Azide Functionalization of Carbon Materials for the Immobilization of Molecular Electrocatalysts

AZIDE FUNCTIONALIZATION OF CARBON MATERIALS FOR THE IMMOBILIZATION OF MOLECULAR ELECTROCATALYSTS

A DISSERTATION SUBMITTED TO THE DEPARTMENT OF CHEMISTRY AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIRMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

Eric Dean Stenehjem May 2014

This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/kk045gz8587

ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Daniel Stack, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Christopher Chidsey

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Edward Solomon

Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost for Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives.

iii

Abstract

Development of molecular electrocatalysts for the efficient interconversion between stored chemical energy and electrical energy would provide a pathway towards a sustainable energy future. For use of a molecular electrocatalyst, covalent immobilization at an electrode surface is highly advantageous to provide fast electron transport and prevent catalyst loss. The copper-catalyzed azide-alkyne cycloaddition

(CuAAC) reaction is desirable for such immobilization as it is selective, modular, and high yielding. Use of carbon materials for immobilization leverages their prevalent use as electrodes for energy applications.

Chapter 2 develops a novel preparation of gaseous iodine azide and subsequent use for the azide modification of glassy carbon surfaces. Generating dilute gaseous iodine azide in a nitrogen carrier gas stream from the flow of iodine monochloride vapor over a column of sodium azide provides a safe and convenient method to prepare, handle, and use this potentially explosive reagent. Its immediate use to treat hydrogen terminated glassy carbon was highly reproducible and chemically specific for azide functionalization to produce surfaces containing azides as the sole nitrogen species with coverage of 3.4 × 1014 molecules cm-2, ca. a quarter of a densely packed azide monolayer. Coupling ethynylferrocene to azide-modified glassy carbon via a

CuAAC reaction formed a 1,2,3-triazole linker with a coverage of 8 × 1013 molecules cm-2, a third of a densely packed ferrocene monolayer. Using X-ray photoelectron spectroscopy, the 1,2,3-triazole linker was observed to be hydrolytically stable in aqueous 1 M HClO4 or 1 M NaOH for at least 12 h at 100 °C.

Chapter 3 expands the gas-phase azide functionalization methodology towards high surface area mesoporous Vulcan XC-72R carbon powder. The increased surface area necessitated improving generation of gaseous iodine azide to maximize yields reaching the carbon surface. A 6-fold improvement in iodine azide yield was achieved by reducing decomposition, likely due to trace water, by thoroughly drying and maintaining anhydrous conditions. Treatment of XC-72R with gaseous iodine azide results in highly reproducible and chemically specific azide functionalization to produce surfaces containing azides as the sole nitrogen species with coverage of

1.4 × 1014 molecules cm-2. Coupling ethynylferrocene achieved coverage of 1.6 × 1013 molecules cm-2. Quantitative X-ray photoelectron spectroscopy indicates that all ferrocene molecules are bonded through a 1,2,3-triazole linker with no detectable physisorbed species.

Chapter 4 applies azide-modified surfaces in the development of an immobilized ruthenium electrocatalyst for the oxidation of benzyl alcohol and methanol with a 550 mV vs. NHE catalytic onset potential, a significant attenuation in potential (>300 mV) from reported immobilized ruthenium electrocatalysts. This electrocatalyst exhibits fast reaction kinetics with a TOF greater than 10 s-1 and is robust with 700 2-electron turnovers. The active catalytic species is postulated to be an

IV immobilized [Ru (ethynyl-TPA)(=O)Cl](PF6), formed electrochemically in two

II successive proton-coupled electron transfer steps from a Ru (OH2) species. The

II surface Ru (OH2) species was generated by photoinduced ligand exchange of DMSO

II on an immobilized [Ru (ethynyl-TPA)(DMSO)Cl](PF6) complex with H2O.

Acknowledgments

I would like to express my thanks and gratitude to my advisor Professor Dan

Stack for the opportunity to pursue my doctoral studies. The freedom and independence Dan afforded me along with his advice and guidance has provided me the opportunity to grow as a researcher. Professor Chris Chidsey has also been a valued mentor and collaborator who’s analytical and thoughtful approach to scientific problems has greatly contributed to this work. I’d also like to thank Professor Edward

Solomon for acting as a reader on my thesis committee along with Professor Jennifer

Wilcox for acting as the chair of my thesis defense and Professor Matt Kanan as a non-reader member.

Over the course of my graduate studies I have had the privilege of working with many different colleagues, both at Stanford and other institutions. Current and former members of the Stack and Chidsey group have been a pleasure to work with on a daily basis and have each contributed their individual insights and knowledge both to my research and my growth as a scientist. I’d like to express my gratitude to all of them: Cooper Citek, Thomas Cook, Sam Fretz, Dr. Brannon Gary, Olivia Hendricks,

Chris Lyons, Ross Moretti, Srinivasan Ramakrishnan, Matthijs Van den Berg, Mithi

Adhikari, Paul Alperin, Dr. Katharina Butsch, Dr. Ali Hosseini, Dr. Peng Kang, Dr.

Bolin Lin, Dr. Randy Lowe, Dr. Charles McCrory, Dr. Soushi Miyazaki, Dr. Jun

Nakazawa, Dr. Matt Pellow, Dr. Jonathan Prange, Dr. Brian Smith, Dr. Tim Storr, Dr.

Andrew Thomas, Dr. Pratik Verma, and Dr. Vadim Ziatdinov. My collaborations with researchers at GE Global Research, Lawrence Berkeley National Lab, and Yale

University has also exposed me to individuals and organizations with diverse backgrounds that have been invaluable for my growth as a scientist. In addition, the analytically resources and expertise on campus have been invaluable to the completion of my dissertation work. Specifically I’d like to thank Chuck Hitzman, Doug Turner, and Guangchao Li for their advice in developing and troubleshooting analytical techniques.

I have made many great friends during my time in graduate school and have learned much from them. I thank each of them for their friendship and look forward to continuing our friendship into the future.

Finally, I am deeply grateful for the love and support of my family. Their encouragement and advice, particularly throughout the last year of my graduate studies, has helped me overcome many struggles and challenges I never imagined encountering. I owe a very important debt to them and it is to them I dedicate this work.

vii

Table of Contents

Chapter 1. Surface Modification Methodologies of Graphitic Carbon Applicable for Covalent Immobilization of Molecular Electrocatalysts ...... 1

Abbreviations ...... 2

1.1 Introduction ...... 3

1.2 Chemical Structure of Graphitic Carbon Materials ...... 4

1.3 Surface Modification Methodologies ...... 5

1.3.1 Oxidation ...... 5

1.3.2 Aryl Diazonium Reduction ...... 7

1.3.3 Alkenes and Alkynes Addition ...... 8

1.3.4 Carboxylate Oxidation ...... 10

1.3.5 Chloromethylation ...... 11

1.3.6 Azide Modification for the Cycloaddition of Alkynes ...... 12

1.4 Conclusion ...... 14

1.5 References ...... 14

Chapter 2. Gas-Phase Azide Functionalization of Glassy Carbon ...... 20

Abbreviations ...... 21

Abstract ...... 22

2.1 Introduction ...... 22

2.2 Results ...... 24

2.2.1 Gas-phase Procedure for Azide Functionalization ...... 24

2.2.2 Solution Procedure for Azide Functionalization ...... 29

2.2.3 Surface Immobilization of Ethynylferrocene ...... 29 viii

2.2.4 Stability of the Immobilized Species ...... 32

2.3 Discussion ...... 33

2.4 Conclusion ...... 37

2.5 Experimental Section ...... 38

2.5.1 General Remarks ...... 38

2.5.2 Analytical Instrumentation ...... 38

2.5.3 Preparation of Glassy Carbon ...... 39

2.5.4 Preparation of NaN3(s) Dispersed on Silica ...... 39

2.5.5 Standard Gas-Phase Azide Functionalization Procedure ...... 40

2.5.6 Flow Reactor Design ...... 40

2.5.7 N 1s Peak Fitting ...... 41

2.5.8 Solution Azide Functionalization Procedure ...... 41

2.5.9 Ferrocene-Modified (2a), Heptafluoropropyl-Modified (2b), and

Trifluorotoluene-Modified (2c) Surfaces ...... 42

2.5.10 Stability of Surface Species ...... 42

2.6 Acknowledgments ...... 43

2.7 References ...... 43

Chapter 3. Gas-Phase Azide Functionalization of High Surface Area Carbon .. 47

Abbreviations ...... 48

Abstract ...... 49

3.1 Introduction ...... 49

3.2 Results ...... 51

3.2.1 Gas-Phase Azide Functionalization ...... 51

3.2.2 Covalent Surface Immobilization of Ethynylferrocene ...... 55 ix

3.3 Discussion ...... 58

3.4 Conclusion ...... 63

3.5 Experimental Section ...... 64

3.5.1 General Remarks ...... 64

3.5.2 Analytical Instrumentation ...... 64

3.5.3 Synthesis of Iodine Azide (IN3) ...... 66

3.5.4 Synthesis of XC-N3 ...... 68

3.5.5 Determination of Adsorbed I2(s) on Carbon Powder ...... 68

3.5.6 Synthesis of XC-Fc ...... 69

3.5.7 XC-Fc Preparation for ICP-OES ...... 69

3.5.8 Electrochemical Measurements of Ferrocene-modified XC-72R ...... 69

3.6 Acknowledgments ...... 70

3.7 References ...... 70

Chapter 4. Electrocatalytic Oxidization of Alcohols by a Surface Immobilized

Ruthenium Complex ...... 74

Abbreviations ...... 75

Abstract ...... 76

4.1 Introduction ...... 76

4.2 Results ...... 78

4.2.1 Immobilization of [Ru(ethynyl-TPA)(DMSO)Cl](PF6) (1) ...... 78

4.2.2 Characterization of ITO-TPA-Ru and GC-TPA-Ru ...... 79

4.2.3 Electrocatalytic Alcohol Oxidization by GC-TPA-Ru Surfaces ...... 80

4.2.4 Electrocatalytic Alcohol Oxidization by ITO-TPA-Ru Surfaces ...... 85

4.3 Discussion ...... 90 x

4.4 Conclusion ...... 94

4.5 Experimental Section ...... 95

4.5.1 General Remarks ...... 95

4.5.2 Analytical Instrumentation ...... 96

4.5.3 Synthesis of [Ru(ethynyl-TPA)(DMSO)Cl](PF6) (1) ...... 96

4.5.4 Synthesis of [Ru(ethynyl-TPA)(CD3CN)Cl](PF6) ...... 97

4.5.5 Immobilization of Ruthenium Complex 1 ...... 97

4.5.6 Electrocatalytic Alcohol Oxidization ...... 98

4.5.7 Activation of Immobilized Complex 1 ...... 98

4.6 Acknowledgments ...... 98

4.7 References ...... 98

List of Tables

Table 2.1. XPS Surface Atomic Concentrations of Carbon Samples after

Indicated Procedure ...... 27

Table 2.2. XPS Atomic Concentrations for Ferrocene-Modified Surfaces (2a) ...... 30

Table 2.3. XPS Sensitivity and Fitting Parameters ...... 39

Table 3.1. Characterization of Modified XC-72R Carbon Powder ...... 53

Table 3.2. Characterization of XC-Fc ...... 56

Table 3.3. XPS Sensitivity and Fitting Parameters ...... 65

xii

List of Figures

Figure 1.1. Examples of surface oxygenated species on the edge plane of carbon ...... 5

Figure 1.2. Immobilization of molecular species by coupling with (a) carboxylic acids

or (b) phenols on a carbon surface ...... 6

Figure 1.3. Modification of carbon surfaces by reduction of an aryl diazonium salt .... 7

Figure 1.4. Alkene and alkyne addition to the edge plane of carbon ...... 9

Figure 1.5. Modification of a carbon surface by oxidation of a carboxylated ligand

and use in the formation of an immobilized ruthenium electrocatalyst ...... 10

Figure 1.6. Chloromethylation of carbon followed by nucleophilic displacement of

the benzyl chloride ...... 11

Figure 1.7. Immobilization of an ethynylated Cu electrocatalyst onto azide-modified

carbon along with the proposed Cu2O2 intermediate species for the 4-electron

reduction of O2 to H2O ...... 12

Figure 2.1. Gas-phase azide functionalization of carbon occurs in a two-step

procedure: (1) Hydrogenation of carbon surfaces under forming gas.

(2) Azide functionalization using IN3(g) ...... 25

Figure 2.2. Integrated IN3(g) absorbance of the effluent gas stream through a

10 cm path length. Insert shows 2055 and 2065 cm-1 IR absorption peaks

characteristic of IN3(g) (shaded) ...... 25

Figure 2.3. UV/vis absorbance of the effluent gas stream through a 10 cm path length.

Insert shows UV/vis spectrum at maximum IN3(g) absorbance ...... 26

xiii

Figure 2.4. XPS spectra of (a) hydrogenated and (b) gas-phase azide–modified

glassy carbon ...... 27

Figure 2.5. XPS spectra of N 1s region with fitted spectrum and residuals after

subtraction of Shirley background for (a) gas-phase azide-modified surfaces

and (b) ferrocene-modified surfaces with an insert of the Fe 2p region.

Dotted lines are individual fitting components ...... 28

Figure 2.6. Surface Immobilization on Carbon ...... 31

Figure 2.7. Cyclic voltammetry for gas-phase ferrocene-modified surface.

Integration shown by shaded region ...... 32

Figure 2.8. Stability of heptafluoropropyl-modified (2b) and trifluorotoluene-modified

surfaces (2c) in both 1 M HClO4 and 1 M NaOH at 100 °C measured by XPS.

Error bars at 95% confidence interval ...... 33

Figure 3.1. Surface modification of XC-72R carbon powder ...... 52

Figure 3.2. XPS spectra of (a) XC-H and (b) XC-N3 surfaces ...... 53

Figure 3.3. XPS spectra of the N 1s region with fitted spectrum and residuals after

subtraction of Shirley background for (a) XC-N3 and (b) XC-Fc with insert of

Fe 2p region. Dotted lines are individual fitting components ...... 54

Figure 3.4. Cyclic voltammetry for XC-Fc with peak maximums (□) and

linear background (dotted line) ...... 58

Figure 3.5. Diagram of the flow reactor ...... 67

Figure 3.6. UV/vis absorbance of the effluent gas through a 10 cm path length for the

individual components of IN3(g) at 320 nm (blue solid line), ICl(g) at 467 nm

(red long dashed line), and I2(g) at 530 nm (purple short dashed line). 20 µl of xiv

ICl(l) was injected onto the glass wool column and carried over a dried column

6 of (a) NaN3(s) on silica as previously reported, (b) NaN3(s), showing

unconsumed ICl(g), and (c) Finely ground NaN3(s). Optimized synthesis (c)

achieved a 10-fold greater max concentration and 6-fold improved yield of

IN3(g) over previously report synthesis (a) ...... 67

Figure 3.7. UV/vis spectrum of XC-N3 in 10 mL toluene ...... 68

Figure 4.1. Isomers of the ruthenium complex 1 ...... 79

Figure 4.2. Surface immobilization of the ruthenium complex 1 ...... 79

Figure 4.3. CV of ITO-TPA-Ru in (a) acetonitrile and (b) pH 1 aqueous solution ... 80

Figure 4.4. CV of GC-TPA-Ru in pH 1 aqueous solution ...... 80

Figure 4.5. CV of GC-TPA-Ru before (dashed line) and after light exposure

(solid line) in a pH 7 phosphate buffer with (a) benzyl alcohol and (b) methanol.

