Metal-Free Heteroatom Doped-Carbon Nanomaterials for Energy Conversion and Storage

Home , Heteroatom

METAL-FREE HETEROATOM DOPED-CARBON NANOMATERIALS FOR ENERGY CONVERSION AND STORAGE

MIN WANG

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Dissertation Adviser: Professor Liming Dai

Department of Macromolecular Science and Engineering CASE WESTERN RESERVE UNIVERSITY

May, 2017

CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

Min Wang

Candidate for the degree of Doctor of Philosophy*

(Signed) Professor Liming Dai______(Chair of the committee)

Professor David Schiraldi______(Committee member)

Professor Rigoberto Advincula______(Committee member)

Professor Clemens Burda______(Committee member)

Date of Defense Dec, 19th, 2016

*We also certify that written approval has been obtained for any proprietary material contained therein. TABLE OF CONTENTS

LIST OF TABELS……………………………………………………………………...………. .I

LIST OF FIGURES…………………………………………………………………………….. II

ACKNOWLEDGMENTS…………………………………………………………..……...XXIII

ABSTRACTS………………………………………….…………………………………… XXIV

CHAPTER I. Introduction…………………………………………………………………….1 1.1 Introduction of Carbon Nanomaterials 1.1.1 Carbon Nanotubes 1.1.2 Graphenes 1.1.3 Porous Carbon Nanomaterials 1.2 Oxygen Reduction and PEMFC for Energy Conversion 1.2.1 Electrochemical Oxygen Reduction Reaction (ORR) 1.2.2 Current State of ORR Electrocatalysts 1.3 Metal-free Carbon Based Electrocatalysts for ORR 1.3.1 Heteroatom-doped Carbon Nanomaterials for ORR 1.3.2 Role of Heteroatom Doping 1.3.3 Other Strategies for ORR Electrocatalysis 1.4 Research Objectives

CHAPTER II………………………………………………………………………………….34 Heteroatoms-codoped Carbon Nanotubes as High-Performance Metal-free Electrocatalysts for Energy Conversion

CHAPTER III…………………………………………………………………………………72 Graphitic Carbon Nitrides Supported by Nitrogen-doped Graphene as Efficient Metal-free Electrocatalysts for Oxygen Reduction

CHAPTER IV…………………………………………………………………………………95 Rational Design of 3 Dimensional Metal-free Nitrogen-doped Graphene / Carbon Nanotube Composites for Energy Conversion and Storage

CHAPTER V……………………………………………………………………………...…...132 Metal-free Nitrogen-doped Hierarchical Porous Carbons derived from Biomass for Energy Conversion and Storage

REFERENCES………………………………………………………………………………...163 LIST OF TABLES

CHAPTER I. Introduction

Society. ………………………..……...………………………..……...………………………..32

CHAPTER II. Heteroatoms-codoped Carbon Nanotubes for Electrocatalysis

Table 2.1 XPS analysis of heteroatoms doped carbon nanotubes. ………………………..……...55

CHAPTER III Graphitic Carbon Nitrides Supported by Nitrogen-doped Graphene as

Efficient Metal-free Electrocatalysts for Oxygen Reduction

Table 3.1 Preparation methods and electrocatalytic performance of C3N4 based electrocatalysts

for ORR in 0.1M O2-saturated KOH. ………………………..……...………………………..…94

CHAPTER IV Rational Design of 3D Metal-free Nitrogen-doped Graphene / Carbon

Nanotubes Composites for Energy Conversion and Storage

Table 4.1 XPS results of N-CNT, N-G, G-CNT and N-G-CNT. ………………………..……...104

Table 4.2 BET results of N-CNT, N-G, G-CNT and N-G-CNT. ………………………..…....107

LIST OF FIGURES

CHAPTER I. Introduction

Figure 1.1 Structures of fullerene (0D), CNT (1D), graphene (2D), and 3D carbon nanomaterials,

indicating that the 2D graphene is a building block for other graphitic carbon materials. Reprinted

Chemical Society. ………………………..……...…………………………..……………………2

Figure 1.2 Important synthesis methods of graphene. (a). Chemical vapor deposition (CVD)

(b). Mechanical exfoliation (c). Liquid phase exfoliation. Reprinted from Ref 143 “Nature 490,

192-200 (2012)” with permission from the Nature Publishing Group. (d). Reduction of graphite oxide (GO). Graphite is oxidized to GO and thermal reduction of GO yields reduced graphene oxide (R-GO). Reproduced from Ref. 144 with permission from the PCCP Owner Societies. (e).

Schematic representation of the mechanochemical reaction between in situ generated active carbon species and reactant gases in a sealed ball-mill crusher. The cracking of graphite by ball milling in the presence of corresponding gases and subsequent exposure to air moisture resulted in the formation of EFGnPs. Reprinted with permission from Ref 145. Copyright 2013 American

Chemical Society. ………………………..……...…………………………..……………………6

Figure 1.3 Advanced carbon nanomaterials from different synthesis routes. (a). Template- free fabrication of hierarchical porous carbon by constructing carbonyl crosslinking bridges between polystyrene chains. Adapted from Ref. 162 with permission from Copyright 2010 Royal

Society of Chemistry. (b). Chemically bonded graphene/carbon nanotube composites as flexible

II supercapacitor electrode materials are synthesized by amide bonding. Reprinted with permission from Ref 163. Copyright 2013 Wiley-VCH. (c). Hierarchical porous carbons from carbonized metal-organic frameworks for lithium sulphur batteries. Adapted from Ref. 164 with permission from Copyright 2013 Royal Society of Chemistry. (d). N and P co-doped porous carbons (NPMC): schematic illustration of the preparation process for the NPMC foams. An aniline-phytic acid complex is formed, followed by oxidative polymerization into a 3D PANi hydrogel crosslinked with phytic acids. The PANi hydrogel is freeze-dried into an aerogel and pyrolysed in Ar to produce the NPMC. Reprinted from Ref. 165 with permission from the Nature Publishing Group.

Figure 1.4 (a) Schematics of a fuel cell. Reprinted with permission from Ref. 15 Chem. Rev.,

2015, 115 (11), 4823-4892. Copyright 2015 American Chemical Society. (b). Typical polarization curve of PEMFCs. The typical polarization curve describes the relationship between cell voltage and current density used to evaluate cell performance. Reprinted with permission from Ref. 174

Figure 1.5 (a). The process and thermodynamic electrode potentials of electrochemical oxygen reduction reaction in acid and alkaline media, respectively. Adapted from Ref.177 with permission of Springer. (b). LSV curves of electrocatalysts in oxygen-saturated electrolyte with different rotating rates. (c) Oxygen reduction curves on the disc and ring electrodes of RRDE at 5 mV s−1 scan rate at 1600 rpm, respectively. Reprinted with permission from Ref. 174 Copyright 2015

AAAS. ………………………..……...……………………………………………..……………14

III

Figure 1.6 Metal-free carbon nanomaterials for oxygen reduction. (a). Vertically aligned nitrogen-doped carbon nanotubes (VA-NCNT): (1). Scanning electron microscopy (SEM) image of the as-synthesized VA-NCNTs on a quartz substrate. (2). TEM image of the electrochemically purified VANCNTs. (3). CVs for oxygen reduction at the unpurified (upper) and electrochemically purified (bottom) VA-NCNT/GC electrodes in the argon-protected (dotted curves) or air-saturated

0.1 M KOH (solid red curves) at the scan rate of 100 mVs−1. Reproduced with permission from

transparent N-doped graphene film floating on water after removal of the nickel layer by dissolving

in an aqueous acid solution. (2) AFM images of the N-doped graphene film. (3). RRDE voltammograms for the ORR in air-saturated 0.1 M KOH at the C-graphene electrode (red line),

Pt/C electrode (green line), and N-doped graphene electrode (blue line). Reprinted with permission from Ref. 37 Copyright 2010 American Chemical Society. (c). BCN graphene: (1). BCN graphene model, C gray, H white, B pink, N blue. (2). SEM image of BCN graphene, scale bar: 500nm. (3).

LSV curves of ORR on BCN graphene with different compositions in O2-saturated 0.1 M KOH

solution at 10 mV s-1 and compared with the commercial Pt/C electrocatalyst. Reprinted with

TEM (2) images of VA-BCN nanotubes. (3). LSV curves of various electrodes in O2-saturated 0.1

M KOH electrolyte at a scan rate of 10 mVs-1 at a rotation rate of 1000 rpm. Reprinted with

(1). Schematic representation of the synthesis of COP-4 through monomers tris(4-

bromophenyl)amine (TBA), 2,4,6-tris(4-bromo-phenyl)-[1,3,5] triazine (TBT), (4’-bromo- biphenyl-4-yl)-porphyrine (TBBPP), and 2,4,6-tris(5-bromothiophen-2-yl)-1,3,5-triazine (TBYT),

respectively, using nickel-catalyzed Yamamoto-type Ullmann cross-coupling reaction. (2). LSV

−1 curves of COP graphene in O2-saturated 0.1 M KOH at 1600 rpm at a sweep rate of 5 mV s .

graphene quantum dots: (1). Synthesis of Quantum Dots 1-3. (2). CV curves (50 mV s−1) of N-

−1 GQD-1 on a RDE in a N2- and an O2-saturated 0.1 M KOH solution. (3). LSV curves (10 mV s ) for N-GQD-1, N-GQD-2, GQD-3 and Pt/C on a RDE (1600 rpm) in an O2-saturated 0.1 M KOH solution. Also shown is the LSV curve for 11, a much smaller N-substituted heterocycle with structure shown in the inset. Reprinted with permission from Ref. 193 Copyright 2012 American

Chemical Society. ………………………………………………………………………………19

Figure 1.7 Metal-free edge-functionalized graphene via ball milling for oxygen reduction. (a).

