi, ACSNano 2013, 7, 9366–9374.

[377] Z. Xu, Z. Li, C. M. Holt, X. Tan, H. Wang, B. S. Amirkhiz, T. Stephenson,D. Mitlin, The Journal of Physical Chemistry Letters 2012, 3, 2928–2933.

[378] G. Wang, X. Sun, F. Lu, H. Sun, M. Yu, W. Jiang, C. Liu, J. Lian, Small2012, 8, 452–459.

[379] Io-li-tec, Graphene Materials, http://www.nanomaterials.iolitec.de/Nanomaterialien/graphit-und-graphen.html, accessed: 2016-03-05.

[380] Vorbeck Materials, Graphene Ink Product List, http://cdn.shopify.com/s/files/1/0180/9087/files/Vor-Ink_Product_List_4_14.pdf?111,accessed: 2016-03-20.

[381] Graphene Supermarket, Conductive Graphene Dispersion, https://graphene-supermarket.com/Conductive-Graphene-Dispersion-100.html, accessed: 2016-03-05.

[382] Angstron Materials, Custom Graphene Dispersion, https://www.angstronmaterials.com/shop/custom-graphene-dispersions/, accessed:2016-03-05.

[383] D.-H. Kim, N. Lu, R. Ma, Y.-S. Kim, R.-H. Kim, S. Wang, J. Wu, S. M. Won,H. Tao, A. Islam, K. J. Yu, T.-i. Kim, R. Chowdhury, M. Ying, L. Xu, M. Li,H.-J. Chung, H. Keum, M. McCormick, P. Liu, Y.-W. Zhang, F. G. Omenetto,Y. Huang, T. Coleman, J. A. Rogers, Science 2011, 333, 838–843.

[384] S. Park, G. Wang, B. Cho, Y. Kim, S. Song, Y. Ji, M.-H. Yoon, T. Lee, NatureNanotechnology 2012, 7, 438–442.

lv

REFERENCES

[385] J. A. Rogers, T. Someya, Y. Huang, Science 2010, 327, 1603–1607.

[386] C. Hu, L. Song, Z. Zhang, N. Chen, Z. Feng, L. Qu, Energy & EnvironmentalScience 2015, 8, 31–54.

[387] Y. J. Kang, H. Chung, C.-H. Han, W. Kim, Nanotechnology 2012, 23, 065401.

[388] S. Hu, R. Rajamani, X. Yu, Applied Physics Letters 2012, 100, 104103.

[389] Z. Weng, Y. Su, D.-W. Wang, F. Li, J. Du, H.-M. Cheng, Advanced EnergyMaterials 2011, 1, 917–922.

[390] B. G. Choi, J. Hong, W. H. Hong, P. T. Hammond, H. Park, ACS Nano 2011,5, 7205–7213.

[391] B. G. Choi, S.-J. Chang, H.-W. Kang, C. P. Park, H. J. Kim, W. H. Hong,S. Lee, Y. S. Huh, Nanoscale 2012, 4, 4983–4988.

[392] H. Liu, Y. Liu, J. Li, Physical Chemistry Chemical Physics 2010, 12, 1685–1697.

[393] V. Khomenko, E. Raymundo-Pinero, E. Frackowiak, F. Beguin, Applied PhysicsA 2006, 82, 567–573.

[394] A. Lewandowski, I. Stepniak, Physical Chemistry Chemical Physics 2003, 5,4215–4218.

[395] W. Lu, K. Henry, C. Turchi, J. Pellegrino, Journal of the ElectrochemicalSociety 2008, 155, A361–A367.

[396] C. Largeot, C. Portet, J. Chmiola, P.-L. Taberna, Y. Gogotsi, P. Simon, Journalof the American Chemical Society 2008, 130, 2730–2731.

[397] T. Y. Kim, H. W. Lee, M. Stoller, D. R. Dreyer, C. W. Bielawski, R. S. Ruoff,K. S. Suh, ACS Nano 2010, 5, 436–442.

[398] C. Fu, Y. Kuang, Z. Huang, X. Wang, Y. Yin, J. Chen, H. Zhou, Journal ofSolid State Electrochemistry 2011, 15, 2581–2585.