Background CV of light-exposed GC-TPA-Ru in a pH 7 phosphate buffer

(dotted line) ...... 81

Figure 4.6. CV of GC-TPA-Ru before (dashed line) and after light exposure

(solid line) in a pH 7 phosphate buffer ...... 82

Figure 4.7. CV of light-exposed GC-TPA-Ru in a pH 7 phosphate buffer (solid line)

and pH 1 aqueous solution (dotted line) aqueous solutions

with benzyl alcohol ...... 82

Figure 4.8. Variance in CV catalytic oxidation currents of benzyl alcohol in a pH 7

phosphate buffer between three separate light-exposed GC-TPA-Ru surfaces

prepared in the same batch under identical conditions ...... 83

Figure 4.9. CV in pH 7 phosphate buffer with benzyl alcohol of a GC-TPA-Ru

surface exposed to light for 1 hr (solid line), 2 hr (dashed line), and 4 hr

(dotted line) ...... 83

Figure 4.10. CV in a pH 7 buffer with benzyl alcohol of 1 immobilized on glassy

carbon from modified CuAAC reaction, without DMSO, before (dashed line),

after 1 hr light exposure in acetonitrile (solid line), and after 1 hr acetonitrile

immersion in the dark (dotted line) ...... 84

Figure 4.11. CV of 1 immobilized from DCM (solid line) and DMF (dashed line)

CuAAC reactions in (a) pH 7 phosphate buffer with benzyl alcohol and

(b) pH 1 aqueous solution ...... 85

Figure 4.12. CV of ITO-TPA-Ru before (dashed line) and after light exposure

(solid line) in a pH 7 phosphate buffer with benzyl alcohol. Light-exposed

ITO-TPA-Ru in a pH 1 aqueous solution with benzyl alcohol (dotted line) .... 86

Figure 4.13. CV of an ITO-TPA-Ru surface exposed to light for 1 hr (solid line),

4 hr (dashed line), and 20 hr (dotted line) in a (a) pH 1 aqueous solution and

(b) pH 7 phosphate buffer with benzyl alcohol ...... 87

Figure 4.14. CV of immobilized [Ru(ethynyl-TPA)(CD3CN)Cl](PF6) with

impurity of complex 1 ...... 87

Figure 4.15. Chronoamperometry at 1.2 V of light-exposed ITO-TPA-Ru (solid line)

and ITO (dashed line) in pH 1 aqueous solution with benzyl alcohol ...... 88

Figure 4.16. CV of ITO-TPA-Ru after holding at 1.2 V for 0 s (solid line),

1000 s (dashed line), and 10000 s (dotted line) in a pH 1 aqueous solution ...... 89

xvi

Figure 4.17. CV of light-exposed ITO-TPA-Ru immersed in H2O (solid line) and

acetonitrile (dashed line) for the oxidation of benzyl alcohol in pH 7

phosphate buffer ...... 89

Figure 4.18. CV of light-exposed ITO-TPA-Ru in pH 1 aqueous solution (solid line)

and pH 7 phosphate buffer (dashed line) ...... 89

xvii

Chapter 1

Surface Modification Methodologies of Graphitic Carbon

Applicable for Covalent Immobilization of Molecular

Electrocatalysts

Chapter 1

List of Abbreviations

CuAAC Copper-catalyzed azide-alkyne cycloaddition

HOPG Highly ordered pyrolytic graphite

Me-bpy-Ac 4’-methyl-(2,2’-bipyridine)-4-acetate

NHE Normal hydrogen electrode

PPF Pyrolyzed photoresist film tpy 2,2′:6′,2′′-terpyridine

UV Ultraviolet

XPS X-ray photoelectron spectroscopy

Chapter 1

1.1 Introduction

Transitioning away from a society where the energy supply is based primarily on fossil fuel consumption to sustainable sources requires significant advances in materials for energy conversion and storage. Graphitic carbon materials hold promise for such advancement. Currently graphitic materials have prominent use for energy applications including graphite as anode materials in lithium ion batteries, activated carbon for supercapacitors, and carbon black as a support for electrocatalysis in fuel cells. This prevalence is due to the low-cost, abundance, minimal environmentally impact, and range of structural forms (thin films, felts, fibers, foams, glasses, and powders) of carbon materials. Surface modification of carbon materials1-4 allows additional control of the interfacial properties through which to introduce new functionality critical to tailoring these materials for energy conversion and storage applications.

Immobilization of metal complexes, specifically molecular electrocatalysts, provides additional functionality that has potential for use in the interconversion between stored chemical energy in covalent bonds and electrical energy. Research to develop efficient molecular electrocatalysts is ongoing and is discussed elsewhere.5

Robust covalent immobilization of a molecular electrocatalyst to an electrode surface is necessary to provide fast electron transfer kinetics along with preventing dissolution of the electrocatalyst into the fuel. In addition, the full control of the speciation for the immobilized catalyst is possible by using an applied potential and by preventing deactivation through oligomerization by site isolation. Finally, immobilization provides fuel flexibility as the catalyst no longer needs to be solubilized. 3

Chapter 1

This review highlights surface modification methodologies of graphitic carbon materials for the covalent immobilization of discrete molecular species. Where these methodologies have been applied for the immobilization of molecular electrocatalysts, reactivity of these systems is discussed in greater detail.

1.2 Chemical Structure of Graphitic Carbon Materials.

Graphitic carbon materials have a range of structural forms including thin films, felts, fibers, foams, glasses, and powders. Although seemingly dissimilar, this variety of material results from the unique spatial arrangement and relative abundance of basal plane and edge plane carbon sites.6 The arrangement and abundance also determines the bulk properties of carbon materials including porosity, hardness, thermal and electrical conductivity, and surface reactivity. Highly ordered pyrolytic graphite (HOPG) can be prepared to expose primarily either its basal plane or edge plane sites, but the surfaces of other carbon materials, including glassy carbon, pyrolyzed photoresist film (PPF), and carbon black, contain a mixture of these sites.

Carbon surfaces are primarily reactive towards modification at at edge plane sites and basal plane defects, whereas the basal plane is relatively unreactive except at high temperatures.4

Surface exposed edge plane sites are terminated natively with oxygenated species including carboxylic acids, anhydrides, lactones, lactols, phenols, carbonyls, and quinones formed through the reaction with dioxygen and water (Figure 1.1).6

Determining the type and amount of each oxygenated species is possible using a combination of X-ray photoelectron spectroscopy (XPS), thermal desorption mass spectrometry, chemical titration, and infrared spectroscopy.7 The amount of each 4

Chapter 1

oxygenated species is not only dependent on the surface structure of the carbon material but also upon its handling and treatment. Removal of surface oxygenated species can be accomplished by treatment with H2 plasma or forming gas (5% H2 in

8 N2) at 1000 °C, which creates a hydrogen terminated surface. Carbon surfaces containing oxygenated species are used primarily for surface modification, as opposed to hydrogen terminated surfaces that are more difficult to handle.

Figure 1.1. Examples of surface oxygenated species on the edge plane of carbon.

1.3 Surface Modification Methodologies

1.3.1 Oxidation. Molecular species can be covalently attached to carbon surfaces by coupling with carboxylic acids, anhydrides, or phenols. Conversion of surface carboxylic acid and anhydride groups to acyl chlorides is possible using thionyl chloride followed by a reaction with an amine or hydroxyl containing species couples discrete species to the surface through an amide or ester linkage, respectively

(Figure 1.2a).6 Surface acyl chlorides have been used for the immobilization of CuII and MnIII hydroxyl-modified salen complexes on a variety of mesoporous carbon materials.9-12 Surface phenol groups react with acyl chlorides13 or cyanuric chloride14 to create ester or ether bonds, respectively (Figure 1.2b). Attachment of cyanuric chloride creates either 1 or 2 ether bonds to the carbon surface as attachment points for

Chapter 1

added amine or hydroxyl containing species.14 A NiII Schiff base complex containing a hydroxyl group has been immobilized by reacting with surface attached cyanuric chloride.15 Coverage of covalently immobilized species is generally low, less than

10% of a densely packed monolayer,2 as surfaces contain a distribution of oxygenated species. Different oxidative treatments of the carbon material affect this distribution.

Two commonly used treatments are 1) dilute nitric acid, which favors creation of surface carboxylic acids along with some phenol and carbonyl groups and 2) dilute O2

(typically, 5% in N2) at 350 – 450 °C, which mainly forms phenols and carboxylic

12,16 acids. Treatment of oxidized carbon surfaces with LiAlH4 can reduce many oxygenated species to surface phenols.14

Figure 1.2. Immobilization of molecular species by coupling with (a) carboxylic acids or (b) phenols on a carbon surface.

Immobilization of metal complexes through coupling with surface oxygenated species on carbon has been used for the heterogenization of discrete homogeneous catalysts for various non-electrocatalytic reactions.17 A primary problem with this methodology is distinguishing between complexes covalently bonded to the surface 6

Chapter 1

and those which are physisorbed strongly,18 as unambiguous spectroscopic differences are lacking.12 This ambiguity, along with the low coverage, can make characterization difficult for complexes immobilized by this methodology.

1.3.2 Aryl Diazonium Reduction. Covalent attachment of aryl species onto carbon surfaces can be accomplished by the addition of aryl radicals, which are formed by the one electron reduction, with loss of N2, of a diazonium salt either electrochemically19,20 or spontaneously for some mesoporous carbon materials

(Figure 1.3).21-23 Irreversible electrochemical reduction of aryl diazonium salts commonly ranges between 80 – 400 mV vs NHE.19 This low reduction potential explains why diazonium salts can spontaneously add to some carbon materials; carbon materials are slightly reducing.21 Addition of an aryl radical to the surface creates a robust C–C bond, stable to at least 500 °C under vacuum.24 Due to the reactivity of this aryl radical intermediate, ill-defined multilayer structures are formed, up to 15 nm thick.19 Multilayer structures are problematic as introduced functionality from the aryl species is buried along with a dramatic reduction of the pore diameter in porous materials. Formation of a monolayer is possible through use of steric blocking groups,25,26 degradation of the multilayer structure,27 and localized electrogeneration of the diazonium.28

Figure 1.3. Modification of carbon surfaces by reduction of an aryl diazonium salt.

Chapter 1

Surface modification using aryl diazoniums salts is tolerant to a variety of functional groups in the para-position, which provides a chemical handle for further surface modification via a number of different coupling reactions.3 For the immobilization of metal complexes the use of a copper-catalyzed azide-alkyne cycloaddition (CuAAC) is prominent because of its selectivity, high yield, and modularity. Both the diazonium salts of 4-azidobenzene and 4-ethynylbeneze have been used to create azidobenzene-modified and ethynylbenzene-modified carbon surfaces, respectively.29 Immobilization of ethynylferrocene and azidoferrocene on their complementary surfaces via a CuAAC reaction achieved 70% and 50% of a densely packed monolayer, respectively. Using a sterically bulky alkyne protecting group, a monolayer of ethynylbenzene-modified surface was formed, which when coupled with azidoferrocene achieved 95% of a densely packed monolayer.26 The higher ferrocene coverage achieved on monolayer ethynylbenzene-modified surfaces compared to multilayer surfaces suggests a multilayer structure may sterically hinder the CuAAC reaction.

1.3.3 Alkenes and Alkynes Addition. Addition of alkenes and alkynes to graphitic carbon surfaces occurs photochemically (254 nm)30-32 and at elevated temperatures (175-200 °C).33-35 In both methodologies covalent surface attachment is proposed to occur through a [4+2] Diels-Alder or a [2+2] cycloaddition to edge plane sites (Figure 1.4). Similar attachment to the carbon basal plane is not observed.35

Atomic force microscopy images for the photochemical attachment of alkenes and alkynes indicate sub-monolayer coverage, which was quantified for ethynylferrocene immobilized on a PPF at 16% of a densely packed monolayer by voltammetry.31 8

Chapter 1

Coverages ranging between 1 x 1013 and 1 x 1014 molecules/cm2, less than a third of a monolayer at most, were also measured for the thermal immobilization of alkene and alkyne derivatives of ferrocene and Zn porphyrin onto glassy carbon and PPF.33

Figure 1.4. Alkene and alkyne addition to the edge plane of carbon.

No reported examples of immobilized electrocatalytic complexes by either photochemical or thermal methodologies exist but such reactions should be possible as immobilized alkene and alkyne derivatives of ferrocene and Zn porphyrin have been achieved.31,33 However, the complex must be stable towards the UV irradiation or high temperatures necessary under these coupling conditions. While the photochemical methodology is applicable to planar carbon materials, difficulties might arise in mesoporous carbon materials as UV illumination would not penetrate past the exposed surface.

Chapter 1

1.3.4 Carboxylate Oxidation. Oxidation of benzyl36-38 and alkyl

39 carboxylates release CO2 to generate a carbon radical through loss of CO2, which can couple to a carbon surface (Figure 1.5). Such radical species are short lived and must be formed at an electrode surface as further oxidation to the carbocation species is more favorable.40 Typically this is performed through repeated cyclic voltammetry

red scans to high oxidation potentials (Ep > 1.4 V vs NHE), with subsequent scans showing less oxidative current due to the formation of passivating multilayer structure.

37 Multilayer thickness up to 13 nm were measured for a naphthyl–CH2– film. These multilayer structures can bury introduced functionality along with reducing pore diameter. The latter will result in slower diffusion through porous materials in any application. In addition, the high oxidation potential needed to generate these radical species limits compatible functional groups.

Figure 1.5. Modification of a carbon surface by oxidation of a carboxylated ligand and use in the formation of an immobilized ruthenium electrocatalyst.

Currently no reports of direct immobilization of a metal electrocatalyst from this methodology exist in the literature. However, a carboxylate modified ligand, lithium 4’-methyl-(2,2’-bipyridine)-4-acetate (Me-bpy-Ac), was attached to a high surface area carbon felt electrode by anodic oxidation and subsequently metallated

II 2+ with a ruthenium complex to form an immobilized [Ru (tpy)(Me-bpy-Ac)(OH2)]

[typ = 2,2′:6′,2′′-terpyridine] electrocatalyst (Figure 1.5).38 Coverage of this RuII

Chapter 1

electrocatalyst was estimated at 2.4 x 1012 molecules/cm2, 1–2% of a dense monolayer. Bulk electrocatalytic oxidation of benzyl alcohol was performed using this immobilized ruthenium complex at 0.94 V vs NHE in a phosphate buffer (pH 7.2).41

After 4 hr, benzaldehyde (13%) and benzoic acid (85%) products were detected with a total of 183 two-electron turnovers and 28% current efficiency. The low current efficiency is due to extensive water oxidation from the carbon felt electrode.