Edge-halogenated graphene: (1). A schematic representation for the edge expansions of XGnPs caused by the edge-halogens. (2). CV curves of pristine graphite, ClGnP, BrGnP, IGnP, Pt/C electrodes in N2- and O2-saturated 0.1 M KOH solution with a scan rate of 0.1 V/s. (3). LSV curves

of pristine graphite, ClGnP, BrGnP, IGnP, Pt/C electrodes in an O2-saturated 0.1 M KOH solution with a rotation rate of 1600 rpm. Reprinted from Ref. 58 with permission from the Nature

Publishing Group. (b). Edge-sulfurized graphene: (1). Proposed chemical structure of SGnP. (2).

LSV curves of graphite, SGnP and Pt/C electrodes in an O2-saturated 0.1 M KOH solution at a scan rate of 10 mV s −1 with a rotation rate of 1600 rpm. Reprinted with permission from Ref.197

Copyright 2013 Wiley-VCH. (c). Edge sulfonic acid functionalized graphene (SO3-GnP): (1). A schematic representation for the edge-delamination of SO3-GnP in KOH electrolyte. (2). CV curves of SO3-GnP electrodes in N2- and O2-saturated 0.1 M KOH solution Reprinted with permission from Ref. 57 Copyright 2013 American Chemical Society. …………………………24

Figure 1.8 (a). Main bonding configurations for N in CNTs: pyridinic N, pyrrolic N, and graphitic

CNT/GC (curve 2), and VA-NCNT (curve 3) electrodes. (c). Calculated charge density distribution for the NCNTs (left) and schematic representations of possible adsorption modes of an oxygen molecule at the CNTs (right, top) and NCNTs (right, bottom). The C atoms around the pyrrolic nitrogen could possess much higher positive charges than do the C atoms around the pyridinic nitrogen. Reproduced with permission from Ref.91 Copyright 2009 AAAS. (d). LUMO/HOMO and (e). Charge distribution of (5,5)-12.9 (length) CNT and NCNT. Reprinted with permission from Ref.198 Copyright 2010 American Chemical Society. ……………………………………26

Figure 1.9 Metal-free electrocatalysis via intermolecular charge transfer. (a). PDDA functionalized graphene: (1). Schematic illustration of the electron-withdrawing from graphene by

PDDA to facilitate the ORR process. (2). CV curves of oxygen reduction on the graphene and

PDDA-graphene electrodes in an O2-saturated 0.1 M KOH solution. (3). LSV curves for oxygen reduction on the graphene, PDDA-graphene, and Pt/C electrodes in an O2-saturated 0.1 M KOH solution. Reprinted with permission from Ref. 200. Copyright 2011 American Chemical Society.

(b). PDDA functionalized ACNT: (1).Illustration of charge transfer process and oxygen reduction reaction on PDDA-CNT. (2).CV curves of ORR on PDDA-ACNT in N2 (black curve) and O2 (red curve)-saturated 0.1 M KOH solutions at a scan rate of 50 mV s-1. (3). Linear sweep voltammetry

CHAPTER II. Heteroatoms-codoped Carbon Nanotubes for Electrocatalysis

Figure 2.1 Morphology of BN-CNT. (a). SEM image of freeze dried BN-CNT foam, the scale bar is 10 μm. (b), (c) SEM images of BN-CNT sheets at different magnification. The scale bar is

(b). 5 μm and (c). 300nm, respectively. …………………………………………………………41

Figure 2.2 Morphology of BN-CNT. (a), (b). TEM images of BN-CNT, the scale bar is 200nm and 20nm for (a) and (b), respectively. (c), (d) TEM images of individual nanotube structure of

BN-CNT. The scale bar is 10nm. ………………………………………………………………..42

Figure 2.3 Structure analysis of BN-CNT. (a).Raman spectra and (b).XRD patterns of B-CNT,

N-CNT, BN-CNT and r-CNT. (c). XPS survey spectra of B-CNT, N-CNT and BN-CNT. (d). BET

of BN-CNT, insert: pore size distribution of BN-CNT. …………………………………………44

Figure 2.4 XPS analysis of BN-CNT. (a) High resolution XPS C1s deconvoluted spectrum of

BN-CNT; (b) High resolution XPS B1s deconvoluted spectrum of BN-CNT; (c). High resolution

XPS N1s deconvoluted spectrum of BN-CNT and (d). High resolution XPS O1s deconvoluted

spectrum of BN-CNT. …………………………………………………………………...………45

Figure 2.5 XPS analysis of N-CNT and B-CNT. High resolution XPS C1s deconvoluted

spectrum of (a) N-CNT and (b). B-CNT; (c) High resolution XPS N1s deconvoluted spectrum of

N-CNT; (d) High resolution XPS B1s deconvoluted spectrum of B-CNT. High resolution XPS

C1s deconvoluted spectrum of (e) N-CNT and (f). B-CNT. ……………………………………47

VII

Figure 2.6 Proposed chemical structure and electrocatalytic process of oxygen reduction and

hydrogen evolution on BN-CNT. ……………………………………………………………….48

Figure 2.7 Catalytic activity towards electrochemical reduction of oxygen in basic electrolyte

at room temperature. (a) Cyclic voltammetry (CVs) of BN-CNT in O2-saturated and N2-saturated

0.1M KOH obtained at a sweep rate of 50 mV s −1; (b) LSVs of BN-CNT on the RRDE at 1600rpm

in O2-saturated 0.1M KOH, the insert: electron transfer number estimated from the ring and disk

currents; (c). LSVs of BN-CNT at different rotating speeds (400rpm, 600rpm, 900rpm, 1200rpm

−1 and 1600rpm) at a sweep rate of 5 mV s in 0.1M O2-saturated KOH; (d). Koutecky-Levich plots

of BN-CNT derived from Figure 2.7 (c); (e). Durability curves (i-t) of BN-CNT and Pt/C obtained in at -0.3 V versus SCE at a rotation rate of 1000 rpm. …………………………………………49

Figure 2.8 Catalytic activity towards electrochemical reduction of oxygen in basic electrolyte at room temperature. (a) CVs of B-CNT, N-CNT, BN-CNT and r-CNT in O2-saturated 0.1M

KOH obtained at a sweep rate of 50 mV s −1; (b) LSVs of BN-CNT on the RRDE at 1600rpm in

O2-saturated 0.1M KOH, the insert: electron transfer number estimated from the ring and disk

currents; LSVs of (b). B-CNT (c). N-CNT and (d). r-CNT at different rotating speeds (400rpm,

−1 600rpm, 900rpm, 1200rpm and 1600rpm) at a sweep rate of 5 mV s in 0.1M O2-saturated KOH;

(e). LSVs of B-CNT, N-CNT, BN-CNT, r-CNT and Pt/C electrodes obtained at a rotation speed

−1 of 1600rpm at a sweep rate of 5 mV s in 0.1M O2-saturated KOH. ……………………………52

VIII

Figure 2.9 LSVs of (a). BN-CNT 100/5 and (b). BN-CNT 100/25 at different rotating speeds

−1 (400rpm, 600rpm, 900rpm, 1200rpm and 1600rpm) at a sweep rate of 5 mV s in 0.1M O2-

saturated KOH; (c). LSVs of BN-CNT 100/5, BN-CNT 100/10 and BN-CNT 100/25 electrodes

−1 obtained at a rotation speed of 1600rpm at a sweep rate of 5 mV s in 0.1M O2-saturated KOH.

(d). XPS survey spectra of BN-CNT 100/5, BN-CNT 100/10 and BN-CNT 100/25. ………….54

Figure 2.10 Catalytic activity towards electrochemical oxygen evolution in basic electrolyte at room temperature. (a). LSVs of B-CNT, N-CNT, BN-CNT, r-CNT and Pt/C electrodes

−1 obtained at a rotation speed of 1600rpm at a sweep rate of 5 mV s in 0.1M O2-saturated KOH.

(b). LSVs of BN-CNT 100/5, BN-CNT 100/10 and BN-CNT 100/25 electrodes obtained at a

−1 rotation speed of 1600rpm at a sweep rate of 5 mV s in 0.1M O2-saturated KOH. (c).

Summarization of ORR and OER performance of B-CNT, N-CNT, BN-CNT with different doping

level, r-CNT and Pt/C electrodes. All the potentials in the table are vs. RHE. ………………….56

Figure 2.11 Electrocatalytic activity towards hydrogen evolution reaction. LSVs of BN-CNT,

−1 N-CNT, B-CNT, r-CNT and Pt/C obtained at a sweep rate of 10 mV s in (a). 0.5M N2 saturated

H2SO4 and (b). 0.1M N2 saturated KOH; Tafel plots of BN-CNT, N-CNT, B-CNT, r-CNT and

Pt/C obtained in (c). 0.5M N2 saturated H2SO4 and (d). 0.1M N2 saturated KOH; Potential value

of BN-CNT, N-CNT, B-CNT, r-CNT and Pt/C when the current density is 10mA/cm2 in (e). 0.5M

N2 saturated H2SO4 and (f). 0.1M N2 saturated KOH. …………………………………………..58

Figure 2.12 Electrocatalytic activity towards hydrogen evolution reaction. (a). LSVs of BN-

CNT in different PH-valued electrolytes; (b). Overpotential at 20mA cm-2 in different PH-valued

electrolytes on BN-CNT electrode; (c). LSVs of BN-CNT obtained at different sweep rates (5, 10,

−1 20, 50 mV s ) in 0.5M N2 saturated H2SO4 and 0.1M N2 saturated KOH; (d). Stability of BN-

CNT electrode in in 0.5M N2 saturated H2SO4 and 0.1M N2 saturated KOH. ………………….61

Figure 2.13 Electrocatalytic activity of BN-CNT towards hydrogen evolution reaction. LSVs

of BN-CNT with different doping levels (BN-CNT 100/5, BN-CNT 100/10 and BN-CNT 100/25)

in (a). 0.5M N2 saturated H2SO4 and (c). 0.1M N2 saturated KOH; Tafel plots of B, N-CNT with

different doping levels in (b). 0.5M N2 saturated H2SO4 and (d). 0.1M N2 saturated KOH. …….62

Figure 2.14 (a), (b). SEM images of N, F-CNTs, the scale bar is 20 μm and 1 μm for (a) and (b), respectively. (c), (d). TEM images of N, F-CNTs at different magnification. The scale bar is (c).