[399] Z. Lei, Z. Liu, H. Wang, X. Sun, L. Lu, X. Zhao, Journal of Materials ChemistryA 2013, 1, 2313–2321.

lvi

APPENDIX

[400] Q. Cheng, J. Tang, J. Ma, H. Zhang, N. Shinya, L.-C. Qin, Physical ChemistryChemical Physics 2011, 13, 17615–17624.

[401] D. Pech, M. Brunet, H. Durou, P. Huang, V. Mochalin, Y. Gogotsi, P.-L.Taberna, P. Simon, Nature Nanotechnology 2010, 5, 651–654.

[402] Z. Niu, J. Chen, H. H. Hng, J. Ma, X. Chen, Advanced Materials 2012, 24,4144–4150.

[403] X. Cao, Y. Shi, W. Shi, G. Lu, X. Huang, Q. Yan, Q. Zhang, H. Zhang, Small2011, 7, 3163–3168.

[404] H. Nishide, K. Oyaizu, Science 2008, 319, 737–738.

[405] J. R. Miller, Science 2012, 335, 1312–1313.

[406] J. Taurozzi, V. Hackley, M. Wiesner, NIST Special Publication 2012, 1200, 2.

[407] Y. Zhu, Z. Sun, Z. Yan, Z. Jin, J. M. Tour, ACS Nano 2011, 5, 6472–6479.

[408] P. T. Hang, G. Brindley, Clays and Clay Minerals 1970, 18, 203–212.

[409] R. S. Rubino, E. S. Takeuchi, Journal of Power Sources 1999, 81, 373–377.

[410] M. Choucair, P. Thordarson, J. A. Stride, Nature Nanotechnology 2009, 4,30–33.

[411] Cell Coin Assembly, http://www.mtixtl.com/productimages/battery-equipment/spring%20and%20spacer%20structure-2.jpg, ac-cessed: 2016-01-23.

[412] S. Piscanec, M. Lazzeri, F. Mauri, A. Ferrari, The European Physical Journal-Special Topics 2007, 148, 159–170.

[413] A. B. Bourlinos, V. Georgakilas, R. Zboril, T. A. Steriotis, A. K. Stubos, Small2009, 5, 1841–1845.

lvii

A Appendix

Raw Graphite Starting Materials

REM Measurements of GFG 350

Figure A.1: Scanning electron image of the graphite starting material GFG 350.

Raman Data of GFG 350

Figure A.2 shows the Raman spectrum of the starting material GFG 350 obtained bynon-resonant Raman spectroscopy, namely 532 nm laser excitation. The spectrum isdominated by two intense features. The G peak at 1587 cm−1 and the 2D peak at2749 cm−1. Other Raman bands in the spectrum are seen at 2464 and 3283 cm−1

(2D’ mode), respectively. Notably, no D-band is observed, which would occur atapproximately 1350 cm−1. Since zone-boundary phonons, which give rise to theD-band, do not satisfy the Raman fundamental selection rule, they are not seen inthe first-order Raman spectrum of defect-free graphite.[412] Conclusively, the absenceof the D-band indicates that the starting material possesses a high structural qualitywith no defect content that is appreciable in the Raman spectrum. The 2D peak ofthe bulk graphite consists of two components 2D1 and 2D2, as shown in the inset ofFigure A.2. The 2D-band has half the height of the G peak and its full width half

lviii

APPENDIX

maximum (FWHM) is about 88 cm−1.

Figure A.2: Raman spectrum of the graphitic starting material GFG 350. The samplewas put onto a Si/SiO2 wafer and excited at 532 nm. Inset: Fitting ofthe 2D-band. Black dots - experimental data; blue curve - fitting.

Zeta Potential

a) b)

Figure A.3: Electrophoretic mobility distribution of a) G-1 and b) G-7 dispersionsafter centrifugation, respectively. The zeta potential was calculatedaccording to Chapter 7. The electrical field was 16.2 V cm−1.

lix

APPENDIX

a) b)

c) d)

e)

Figure A.4: Electrophoretic mobility distribution of a) G-1*-1, b) G-NaOH*-1, c)G-I1-1, d) G-I1*-1, and e) G-I2*-1 dispersions after centrifugation,respectively. The zeta potential was calculated according to Chapter 7.The electrical field was 16.2 V cm−1.

lx

APPENDIX

Figure A.5: Electrophoretic mobility distribution of G-VBTA*-1 dispersions aftercentrifugation. The zeta potential was calculated according to Chapter 7.The electrical field was 16.2 V cm−1.