II 2+ Immobilized [Ru (tpy)(Me-bpy-Ac)(OH2)] also showed electrocatalytic activity towards other primary and secondary benzyl alcohol species.38

1.3.5 Chloromethylation. A mesoporous activated carbon material is susceptible to chloromethylation, conversion of an aromatic C–H to a –CH2Cl group

(Figure 1.6).42 Coverage of 6 x 1013 benzyl chlorides/cm2 is possible.

Chloromethylation has minimal effect on the pore diameter and surface area of the modified material. Further surface functionalization was achieved by nucleophilic displacement of the benzyl chlorides with thiosulfate42 and primary amines.43 These thiosulfate and amine modified-materials have been used as chemical sorbents for heavy metal separations.

Figure 1.6. Chloromethylation of carbon followed by nucleophilic displacement of the benzyl chloride.

While chloromethylated carbon has not been used for the immobilization of electrocatalysts, this methodology holds promise for electrocatalytic applications as it achieves good coverages of benzyl chlorides, is susceptible to nucleophilic

Chapter 1

displacement, potentially by electrocatalytic complexes, and minimally affects the pore diameter of porous carbon materials as functionality is attached directly to the surface.

1.3.6 Azide Modification for the Cycloaddition of Alkynes. Azide modification of carbon materials followed by subsequent coupling with terminal alkynes via a CuAAC reaction44-46 represents a versatile strategy for surface modification (Figure 1.7). The CuAAC reaction is tolerant of many functional groups,47,48 therefore using this methodology a wide variety of alkynes can be coupled to azide-modified carbon to produce a range of functionalized surfaces. This introduced functionality is covalently linked to the surface through a 1,2,3-triazole linker, which is highly robust and hydrolytically stable remaining unchanged in both

49 aqueous 1 M HClO4 and 1 M NaOH for at least 12 h at 100 °C.

Figure 1.7. Immobilization of an ethynylated Cu electrocatalyst onto azide-modified carbon along with the proposed Cu2O2 intermediate species for the 4-electron reduction of O2 to H2O.

Azide modification has been accomplished with hydrogen terminated carbon

50 materials using IN3 in acetonitrile for PPF and edge-plane graphite, in hexane for glassy carbon,51 and in the gas-phase for glassy carbon49 and mesoporous Vulcan

52 XC-72R. Azide addition to the carbon surface by IN3 is thought to occur by radical substitution of activated C–H bonds and is highly selective in the gas-phase, producing 12

Chapter 1

surfaces containing azides as the sole surface nitrogen species.49 The highest reported

14 2 azide coverage was 4 x 10 azides/cm from the gas-phase IN3 treatment of glassy carbon, approximately 25% of a densely packed azide monolayer.49 The cycloaddition of terminal alkynes to azide-modified surfaces to form the covalent 1,2,3-triazole linker can be detected by XPS from changes to the N 1s region.53 The reaction of ethynylferrocene with azide-modified glassy carbon produced surfaces with a voltammetric coverage of 8 × 1013 molecules/cm2, approximately 33% of a densely packed ferrocene monolayer. These covalently attached ferrocene exhibit fast electron transfer kinetics, greater than 1000 s-1, through the 1,2,3-triazole into the carbon electrode surface.49

Electrocatalytic 4-electron O2 reduction to H2O was observed for a

Cu(3-ethynyl-phenanthroline) complex covalently immobilized on azide-modified

51 glassy carbon via a CuAAC reaction. The O2 reduction current was second order in copper coverage, suggesting a Cu2O2 intermediate species for the 4-electron reduction of O2 (Figure 1.7). At low copper coverage, 2-electron O2 reduction occurs with production of H2O2, presumably from a site-isolated mononuclear Cu complex.

Control of the surface coverage of the immobilized Cu complex allowed inference of a

Cu2O2 intermediate not observed previously for the electrocatalytic O2 reduction by a physisorbed Cu(phenanthroline) complex.54,55 An ethynylated ruthenium complex has also been developed and immobilized via a CuAAC reaction for the electrocatalytic oxidation of alcohols, as discussed in Chapter 4 of this work.56 This methodology promises to be generalizable for the immobilization of any terminal alkynes species.

Chapter 1

1.4 Conclusion

In summary, diverse methodologies exist for the covalent attachment of molecular species to graphitic carbon surfaces. Of these methodologies, aryl and azide modifications have been applied to a wide range of carbon materials, including mesoporous powders, to covalently attach discrete molecular species. Immobilization of metal complexes commonly relies upon further synthetic coupling to these introduced molecular species, of which CuAAC is highly promising. While the feasibility of covalently immobilized electrocatalysts on graphitic materials has been demonstrated for select energy applications, ongoing work is need both in electrocatalyst design and their immobilization to achieve the significant materials advancements necessary for a sustainable energy future.

1.5 References

(1) Liang, C.; Li, Z.; Dai, S. Angew. Chem., Int. Ed. 2008, 47, 3696-3717.

(2) McCreery, R. L. Chem. Rev. 2008, 108, 2646-2687.

(3) Mahouche-Chergui, S.; Gam-Derouich, S.; Mangeney, C.; Chehimi, M. M.

Chem. Soc. Rev. 2011, 40, 4143-4166.

(4) Barriere, F.; Downard, A. J. J. Solid State Electrochem. 2008, 12, 1231-1244.

(5) DuBois, D. L. Inorg. Chem. 2014, 53, 3935-3960.

(6) Kinoshita, K. Carbon: Electrochemical and Physicochemical Properties; John

Wiley and Sons: New York, 1988.

(7) Boehm, H. P. Carbon 2002, 40, 145-149.

(8) Kuo, T.; McCreery, R. L. Anal. Chem. 1999, 71, 1553-1560.

Chapter 1

(9) Silva, A. R.; Freitas, M. M. A.; Freire, C.; de Castro, B.; Figueiredo, J. L.

Langmuir 2002, 18, 8017-8024.

(10) Silva, A. R.; Vital, J.; Figueiredo, J. L.; Freire, C.; de Castro, B. New J. Chem.

2003, 27, 1511-1517.

(11) , , J.; Freire, C.; de Castro, B. Micropor.

Mesopor. Mater. 2004, 68, 83-89.

(12) Mahata, N.; Silva, A. R.; Pereira, M. F. R.; Freire, C.; de Castro, B.;

Figueiredo, J. L. J. Colloid Interface Sci. 2007, 311, 152-158.

(13) Ray, K. G.; McCreery, R. L. J. Electroanal. Chem. 1999, 469, 150.

(14) Alves, J. A. C.; Freire, C.; de Castro, B.; Figueiredo, J. L. Colloid Surf., A

2001, 189, 75-84.

(15) Silva, A. R.; Freire, C.; de Castro, B.; Freitas, M. M. A.; Figueiredo, J. L.

Micropor. Mesopor. Mater. 2001, 46, 211-221.

(16) Figueiredo, J. L.; Pereira, M. F. R.; Freitas, M. M. A.; Órfão, J. J. M. Carbon

1999, 37, 1379-1389.

(17) Figueiredo, J. L. J. Mater. Chem. A 2013, 1, 9351-9364.

(18) Cardoso, W.; Silva, A. R.; de Castro, B.; Freire, C. Appl. Catal., A 2005, 285,

110-118.

(19) Pinson, J.; Podvorica, F. Chem. Soc. Rev. 2005, 34, 429-439.

(20) Belanger, D.; Pinson, J. Chem. Soc. Rev. 2011, 40, 3995-4048.

(21) Li, Z.; Dai, S. Chem. Mater. 2005, 17, 1717-1721.

(22) Toupin, M.; Bélanger, D. J. Phys. Chem. C 2007, 111, 5394-5401.

Chapter 1

(23) (a) Belmont, J. A. U.S. Patent 5,554,739, September, 10 1996 (Cabot

Corporation). (b) Belmont, J. A. U.S. Patent 5,851,280, December 22, 1998

(Cabot Corporation). (c) Belmont, J. A.; Adams, C. E. U.S. Patent 5,895,522,

April 20, 1999 (Cabot Corporation). (d) Johnson, J. E.; Belmont, J. A. U.S.

Patent 5,803,959, September 8, 1998. (e) Belmont, J. A.; Adams, C. E. U.S.

Patent 5,713,988, February 3, 1998 (Cabot Corporation). (f) Adams, C. E.;

Belmont, J. A. U.S. Patent 5,707,432, January 13, 1998 (Cabot Corporation).

(g) Belmont, J. A. U.S. Patent 5,672,198, September 30, 1997 (Cabot

Corporation). (h) Belmont, J. A.; Amici, R. M.; Galloway, C. P. U.S. Patent

6,042,643, March 28, 2000. (i) Belmont, J. A.; Amici, R. M.; Galloway, C. P.

U.S. Patent 6,494,946, December 17, 2002. (j) Belmont, J. A.; Amici, R. M.;

Galloway, C. P. U.S. Patent 6,740,151, May 25, 2004. Patent

(24) Allongue, P.; Delamar, M.; Desbat, B.; Fagebaume, O.; Hitmi, R.; Pinson, J.;

Saveant, J. M. J. Am. Chem. Soc. 1997, 119, 201.

(25) Combellas, C.; Jiang, D. E.; Kanoufi, F.; Pinson, J.; Podvorica, F. Langmuir

2009, 25, 286-293.

(26) Leroux, Y. R.; Fei, H.; Noel, J. M.; Roux, C.; Hapiot, P. J. Am. Chem. Soc.

2010, 132, 14039-14041.

(27) Nielsen, L. T.; Vase, K. H.; Dong, M.; Besenbacher, F.; Pedersen, S. U.;

Daasbjerg, K. J. Am. Chem. Soc. 2007, 129, 1888-1889.

(28) Kongsfelt, M.; Vinther, J.; Malmos, K.; Ceccato, M.; Torbensen, K.; Knudsen,

C. S.; Gothelf, K. V.; Pedersen, S. U.; Daasbjerg, K. J. Am. Chem. Soc. 2011,

133, 3788-3791. 16

Chapter 1

(29) Evrard, D.; Lambert, F.; Policar, C.; Balland, V.; Limoges, B. Chem. Eur. J.

2008, 14, 9286-9291.

(30) Sun, B.; Colavita, P. E.; Kim, H.; Lockett, M.; Marcus, M. S.; Smith, L. M.;

Hamers, R. J. Langmuir 2006, 22, 9598-9605.

(31) Yu, S. S.; Downard, A. J. Langmuir 2007, 23, 4662-4668.

(32) Wang, X.; Landis, E. C.; Franking, R.; Hamers, R. J. Acc. Chem. Res. 2010,

43, 1205-1215.

(33) Ssenyange, S.; Anariba, F.; Bocian, D. F.; McCreery, R. L. Langmuir 2005,

21, 11105-11112.

(34) Ababou-Girard, S.; Solal, F.; Fabre, B.; Alibart, F.; Godet, C. J. Non-Cryst.

Solids 2006, 352, 2011-2014.

(35) Ababou-Girard, S.; Sabbah, H.; Fabre, B.; Zellama, K.; Solal, F.; Godet, C. J.

Phys. Chem. C 2007, 111, 3099-3108.

(36) Andrieux, C. P.; Gonzalez, F.; Savéant, J.-M. J. Am. Chem. Soc. 1997, 119,

4292-4300.

(37) Brooksby, P. A.; Downard, A. J.; Yu, S. S. C. Langmuir 2005, 21, 11304-

11311.

(38) Geneste, F.; Moinet, C.; Jezequel, G. New J. Chem. 2002, 26, 1539-1541.

(39) Astudillo, P. D.; Galano, A.; González, F. J. J. Electroanal. Chem. 2007, 610,

137-146.

(40) Wayner, D. D. M.; McPhee, D. J.; Griller, D. J. Am. Chem. Soc. 1988, 110,

132-137.

(41) Geneste, F.; Moinet, C. New J. Chem. 2004, 28, 722-726. 17

Chapter 1

(42) Samuels, W. D.; LaFemina, N. H.; Sukwarotwat, V.; Yantasee, W.; Li, X. H.

S.; Fryxell, G. E. Sep. Sci. Technol. 2010, 45, 228-235.

(43) Chouyyok, W.; Shin, Y.; Davidson, J.; Samuels, W. D.; Lafemina, N. H.;

Rutledge, R. D.; Fryxell, G. E.; Sangvanich, T.; Yantasee, W. Environ. Sci.

Techol. 2010, 44, 6390-6395.

(44) Tornoe, C. W.; Christensen, C.; Meldal, M. J. Org. Chem. 2002, 67, 3057-

3064.

(45) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B. Angew. Chem.

Int. Ed. 2002, 41, 2596-2599.

(46) Kolb, H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed. 2001, 40,

2004-2021.

(47) Evans, R. A. Aust. J. Chem. 2007, 60, 384.

(48) Iha, R. K.; Wooley, K. L.; Nystrom, A. M.; Burke, D. J.; Kade, M. J.; Hawker,

C. J. Chem. Rev. 2009, 109, 5620-5686.

(49) Stenehjem, E. D.; Ziatdinov, V. R.; Stack, T. D. P.; Chidsey, C. E. D. J. Am.

Chem. Soc. 2013, 135, 1110-1116.

(50) Devadoss, A.; Chidsey, C. E. D. J. Am. Chem. Soc. 2007, 129, 5370-5371.

(51) McCrory, C. C. L.; Devadoss, A.; Ottenwaelder, X.; Lowe, R. D.; Stack, T. D.

P.; Chidsey, C. E. D. J. Am. Chem. Soc. 2011, 133, 3696–3699.

(52) Stenehjem, E. D. Chapter 3: Gas-Phase Azide Functionalization of High

Surface Area Carbon. Ph. D. Dissertation, Leland Stanford Junior University,

Stanford, CA, May 2014; pp 47 – 73.

Chapter 1

(53) Collman, J. P.; Devaraj, N. K.; Eberspacher, T. P. A.; Chidsey, C. E. D.

Langmuir 2006, 22, 2457-2464.

(54) Zhang, J. J.; Anson, F. C. Electrochim. Acta 1993, 38, 24232429.

(55) Lei, Y. B.; Anson, F. C. Inorg. Chem. 1994, 33, 5003-5009.

(56) Stenehjem, E. D. Chapter 4: Electrocatalytic Oxidation of Alcohols by a

Surface Immobilized Ruthenium Complex. Ph. D. Dissertation, Leland Stanford

Junior University, Stanford, CA, May 2014; pp 74 – 100.