20nm and (d). 100nm. (e). Proposed chemical structure of N, F-CNTs. ……………………...... 64

Figure 2.15 Structure analysis of N, F-CNTs. (a). XRD patterns, (b). Raman spectra and (c).

TGA curves of MWCNTs, BM-MWCNTs and N,F-CNTs. (d). XPS survey spectrum of N, F-

CNTs; (d). High resolution XPS N1s deconvoluted spectrum of N, F-CNTs; (e) High resolution

XPS F1s spectrum of N, F-CNTs. ………………………………………………………………67

Figure 2.16 Electrocatalytic activity towards oxygen reduction and PEMFC. (a) CVs of N, F-

-1 CNTs and BM-MWCNTs obtained at a sweep rate of 50 mV s in 0.1M O2-saturated KOH. (b).

LSVs of N, F-CNTs at different rotating speeds (400rpm, 900rpm, 1200rpm and 1600rpm) at a

−1 sweep rate of 5 mV s in 0.1M O2-saturated KOH; (c) LSVs of N,F-CNTs, BM-MWCNTs and

-1 Pt/C electrodes obtained at a rotation speed of 1600rpm at a sweep rate of 5 mV s in 0.1M O2

saturated KOH. (d). Durability curves (i-t) of N, F-CNTs and Pt/C obtained in at -0.3 V versus

SCE at a rotation rate of 1000 rpm. (e). Polarization curve of the MEAs fabricated with of N,F-

CNTs and BM-MWCNTs as cathode electrodes for H2/O2 at 80˚C, DuPont Nafion 211 membrane,

30/30 psi anode and cathode back pressure. Anode electrodes were Pt coated electrode with loading amount of 1.0 mg/cm2; (f). Polarization and power density of N, F-CNTs. …………….68

Figure 2.17 Electrocatalytic activity towards oxygen reduction and evolution reactions. (a)

ORR LSVs and (b). OER LSVs of N, F-CNTs with different doping levels (N, F-CNTs-A, N, F-

CNTs-B, N, F-CNTs-C, N, F-CNTs-D and BM-MWCNTs) obtained at a rotation speed of

-1 1600rpm at a sweep rate of 5 mV s in 0.1M O2 saturated KOH. (c) ORR LSVs and (d). OER

LSVs of N, F-CNTs with different ball milling time (N, F-CNTs-24h, N, F-CNTs-48h, N, F-CNTs-

72h and BM-MWCNTs-48h) obtained at a rotation speed of 1600rpm at a sweep rate of 5 mV s-1

in 0.1M O2 saturated KOH. ………………………………………………………………………70

CHAPTER III Graphitic Carbon Nitrides Supported by Nitrogen-doped Graphene as

Efficient Metal-free Electrocatalysts for Oxygen Reduction

Figure 3.1 (a), (b). SEM images of N-G, the scale bar is 100 μm and 5 μm, respectively; (c).

Nitrogen adsorption-desorption isotherm of the N-G; (d). The corresponding DFT incremental pore size distribution curve of the N-G. …………………………………………………………78

Figure 3.2 (a). Raman spectra and (b). XPS survey spectra of N-G and r-G; (c) Thermogravimetric analysis (TGA) of N-G and r-G in air condition. …………………………………………………79

Figure 3.3 (a). TEM image of C3N4@N-G, the scale bar is 200nm; (b). TEM image of C3N4@N-

G, the scale bar is 100nm; (c). TEM image of g-C3N4, the scale bar is 100nm; (d). TEM image of

N-G, the scale bar is 200nm. ……………………………………………………………………..81

Figure 3.4 (a), (b). SEM image of C3N4@N-G, the scale bar is 10 μm; (c). SEM image of g-C3N4,

the scale bar is 2 μm; (d). SEM image of N-G, the scale bar is 10 μm. ………………………….82

Figure 3.5 (a). XRD patterns of g-C3N4, N-G and g-C3N4@N-G; (b) Raman spectra of N-G and

g-C3N4@N-G; (c). TGA curves of g-C3N4, N-G and g-C3N4@N-G under air atmosphere; (d) XPS

full spectrum of C3N4@N-G; (e) High resolution XPS C1s deconvolution spectra of C3N4@N-G;

(f). High resolution XPS N1s deconvolution spectra of C3N4@N-G. ……………………………83

Figure 3.6 (a) XPS full spectrum of g-C3N4; (b) XPS survey spectrum of N-G; (c) High resolution

XPS C1s deconvoluted spectrum of g-C3N4; (d) High resolution XPS C1s deconvoluted spectrum

of N-G; (e) High resolution XPS N1s deconvoluted spectrum of g-C3N4; and (f) High resolution

XPS N1s deconvoluted spectrum of N-G. ………………………………………………………86

Figure 3.7 Catalytic activity towards electrochemical reduction of oxygen in 0.1M O2-

saturated KOH aqueous solution at room temperature. (a) Cyclic voltammetry (CVs) of g-

−1 C3N4, N-G and g-C3N4@N-G obtained at a sweep rate of 50 mV s ; (b) Comparison of linear

sweep voltammograms (LSVs) of g-C3N4, N-G and g-C3N4@N-G at different rotating speeds

(400rpm, 600rpm, 900rpm, 1200rpm and 1600rpm) at a sweep rate of 5 mV s −1; (c). LSVs of g-

C3N4@N-G on the RRDE at 1600rpm, the insert: electron transfer number of g-C3N4@N-G

XII

estimated from the ring and disk currents; (d). LSVs of g g-C3N4@N-G at different rotating speeds

−1 at a sweep rate of 5 mV s in O2-saturated and N2-saturated KOH; (e).Koutecky-Levich plot for g-C3N4@N-G obtained from LSVs in (d); (f). Electron transfer number of g-C3N4, N-G and g-

C3N4@N-G calculated from Koutecky-Levich equations. ………………………………………88

Figure 3.8 (a) LSVs of g-C3N4, r-G, N-G, g-C3N4@r-G, g-C3N4@N-G and Pt/C obtained at a

−1 rotation rate of 1600rpm obtained at a sweep rate of 5 mV s in 0.1M O2-saturated KOH aqueous

solution at room temperature; (b). The current-time (i-t) chronoamperometric responses for ORR

at the g-C3N4@N-G and Pt/C electrodes in 0.1M O2-saturated KOH aqueous solution at -0.3 V

versus SCE, and 3.0 M methanol was added at around 200 s; (c). Durability curves (i-t) of g-

C3N4@N-G and Pt/C obtained in at -0.3V versus SCE at a rotation rate of 1000 rpm………….91

Figure 3.9 Catalytic activity towards electrochemical reduction of oxygen in 0.1M O2-

saturated KOH aqueous solution at room temperature. (a) Cyclic voltammetry (CVs) of g-

C3N4, N-G, g-C3N4 & N-G mixture and ball milled g-C3N4@N-G obtained at a sweep rate of 50

−1 mV s ; (b) LSVs of g-C3N4, N-G, g-C3N4 & N-G mixture and ball milled g-C3N4@N-G obtained

−1 at a rotation speed of 1600rpm at a sweep rate of 5 mV s ; Koutecky-Levich plot for (c). g-C3N4

and (d). N-G obtained from LSVs in Figure 3(b); (e). Cyclic voltammetry (CVs) of g-C3N4@N-G

in N2 and O2 saturated KOH solution; (f). Enlarged ring current density of C3N4@N-G obtained at

a rotation speed of 1600rpm in Figure 4.5 (c). ……………………………………………………92

CHAPTER IV Rational Design of 3D Metal-free Nitrogen-doped Graphene / Carbon

Nanotubes Composites for Energy Conversion and Storage

XIII

Figure 4.1 (a) SEM and (b) TEM images of N-CNT bundles. (c) SEM and (d) TEM images of the

N-G-CNT sheets. Top view of (e) the N-G-CNT and (f) the N-G films made by dispersing the materials in isopropanol uniformly, dropping the dispersions onto two Al foils and then drying the films. …………………………………………………………………………………………....103

Figure 4.2 (a). XRD patterns, (b). Raman spectra, (c). XPS spectra and (d). Ratio of C, O and N of N-CNT, N-G, G-CNT and N-G-CNT. ………………………………………………………104

Figure 4.3 High resolution XPS spectra. (a) C 1S, (b) N 1S of N-G-CNT; (c). N 1S of N-CNT;

(d). N 1S of N-G. …………………………………….…………………………………………106

Figure 4.4 Nitrogen sorption isotherms of (a) N-CNT, (b) N-G, (c) G-CNT and (d) N-G-CNT.