XRD Measurements

Figure A.6: XRD spectra of raw graphite and pretreated graphite in tetraethylam-monium hydroxide (I1).

lxi

APPENDIX

Raman Analysis

Figure A.7: Histogram resulting from the Raman mapping of G-iPVBTA, showingcounts versus 2D-FWHM and the corresponding log-normal distribution,respectively. The sample was dip coated from dispersion onto a Si/SiO2wafer and excited at 532 nm.

Figure A.8: Histogram resulting from the Raman mapping of G-DMP, showingcounts versus 2D-FWHM and the corresponding log-normal distribution,respectively. The sample was dip coated from dispersion onto a Si/SiO2wafer and excited at 532 nm.

lxii

APPENDIX

Figure A.9: Histogram resulting from the Raman mapping of G-EG, showing countsversus 2D-FWHM and the corresponding log-normal distribution, re-spectively. The sample was dip coated from dispersion onto a Si/SiO2wafer and excited at 532 nm.

Figure A.10: Histogram resulting from the Raman mapping of G-NMP, showingcounts versus 2D-FWHM and the corresponding log-normal distribution,respectively. The sample was dip coated from dispersion onto a Si/SiO2wafer and excited at 532 nm.

lxiii

APPENDIX

Surface Tension Measurements

a) b)

c)

Figure A.11: Surface tension of water-alcohol mixture at various weight fraction at25 ◦C. Tested solvents: a) acetone/water, b) isopropyl alcohol/water,and c) ethanol/water, respectively.

lxiv

APPENDIX

Solvents for the Liquid-Phase Exfoliation of Graphite

Solvent Surface tension /

mN m−1

Boiling point /

◦C

Graphene concentration /

mg mL−1

Ref.

1,3-Dimethyl-2-imidazolidone 33.60 225 0.0054 [343]

1,4-Dioxane 33.00 101 0.0028 [343]

1-Dodecyl-2-pyrrolidone 30.70 202 0.0021 [343]

1-Vinylpyrrolidin-2-one 42.70 92 0.0055 [343]

Acetone 25.20 56 0.0012 [21,116,343]

Benzonitrile 39.52 188 0.0048 [343]

Benzyl benzoate 45.95 323 0.0047 [21,343]

Bromobenzene 36.25 156 0.0051 [343]

Chlorobenzene 33.60 131 0.0029 [343]

Chloroform 27.50 61 0.0034 [116,343]

Cyclohexanone 34.40 156 0.0073 [116,343]

Cyclopentanone 33.31 131 0.0085 [343]

Dibenzyl ether 39.80 135 0.0035 [343]

Dimethyl sulfoxide 42.98 189 0.0037 [343]

Dimethyl phthalate 40.50 283 0.0018 [343]

Ethanol 22.10 78 0.0016 [343]

Ethyl acetate 23.52 77 0.0026 [343]

Ethylene glycol 47.70 197 0.0010 [343]

γ-Butyrolactone 46.50 204 0.0041 [343]

Hexafluorobenzene 21.80 80 0.1 [413]

Isopropyl alcohol 23.00 83 0.0031 [21,116,343]

N,N-Dimethylacetamide 36.70 165 0.0039 [343]

N,N-Dimethylformamide 37.10 152 0.0041 [343]

N-methyl-pyrrolidone 40.79 202 2 [21,285,343]

Octafluorotoluene 22.29 103 0.05 [413]

ortho-Dichlorobenzene 36.60 180 0.03 [363]

Pyridine 38.00 115 0.3 [413]

Water 72.80 100 0.0011 [343]

lxv

APPENDIX

AFM Analysis

Figure A.12: AFM image of a SiO2 which was dip coated in a solution of 1 followedby rinsing with millipore water and absolute ethanol.

Electrochemical Characterization

Figure A.13: Cyclic voltammetry profile of an G-I1*-7 supercapacitor, a 7 super-capacitor without any graphitic active material, and a symmetricallyassembled supercapacitor from uncoated Au current collector foils at100 mV s−1, respectively.

lxvi

Liquid-Phase Exfoliated Graphene for Supercapacitors · Liquid-Phase Exfoliated Graphene for Supercapacitors Der Naturwissenschaftlichen Fakult¨at der Friedrich ... graphene through

Documents