Chapter 2

Gas-Phase Azide Functionalization of Glassy Carbon

Reproduced with permission from Stenehjem, E.D.; Ziatdinov, V.R.; Stack, T.D.P.; Chidsey,

C.E.D. “Gas-Phase Azide Functionalization of Carbon” Journal of the American Chemical

Chapter 2

List of Abbreviations

At. % Atomic percent

CuAAC Copper-catalyzed azide-alkyne cycloaddition

DMSO Dimethyl sulfoxide

Fc Ferrocene

IR Infrared

TTMA Tris-(ethylacetyltriazolyl)methylamine

UV/vis Ultraviolet–visible

XPS X-ray photoelectron spectroscopy

Chapter 2

Abstract

Tailoring the surface and interfacial properties of inexpensive and abundant carbon materials plays an increasingly important role for innovative applications including those in electrocatalysis, energy storage, gas separations, and composite materials. Described here is the novel preparation and subsequent use of gaseous iodine azide for the azide modification of carbon surfaces. In-line generation of gaseous iodine azide from iodine monochloride vapor and solid sodium azide is safe and convenient. Immediate treatment of carbon surfaces with this gaseous stream of iodine azide provides a highly reproducible, selective, and scalable azide functionalization that minimizes waste and reduces deleterious side reactions. Among the possible uses of azide-modified surfaces, they serve as versatile substrates for the attachment of additional functionality by coupling with terminal alkynes under the mild copper-catalyzed azide-alkyne cycloaddition (CuAAC) “click” reaction. For instance, coupling ethynylferrocene to azide-modified glassy carbon surfaces achieves ferrocene coverage up to 8 × 1013 molecules/cm2 by voltammetric and XPS analyses.

The 1,2,3-triazole linker formed during the CuAAC reaction is robust and hydrolytically stable in both aqueous 1 M HClO4 and 1 M NaOH for at least 12 h at

100 °C.

2.1 Introduction

Carbon materials have found prominent use across a variety of applications including those in electrocatalysis, energy storage, gas separations, and composite materials. This extensive use results from not only the abundance and low-cost of carbon but also its desirable physical and chemical properties, such as high electrical 22

Chapter 2

and thermal conductivity, chemical stability, and range of structural forms (felts, fibers, foams, glasses, and powders). Furthermore, surface modification of carbon materials1,2 allows for the introduction of additional functionality to tailor its surface and interfacial properties for the desired application. Surface modification strategies for the creation of these functional materials should ideally (1) have a robust linker between the introduced functionality and the surface, best provided by hydrolytically stable covalent bond formation3,4 and (2) undergo selective transformations to create chemically well-defined surfaces, thus allowing for the rational design of materials.

A commonly used carbon surface functionalization methodology involves coupling with aryl radicals, which are formed in solution by the reduction of aryl diazonium salts, resulting in a robust and hydrolytically stable C–C bond to the surface.5,6 However, the promiscuous reactivity of this aryl radical intermediate commonly leads to the formation of ill-defined multilayer structures up to 15 nm thick.7 In select cases a monolayer has been achieved by degradation of the multilayer structure above,8 use of steric blocking groups,9,10 and localized electrogeneration of the diazonium.11 Other methods for robust surface functionalization include radical coupling of amines,12 coupling of alkynes at high temperatures,13 and photochemical coupling with alkenes.14,15 These relatively nonselective methods, while producing robust C–C or C–N bonds, are limited in scope to molecules tolerant to the harsh coupling conditions.

Another surface functionalization strategy is the introduction of a surface group onto which additional functionality can be coupled in subsequent chemical steps. Surface azides provide such a functional group that couples with terminal 23

Chapter 2

alkynes via a copper-catalyzed azide-alkyne cycloaddition (CuAAC) reaction16,17 through formation of a 1,2,3-triazole linker, a “click” reaction.18 The mild conditions of the CuAAC reaction tolerate many functional groups, including those in protein and nucleic acids, allowing a wide range of species to be coupled to azide-modified surfaces through attachment of a terminal alkyne group.19,20 This strategy has been applied to carbon materials by using iodine azide (IN3) solutions to produce azide- modified surfaces.21,22

Presented here is an alternative, simple, scalable, and chemically specific method for the highly reproducible azide functionalization of glassy carbon surfaces using IN3 in the gas-phase. The procedure allows for IN3(g) to be prepared in-line safely and directly from commercially available chemicals along with minimizing waste and providing greater control of the reaction environment to reduce deleterious side reactions compared to solution iodine azide functionalization methodologies.

Glassy carbon was chosen as a representative carbon material because of its favorable properties for electrochemical and surface analyses. These azide-modified surfaces can be coupled with terminal alkynes to form robust triazole linkers to the surface that are hydrolytically stable at elevated temperatures in both highly acidic and basic aqueous solutions.

2.2 Results

2.2.1 Gas-phase Procedure for Azide Functionalization. As illustrated in

Figure 2.1, polished glassy carbon disks were first hydrogenated under forming gas

(5% H2/N2) at 1000 °C. After 90 min, the furnace was cooled to room temperature.

ICl(l) was dispensed onto a bed of glass wool in the upper chamber of a flow reactor 24

Chapter 2

and its vapors were carried by N2(g) through a column of NaN3(s) dispersed on silica to form IN3(g) in the effluent gas stream. IN3(g) was identified by characteristic IR

-1 23 24 (2055 and 2065 cm ) and UV/vis (320 and 243 nm) absorbances along with I2(g)

(530 nm);25 no other species in appreciable quantities were detected (Figures 2.2 and

26 2.3). This gas steam containing IN3(g) was passed over the hydrogenated carbon surfaces in the dark. The furnace tube was then purged with N2(g) before the glassy carbon was removed for analysis or further surface modification.

Figure 2.1. Gas-phase azide functionalization of carbon occurs in a two-step procedure: (1) Hydrogenation of carbon surfaces under forming gas. (2) Azide functionalization using IN3(g).

Figure 2.2. Integrated IN3(g) absorbance of the effluent gas stream through a 10 cm path length. Insert shows 2055 and 2065 cm-1 IR absorption peaks characteristic of IN3(g) (shaded).

Chapter 2

Figure 2.3. UV/vis absorbance of the effluent gas stream through a 10 cm path length. Insert shows UV/vis spectrum at maximum IN3(g) absorbance.

Treatment of hydrogenated carbon with this gas stream results in the addition of nitrogen and iodine to the surface as shown by X-ray photoelectron spectroscopy

(XPS) (Figure 2.4). Atomic concentrations obtained from these spectra were highly reproducible between preparation cycles (Table 2.1, entry 2). Two peaks with binding energies of 404 and 400 eV appear in the XPS N 1s region (Figure 2.5a). An azide group is expected to have two peaks with a ratio of 1:2 in this region; the central nitrogen exhibits a higher binding energy than the two outer nitrogens.27 A two- component fit suffices to model these peaks, one for the distinct higher binding energy central azide nitrogen and the other representative of the outer two nitrogens of the azide plus other nitrogen species. The higher energy component was fixed at 403.9 eV with a width of 2.0 eV, while the other component was left unconstrained. Fitting the

N 1s region (Figure 2.5a) for gas-phase azide-modified surfaces results in a ratio of

1:2.0, consistent with azides as the sole nitrogen species (Table 2.1, entry 2).

Chapter 2

Figure 2.4. XPS spectra of (a) hydrogenated and (b) gas-phase azide–modified glassy carbon.

Table 2.1. XPS Surface Atomic Concentrations of Carbon Samples after Indicated Procedure. entry procedure C at. %a N at. %a Azide at. %a O at. %a I at. %a Cl at. %a 1 hydrogenation 98(2) 0.0(3) 0.0(1) 2(1) 0.03(3) 0.00(5) 2 std. gas-phase 93.6(5) 5.1(2) 5.1(3) 0.9(4) 0.41(8) 0.00(5) 3 std. w/o ICl 98.4 0.0 0.0 1.4 0.02 0.00

4 std. w/o NaN3 97.6 0.0 0.0 1.1 0.39 0.97 5 std. w/o drying silica 98.8 <0.1 0.0 0.9 0.33 0.00 6 std. w/o drying N2(g) 95.0 3.5 3.7 1.1 0.44 0.00 7 std. w/ 10% O2/Ar(g) 94.3(4) 4.5(4) 4.5(4) 0.9(4) 0.32(8) 0.00(5) 8 std. w/ light 93.8(6) 4.9(2) 4.9(3) 0.9(4) 0.40(8) 0.00(5)

9 I2(g) 98.2 0.0 0.0 1.5 0.30 0.00 10 hexane 90(2) 6.5(4) 6.0(4) 2(1) 0.6(3) 0.33(7) 11 acetonitrile 91(2) 5.1(4) 3.9(4) 3(1) 0.4(3) 0.6(1) aAveraged value from three or more samples. Uncertainty of last significant figure, located in parentheses, is reported at 95% confidence interval with three or more independent data sets.

Chapter 2

Figure 2.5. XPS spectra of N 1s region with fitted spectrum and residuals after subtraction of Shirley background for (a) gas-phase azide-modified surfaces and (b) ferrocene-modified surfaces with an insert of the Fe 2p region. Dotted lines are individual fitting components.

The IN3 gas-phase procedure was varied to determine factors critical for reproducible preparations. Omission of either ICl(l) or NaN3(s) resulted in no formation of IN3(g) and no nitrogen addition to the carbon surface (Table 2.1, entries 3

28 and 4). In the omission of NaN3(s), ICl(g) (467 nm) was observed by UV/vis in the effluent gas stream, which correlated to the addition of iodine and chlorine to the carbon surfaces (Table 2.1, entry 4). The absence of a peak at 467 nm in the effluent gas stream during the standard gas-phase procedure suggests complete consumption of

ICl. Minimization of water, including the use of dried N2 carrier gas, is critical for the reproducibility of this gas-phase procedure. Hydrated NaN3(s) dispersed on silica

(~5% water by mass) led to surfaces with less than 0.1 atomic % nitrogen (Table 2.1, entry 5). Without drying the N2 carrier gas, the IN3(g) concentration by UV/Vis and the surface nitrogen concentration (Table 2.1, entry 6) are reduced. Replacing the N2 carrier gas with dried 10% O2/Ar decreases the surface nitrogen concentration to a

Chapter 2

small but statistically significant extent (Table 2.1, entry 7). Performing the gas-phase procedure under ambient lighting produced samples with similar elemental composition as in the dark (Table 2.1, entry 8). Because I2(g) was observed in the carrier gas stream, reduced glassy carbon was exposed independently to an I2(g) saturated nitrogen stream. This exposure to I2(g) resulted in iodine addition to the surface (Table 2.1, entry 9) with a concentration similar to that of the standard gas-phase procedure.

2.2.2 Solution Procedure for Azide Functionalization. Azide-modified glassy carbon surfaces were also prepared using previously reported solution

21,22 methodologies. Hydrogenated carbon surfaces react with IN3 dissolved in hexane or acetonitrile, generated in situ from NaN3(s) and ICl(l). Hexane solutions produced surfaces with a higher azide surface concentration (6.0 atomic %), compared to the gas-phase procedure (5.1 atomic %), but contained some non-azide nitrogen species

(0.5 atomic %) (Table 2.1, entry 10). Acetonitrile solutions produced the lowest surface azide concentration (3.9 atomic %) as well as significantly more non-azide nitrogen species (1.2 atomic %) (Table 2.1, entry 11). Measurable chlorine and higher oxygen concentrations were also observed on surfaces from solution preparations.

2.2.3 Surface Immobilization of Ethynylferrocene. Gas-phase azide- modified surfaces were treated overnight with 1.0 mM ethynylferrocene (1a), 1.0 mM tris(ethylacetyltriazolyl)methylamine-Cu(II) nitrate, and 2.0 mM ascorbic acid in a

N2(g)-sparged 1:1 vol:vol mixture of dimethyl sulfoxide (DMSO) and H2O to form a

1,2,3-triazole linkage between ferrocene and the surface (2a) through a CuAAC reaction. These surfaces were cleaned by sonication in DMSO. Fitting the XPS N 1s 29

Chapter 2

region separates the azide and non-azide nitrogen concentrations (Figure 2.5b). The stoichiometric 3:1 ratio of non-azide nitrogen to iron is consistent with formation of triazole linkage anticipated for 2a (Table 2.2, entry 1). However, during the CuAAC reaction, loss of approximately half of the surface nitrogen occurred concurrently with triazole formation. Because all non-azide nitrogen is accounted for by triazole formation, this nitrogen loss is attributed to the loss of azide groups (N3) from the surface. Azide loss was also observed during aqueous control experiments in

1:1 DMSO:H2O the azide concentration decreased from 5.1 ± 0.3 to 2.6 ± 0.5 atomic % in 2 h and was stable thereafter; no non-azide species were observed. In addition, the surface oxygen concentration increases from 1.1 ± 0.4 to 2.5 ± 0.7 atomic % over this time. In a separate control experiment, no changes in nitrogen or oxygen concentrations were observed for azide-modified surfaces under dry N2(g) overnight. Loss of approximately half the surface nitrogen also occurred under reducing conditions with 1.0 mM tris(ethylacetyltriazolyl)methylamine-Cu(II) nitrate and 2.0 mM ascorbic acid in 1:1 DMSO:H2O without an alkyne present. Similar to the previous control in DMSO:H2O, this loss occurred over the initial 2 h and was stable thereafter. However, in contrast to the previous control, non-azide nitrogen species

(1.2 atomic %) were observed along with unreacted azides (1.4 atomic %).

Table 2.2. XPS Atomic Concentrations for Ferrocene-Modified Surfaces (2a). coveragea,b entry azide prep. C at. %a N at. %a azide at. %a O at. %a Fe at. %a molecules/cm2 1 gas-phase 94(1) 2.2(4) 1.0(3) 3.6(8) 0.37(4) 8(1) × 1013 2 hexane 93(1) 2.5(5) 1.3(6) 3.8(6) 0.27(8) 6(1) × 1013 3 acetonitrile 94(1) 1.8(2) 0.7(5) 4(1) 0.22(5) 3(1) × 1013 aAveraged value from three or more samples. Uncertainty of last significant figure, located in parentheses, is reported at 95% confidence interval with three or more independent data sets. bCoverage determined by electrochemical methods. 30

Chapter 2

Figure 2.6. Surface Immobilization on Carbon.

Cyclic voltammetry of the ferrocene-modified carbon surfaces (2a) in 1 M

0 HClO4 displays a reversible peak at E = 510 ± 10 mV vs NHE (Figure 2.7). The peak current varies linearly with scan rates from 0.5 to 10 V/s, consistent with surface- immobilized ferrocene. Electrochemical surface coverage of 8 ± 1×1013 ferrocene/cm2 was obtained (Figure 2.7). The 170 ± 10 mV full-width at half-maximum of the cathodic peak, nearly twice the ideal Nernstian width, suggests heterogeneity or interaction among the ferrocene sites.29 A standard electron-transfer rate of greater than 1000 s-1 is inferred from the 15 mV peak-to-peak splitting at a scan rate of

10 V/s.30

Coupling of ethynylferrocene (1a) onto azide-modified surfaces prepared from

IN3 in solution led to statistically significantly lower ferrocene coverage as determined by both XPS and electrochemical measurements (Table 2.2, entries 2 and 3).31 This is counterintuitive given the higher initial surface azide concentration from IN3 in hexane solutions and is suggestive that gas-phase azide-modified surfaces have a greater propensity toward triazole formation.