Inset of each isotherm: its corresponding pore size distribution. ……………………………….107

Figure 4.5 SEM images of catalyst layer cross-sections used in RDE measurements. (a), (b)

N-G; (c), (d) N-CNT; and (e), (f) N-G-CNT. Catalyst layers contain 5 wt.% Nafion binder loaded on the glass carbon electrode. ………………………………………………………………….109

Figure 4.6 Catalytic activity of N-G-CNT towards electrochemical reduction of oxygen in

0.1M O2-saturated KOH aqueous solution at room temperature. (a) Cyclic voltammetry (CVs)

of N-G-CNT in N2 and O2 saturated KOH solution; (b) LSVs of N-G-CNT and commercial Pt/C obtained at a rotation speed of 1600rpm at a sweep rate of 5 mV s −1; (c). LSVs of N-G-CNT

XIV

obtained at different rotation speeds (400rpm, 600rpm, 900rpm, 1200rpm and 1600rpm) at a sweep

rate of 5 mV s −1; (d). Koutecky-Levich plots of N-G-CNT. ……………………………………111

Figure 4.7 (a). CVs and of N-G, N-CNT, N-G-CNT and G-CNT obtained at a sweep rate of 50

mV s −1; (b). ORR LSVs of N-G, N-CNT, N-G-CNT and G-CNT obtained at a rotation speed of

1600rpm at a sweep rate of 5 mV s −1; (c). OER LSVs of N-G, N-CNT, N-G-CNT, G-CNT (d).

OER LSVs of N-G-CNT compared with 20% Pt/C obtained at a rotation speed of 1600rpm at a sweep rate of 5 mV s −1. …………………………………………………………………………114

Figure 4.8 LSVs of (a) N-CNT, (c) N-G, and (e) G-CNT obtained at different rotation speeds at a sweep rate of 5 mV s −1; Koutecky-Levich plots of (b) N-CNT, (d) N-G, and (f) G-CNT. ……115

Figure 4.9 Long time stability and tolerance to methanol/carbon monoxide of metal-free

catalyst N-G-CNT. The relative ORR cathodic current as the function of time for the N-G-CNT

and 20% Pt/C before and after adding (a) 3.0 M methanol, and (b) CO into the O2-saturated 0.1 M

KOH. (c) The normalized ORR cathodic current-time response of the N-G-CNT and 20% Pt/C in

O2-saturated 0.1 M KOH for 50000 s. …………………………………………………………..116

Figure 4.10 Electrocatalytic activities of the carbon-based metal-free N-G-CNT catalysts in acidic electrolyte (O2-saturated 0.1 M HClO4) by half-cell tests. (a) LSV curves of the N-G-

CNT compared with N-G and N-CNT electrocatalysts by RDE at scan rate of 10 mV s-1 and a rotation speed of 1600 rpm. (b). LSV curves and (c). Electron-transfer number of the N-G-CNT

compared with Fe/N/C and Pt/C(20%) electrocatalysts by RDE (d). Long time stability, and

tolerance to (e). methanol and (f). Carbon monoxide of metal-free catalyst N-G-CNT compared with Fe/N/C and Pt/C(20%) electrocatalysts at 0.5 V (vs. RHE). CO ( flow 100 mL s-1) was injected into the electrolytes (100 mL) at the time of 200 s and stopped at the 500 s. Methanol (10 mL) was injected into the electrolytes (100 mL) at the time of 200 s. …………………………118

Figure 4.11 Power and durability performance of N-G-CNT based MEAs in PEM fuel cells.

(a). Polarization curves of the N-G-CNT with or without carbon black (KB) at the loading of 2 mg cm-2 for each catalyst layer composition. The weight ratio of Carbon (N-G-CNT+KB) /Nafion = 1

/1. (b). Cell polarization and power density as the function of gravimetric current for the N-G-

CNT/KB (0.5 / 2 mg cm−2) with the weight ratio of (N-G-CNT+KB)/Nafion = 1/1. (c). Durability of the metal-free N-G-CNT in a PEM fuel cell measured at 0.5 V compared with a Fe/N/C catalyst.

Catalyst loading of N-G-CNT/KB (0.5 mg cm−2) and Fe/N/C (0.5 and 2 mg cm−2). Test condition:

H2/O2: 80°C, 100% relative humidity, 2-bar back pressure. ……………………………………120

Figure 4.12 Typical cross-section SEM images of the GDLs with the MEAs of (a-c) N-G-CNT (2 mg cm-2) and (d-f) N-G-CNT + KB (0.5 + 2 mg cm-2) as the cathode catalyst layers, respectively.

Arrows point several N-G-CNT sheets separated by KB agglomerates in (f). Nafion membrane

(N211) as the separator, and Pt/C as the anode. A piece of carbon paper with a carbon black layer

(ElectroChem Inc, Carbon Micro-porous Layer (CMPL)) was used as the gas diffusion layer

(GDL). (g). Schematic drawings of the MEA catalyst layer cross section, showing that O2

efficiently diffused through the carbon black separated N-G-CNT sheets but not the densely

packed N-G-CNT sheets. ………………………………………………………………………121

XVI

Figure 4.13 SEM images of electrodes cross-sections composed of (a)~(d) N-G-CNT sheets

separated by KB nanoparticles. Yellow arrows indicate the separated N-G-CNT sheets. (e), (f)

Densely packed N-G-CNT electrode without adding KB. ……………………………………124

Figure 4.14 SEM images of graphene and MWCNTs hybrids at different ratio (G:CNT= 3:1,

1:1, 1:3). (a), (b). N-3G-CNTs; (c), (d). N-G-CNT foam treated at 800oC for 3h; (e), (f). N-G-

3CNT 3h at different magnifications. …………………………………………………………125

Figure 4.15 (a). CV curves of N-3G-CNT at different scan rates (10mV/s, 20mV/s, 50mV/s,

100mV/s); (b). Charge-discharge curves of N-3G-CNT at different current densities (1.0A/g,

2.0A/g, 4.0A/g); (c). Specific capacitance of N-3G-CNT at different current densities; (d). CV curves of N-3G-CNT and 3G-CNT at a scan rate of 50 mV/s. …………………………………127

Figure 4.16 (a). CV curves of 3G-CNT at different scan rates (10mV/s, 20mV/s, 50mV/s,

100mV/s); (b). Charge-discharge curves of 3G-CNT at different current densities (1.0A/g, 2.0A/g,

4.0A/g); (c). CV curves of 3G-CNT, G-CNT, G-3CNT, r-G and r-CNT at a scan rate of 50 mV/s;

(d). Specific capacitance of 3G-CNT, G-CNT, G-3CNT, r-G and r-CNT at different current densities (1A/g, 2A/g, 4A/g, 10A/g). …………………………………………………………129

Figure 4.17 (a). CV curves of N-3G-CNT, N-G-CNT, N-G-3CNT, N-G and N-CNT at a scan rate of 50 mV/s; (b). Specific capacitance of N-3G-CNT, N-G-CNT, N-G-3CNT, N-G and N-CNT at different current densities (1A/g, 2A/g, 4A/g, 10A/g); (c). Specific capacitance of nitrogen doped

XVII

and H2 reduced graphene and CNT hybrids with different composition at a current density of

1.0A/g. ………………………………………………………………………………………….130

CHAPTER V. Metal-free Nitrogen-doped Hierarchical Porous Carbons derived from

Biomass for Energy Conversion and Storage

Figure 5.1 SEM images of (a). (b), (c) as-synthesized Daikon-NH3-900 without HCl washing treatment and (d). Daikon-NH3-900 after HCl washing treatment. The scale bar is (a). 200μm, (b).

50μm, (c). 10μm, and (d). 2μm, respectively. …………………………………………………138

Figure 5.2 TEM images of Daikon-NH3-900 in different magnifications. The scale bar is (a).

200nm, (b). 200nm, (c). 20nm, and (d). 10nm, respectively. …………………………………139

Figure 5.3 TEM images of (a), (b). Daikon-NH3-900 and (c), (d). Daikon-Ar-900. The scale bar

is (a). 200nm, (b). 20nm, (c). 20nm, and (d). 10nm, respectively. (e). Pore size distribution of

Daikon-NH3-900. ………………………………………………………………………………141

Figure 5.4 Structure characterizations of Daikon-NH3-900 and Daikon-Ar-900. (a). Raman spectra; (b). XRD patterns; (c). Nitrogen sorption isotherms and (d). Thermogravimetric analysis

(TGA) in air condition. …………………………………………………………………………143

Figure 5.5 XPS analysis of Daikon-NH3-900. (a). XPS full spectrum of Daikon-NH3-900 and

Daikon-Ar-900; (b). High resolution XPS C1s deconvoluted spectrum of Daikon-NH3-900; (c)

XVIII

High resolution XPS N1s deconvoluted spectrum of Daikon-NH3-900 and (d) High resolution XPS

O1s deconvoluted spectrum Daikon-NH3-900. ………………………………………………..144

Figure 5.6 Catalytic activity towards electrochemical reduction of oxygen in acidic electrolyte at room temperature. (a) Cyclic voltammetry (CVs) of Daikon-NH3-900 in O2-

−1 saturated and N2-saturated 0.5M H2SO4 obtained at a sweep rate of 50 mV s ; (b) Linear sweep voltammograms (LSVs) of Daikon-NH3-900 on the RRDE at 1600rpm in 0.5M O2-saturated

H2SO4 , the insert: electron transfer number of Daikon-NH3-900 estimated from the ring and disk currents; (c). LSVs of Daikon-NH3-900 at different rotating speeds (400rpm, 600rpm, 900rpm,

−1 1200rpm and 1600rpm) at a sweep rate of 5 mV s in 0.5M O2-saturated H2SO4; (d). Durability curves (i-t) of Daikon-NH3-900 and Pt/C obtained in at 0.3 V versus Ag/AgCl at a rotation rate of

1000 rpm; The current-time (i-t) chronoamperometric responses for ORR at the Daikon-NH3-900 and Pt/C electrodes in 0.5M O2-saturated H2SO4 aqueous solution at 0.3 V versus Ag/AgCl, (e).