Chapter 2

Figure 2.7. Cyclic voltammetry for gas-phase ferrocene-modified surface. Integration shown by shaded region.

2.2.4 Stability of the Immobilized Species. Under normal laboratory handling conditions, no nitrogen or iron loss was observed for ferrocene-modified surfaces (2a).

To investigate the stability of immobilized species under harsher conditions, two fluorinated species, heptafluoropent-1-yne (1b) and 4-ethynyl-α,α,α-trifluorotoluene

(1c), were immobilized on gas-phase azide-modified surfaces in a manner similar to that for ethynylferrocene (1a). These species are expected to be more robust than ferrocene while providing a high sensitivity XPS fluorine tag. The F 1s peak at 688 eV confirms immobilization of these species with fluorine concentrations of 2.4 ± 0.3 atomic % for heptafluoropropyl-modified surfaces (2b) and 0.9 ± 0.1 atomic % for trifluorotoluene-modified surfaces (2c) relative to all other surface atoms. As observed with ferrocene-modified surfaces (2a), nitrogen was lost during the CuAAC reactions for 2b and 2c. Non-azide nitrogen concentrations of 1.2 ± 0.2 atomic % and 0.9 ± 0.2 atomic % for 2b and 2c surfaces, respectively, are as expected from measured surface fluorine and primarily attributed to triazole formation.

Chapter 2

The fluorine, azide nitrogen, and non-azide nitrogen surface concentrations in both 1 M HClO4 and 1 M NaOH at 100 °C over a 12 h period were monitored by XPS.

No perceivable change in non-azide nitrogen occurs for heptafluoropropyl-modified surfaces (2b) under acidic or basic conditions. However, fluorine and azide nitrogen are reduced (Figure 2.8 a,b). Retention of fluorine is substantially greater under acidic than under basic conditions. Similar stability of surface species is observed for trifluorotoluene-modified surfaces (2c) where non-azide nitrogen is conserved but fluorine and azides are lost (Figure 2.8 c,d). 2c surfaces show a greater percent loss of fluorine than 2b surfaces but no differentiation of fluorine stability under acidic or basic conditions.

Figure 2.8. Stability of heptafluoropropyl-modified (2b) and trifluorotoluene- modified surfaces (2c) in both 1 M HClO4 and 1 M NaOH at 100 °C measured by XPS. Error bars at 95% confidence interval.

2.3 Discussion

32,33 Azide functionalization using IN3 has been studied extensively in solution, but its gas-phase reactivity has not previously been demonstrated. This may partially be due to the explosive dangers from the literature-reported preparation of gaseous IN3 33

Chapter 2

34,35 using AgN3(s), a very reactive azide reagent. In contrast, contemporary solution

36 methodologies pioneered by Hassner use the in situ preparation of IN3 from ICl(l) and the much more stable azide reagent, NaN3(s). The work here reports a convenient procedure for the preparation of gaseous IN3(g) by flowing ICl vapors over a column of NaN3(s) dispersed on silica. This procedure mitigates explosive risks by (i) using the more stable NaN3(s) reagent, (ii) minimizing the amount of IN3(g) prepared,

(iii) limiting the rate of formation of IN3(g) by capitalizing on the low vapor pressure and rate of delivery of ICl(l) at room temperature, and (iv) diluting the prepared IN3(g) in a carrier gas stream. With these precautions in place, IN3(g) was prepared safely and used in the gas-phase azide functionalization of carbon surfaces.

Gas-phase reactivity of IN3(g) is inferred to proceed through a radical pathway by analogy to its solution reactivity, which favors radical pathways in low dielectric

36 environments. Reported radical reactivity of IN3 in solution includes radical addition to olefins36 and radical substitution of benzylic hydrogens.37 Radical substitution reactivity would account for greater incorporation of azide than iodine, as is observed for azide-modified carbon in this study. In addition, much of the observed surface iodine concentration may be from the reaction of the I2 byproduct with the reduced surface, as seen from the I2(g) control experiment (Table 2.1, entry 9).

The high selectivity and reproducibility of the gas-phase azide functionalization of glassy carbon, which produces surfaces containing azides as the sole nitrogen species, may stem from the ease with which the reaction environment can be controlled to reduce undesired side reactions. Decomposition of IN3(g) by water35 is found to be detrimental to obtaining reproducibly functionalized surfaces. 34

Chapter 2

Thoroughly drying the N2 carrier gas and the column of NaN3(s) dispersed on silica is essential. Additionally, protection of the hydrogenated carbon surfaces from atmospheric oxygenation2 before azide functionalization is important. Gas-phase azide-modified surfaces have lower measurable oxygen concentrations than either solution azide-modified surfaces or hydrogenated surfaces after atmospheric exposure, suggesting that surface sites susceptible to oxygenation are instead blocked by azide functionalization. Because azide functionalization is observed to be at most slightly sensitive to dioxygen when dried 10% O2/Ar is used as the carrier gas, it can be inferred that exclusion of water, not O2(g), is necessary to prevent surface oxygenation prior to azide functionalization. Another valuable aspect of this gas-phase procedure is the complete reaction of ICl as it moves through the column of NaN3(s), which prevents subsequent competition of ICl with IN3 for sites on the hydrogenated carbon surface.

One use of azide-modified surfaces is as an effective handle for further surface functionalization by coupling to a variety of terminal alkyne species under the mild

CuAAC reaction. Ferrocene-modified surfaces (2a) had coverages of 8 ± 1×1013 ferrocene/cm2, approximately a third of monolayer for densely packed ferrocene.38

Similar coverage was achieved for heptafluoropropyl-modified (2b) and trifluorotoluene-modified (2c) surfaces as determined by XPS. An initial azide coverage before the CuAAC reaction of 3-4 • 1014 azides cm-2 is extrapolated from the electrochemical coverage and XPS concentrations of ferrocene-modified surfaces.

This azide coverage corresponds to a quarter of a densely packed monolayer on the

Chapter 2

zigzag edge of graphite. In actuality, because of the surface roughness, this is an overestimate and less than a quarter of a monolayer is actually azide functionalized.

Quantitative XPS analysis and modeling allows for the fate of surface azides to be followed through chemical transformations. During the CuAAC reaction and

DMSO:H2O control, azide loss was observed. This loss is postulated to occur by hydrolysis of the azide group from the surface because no evidence for other loss mechanisms, such as N2(g) loss upon azide reduction that should leave non-azide nitrogen on the surface, was observed. Hydrolysis of the azide group is also consistent with the measured increase in the surface oxygen concentration. The fact that approximately half of the surface azides of gas-phase modified surfaces are lost under aqueous conditions, while the remainder is stable, demonstrates a heterogeneous azide population. The origin of this azide heterogeneity is presumably a reflection of the underlying heterogeneity of the glassy carbon surface.

Azide reduction is inferred to occur under Cu(I) conditions without an alkyne present because nitrogen loss is accompanied by an increase in non-azide nitrogen species. These observations are consistent with the reduction of azide to an amine and

39 loss of N2, which is known to occur for various azides using a Cu(I) reductant. As the entirety of nitrogen loss under these Cu(I) conditions can be accounted for by the

N2 loss upon azide reduction, reduction must occur faster than hydrolysis. The apparent lack of this reduction reaction during the CuAAC reaction suggests that interactions between the Cu(I) complex and an alkyne species disfavor the reductive pathway.

Chapter 2

The triazole linker is stable toward hydrolytic cleavage in both 1 M acid and

1 M base for at least 12 h at 100 °C because no change was observed in non-azide nitrogen by XPS. However, under these harsh conditions, fluorine loss was observed for 2b and 2c surfaces. Fluorine loss is proposed to occur through hydrolysis of the

CF2 group alpha to the triazole; C–F bonds alpha to aromatic systems are known to undergo hydrolysis under strongly acidic and basic conditions.40 C–F bond hydrolysis is observed to extend along the aliphatic CF2 chains of the 2b surfaces, consistent with literature precedent.40

2.4 Conclusion

In summary, the use of gaseous IN3 under a controlled environment allows for the simple, scalable, reproducible, and chemically specific azide functionalization of carbon. The IN3(g) reagent can be safely prepared in-line from commercially available chemicals for immediate use in this gas-phase procedure. The controlled nature of this procedure reduces undesirable side reactions to produce azide-modified surfaces that contain no other nitrogen species. These azide-modified surfaces provide a handle for further functionalization by the coupling with terminal alkynes in a CuAAC reaction to create a hydrolytically robust triazole linker. The coverages of such immobilized species are improved relative to previous azide functionalization methods.

This procedure has been expanded beyond glassy carbon for the azide functionalization of other carbon materials, including several carbon black powders.

These azide-modified materials are versatile substrates that are being modified for the rational design of functional materials, including the development of electrocatalytic surfaces. 37

Chapter 2

2.5 Experimental Section

2.5.1 General Remarks. All chemicals were used as received unless otherwise stated. Hexane, acetonitrile (HPLC grade), dimethyl sulfoxide (BioReagents), ethanol

(200 proof), ethynylferrocene, sodium hydroxide, perchloric acid, and iodine monochloride were obtained from Fisher Scientific. Copper(II) nitrate was obtained from Strem Chemicals, sodium azide from TCI America, L-(+)-ascorbic acid from

J.T. Baker, 3,3,4,4,5,5,5-Heptafluoropentyne from SynQuest, 4-Ethynyl-α,α,α- trifluorotoluene from Sigma-Aldrich, and silia-P flash silica gel from Silicycle. All water used was deionized. Tris-(ethylacetyltriazolyl)methylamine (TTMA) was

41-43 prepared by a literature procedure. All gases were obtained from Praxair. N2 and

10% O2/Ar were dried by passing over a drierite/5 Å molecular sieves column.

Sigradur G glassy carbon disks with a 0.195 cm2 geometric surface area were obtained from HTW Hochtemperatur-Werkstoffe GmbH. Wet/dry silicon carbide sandpaper

600 and 3000 grit were obtained from Partsmaster and USA1 Abrasives, respectively.

The furnace tube is made of fused silica.

2.5.2 Analytical Instrumentation. X-ray photoelectron spectroscopy (XPS) was carried out on a Physical Electronics PHI 5000 Versaprobe spectrometer with Al

Kα radiation (1486 eV) at an angle of 45°. The CasaXPS software package was used for analysis and fitting of the spectra. Elemental composition was determined by the integration of all chemical species detected by XPS using the parameters for the sensitivity factors and starting/ending points with an averaging width of ±2.5 eV for the Shirley baseline in Table 2.3.

Chapter 2

Table 2.3. XPS Sensitivity and Fitting Parameters. region sensitivity start / eV end / eV C 1s 0.314 297 279 N 1s 0.499 410 394 O 1s 0.733 538 525 I 3d5/2 6.302 625.6 613.6 Cl 2p 0.954 206 193 Fe 2p3/2 1.964 713 703 F 1s 1.000 693 683

Electrochemical measurements were recorded on a CV50W with a Pt counter electrode and an Ag/AgCl/KCl(sat) reference electrode. A Teflon cell with a face seal having an area of 0.1237 cm2 was used to analyze the glassy carbon disks. The

CV50W software package was used to record data. The reference was calibrated against both an Accumet Glass Body Ag/AgCl/4.0 M KCl(aq) electrode and an

Accumet Glass Body saturated calomel electrode (SCE). All reported values are referenced to the normal hydrogen electrode (NHE). The aqueous electrolyte solution used for all electrochemical measurements was 1 M HClO4.

2.5.3 Preparation of Glassy Carbon. Prior to each use, the carbon surfaces were prepared by sanding with 600 grit sandpaper for 30 s followed by polishing with

3000 grit sandpaper until surfaces appeared unmarred, approximately 90-120 s. Disks were cleaned by rinsing with water followed by sonication in water for 30 s to remove loose particles and dried in a stream of N2(g).

2.5.4 Preparation of NaN3(s) Dispersed on Silica. A 3.0 g of NaN3(s) was dissolved in 100 mL of 60% EtOH:H2O solution. A 9.25 g amount of silica was added

(5 mmol NaN3/g silica) and evaporated to a free-flowing powder under vacuum at

40 °C. The resultant NaN3(s) dispersed on silica was further dried at 120 °C under vacuum overnight, losing an additional 5% weight.

Chapter 2

2.5.5 Standard Gas-Phase Azide Functionalization Procedure. Warning:

Azides are explosive in nature and proper precautionary steps should be undertaken.

This gas-phase procedure was designed to mitigate the risks involved with the preparation and use of azide reagents. Care should be taken that the furnace is plumbed in stainless steel so that explosive metal azides do not build up as may occur in copper or lead plumbing.44

The polished glassy carbon disks were placed into a furnace and purged with forming gas (5% H2 /N2) at 2 L per min (LPM) for 5 min. Under a flow of 2 LPM of forming gas, the furnace was heated to 1000 °C for 90 min to hydrogenate the carbon surfaces and then the furnace was cooled to room temperature over 1 h. A 0.5 g amount of NaN3(s) dispersed on silica (5 mmol/g) was poured freely into the stainless steel lower chamber of a flow reactor, which was purged with dried N2(g) for 10 min at 1 LPM. During this purging the flow reactor was heated to remove trace water. The reactor was placed in-line with the furnace at a N2(g) flow rate of 1 LPM and 10 µL of

ICl(l) was injected into the upper chamber under positive N2(g) pressure through an access port. ICl vapors were carried by N2(g) and reacted with NaN3(s) dispersed on silica to form IN3(g). The gas stream containing IN3(g) in turn reacts with the hydrogenate glassy carbon surfaces. Ten min after the injection of ICl(l), the furnace was purged at 2 LPM with N2(g) for 1 min to remove any remaining IN3(g) before removing the azide-modified glassy carbon. The functionalized glassy carbon was analyzed or used immediately.

2.5.6 Flow Reactor Design. The flow reactor consists of a ¼ in. stainless steel upper and lower chamber connected by a 1/16 in. restriction (Figure 2.1). The lower 40

Chapter 2

chamber is 75 mm tall with a plug of glass wool supporting the NaN3(s) dispersed on silica. The upper chamber contains a small bed of glass wool to absorb ICl(l) and is connected directly to a capped Swagelok t-joint to allow injection of ICl(l) under positive pressure.

2.5.7 N 1s Peak Fitting. The N 1s region was fit with two components. A higher energy component fixed at 403.9 eV with a full width at half-maximum of

2.0 eV and an unconstrained component around 400 eV. The parameters for the higher energy component were determined by fitting gas-phase azide-modified surfaces with a constrained two- or three- peak model with fixed areas to represent the distinct nitrogens of an azide.