CO and (f). 3.0 M methanol was added at around 200 s. ………………………………………147

Figure 5.7 (a). LSVs of Daikon-Ar-900, Daikon-NH3-800, Daikon-NH3-900, Daikon-NH3-1000

−1 and Pt/C electrodes in 0.5M O2-saturated H2SO4 obtained at a sweep rate of 5 mV s ; (b) The electron transfer number and (c). The ring current of Daikon-NH3-800, Daikon-NH3-900 and

Daikon-NH3-1000 electrodes in 0.5M O2-saturated H2SO4 obtained on RRDE at a rotation speed of 1600 rpm; (d). Raman of Daikon-NH3-800, Daikon-NH3-900 and Daikon-NH3-1000……..148

−1 Figure 5.8 (a) CVs of Daikon-NH3-900 obtained at a sweep rate of 50 mV s in O2- and N2- saturated 0.1M KOH aqueous solution; (b) LSVs of Daikon-NH3-900 on the RRDE at 1600rpm in

XIX

O2 saturated 0.1M KOH, the insert: electron transfer number of Daikon-NH3-900 estimated from

the ring and disk currents; (c). LSVs of Daikon-Ar-900, Daikon-NH3-900 and Pt/C obtained at a

−1 rotation rate of 1600rpm obtained at a sweep rate of 5 mV s in 0.1M O2-saturated KOH; (d).

2 Polarization curve and power density of the MEA fabricated with of Daikon-NH3-900 (3.0 mg/cm )

as cathode electrode for H2/O2 at 80˚C, DuPont Nafion 211 membrane, 30/30 psi anode and

cathode back pressure. Anode electrode was Pt coated electrode with loading amount of 1.0

mg/cm2. ………………………………………………………………………………………...150

Figure 5.9 (a). LSVs of Daikon-NH3-900 at different rotating speed (400rpm, 900rpm, 1200rpm

−1 and 1600rpm) at a sweep rate of 5 mV s in 0.1M O2-saturated KOH aqueous solution. (b). RRDE

LSVs of Daikon-NH3-800, Daikon-NH3-900 and Daikon-NH3-1000 obtained at a rotation rate of

1600rpm in 0.1M O2-saturated KOH; (e). The corresponding ring current of Figure 5.9 (b); (d).

The electron transfer number of Daikon-NH3-800, Daikon-NH3-900 and Daikon-NH3-1000

estimated from Figure 5.9 (b) and (c). …………………………………………………………152

Figure 5.10 (a). The current-time (i-t) chronoamperometric responses for ORR at the Daikon-

NH3-900 and Pt/C electrodes in 0.1M O2-saturated KOH aqueous solution at -0.3 V versus SCE, and CO was added at around 200 s; (b). The current-time (i-t) chronoamperometric responses for

ORR at the Daikon-NH3-900 and Pt/C electrodes in 0.1M O2-saturated KOH aqueous solution at

-0.3 V versus SCE, and 3.0 M methanol was added at around 200 s; (c). Durability curves (i-t) of

Daikon-NH3-900 and Pt/C obtained in at -0.3V at a rotation rate of 1000 rpm………………….153

Figure 5.11. (a). Polarization curve and (b). Power density of the MEAs fabricated with of Daikon-

2 2 NH3-900 (3.0 mg/cm ) and Daikon-Ar-900 (3.0 mg/cm ) as cathode electrodes for H2/O2 at 80˚C,

DuPont Nafion 211 membrane, 30/30 psi anode and cathode back pressure. Anode electrodes were

Pt coated electrode with loading amount of 1.0 mg/cm2. ………………………………………155

Figure 5.12 Electrochemical capacitance results of Daikon-NH3-900 in 1.0 M H2SO4. (a). CV

curves of Daikon-NH3-900 at different scan rates (10mV/s, 20mV/s, 50mV/s, 100mV/s, 200mV/s);

(b). Charge-discharge curves of Daikon-NH3-900 at different current densities (1.0A/g, 2.0A/g,

5.0A/g and 10A/g); (c). Specific capacitance of Daikon-NH3-900 at different current densities; (d).

Cycling stability of Daikon-NH3-900 at a current density of 10.0A/g up to 4000 cycles………156

Figure 5.13 Electrochemical capacitance results in 1.0 M H2SO4 of daikon-derived porous carbons at different temperatures. (a). CV curves of Daikon-NH3-800 at different scan rates

(10mV/s, 20mV/s, 50mV/s, 100mV/s, 200mV/s); (b). Charge-discharge curves of Daikon-NH3-

800 at different current densities (1.0A/g, 2.0A/g, 5.0A/g and 10A/g); (c). CV curves of Daikon-

NH3-1000 at different scan rates (10mV/s, 20mV/s, 50mV/s, 100mV/s, 200mV/s); (d). Charge-

discharge curves of Daikon-NH3-1000 at different current densities (1.0A/g, 2.0A/g, 5.0A/g and

10A/g). …………………………………………………………………………………………158

Figure 5.14 Electrochemical capacitance comparison of daikon-derived porous carbons at different temperatures in 1.0 M H2SO4. (a). CV curves of Daikon-NH3-800, Daikon-NH3-900

and Daikon-NH3-1000 at different scan rate of 50mV/s; (b). Charge-discharge curves of Daikon-

NH3-800, Daikon-NH3-900 and Daikon-NH3-1000 at current density of 1.0A/g; (c). Specific

XXI

capacitance of Daikon-NH3-800, Daikon-NH3-900 and Daikon-NH3-1000 at different current

densities. ………………………………………………………………………………………..159

Figure 5.15 Electrochemical capacitance results in 1.0 M H2SO4 of daikon-derived porous carbons with and without nitrogen doping. (a). CV curves of Daikon-Ar-900 at different scan rates (10mV/s, 20mV/s, 50mV/s, 100mV/s, 200mV/s); (b). Charge-discharge curves of Daikon-

Ar-900 at different current densities (1.0A/g, 2.0A/g, 5.0A/g and 10A/g); (c). CV curves of

Daikon-NH3-900 and Daikon-Ar-900 at a scan rate of 50 mV/s; (d). Charge-discharge curves of

Daikon-NH3-900 and Daikon-Ar-900 at a current density of 1A/g; (e). Specific capacitance of

Daikon-NH3-900 and Daikon-Ar-900 at different current densities. …………………………...161

XXII

ACKNOWLEDGMENTS

I would like to convey my deepest gratitude to my advisor, Professor Liming Dai, for all his

support and guidance throughout my Ph.D. study. I am fortunate enough in having this opportunity

to work with him. His scientific insight and continued motivation have ensured me to successfully

pursue my Ph.D. research. I would also like to extend my appreciation to my defense committee

members, Professor David Schiraldi, Professor Rigoberto Advincula and Professor Clemens Burda.

It’s my greatest pleasure to meet and work together with all the present and previous group members in Prof. Dai’s research group in Case Western Reserve University. I convey my sincere

thanks to Dr. Song Liu, Dr. Yuhua Xue, Dr. Jun Liu, Dr. Ming Xu, Ms. Xueliu Fan, Dr. Jiantie

Xu, Dr. Zhonghua Xiang, Mr. Xiaoyi Chen, Dr. Dan Wang, Dr. Jintao Zhang, Dr. Jiangyong Liu,

Dr. Fangxin Hu, Dr. Yanrong He, Dr. Yonghua Chen, Dr. Shungang Song, Dr. Cailiang Zhang,

Dr. Naisheng Jiang, Dr. Xuli Chen, Dr. Chunxian Guo, Dr. Mingdao Zhang, Dr. Long Qie for their help and discussions. My Special thanks to Dr. Jianglan Shui and Dr. Chuangang Hu for excellent

mentorship and help on my research work and thesis writing. I would also like to thank Dr. Qiong

Wu in this Department of Macromolecular Science and Engineering, Dr. Hao Qu and Dr. Anand

Murugaiah from Momentive and all other friends at Case Western for all of their help.

I am deeply thankful to my friend Mr. Liang Yue, dear roommate Ms. Rao Fu, close friends

Ms. Yuan Tao for their supports during all these years.

Last but not the least, my deepest gratitude to my parents and family members for all their

encouragement in my life. I won’t have made it so far without their support.

XXIII

Metal-free Heteroatom Doped-Carbon Nanomaterials for Energy Conversion and Storage

MIN WANG

ABSTRACT

Fuel cell is considered as one of the most environmentally friendly devices for sustainably converting chemical energy to electricity. Till now, its large-scale application has been hindered by the high-cost of platinum catalysts required for oxygen reduction reaction (ORR) at the cathode.

Although the amount of noble metal needed for the desired catalytic effect could be reduced by using nonprecious-metal catalysts, they are still too expensive for the commercial mass production of clean energy, or their energy conversion efficiency is too low.

Owing to the related low cost, large surface area, high electrical conductivity, rich electrocatalytic active sites, and corrosion resistant properties, a new class of carbon-based, metal- free catalysts has been developed which could dramatically reduce the cost and increase the efficiency of fuel cells when used as alternative ORR catalysts. For example, various heteroatom- doped carbon nanomaterials have been developed as efficient metal-free catalysts for ORR.

XXIV However, their performance still needs to be further improved to pave the way for technological

advancement, particularly in the proton exchange membrane fuel cell (PEMFC). Another problem

is that most of the ORR studies on carbon-based electrocatalysts have been conducted using a half

electrochemical cell with alkaline medium, the electrochemical performance of carbon-based

metal-free nanomaterials in acidic medium is rarely, let alone the single fuel cell.

In this study, various metal-free carbon-based catalysts have been developed by rational

heteroatom doping coupled with super advanced structures (by intrinsic and macroscopic tune).

Different units, including carbon nanotubes (CNTs), graphene, graphitic carbon nitrides, and

biomass, were selected for constructing novel carbon-based nanomaterials with highly porous

structure and abundant electrochemically active sites. Heteroatoms doped or co-doped carbon

nanotubes (B,N-CNT and N,F-CNTs) , carbon nitride decorated nitrogen doped graphene

(C3N4@N-G) composites, N-doped graphene/CNT hybrids (N-G-CNT), and N-doped hierarchical porous carbon have been successfully developed as high efficient catalysts. The ORR performances of these carbon-based catalysts were investigated in three-electrode electrochemical setup in both alkaline and acidic mediums. PEMFCs utilizing the carbon-based catalysts as cathode catalyst for ORR were also fabricated and studied in detail. Especially, the rationally designed N-

G-CNT exhibited excellent catalytic activities and selectivity for ORR, even with good single cell performance.

XXV

CHAPTER I. Introduction

1.1 Introduction of Carbon Nanomaterials

Carbon is one of the most abundant elements in the biosphere. Among carbon based

materials, graphite and diamond, are the most popular ones in nature due to their useful properties.

Besides, there are many allotropes of carbon with different molecular structures as presented in

Figure 1.1. For instance, buckminsterfullerene (C60) was first realized in 1985 by Harold Kroto,

Robert Curl, and Richard Smalley, who were later awarded with the 1996 Nobel Prize in

Chemistry.1 As can be seen, buckminsterfullerene, a football-like zero-dimensional (0D) structure,

is consisted of twenty hexagons and twelve pentagons, with one carbon atom at each vertex of each polygon. Subsequently, Japanese scientist Sumio Iijima in NEC Corporation first observed

multiwall nanotubes formed in the soot of arc discharge in 1991.2 Two years later, Iijima3 and

Donald S. Bethune at IBM4 independently discovered single-wall carbon nanotubes (SWCNTs),

and developed methods to produce SWCNTs using transition-metal catalysts. Dr. Sumio Iijima

was awarded the 2008 Kavli Prize in Nanoscience for his contributions in the development of one- dimensional (1D) needle-like carbon nanotubes, which opened up a new field in materials science and nanotechnology.5,6 In 2004, Andre Geim and Kostya Novoselov at the university of

Manchester used a Scotch tape technique to extract the single-atom-thick carbon lattice, called graphene layer, from bulk graphite.7 Graphene is a two-dimensional (2D) planar structure of

carbon, which is composed of honeycomb lattice with one carbon atom at each vertex. Due to its

excellent properties for various potential applications, graphene has been extensively studied since

its discovery. Geim and Novoselov were awarded the 2010 Nobel Prize in Physics for their

groundbreaking experiment to produce the free-standing graphene sheet.8 The discovery of CNTs

and graphene have enormously attracted researchers’ attention to develop more advanced carbon-

based nanomaterials, including three dimensional (3D) carbon structures.

Diamond and graphite are two naturally available three-dimensional (3D) structures of carbon. Graphite, a soft and black mineral, has the lowest energy state among all allotropes of carbon at ambient temperature and pressure.9 Graphite has a layered, planar 3D structure. The

carbon atoms in the graphitic plane are covalently bonded, while the bonding between the layers

is through relatively weak van der Waals interaction.9,10 On the other hand, Diamond, another 3D

carbon material, is hard and transparent with each carbon atom bonding to four other carbon atoms in a regular repetitive pattern called hexagonal closed packing.11 More recently, scientists around

the world have made great efforts to develop novel 3D carbon nanomaterials, such as the well- ordered 3D graphene-CNT pillars (Figure 1).12-14

Figure 1.1 Structures of fullerene (0D), CNT (1D), graphene (2D), and 3D carbon nanomaterials,

indicating that the 2D graphene is a building block for other graphitic carbon materials. Reprinted

Chemical Society.15

Carbons with different structure showed different properties. Up to now, carbon

nanomaterials have been developed for a variety of potential applications.16,17 Generally speaking, carbon nanomaterials have a high surface area,18 tunable pore structure19 and easy accessibility for

reactant atoms or molecules or ions20. They also exhibit high electron mobility, phonon and heat

transport capability.21 Thus, the application of nanocarbons for energy storage and conversion has become one of the most popular research areas.22-25 Indeed, carbon nanomaterials, including

graphene and CNTs, have widely been used as electrode in metal ion batteries,26 such as lithium

ion27 and sodium ion batteries28. Carbon nanomaterials are also promising electrodes for

supercapacitors29 owing to their good conductivity and high double-layer capacitance.18 More importantly, functionalized carbon nanomaterials exhibit excellent electrocatalytic activities towards different reactions15, 17, 30-36, 91, 93 , and nanocarbons are becoming cost-effective electrocatalysts in fuel cells37,38, water splitting system and metal air batteries for energy

conversion.39-60 Nanocarbons are also used in solar cells as light harvesting media, photoactive

layer61,62 and transparent electrodes63,64.26,65,66 Other applications of nanocarbons include

adhesive67, advanced high-performance composites (golf shafts, tennis rackets, etc.),68

biosensors69,70, thermal management systems71, conductive printing inks72 and drug-delivery

systems73,74. Carbon nanomaterials are also widely used in water purification systems75, gas

76 77,78 separation and storage like CO2 , H2 capture .

1.1.1 Carbon Nanotubes

Carbon nanotubes have attracted extensive research intentions since its discovery in 1991.

Unlike C60, Carbon nanotube has a needle-like carbon architecture, consisting of carbon hexagons

arranged in a concentric way with both ends normally capped by fullerene-hemispheres.79

Considered as very typical 1D materials.80 CNTs are divided into two different types of CNTs with distinct properties: single-walled carbon nanotubes (SWCNTs) and multi-walled carbon nanotubes

(MWCNTs). SWCNTs can be considered as the structure of wrapping one layer graphene into a seamless cylinder with a diameter normally around 1 nm. MWCNTs have more than one concentrically rolled-up graphene layers constituting their walls. The interlayer distance in

MWCNTs is about 3.4 Å, which is similar to that of graphite.81 The special structures of CNTs

result in many intriguing electrical, mechanical and thermal properties. For example, the high

specific surface area (~1315 m2 g−1 for SWCNTs) is one important properties of CNTs, which

allows them to be applicable as membranes75 and in gas storage systems77. Moreover, SWCNTs

can either be semiconducting or metallic, depending on the diameter and helicity of the hexagonal

carbon rings on its walls.15 Thus, CNTs are widely applied in the energy-related areas like field

effect transistor (FET)82, supercapacitors83,84, batteries85,86 and solar cells64,87. Owing to CNTs’ excellent mechanical properties, such as high tensile strength and elastic modulus, they are widely used as nanofillers in high-performance polymer composites.88,89 Last but not least, functionalized

CNTs and their derivatives are promising electrocatalysts towards oxygen reduction reaction and other reactions.15, 90, 91 Functionalization strategies include heteroatom doping91,92, adsorption of polymers93, incorporation with graphene94,95, introduction of metals96, attaching functional groups

and so on. The functionalized CNTs with high electrocatalytic activity are also good candidates

for fuel cells97 and metal air batteries44,98,99. This thesis work focuses on MWCNTs doped with different heteroatoms (B, N, F), for electrochemical catalysis.

1.1.2 Graphenes

Graphene is the one-atom-thick planar sheets consisting of sp2-hybridized carbon atoms.100

Since its discovery in 2004, graphene has attracted worldwide attention due to its intriguing

properties and great potential in various applications. Graphene has many excellent properties,

including a very large theoretical specific area of 2630 m2 g−1, 101 high thermal conductivity (up to

5000 W m−1 K−1),102 high transparency (~97.7% to the visible light for a single layer graphene)103,

high Young’s modulus (~1.0 TPa)104 and outstanding electrical conductivity and charge

mobility105. Possessing the above mentioned properties, graphene shows a great potential in many

applications, including composites, thermal devices, membranes106,107, sensors108-110 and energy

related systems24,111,112. In particular, graphene has been emerging as an excellent candidate for various energy storage and conversion. For example, graphene has been used to develop

transparent and flexible electrodes to revolutionize the conventional concept of advancement for

electronic devices.113-116 Besides, graphene has also been widely utilized in FETs117,118, solar

cells119-121 and metal-ion based batteries122-126. Furthermore, the large specific surface area and

high electrical as well as thermal conductivities of graphene have made it suitable for

electrochemical energy storage applications, including as, electrodes in supercapacitors127 and electrocatalysis32,49,128-132.

Graphene can be prepared either from bottom-up or top-down approaches. Figure 1.2 summarizes several important techniques for graphene synthesis.133 In this respect, chemical vapor

deposition (CVD) is a widely used bottom-up method to prepare graphene thin films.134 CVD

technique usually requires methane or acetylene as carbon sources and the graphene layers are

deposited on a metal substrates, like copper or nickel, as catalysts. 135 Graphene, grown by CVD

method, usually possesses perfect honeycomb lattice structure with a negligible amount of defects attractive for advanced electronic and photonic devices, in which transparent and defect free highly-crystalline graphene films are essential.136,137 The mechanical exfoliation technique which

is a top-down approach, can produce single or few layered graphene sheets from bulk graphite, as

exemplified by the scotch tape technique developed by the Manchester group.138,139 Liquid phase

exfoliation of graphite140 in proper solvents, such as N-methyl-pyrrolidone (NMP),141 can also

yield monolayer and few layers of graphene.142

Figure 1.2 Important synthesis methods of graphene. (a). Chemical vapor deposition (CVD)

(b). Mechanical exfoliation (c). Liquid phase exfoliation. Reprinted from Ref 143 “Nature 490,

192-200 (2012)” with permission from the Nature Publishing Group.143 (d). Reduction of graphite

oxide (GO). Graphite is oxidized to GO and thermal reduction of GO yields reduced graphene

oxide (R-GO). Reproduced from Ref. 144 with permission from the PCCP Owner Societies. 144

(e). Schematic representation of the mechanochemical reaction between in situ generated active carbon species and reactant gases in a sealed ball-mill crusher. The cracking of graphite by ball milling in the presence of corresponding gases and subsequent exposure to air moisture resulted in the formation of EFGnPs. Reprinted with permission from Ref 145. Copyright 2013 American

Chemical Society.145

All the above mentioned methods produce very limited quantity of graphene. Large quantity

of graphene can be synthesized by solution chemistry as presented in Figure 1.2 (d).146,147 Briefly,

graphite powder or flakes are used as raw material which is reacted with strong oxidants like

potassium permanganate, concentrated sulfuric acid to introduce oxygen containing groups,

yielding the graphite oxide (GO).148 This technique is widely known as Hummer’s method.149 The

interlayer van der Waals force of graphite is greatly interrupted and weaken during the oxidation

procedure, resulting in an increased d-spacing in the GO. Although it is difficult to determine exact

structure of GO, it is rich of epoxide, hydroxyl, ketone, carbonyl, and carboxylic groups in both

basal plane and at the edges.150 Containing all these oxygen-containing groups, the bulk oxidized

graphite can easily be exfoliated into single GO layers. Subsequent reduction of GO can in turn

produce graphene in large quantities. A variety of reduction methods, such as thermal reduction151

and chemical reduction,152 have been developed and extensively studied.153 It should be noted that,

GO with rich oxygen-containing groups like carboxylic groups, is a very useful precursor for

functionalization of graphene with different moieties.