Using the integrated area of the higher energy component and the total integrated nitrogen area the surface atomic concentration of azide can be calculated by:

zide at. at.

The uncertainty in peak fitting was determined using Monte Carlo simulations in CasaXPS and is reported at the 95% confidence level. The uncertainty in the surface atomic concentration of azide is then given by the formula:

azide at. at.

2.5.8 Solution Azide Functionalization Procedure. The polished carbon disks were placed into a furnace and purged with forming gas (5% H2 /N2) at 2 LPM for 5 min. Under a flow of 2 LPM of forming gas, the furnace was heated to 1000 °C 41

Chapter 2

for 90 min to hydrogenate the carbon surfaces, and then the furnace was cooled to room temperature over 1 h. Preparation of IN3 solutions was adapted from literature

21,22 procedures. A 10 μL amount of ICl(l) was added to 0.1 g of NaN3 in 10 mL of hexanes or dry acetonitrile. Significant solid, unreacted NaN3 and NaCl product were left undissolved at the bottom of the vial. These solutions were used immediately.

Hydrogenated glassy carbon disks were placed into the solution for 1 h before being removed and rinsed with methanol.

2.5.9 Ferrocene-Modified (2a), Heptafluoropropyl-Modified (2b), and

Trifluorotoluene-Modified (2c) Surfaces. An appropriate mass of the terminal-alkyne (1a-c) was dissolved in 5 mL of DMSO to produce a 2 mM stock solution. Five milliliters of a 2.0 mM tris(ethylacetyltriazolyl)methylamine-Cu(II) nitrate aqueous solution, prepared in a 1:1 molar ratio of tris(ethylacetyltriazolyl)methylamine and Cu(II) nitrate, was added to the DMSO solution. The 1:1 DMSO:H2O solution was sparged with N2(g) for 10 min. A 100 µl amount of a 200 mM aqueous ascorbic acid solution was added to the DMSO:H2O solution. Azide-modified glassy carbon was immersed in this solution for 14 h at room temperature before being cleaned by sonication for 10 min in DMSO followed by rinsing with water and drying in a stream of N2(g).

2.5.10 Stability of Surface Species. Heptafluoropropyl-modified (2b) and trifluorotoluene-modified (2c) surfaces were placed in both 1 M HClO4 and 1 M

NaOH for 0 to 12 h at 100 °C. At set time points, disks were removed and rinsed with water before being analyzed by XPS.

Chapter 2

2.6 Acknowledgments

This work was supported by the US Department of Energy, Office of Basic

Energy Sciences, as part of the Center for Electrocatalysis, Transport Phenomena and

Materials for Innovative Energy Storage (CETM) Energy Frontier Research Center

(EFRC) project number DE-SC00001055.

2.7 References

(1) Liang, C.; Li, Z.; Dai, S. Angew. Chem., Int. Ed. 2008, 47, 3696-3717.

(2) McCreery, R. L. Chem. Rev. 2008, 108, 2646-2687.

(3) Yang, W.; Auciello, O.; Butler, J. E.; Cai, W.; Carlisle, J. A.; Gerbi, J. E.;

Gruen, D. M.; Knickerbocker, T.; Lasseter, T. L.; Russell, J. N.; Smith, L. M.;

Hamers, R. J. Nat. Mater. 2002, 1, 253-257.

(4) Lasseter, T. L.; Clare, B. H.; Abbott, N. L.; Hamers, R. J. J. Am. Chem. Soc.

2004, 126, 10220-10221.

(5) Downard, A. J. Electroanalysis 2000, 12, 1085-1096.

(6) Barriere, F.; Downard, A. J. J. Solid State Electrochem. 2008, 12, 1231-1244.

(7) Pinson, J.; Podvorica, F. Chem. Soc. Rev. 2005, 34, 429-439.

(8) Nielsen, L. T.; Vase, K. H.; Dong, M.; Besenbacher, F.; Pedersen, S. U.;

Daasbjerg, K. J. Am. Chem. Soc. 2007, 129, 1888-1889.

(9) Combellas, C.; Jiang, D. E.; Kanoufi, F.; Pinson, J.; Podvorica, F. Langmuir

2009, 25, 286-293.

(10) Leroux, Y. R.; Fei, H.; Noel, J. M.; Roux, C.; Hapiot, P. J. Am. Chem. Soc.

2010, 132, 14039-14041.

Chapter 2

(11) Kongsfelt, M.; Vinther, J.; Malmos, K.; Ceccato, M.; Torbensen, K.; Knudsen,

C. S.; Gothelf, K. V.; Pedersen, S. U.; Daasbjerg, K. J. Am. Chem. Soc. 2011,

133, 3788-3791.

(12) Downard, A. J.; Garrett, D. J.; Tan, E. S. Langmuir 2006, 22, 10739-10746.

(13) Ssenyange, S.; Anariba, F.; Bocian, D. F.; McCreery, R. L. Langmuir 2005,

21, 11105-11112.

(14) Sun, B.; Colavita, P. E.; Kim, H.; Lockett, M.; Marcus, M. S.; Smith, L. M.;

Hamers, R. J. Langmuir 2006, 22, 9598-9605.

(15) Yu, S. S.; Downard, A. J. Langmuir 2007, 23, 4662-4668.

(16) Tornoe, C. W.; Christensen, C.; Meldal, M. J. Org. Chem. 2002, 67, 3057-

3064.

(17) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B. Angew. Chem.,

Int. Ed. 2002, 41, 2596-2599.

(18) Kolb, H. C.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed. 2001, 40,

2004-2021.

(19) Evans, R. A. Aust. J. Chem. 2007, 60, 384-395.

(20) Iha, R. K.; Wooley, K. L.; Nystrom, A. M.; Burke, D. J.; Kade, M. J.; Hawker,

C. J. Chem. Rev. 2009, 109, 5620-5686.

(21) McCrory, C. C. L.; Devadoss, A.; Ottenwaelder, X.; Lowe, R. D.; Stack, T. D.

P.; Chidsey, C. E. D. J. Am. Chem. Soc. 2011, 133, 3696–3699.

(22) Devadoss, A.; Chidsey, C. E. D. J. Am. Chem. Soc. 2007, 129, 5370-5371.

(23) Schulz, A.; Tornieporth-Oetting, I. C.; Klapoetke, T. M. Inorg. Chem. 1995,

34, 4343-4346. 44

Chapter 2

(24) Engelhardt, U.; Feuerhahn, M.; Minkwitz, R. Z. Anorg. Allg. Chem. 1978, 440,

210-216.

(25) Tellinghuisen, J. J. Chem. Phys. 1973, 58, 2821-2834.

(26) Variance in IN3(g) onset for the IR and UV/vis gas-cells were seen and

postulated to be dependent on laboratory temperature and NaN3(s) dispersed

on silica column packing but was not further investigated.

(27) Collman, J. P.; Devaraj, N. K.; Eberspacher, T. P. A.; Chidsey, C. E. D.

Langmuir 2006, 22, 2457-2464.

(28) Qu, Z.; Yan, N.; Liu, P.; Jia, J.; Yang, S. J. Hazard. Mater. 2010, 183, 132-

137.

(29) Bard, A. J.; Faulkner, L. R. Electrochemical Methods: Fundamentals and

Applications; 2nd ed.; John Wiley & Sons, Inc.: New York, 2001.

(30) Laviron, E. J. Electroanal. Chem. 1979, 101, 19-28.

(31) This lower ferrocene coverage is not due to differences in azide loss; azide

concentrations of acetonitrile and hexane azide-modified surfaces decreased

from 3.9 ± 0.4 to 3.7 ± 0.4 atomic % and 6.0 ± 0.4 to 4.1 ± 0.5 atomic % in

DMSO:H2O over 2 h respectively, significantly less than for gas-phase azide-

modified surfaces.

(32) Dehnicke, K. Angew. Chem., Int. Ed. 1979, 18, 507-514.

(33) Bräse, S.; Gil, C.; Knepper, K.; Zimmermann, V. Angew. Chem., Int. Ed. 2005,

44, 5188-5240.

(34) Hargittai, M.; Molnar, J.; Klapoetke, T. M.; Tornieporth-Oetting, I. C.;

Kolonits, M.; Hargittai, I. J. Phys. Chem. 1994, 98, 10095-10097. 45

Chapter 2

(35) Munz, H. J. Mol. Struct. 2004, 695, 189-202.

(36) Hassner, A. Acc. Chem. Res. 1971, 4, 9-16.

(37) Viuf, C.; Bols, M. Angew. Chem., Int. Ed. 2001, 40, 623-625.

(38) Chidsey, C. E. D.; Bertozzi, C. R.; Putvinski, T. M.; Mujsce, A. M. J. Am.

Chem. Soc. 1990, 112, 4301-4306.

(39) Xia, Y.; Li, W.; Qu, F.; Fan, Z.; Liu, X.; Berro, C.; Rauzy, E.; Peng, L. Org.

Biomol. Chem. 2007, 5, 1695-1701.

(40) Hudlicky, M. Chemistry of Organic Fluorine Compounds: a Laboratory

Manual with Comprehensive Literature Coverage, 2nd ed.; Ellis Horwood:

New York, 1992.

(41) Zhou, X.; Chang, Y.-C.; Oyama, T.; McGuire, M. J.; Brown, K. C. J. Am.

Chem. Soc. 2004, 126, 15656-15657.

(42) Chan, T. R.; Hilgraf, R.; Sharpless, K. B.; Fokin, V. V. Org. Lett. 2004, 6,

2853-2855.

(43) Wang, Q.; Chan, T. R.; Hilgraf, R.; Fokin, V. V.; Sharpless, K. B.; Finn, M. G.

J. Am. Chem. Soc. 2003, 125, 3192-3193.

(44) CRC Handbook of Chemistry and Physics, 92nd ed.; CRC Press: Boca Raton,

FL, 2011/2012.

Chapter 3

Gas-Phase Azide Functionalization of High

Surface Area Carbon

Chapter 3

List of Abbreviations

At. % Atomic percent

CuAAC Copper-catalyzed azide-alkyne cycloaddition

DMSO Dimethyl sulfoxide

Fc Ferrocene

IR Infrared

TTMA Tris-(ethylacetyltriazolyl)methylamine

UV/vis Ultraviolet–visible

XPS X-ray photoelectron spectroscopy

Chapter 3

Abstract

Azide functionalization of mesoporous Vulcan XC-72R carbon powder was accomplished from a gas-phase process using IN3(g). This process is selective for the direct attachment of azides to the carbon surface, minimally perturbing the pore structure. A relatively high coverage of surface azides was achieved at 1.4 ± 0.3 × 1014 azides/cm2. Furthermore, azide-modified carbon is a versatile substrate for the immobilization of additional functional groups by coupling with terminal alkynes via a copper-catalyzed azide-alkyne cycloaddition (CuAAC), a “Click” reaction. This surface chemical transformation can be detected and quantified by X-ray photoelectron spectroscopy to provide direct evidence for covalent bond formation.

The 1,2,3-triazole linker formed during the CuAAC reaction provides good electrical conductivity between the immobilized species and carbon with a standard electron transfer rate greater than 100 s-1 measured for covalently immobilized ferrocene.

3.1 Introduction

High surface area porous carbon materials are well-known for their applications in electrochemical systems, energy storage, catalyst support, chemical sorbents, and composite materials. Surface functionalization of these carbon materials using discrete molecular species allows for the introduction of desired functionality to tailor the surface and interfacial properties for specific applications.1-3 To create a chemically well-defined surface for the rational-design of functionalized materials the surface immobilization of these molecular species should (1) occur through a robust covalent linker, and (2) provide a distinct spectroscopic probe for the characterization of surface chemical transformations.

Chapter 3

This is possible for planar carbon materials through coupling terminal alkyne

4,5 species with azide-modified carbon surfaces produced using IN3 both in solution and

6 the gas-phase. Azide functionalization using IN3 is highly selective, particularly in the gas-phase, producing surface-bound azides as the sole surface nitrogen species.6

Further functionalization of these azide-modified surfaces can be accomplished by coupling with terminal alkynes via a copper-catalyzed azide-alkyne cycloaddition

(CuAAC) reaction7,8 through formation of a covalent and hydrolytically robust6

1,2,3-triazole linker. This chemical transformation is followed readily by the characteristic spectroscopic features of azide and triazole in the X-ray photoelectron spectrum (XPS).9 In addition, the CuAAC reaction is highly tolerant of a diverse set of functional groups10,11 placing minimal restrictions on chemical functionality that can be introduced.

A prevalent method for the functionalization of high surface area carbon materials is through oxidative treatment non-selectively generating various surface oxygenated functionalities.12 These surface oxygenated functionalities are chemical handles for further covalent immobilization of organic species and metal complexes3 to the surface, using established chemical reactions. Despite the prevalence of these oxidative methods, characterization to determine covalent bond formation is difficult due to lack of characterizable spectroscopic signals13 coupled with strong competing physisorption of species.14 Chloromethylated activated carbon has been used for the immobilization of small organic nucleophiles, but differentiation between covalently attached and physisorbed species is difficult.15,16

Chapter 3

Alternatively, immobilization of aryl species onto carbon powders through coupling with aryl radicals, formed spontaneously by the reduction aryl diazonium salts with some carbon materials, are patented by the Cabot Corporation.17 Aryl radicals form robust C–C bonds to the surface but also commonly led to the formation of ill-defined multilayer structures up to 15 nm thick.18-20 Multilayer structures reduced pore diameters thus slowing diffusion, which may be disadvantageous for some applications.

Presented here is the reproducible and chemically specific azide functionalization of high surface area carbon Vulcan XC-72R powder by treatment with gaseous IN3. Vulcan XC-72R was chosen as a representative high surface area mesoporous carbon material because of its prevalent use, particularly in fuel cell applications. The azide-modified XC-72R contained azides as the only nitrogen species and achieved a relatively high surface coverage of 1.4 ± 0.3 × 1014 azides/cm2.

Subsequent derivatization with terminal alkyne containing species, including metal complexes, leads to covalent immobilization through a 1,2,3-triazole linker onto the surface. This chemical transformation is followed readily by XPS to provide the first direct evidence for covalent immobilization of a metal complex onto a carbon powder.

Additionally, the 1,2,3-triazole linker provides fast electron-transfer kinetics between the immobilized metal species and the bulk carbon material.

3.2 Results

3.2.1 Gas-Phase Azide Functionalization. A bed of 150 mg Vulcan XC-72R carbon powder was hydrogenated under forming gas (5% H2/N2) at 1000 °C for

90 min then cooled to room temperature. A flow reactor containing two chambers, one

Chapter 3

a column of NaN3(s) and the other a column of glass wool (Figure 3.5), was dried under a N2(g) stream at 220 °C for 90 min. The NaN3(s) column was cooled to room temperature while the glass wool column was cooled and maintained at 35 – 40 °C.