Very recently, researchers have developed another top-down approach, known as ball

milling technique as depicted in Figure 1.2 (e), to synthesize graphene nanoplatelets in large

quantities. Starting with raw-graphite, edge-carboxylate graphene sheets have been produced by

introducing dry ice (CO2) as reagent in the ball milling system. The ball milling technique

produced conductive and soluble edge-functionalized graphene sheets promising for

electrochemical and electronic applications. By selectively changing the reagent gas, ball milling

technique has been generalized to be an efficient approach to exfoliate graphite and simultaneously

achieve edge functionalization with various heteroatoms. When halogens are introduced as the

reagent with graphite in the ball milling system, a series of edge-halogenated graphene

nanoplatelets (ClGnP, BrGnP and IGnP) with a large BET surface area are successfully

produced.58

1.1.3 Porous Carbon Nanomaterials

Beside 1D CNTs and 2D graphene, the other interesting carbon allotropic nanomaterials

include fullerene and graphene quantum dots154,155 which are basically 0D carbons, and also graphene nanoribbons156 which are categorized as 1D carbon material with tunable band gap

energy157. Graphite and diamond are very well known as traditional 3D carbon nanomaterials.

Except these carbon allotropes, researchers have also been making great efforts to develop novel

porous carbon nanomaterials (PCMs) using different methods. The structural details of porous

nanocarbons are very critical but important factor in order to dictate their properties and applications. According to the distribution of pore size, microporous carbon materials have pore diameter lesser than 2 nm, while mesoporous carbon is the PCM has pore diameter in the range of

2~50 nm, and macroporous carbon contains pores of more than 50 nm in diameter.158 Basically,

in hierarchical porous structure of nanocarbons, the diffusive resistance of ions and atoms is

minimized due to easier mass transport through macropores. Furthermore, the high specific surface

area, together with abundant active sites imparted into the micropores and/or mesopores, is useful for various advanced applications in catalysis,159 energy storage,18 adsorption,160 and many other

sysytems.161

Figure 1.3 Advanced carbon nanomaterials from different synthesis routes. (a). Template-

free fabrication of hierarchical porous carbon by constructing carbonyl crosslinking bridges

Society of Chemistry.162 (b). Chemically bonded graphene/carbon nanotube composites as flexible supercapacitor electrode materials are synthesized by amide bonding. Reprinted with permission from Ref 163. Copyright 2013 Wiley-VCH.163 (c). Hierarchical porous carbons from carbonized

metal-organic frameworks for lithium sulphur batteries. Adapted from Ref. 164 with permission

(NPMC): schematic illustration of the preparation process for the NPMC foams. An aniline-phytic acid complex is formed, followed by oxidative polymerization into a 3D PANi hydrogel

crosslinked with phytic acids. The PANi hydrogel is freeze-dried into an aerogel and pyrolysed in

Ar to produce the NPMC. Reprinted from Ref. 165 with permission from the Nature Publishing

Group 165 (e). Schematic diagrams showing the synthesis and microstructures of a 3D graphene-

radially aligned CNT fiber. Insert: featured structure of advanced carbon nanomaterials.

Figure 1.3 presents several advanced porous-carbon nanomaterials and their preparation

routes. For example, hierarchical porous carbons (HPCs) are synthesized by constructing carbonyl

crosslinking bridges between polystyrene (PS) chains in solution without any templates. 3D porous

carbons can also be constructed by 2D graphene and 1D carbon nanotubes. As seen in Figure 1.3

(b), graphene oxide and amide functionalized CNTs are used as building blocks to form the

chemically bonded 3D graphene/carbon nanotube composites. The porous 3D nanocarbons are

idea flexible electrodes for supercapacitors. Carbonization is also an important technique to

synthesize carbon materials.167 In order to obtain porous carbon structure, it is critical to select

excellent precursors, such as metal-organic frameworks for the carbonization.53, 164 MOFs usually exhibit extremely large surface areas due to their porous structures. As presented in Figure 1.3 (c),

HPCs obtained from carbonized MOFs show tunable hierarchical porous morphologies with exceptionally large specific surface areas (up to 4793 m2/g) and high pore volumes. In another

work, 3D polyaniline (PANi) aerogel crosslinked with phytic acid were carbonized to produce N,

P codoped mesoporous carbons (NPMC) as shown in Figure 1.3 (d). The NPMC with large specific

surface area and rich heteroatom induced active sites were demonstrated to be excellent metal-free

bifunctional electrocatalysts and promising electrode materials for primary and rechargeable Zn-

air battery. More recently, researchers also rationally designed novel graphene-nanotube 3D

architectures through seamless nodal junction formation (Figure 1.3 (e)), which was synthesized

by CVD technique using surface anodized aluminum (AAO) wire as the hard template.166 This 3D

porous graphene-nanotube fiber was applied as the electrode for solid wire supercapacitors, and

the counter electrode for wire-shaped dye-sensitized solar cells (DSSCs).

Functionalized carbon nanomaterials, such as graphene, carbon nanotubes, and other novel porous carbon nanomaterials, having good transport conductivity, high specific surface area, and large amounts of active sites are very promising electrocatalysts for various electrochemical reactions, including oxygen reduction39,168, oxygen evolution169,170, and hydrogen evolution.171,35

These carbon nanomaterials are extremely promising for energy conversion and storage systems,

such as solar cells, batteries, and fuel cells, to fulfill the soaring energy demands.

1.2 Oxygen Reduction and PEMFC for Energy Conversion

The use of traditional nonrenewable fossil fuels often cause the CO2 emission and other

industrial wastes as environmental pollutants. In this regard, it is highly desirable to develop

renewable and environmentally friendly energy resources. Therefore, researchers worldwide have

been devoting enormous efforts to develop advanced solar cells, fuel cells, Li-ion batteries, and

other renewable energy devices. Particularly, fuel cells are very promising green and efficient

energy conversion devices, which can be used for electric vehicles and as stationary electric power.

Figure 1.4 (a) presents the typical working principle for a proton exchange membrane fuel

cell (PEMFC). PEMFCs can convert chemical energy into electrical by consuming hydrogen and

oxygen, producing water without CO2 emissions. In a single fuel cell, there are mainly two parts:

bipolar plates and membrane electrode assembly (MEA).172 Bipolar plates enable the introduction

of gas, the collection of electrical current and also act as mechanical support of the cell. A MEA

consists of three parts including the porous gas diffusion layer, the catalyst layer (active reaction

zone, interface between the electrode and the membrane) and the Nafion membrane.173 The MEA

is the most important part of a fuel cell. It is fed with hydrogen and oxygen generating electrical

power.

Figure 1.4 (a) Schematics of a fuel cell. Reprinted with permission from Ref. 15 Chem. Rev.,

2015, 115 (11), 4823-4892. Copyright 2015 American Chemical Society.15 (b). Typical polarization curve of PEMFCs. The typical polarization curve describes the relationship between cell voltage and current density used to evaluate cell performance. Reprinted with permission from

Figure 1.4 (b) shows the typical steady-state polarization curve of PEMFCs, describing the

relationship between the electrode potential and the current density. It can be seen that the fuel cell voltage is significantly deviated from the theoretical potential 1.23 V. The highest potential in a

PEMFC (the open cell potential), is achieved when an extremely low current passes through the

cell. However, the corresponding voltage drops down with increasing current density, which is

known as polarization. At a low current, the loss is due to the activation polarization related to slow reaction rate on the surface of the electrode, whereas, in the intermediate current range with

its maximum value corresponding to the maximum power density, the loss is generally attributed

to the Ohmic polarization associated to the resistance against the electron flow. On the other hand, at higher currents, above the maximum power density, the loss is related to the concentration polarization, resulting from the change in concentration of the reactants at the electrode surface.175

1.2.1 Electrochemical Oxygen Reduction Reaction

The oxygen reduction reaction (ORR) occurs on the cathode of fuel cells. Figure 1.5 (a) shows the process and thermodynamic electrode potentials of electrochemical ORR in acid and alkaline media, respectively. The oxygen can be directly reduced to water through a desired four electron pathway, or go through a less efficient two electron process with H2O2 as intermediate

product. The mechanism of the electrochemical O2 reduction reaction is complicated consisting of

multiple adsorption/desorption and reaction steps involving oxygen-containing species such as O,

− − 176 OH, O2 , HO2 , and H2O2.