Under a positive N2(g) pressure, 60 µL of ICl(l) was injected onto the column of glass wool. N2(g) flow through the glass wool column carries ICl vapor over the NaN3(s)

6 column to form IN3(g), identifiable by its characteristic UV absorbance at 243 and

32021 nm, in the effluent gas stream. Integration of the 320 nm peak height over the effluent gas flow volume showed a 6-fold increase in IN3(g) yield from the previously reported procedure6 (Figure 3.6), attributed to improved drying of flow reactor components. In addition to IN3(g), trace amounts of unreacted ICl(g) and I2(g) exist in the effluent gas stream. The gas stream containing IN3(g) was passed through a bed of hydrogenated Vulcan XC-72R, XC-H, to produce an azide-treated carbon powder,

XC-N3 (Figure 3.1).

Figure 3.1. Surface modification of XC-72R carbon powder.

Addition of nitrogen, iodine, and chlorine atoms to the carbon surface occurred in the preparation of XC-N3 from XC-H by XPS (Figure 3.2). Atomic concentrations obtained from these spectra showed good reproducibility between independently prepared XC-N3 samples (Table 3.1). In the XPS N 1s region, two peaks at 400 and

Chapter 3

404 eV were observed (Figure 3.3a), with the latter corresponding to the distinct central nitrogen of an azide.9 A two-component-fit is sufficient for determining the

6 mole fraction of azide nitrogen from the total nitrogen (χazide). Fitting the N 1s region for XC-N3 (Figure 3.3a) gave χazide = 1.0 ± 0.1, showing azides as the sole surface nitrogen species.

Table 3.1. Characterization of Modified XC-72R Carbon Powder. XPSa

Material C / at. % N / at. % χazide O / at.% I / at. % Cl / at. % XC-H 96 0.0 - 4 0.05 0.00 XC-N3 94(1) 3.7(3) 1.0(1) 2(1) 0.26(4) 0.22(8)

Combustion Analysisa

Material C / mass % N / mass % N3 / mmol/g XC-H 99.6(3) 0.0 0.0 XC-N3 77(2) 1.8(4) 0.4(1) aAveraged value from three or more measurements. Uncertainty of last significant figure, located in parenthesis, is reported at one standard deviation with four independent samples.

Figure 3.2. XPS spectra of (a) XC-H and (b) XC-N3 surfaces.

Chapter 3

Figure 3.3 XPS spectra of the N 1s region with fitted spectrum and residuals after subtraction of Shirley background for (a) XC-N3 and (b) XC-Fc with insert of Fe 2p region. Dotted lines are individual fitting components.

Nitrogen and carbon mass composition of 1.8 ± 0.4% and 77 ± 2%, respectively, for XC-N3 by combustion analysis is supportive of azide addition as combustion of XC-H has had no detectable nitrogen ( >99% as carbon). The majority of the remaining mass in XC-N3 was adsorbed molecular I2(s), at 16 ± 1% by mass

22 23 (Figure 3.7), a known decomposition product of IN3. As azides are the only detectable nitrogen species, 1.8 ± 0.4% nitrogen corresponds to an azide loading of

14 0.4 ± 0.1 mmol/g XC-N3 (Table 3.1). An azide coverage of 1.4 ± 0.3 × 10 azides/cm2 was calculated by dividing the amount of azides incorporated per gram carbon, determined from the mass percentage of nitrogen and carbon of XC-N3, with the surface area of unmodified XC-72R, 240 m2/g. Azide loadings up to 0.5 mmol/g were possible using a lesser amount of ICl(l) (20 µL vs. 60 µL), but this produced less reproducible results with azide loadings ranging between 0.15 – 0.5 mmol/g. This range of loadings suggests variability in the amount of IN3(g) exiting the NaN3(s) column, which appears to be compensated for by using excess ICl(l). XC-N3 was used

Chapter 3

immediately or stored at −80 °C for analysis. At room temperature under N2(g),

XC-N3 loses ~10% nitrogen over 24 hrs, measured by both XPS and combustion analysis; this nitrogen loss is attributed to loss of azide groups (N3) as χazide = 1.0 is unchanged.

3.2.2 Covalent Surface Immobilization of Ethynylferrocene. 60 mg of

XC-N3 was treated with 0.1 mmol ethynylferrocene (1), 0.02 mmol TTMA-Cu(II) nitrate (2), and 0.05 mmol ascorbic acid dissolved in 6 mL of N2(g)-sparged

2:1 vol:vol dimethyl sulfoxide (DMSO):H2O overnight to form a 1,2,3-triazole covalent linkage between ferrocene and the surface. Excess ascorbic acid was necessary to keep 2 in its catalytically active form, reduced Cu(I), in the presence of oxidizing I2(s) contained within XC-N3. This ferrocene-treated material, XC-Fc, was isolated from solution by filtration on a membrane filter to create a packed bed, which was washed thoroughly with acetonitrile, 0.1 M sodium diethyldithiocarbamate methanol solution, and methanol before being dried at 50 °C under vacuum.

XPS analysis of XC-Fc confirmed iron incorporation along with the appearance of a non-azide nitrogen species (Table 3.2). Fitting the XPS N 1s region

(Figure 3.3b) gave χazide = 0.5 ± 0.2 from which the atomic concentration of non-azide nitrogen was calculated to be 0.6 ± 0.2 atomic %. This non-azide nitrogen had a stoichiometric ratio to iron of 3.1 ± 0.4 in excellent agreement with the expected ratio of 3.0 for the formation of a covalent triazole linkage between ethynylferrocene (1) and surface azides during the CuAAC reaction (Figure 3.1). Consequently, all nitrogen species can be accounted for as either triazoles or unreacted azides.

Chapter 3

XC-Fc iron loading of 0.06 ± 0.01 mmol/g was determined from extracted samples by Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES).

Coverage of covalently attached ferrocene was 1.7 ± 0.3 × 1013 ferrocene/cm2. In control experiments where either surface azides or copper catalyst were excluded, iron loadings were 0.005 and 0.004 mmol/g respectively; iron was not detected by XPS in either control. The control samples contained a slight elevation in iron above the background levels measured in XC-N3 at 0.001 mmol/g that can be attributed to physisorbed ethynylferrocene (1). This physisorbed species was minimal and does not affect the measured iron loadings of XC-Fc within measured uncertainty. Thus, the iron loading can be attributed primarily to covalently bonded ferrocene consistent with the Fe to non-azide nitrogen stoichiometric ratio measured by XPS. Nitrogen loading was determined to be 0.48 ± 0.08 mmol/g by combustion analysis. Comparison to the ferrocene loading attributes 0.18 ± 0.03 mmol/g of the nitrogen to triazole formation during the CuAAC reaction. The remaining nitrogen is presumably unreacted azides,

χazide = 0.6 ± 0.1, consistent with nitrogen speciation determined by XPS that accounts for all nitrogen as either triazoles or azides.

Table 3.2. Characterization of XC-Fc. XPSa

N / at. % χazide Fe / at. % Cu / at. % 1.1(3) 0.5(2) 0.19(3) 0.07(4)

Combustion Analysisa ICP-OESa C / mass % N / mass % Fe / mmol g-1 Cu / mmol g-1 93(2) 0.7(1) 0.06(1) 0.005(3) aAveraged value from three or more measurements. Uncertainty of last significant figure, located in parenthesis, is reported at one standard deviation with four independent samples.

Comparison of XC-Fc to the precursor material, XC-N3, showed a 65 ± 10% decrease in the nitrogen to carbon (N/C) mass ratio by combustion analysis. A similar

Chapter 3

70 ± 10% decrease in the nitrogen atomic concentration was observed by XPS. This nitrogen loss is attributed to azide groups (N3) loss from the surface as all nitrogen species are accounted for as either triazoles or azides by XPS. Azide loss was also observed during control experiments in 2:1 DMSO:H2O where the N/C mass ratio of

XC-N3 decreased by 30% overnight in solution; azides accounted for all nitrogen species before and after the loss. Azide loss was enhanced in DMSO:H2O by the addition of 10 mM potassium iodide as the N/C mass ratio decreased 50% overnight.

Iodide, from the reduction of I2, present during the treatment of XC-N3 with ethynylferrocene presumably contributed to the observed azide loss in XC-Fc. In addition, XC-Fc also showed an increase in oxygen content to 4 ± 2 atomic % comparable the quantity of lost azide groups.

Cyclic voltammetry of a XC-Fc/Nafion film on a glassy carbon electrode in

0 1 M HClO4 showed a broad reversible feature containing three peaks at E = 540, 590, and 700 mV vs NHE (Figure 3.4), suggesting a heterogeneous chemical environment for the immobilized ferrocene species in this film. The area of this reversible feature, along with the peak currents, varied linearly with scan rates from 0.5 to 10 V/s, consistent with a surface-immobilized species. Peak-to-peak splitting averaged 80 mV at a scan rate of 10 V/s, which translates into a standard electron transfer rate of approximately 130 s-1.24 By comparison, electron transfer rates of greater than 1000 s-1 were measured for ferrocene immobilized through a triazole linker on glassy carbon,6 suggesting that the 130 s-1 rate is a bulk property of the XC-Fc/Nafion film.

Chapter 3

Figure 3.4. Cyclic voltammetry for XC-Fc with peak maximums (□) and linear background (dotted line).

3.3 Discussion

Azide functionalization of planar carbon materials including pyrolyzed photoresist, edge-plane graphite, and glassy carbon is possible using IN3 in acetonitrile4 or hexane5 solutions and without a solvent in the gas-phase.6 In all these methodologies IN3 was synthesized from the reaction between NaN3(s) and ICl(l) either in solution or in an in-line flow reactor. However, evidence of immediate IN3 decomposition was observed by the formation of an intensely dark colored species

4-6 attributable to I2. Despite this decomposition IN3 has proven to be highly selective for the azide functionalization of carbon materials,25 particularly in the gas-phase

6 producing surfaces that contain no other nitrogen species. However, IN3 decomposition makes it difficult to provide sufficient reagent necessary for the azide functionalization of high surface area carbon materials. This has prompted efforts to improve IN3 yields by limiting decomposition.

Initial attempts to optimize IN3 synthesis in hexane solution proved unsatisfactory for the azide functionalization of XC-72R. Instead, optimization of the

Chapter 3

previously published gas-phase azide functionalization process6 proved more amenable to limiting IN3 decomposition, resulting in a 6-fold increase of IN3(g) delivered to the carbon material. The decrease in IN3 decomposition is attributed primarily to improved drying of the flow reactor components to eliminate trace water,

23 known to cause decomposition. Despite these changes I2(g) is still produced, albeit with concurrent growth of IN3(g) and I2(g) rather than I2(g) preceding IN3(g) (Figure

3.6). This change in product distributions suggests the optimized synthesis eliminates any limiting reagent, such as water, that causes decomposition. A self-decomposition pathway of IN3(g) is presumably the source of I2(g) in the optimized synthesis. Further studies to limit this decomposition pathway and improve the IN3 yield are in progress.

Gas-phase reactivity of IN3(g) presumably occurs through a radical substitution

6 of activated C–H bonds as previously proposed. Similar radical substitution using IN3 in acetonitrile occurs with benzyl ethers.26 An alternative mechanism is the addition of

27 IN3(g) across an olefin, with equal addition of azide and iodine. The composition of

XC-N3 is most consistent with a radical substitution mechanism, which accounts for the greater incorporation of azide than iodine by XPS. Physisorbed I2 is not detectable by XPS as it presumably desorbs from the surface under the ultra-high vacuum conditions. As trace iodine was observed by XPS, the addition mechanism cannot be ruled out as occurring concurrently to radical substitution. Alternatively, trace iodine along with trace chlorine, may result from the reaction of ICl(g) and I2(g) with the surface; glassy carbon surfaces exhibit such reactivity.6

Highly selective and reproducible azide functionalization of XC-72R was achieved using the optimized IN3(g) synthesis. The azide-modified carbon, XC-N3,

Chapter 3

with an azide loading of 0.4 ± 0.1 mmol/g has no other detectable nitrogen species by

XPS. Typical yields of azide incorporation from limiting ICl(l) were 8 ± 2% when

60 µL of ICl(l) was used; 20 µL of ICl(l) could achieve comparable azide loadings, resulting at best a 30% azide incorporation yield, but also produced materials with a much larger range of azide loadings, between 0.15 and 0.5 mmol/g. This suggests some uncontrolled variable in the amount of IN3(g) reaching the carbon material, possibility due to IN3(g) decomposition. Use of a greater amount of ICl(l), 60 µL, limits the sensitivity in azide functionalization due to variability in IN3(g) synthesis, leading to reproducible azide functionalization of XC-72R.

14 2 Azide coverage of 1.4 ± 0.3 × 10 azides/cm on XC-N3 is a high functional group coverage, corresponding to approximately 1 azide for every 10 surface accessible carbon atoms based on the zigzag edge of graphite. This coverage is more than double that achieved by chloromethylation of activated carbon at 6 × 1013

2 15 -CH2Cl/cm . Azide coverage also compares favorably to the aryl group coverage achieved on carbon powders using aryl diazonium salts, which range between 0.1 and

2.5 × 1014 aryl groups/cm2.17,19,20 However, the aryl radical intermediate formed from the diazonium salt commonly leads to the formation of ill-defined multilayer structures.18-20 Although a multilayer structure can achieve higher surface coverage, it may be disadvantageous, particularly for porous carbon material as a multilayer buries the introduced functionality from the aryl groups and can dramatically reduce the pore diameter slowing mass diffusion through the material. In contrast, direct attachment of a functional group to the carbon surface avoids both of these potential problems.

Chapter 3

A major advantage of azide-modified materials for characterization is that azides provide a quantifiable probe to track surface chemical transformations including spectroscopic evidence of triazole bond formation upon cycloaddition with terminal alkyne species. XPS of XC-Fc, formed by the cycloaddition of ethynylferrocene with XC-N3, showed 3.1 ± 0.4 nitrogens attributable to triazole per ferrocene consistent with covalent immobilization of ferrocene onto the carbon surface with no detectable physisorbed ethynylferrocene by XPS. Bulk control experiments also suggest physisorbed ethynylferrocene to be minimal in XC-Fc (ca. 5% of total iron species). Beyond providing an extremely stable linkage to the carbon surface,6 the triazole also provides good electron-transfer kinetics between immobilized ferrocene and the carbon material. Both the stability and facile electron transfer are important factors for electrochemical device applications.

Quantification of the azide to triazole efficiency for XC-Fc showed that approximately 1/6th of the original azides cyclize to triazoles. Of the remaining azides approximately 2/3 were lost while 1/6 were unreactive. Azide losses of 50% and 30% were also observed for XC-N3 in DMSO:H2O control experiments with and without potassium iodide present. Azide loss observed for modified glassy carbon was previously ascribed to hydrolysis of the azide group.6 A hydrolysis mechanism is consistent for the observed azide loss on XC-N3 both during formation of XC-Fc and the control experiments as no evidence of other nitrogen loss mechanisms exists, such as N2(g) loss upon azide reduction. In addition the 2 atomic % increase of surface oxygen observed in XC-Fc approximately equates to the observed azide loss and may be indicative of a new carbon-oxygen bond formed during hydrolysis. Furthermore,

Chapter 3

the enhancement of azide loss seen in the presents of potassium iodide supports a nucleophilic displacement mechanism such as hydrolysis. Iodide present during formation of XC-Fc, from I2 reduction, presumably contributes to the observed azide loss.

Covalently immobilized ferrocene had a coverage of 1.6 ± 0.3 × 1013 ferrocene/cm2 on XC-Fc, approximately 10% of a densely packed monolayer.28

Previous reports attempting to covalently immobilize metal complexes on high surface area carbon materials have relied primarily on coupling with surface oxygen functionality.3 Coverage of metal complexes from these reports vary widely between

0.8 × 1013 and 13 × 1013 complexes/cm2.13,29 Direct evidence of covalent bond formation for these metal complexes through oxygen coupling reactions are difficult, due to the lack of distinct spectroscopic features.13 In addition, strong physisorption through π–π interactions between metal complexes and carbon π system are indistinguishable from covalently bonded complexes.14 These uncertainties are not seen for immobilization by the cycloaddition of terminal alkynes to the well-defined azide-modified XC-N3, as azides and triazoles can be distinguished and quantified spectroscopically to determine covalently immobilized verse physisorbed species.

3.4 Conclusion

In summary, the reproducible and chemically-specific azide functionalization of high surface area carbon Vulcan XC-72R powder was achieved by treatment with gaseous IN3. IN3(g) selectively adds azides to the carbon surface with no other nitrogen containing species detectable by XPS. Improvements in the yield of the

IN3(g) reagent reaching the carbon surface by attenuating decomposition pathways

Chapter 3

allowed for a respectable surface coverage of 1.4 ± 0.3 × 1014 azides/cm2. The reactivity of the IN3 reagent along with its prior applications towards a variety of carbon materials suggests a generalized application of this reagent towards graphite and amorphous carbon materials. Further surface functionalization of the azide-modified carbon material is possible through cycloaddition with terminal alkyne species through formation of a 1,2,3-triazole linker. This chemical transformation is followed readily spectroscopically to provide direct evidence for covalent immobilization, including the first direct evidence of covalent immobilization of a metal complex, ferrocene, onto a carbon powder. Additionally, the immobilized ferrocene can be addressed electrochemically with facile electron-transfer kinetics between the bulk carbon material and the iron redox center through the 1,2,3-triazole linker.

Further work to ascertain and limit IN3(g) decomposition pathways to optimize yield, along with improving the alkyne-azide cycloaddition efficiency by eliminating azide loss pathways, are ongoing. In addition, applications of azide-modified carbon for the rational design of functional materials are underway.

3.5 Experimental Section

3.5.1 General Remarks. All chemicals and reagents were used as received unless otherwise stated: Vulcan XC-72R (E.T. Horn), dimethyl sulfoxide

(BioReagents Fisher Scientific), methanol (Fisher Scientific), 2-propanol (Fisher

Scientific), perchloric acid (Fisher Scientific), iodine monochloride 99.998% trace metals basis (Sigma-Aldrich), nitric acid TraceMetal grade (Fisher Scientific), ethynylferrocene (Fisher Scientific), copper(II) nitrate (Strem), (+)-Sodium

Chapter 3

L-(+)-ascorbic acid (J.T. Baker), sodium diethyldithiocarbamate (Alfa Aesar), and

Liquion Nafion LQ-1015 1000 EW at 15% weight solution (Ion Power). All water used was deionized. Sodium azide (TCI America) was ground to a fine powder using a mortar and pestle, and dried under vacuum at 220 °C overnight prior to use.

(Tris((1-((O-ethyl)carboxymethyl)-(1,2,3-triazol-4-yl))methyl)amine (TTMA) was

30-32 prepared from literature procedures. All gases were obtained from Praxair. N2 was dried by passing over a drierite/5 Å molecular sieves column and then through an

OMI-2 purifier tube. Sigradur G glassy carbon disks with a 0.195 cm2 geometric surface area were obtained from HTW Hochtemperatur-Werkstoffe GmbH. The furnace tube is made of quartz with a coarse quartz frit in the middle of the tube.

3.5.2 Analytical Instrumentation. X-ray photoelectron spectroscopy (XPS) was carried out on a Physical Electronics PHI 5000 Versaprobe spectrometer with Al

N 1s peaks using a two component system, one fixed at 403.9 eV with a full width at half-maximum of 2.0 eV and an unconstrained component ca. 400 eV, as previously described.6 Inductively Coupled Plasma Spectroscopy (ICP-OES) was conducted on a

Thermo Scientific ICAP 6300 Duo View Spectrometer with a Solid State CID

Detector at the School of Earth Science, Stanford University. Elemental analysis was

Chapter 3

performed by combustion analysis on a Carlo-Erba NA 1500 analyzer at the School of

Earth Science, Stanford University.

Table 3.3. XPS Sensitivity and Fitting Parameters. Region Sensitivity Start / eV End / eV C 1s 0.314 297 279 N 1s 0.499 410 394 O 1s 0.733 538 525 I 3d5/2 6.302 625.6 613.6 Cl 2p 0.954 206 193 Fe 2p3/2 1.964 713 703 Cu 2p3/2 2.626 928 940

Electrochemical measurements were recorded on a WaveNow Potentiostat with a Pt counter electrode and an Ag/AgCl/KCl(sat) reference electrode. A Teflon cell with a face seal having an area of 0.1237 cm2 was used to analyze the glassy carbon disks. The Aftermath software package was used to record and analyze data.

The reference was calibrated against both an Accumet Glass Body

Ag/AgCl/4.0 M KCl electrode and an Accumet Glass Body saturated calomel electrode (SCE). All reported values are referenced to the normal hydrogen electrode

(NHE). The aqueous electrolyte solution used for all electrochemical measurements was 1 M HClO4.

3.5.3 Synthesis of Iodine Azide (IN3). Warning: Azides are explosive in nature and proper precautionary steps should be undertaken while handling them. This procedure was designed to mitigate the risks involved with the synthesis and use of azide reagents. Stainless steel plumbing should be used to avoid explosive metal azides build up as found in copper or lead plumbing.33

Synthesis of IN3(g) was carried out using a two chamber flow reactor as previously described6 with the following modifications for safety and optimization of

IN3(g) yield. The updated flow reactor consists of a 1/4 in. diameter stainless steel

Chapter 3

upstream chamber (35 mm) and downstream chambers (75 mm) connected by a

U-bend with an access port directly above the upstream chamber to allow injection of

ICl(l) (Figure 3.5). Prior to each run, the upstream chamber was packed with glass wool to adsorb the injected ICl(l) and the downstream chamber was loaded with 0.9 g of finely ground, dried NaN3(s). The entire reactor was dried at 220 °C for 90 min under a 50 cc/min flow of dry N2(g) vented to the atmosphere. After 90 min under a static N2(g) atmosphere the downstream chamber (NaN3) was cooled to RT while the upstream chamber was cooled and maintained at 35-40 °C. Under a positive N2(g) pressure, 60 µL of ICl(l) was injected onto the glass wool in the upstream chamber at a N2(g) flow rate of 50 cc/min. ICl(g) vapors were carried by N2(g) and reacted with

NaN3(s) to form IN3(g) in the gas stream. N2(g) was allowed to flow for 2 hr through the reactor to complete the formation of IN3(g).

Figure 3.5. Diagram of the flow reactor.

Chapter 3

Figure 3.6 UV/vis absorbance of the effluent gas through a 10 cm path length for the individual components of IN3(g) at 320 nm (blue solid line), ICl(g) at 467 nm (red long dashed line), and I2(g) at 530 nm (purple short dashed line). 20 µl of ICl(l) was injected onto the glass wool column and carried over a dried column of (a) NaN3(s) on 6 silica as previously reported, (b) NaN3(s), showing unconsumed ICl(g), and (c) finely ground NaN3(s). Optimized synthesis (c) achieved a 10-fold greater maximum concentration and 6-fold improved yield of IN3(g) over the previously report synthesis (a).

3.5.4 Synthesis of XC-N3. A fluidized bed of XC-72R (150 mg) was hydrogenated by heating at 1000 °C for 90 min under a 20 cc/min upward flow of forming gas (5% H2 /N2) in a 1/2 in. OD × 3/8 in. ID vertical quartz tube. The quartz tube contained a coarse frit to support the carbon powder. The hydrogenated XC-72R was cooled to RT over 1 hr and then exposed to a gas stream containing IN3(g) for 2

Chapter 3

hr. After exposure to IN3(g) the furnace was purged with N2(g) at 100 cc/min for 5 min. The XC-N3 was used immediately or stored at −80 °C until analysis.

3.5.5 Determination of Adsorbed I2(s) on Carbon Powder. XC-N3 (2 mg) was transferred to a volumetric flask and diluted to 10 mL with toluene to extract adsorbed I2(s). This mixture was sonicated and filtered through a 0.45 µm membrane filter before UV/vis analysis (Figure 3.7). The extracted I2 was calculated from the literature reported extinction coefficient.34 These extinction coefficients were measured independently and confirmed. A correction between the extracted I2 and total I2 was necessary as 2 mg of XC-72R adsorbed 8 ± 1% of I2 from 10 mL toluene solutions of varying I2 concentration.

Figure 3.7. UV/vis spectrum of XC-N3 in 10 mL toluene.

3.5.6 Synthesis of XC-Fc. Ethynylferrocene (21 mg, 0.1 mmol) was dissolved in 4 mL DMSO along with 2 mL of a 10.0 mM TTMA-Cu(II) nitrate aqueous solution, prepared in a 1:1 molar ratio of TTMA to Cu(II) nitrate. The 2:1 DMSO:H2O solution was sparged with N2(g) for 15 min, followed by the addition of 0.1 mL of a

Chapter 3

0.5 M ascorbic acid (0.05 mmol) aqueous solution and 60 mg of XC-N3. This solution was stirred under N2(g) for 14 hr at RT.

The ferrocene treated carbon, XC-Fc, was isolated from solution by creating a packed bed using centrifugal filtrations through a 0.2 µm PTFE membrane filter. This packed bed was washed using vacuum filtration with 20 mL acetonitrile, 3 mL of a

0.1 M sodium diethyldithiocarbamate methanol solution, and 20 mL methanol. The isolated and washed XC-Fc was dried at 50 °C under vacuum overnight.

3.5.7 XC-Fc Preparation for ICP-OES. A 0.7 mL amount of nitric acid

(70%, TraceMetal) was added to 20 mg of XC-Fc and heated to reflux for 2 min.

After cooling to RT the solution was diluted to 10 mL and filtered through a 0.45 µm polystyrene membrane filter before ICP-OES analysis. While the carbon material remains intact under these conditions, more than 90% of the Fe and Cu was extracted by this procedure as verified by ICP-OES analysis from the difference in Fe and Cu amount between the solution used for immobilization and recovered filtrate from washing.

3.5.8 Electrochemical Measurements of Ferrocene-modified XC-72R.

Isopropyl alcohol (7.5 mL) was added to 25 mg of XC-Fc and 0.1 g of a 15% Nafion solution. This mixture was sonicated for 20 min to make a homogenous suspension.

30 µl of the ink was deposited and dried (3 × 10 µl) on a 0.195 cm2 glassy carbon working electrode for analysis in 1 M HClO4 aqueous solution.

3.6 Acknowledgements

This work was supported by the US Department of Energy, Office of Basic Energy

Sciences as part of the Center for Electrocatalysis, Transport Phenomena and

Chapter 3

Materials for Innovative Energy Storage (CETM) Energy Frontier Research Center

(EFRC) project number DE-SC00001055 and The John Stauffer Stanford Graduate

Fellowship. We thank Dr. G. Li for ICP-OES measurements. XPS measurements performed at the Stanford Nano Shared Facilities.

3.7 References

(1) Liang, C.; Li, Z.; Dai, S. Angew. Chem., Int. Ed. 2008, 47, 3696-3717.

(2) McCreery, R. L. Chem. Rev. 2008, 108, 2646-2687.

(3) Figueiredo, J. L. J. Mater. Chem. A 2013, 1, 9351-9364.

(4) Devadoss, A.; Chidsey, C. E. D. J. Am. Chem. Soc. 2007, 129, 5370-5371.

(5) McCrory, C. C. L.; Devadoss, A.; Ottenwaelder, X.; Lowe, R. D.; Stack, T. D.

P.; Chidsey, C. E. D. J. Am. Chem. Soc. 2011, 133, 3696–3699.

(6) Stenehjem, E. D.; Ziatdinov, V. R.; Stack, T. D. P.; Chidsey, C. E. D. J. Am.

Chem. Soc. 2013, 135, 1110-1116.

(7) Tornoe, C. W.; Christensen, C.; Meldal, M. J. Org. Chem. 2002, 67, 3057-

3064.

(8) Rostovtsev, V. V.; Green, L. G.; Fokin, V. V.; Sharpless, K. B. Angew. Chem.

Int. Ed. 2002, 41, 2596-2599.

(9) Collman, J. P.; Devaraj, N. K.; Eberspacher, T. P. A.; Chidsey, C. E. D.

Langmuir 2006, 22, 2457-2464.

(10) Evans, R. A. Aust. J. Chem. 2007, 60, 384.

(11) Iha, R. K.; Wooley, K. L.; Nystrom, A. M.; Burke, D. J.; Kade, M. J.; Hawker,

C. J. Chem. Rev. 2009, 109, 5620-5686.

Chapter 3

(12) Kinoshita, K. Carbon: Electrochemical and Physicochemical Properties; John

Wiley and Sons: New York, 1988.

(13) Mahata, N.; Silva, A. R.; Pereira, M. F. R.; Freire, C.; de Castro, B.;

Figueiredo, J. L. J. Colloid Interface Sci. 2007, 311, 152-158.

(14) Cardoso, W.; Silva, A. R.; de Castro, B.; Freire, C. Appl. Catal., A 2005, 285,

110-118.

(15) Samuels, W. D.; LaFemina, N. H.; Sukwarotwat, V.; Yantasee, W.; Li, X. H.

S.; Fryxell, G. E. Sep. Sci. Technol. 2010, 45, 228-235.

(16) Chouyyok, W.; Shin, Y.; Davidson, J.; Samuels, W. D.; Lafemina, N. H.;

Rutledge, R. D.; Fryxell, G. E.; Sangvanich, T.; Yantasee, W. Environ. Sci. &

Technol. 2010, 44, 6390-6395.

(17) (a) Belmont, J. A. U.S. Patent 5,554,739, September, 10 1996 (Cabot