To evaluate the electrocatalytic activity of catalysts for ORR in aqueous electrolytes, the

most commonly used techniques include rotating disc electrode (RDE), rotating ring-disc electrode

(RRDE) and linear scan voltammetry (LSV). RDE is a convective electrode system that consists of a rotating shaft and a glass carbon disk electrode. Figure 1.5 (b) represents the typical LSV curves of ORR tested on RDE at various rotating speeds. As can be seen, the current densities increased with increasing rotating speeds, resulting from the enhanced oxygen diffusion and consequent reduction on the electrode surface. Important ORR performance indicators, such as

13 onset potential (Eonset), half-wave potential (E1/2), and the diffusion-limiting current density (jd), can be determined from the LSV curve. Furthermore, the ORR reaction kinetics can also be investigated based on LSVs at different rotation speeds.

Figure 1.5 (a). The process and thermodynamic electrode potentials of electrochemical oxygen reduction reaction in acid and alkaline media, respectively. Adapted from Ref.177 with permission of Springer.177 (b). LSV curves of electrocatalysts in oxygen-saturated electrolyte with different rotating rates. (c) Oxygen reduction curves on the disc and ring electrodes of RRDE at 5 mV s−1

AAAS.174

The overall ORR current density (j) is dependent on the kinetic current (jk) and diffusion-

limiting current (jd). If the applied over-potential is high enough, every atom/ion reaching to the

electrode reacts immediately. The O2 concentration at the electrode is almost zero, leading to a

diffusion limiting plateau. The magnitude of jd is related only to the diffusion rate, which depends

on the rotating speed in the RDE measurements. Theoretically, the overall current (j) is given by the Koutecky-Levich equation178:

0.5 1 / j =1 / jk + 1 / jd =1 / jk + 1 / Bω

where, ω is the electrode rotating rate. B could be determined from the slope of the K-L plots. The electron transfer number can further be determined from the Levich equation as follows179:

2/3 -1/6 B= 0.2 n F (DO2) ν CO2

In the Levich equation, n represents the number of electrons transferred per oxygen molecule,

-1 F is the Faraday constant (F= 96485 C mol ), DO2 is the diffusion coefficient of O2 in 0.1 M KOH

-5 2 -1 2 -1 (~ 1.9×10 cm s ), ν is the kinetic viscosity (0.01 cm s ), and CO2 is the bulk concentration of

-6 -3 O2 (~ 1.2×10 mol cm ). The constant 0.2 is adopted when the rotation speed is expressed in revolution per minute (rpm).180

Beside the RDE measurement, RRDE can also be utilized to investigate the kinetics and related mechanism for ORR. Figure 1.5 (c) shows typical oxygen reduction curves collected on

the disc and ring electrodes in the RRDE measurements. The disc current and ring current are

recorded as a function of the disc electrode potential. Moreover, the oxygen reduction electron

transfer number (n) can be determined by the equation below181,

n= 4jD / (jD + jR/N)

In this equation, jD is the faradic disk current, jR is the faradic ring current, and N is the collection

efficiency (N= 0.3) of the ring electrode.182 Following these techniques, the oxygen reduction electrocatalytic performance of different materials can be evaluated and compared.

1.2.2 Current State of ORR Electrocatalysts

Due to the sluggish kinetic behavior, cathodic oxygen reduction is considered to be a limiting

factor to the fuel cell performance. Thus, electrocatalyst materials need to be modified in order to

enhance the electrochemical reduction of O2 and in turn improve the fuel cell performance. An

ideal ORR electrocatalyst must be able to minimize the over-potential of reaction, reduce the activation energy, increase the interface with reactants, enhance the electrolyte / reactants/ charge transport, and sustain its activity after long-operation time.

The commercially available oxygen reduction catalyst is a carbon-supported platinum with different Pt loading amount.The most common one is of 20 wt. % Pt nanoparticles loaded on carbon black, Vulcan XC-72R, while the Pt/C loading ratio ranges from of 0.2 to 0.4mg/cm2 for

fuel cell.15 However, the high cost related to the noble metal, Pt, hinders commercialization of fuel

cells. Furthermore, the commercial Pt/C catalyst also suffers from other drawbacks, including poor

operation durability, fuel crossover effect, and CO poisoning effect.15

Considering all these disadvantages, researchers have been working to develop alternative

catalysts for the oxygen reduction in fuel cells. The recent studies on ORR electrocatalysts can

mainly be categorized into four types: precious metals, transition metal oxides, carbon-based

transition metal hybrids and metal-free materials.176

Precious metal catalysts mainly include Pt, gold (Au), palladium (Pd), and silver (Ag) with

different morphologies and structures.183,184 Pt-based alloys with low Pt loading have also been studied for oxygen reductions.185

Transition metal spinel oxides, such as Co3O4, MnCo2O4, NiCo2O4 and FeCo2O4, have been

demonstrated to be effective oxygen reduction catalysts.186-188 Furthermore, perovskites, such as

189 LaCoO3 and LaNioO3, also show promising performs for oxygen reduction.

Carbon-based transition metal hybrids, in particular, M-Nitrogen-C (M= Fe, Co etc.), have been considered as a new class of nonprecious metal catalysts for ORR. For example, Wu and co- workers have developed M-N-C electrocatalysts via high temperature synthesis of Fe- and Co- based catalysts in presence of polyaniline (PANI), achieving high activity with remarkable stability for non-precious metal catalysts (700 hours at a fuel cell voltage of 0.4 volts) with an excellent four-electron selectivity.190

1.3 Metal-free Carbon Based Electrocatalysts for ORR

Apart from metal-based materials for ORR catalysis, carbon-based nanomaterials are also potential candidates as metal-free materials for electrocatalysis of oxygen reduction. The interesting characteristics of carbon-based nanomaterials favorable for electrocatalysis include its low cost, environmental acceptability, large specific surface area with various porous structures, high conductivity, corrosion resistance ability, and unique surface and bulk properties. The doping of heteroatoms (e.g., nitrogen) into these carbon structures could modulate the internal electronic configuration, which can influence the catalytic processes. For this purpose, a variety of carbon

nanomaterials have been developed and examined for ORR, opening up a brand new research field in electrocatalysis.

1.3.1 Heteroatom-doped Carbon Nanomaterials for Oxygen Reduction

In order to replace the commercial Pt/C catalyst for energy conversion and storage, our group has made breakthroughs in developing metal-free carbon-based nanomaterials,15, 222 particularly

heteroatom-doped nanocarbons as electrocatalysts for oxygen reduction.91 In this section, several

important metal-free carbon-based ORR electrocatalysts developed mainly by our group will be discussed. These are nitrogen-doped multiwall carbon nanotubes, nitrogen-doped single wall carbon nanotubes, nitrogen-doped graphene thin films, B/N-codoped carbon nanotubes, B/N- codoped graphenes, N-doped holely graphitic carbons, N-doped graphene quantum dots, and edge- functionalized graphene etc. Among different doping strategies involved to develop the above mentioned materials, the in-situ growth by chemical vapor deposition, post treatment in ammonia gas, pyrolysis of dopant-containing precursors, and ball milling techniques are most effective ones.

Figure 1.6 (a) presents the vertically-aligned nitrogen-containing carbon nanotube arrays

(VA-NCNTs) for ORR electrocatalysis in alkaline fuel cells.91 VA-NCNTs are prepared by pyrolyzing iron phthalocyanine at 800-1100°C in NH3 gas, and then purified by electrochemical

method to remove the metal impurities as can be seen in Figure 1.6(a)-3. The VA-NCNT arrays

were applied as a metal-free electrode to catalyze ORR in alkaline medium. The VA-NCNT shows

much higher electrocatalytic activity and long-term operational stability than commercially

available Pt/C electrodes. Moreover, they are free from CO poisoning and methanol cross-over

effects. Quantum mechanical calculations and subsequent experimental observations attribute the

improved catalytic performance to the N doping.91 The nitrogen doping creates a net positive

charge on adjacent carbon atoms in the nanotube carbon plane of VA-NCNTs. This changes the

O2 chemisorption mode, and attracts more electrons from the anode for facilitating the ORR.

0.1 M KOH (solid red curves) at the scan rate of 100 mVs−1. Reproduced with permission from

transparent N-doped graphene film floating on water after removal of the nickel layer by dissolving

in an aqueous acid solution. (2) AFM images of the N-doped graphene film. (3). RRDE voltammograms for the ORR in air-saturated 0.1 M KOH at the C-graphene electrode (red line),

Pt/C electrode (green line), and N-doped graphene electrode (blue line). Reprinted with permission

graphene model, C gray, H white, B pink, N blue. (2). SEM image of BCN graphene, scale bar:

500nm. (3). LSV curves of ORR on BCN graphene with different compositions in O2-saturated

0.1 M KOH solution at 10 mV s-1 and compared with the commercial Pt/C electrocatalyst.

BCN: SEM (1) and TEM (2) images of VA-BCN nanotubes. (3). LSV curves of various electrodes

-1 in O2-saturated 0.1 M KOH electrolyte at a scan rate of 10 mVs at a rotation rate of 1000 rpm.

holey graphitic carbon: (1). Schematic representation of the synthesis of COP-4 through monomers

tris(4-bromophenyl)amine (TBA), 2,4,6-tris(4-bromo-phenyl)-[1,3,5] triazine (TBT), (4’-bromo-

biphenyl-4-yl)-porphyrine (TBBPP), and 2,4,6-tris(5-bromothiophen-2-yl)-1,3,5-triazine (TBYT), respectively, using nickel-catalyzed Yamamoto-type Ullmann cross-coupling reaction. (2). LSV

−1 curves of COP graphene in O2-saturated 0.1 M KOH at 1600 rpm at a sweep rate of 5 mV s .

graphene quantum dots: (1). Synthesis of Quantum Dots 1-3. (2). CV curves (50 mV s−1) of N-

−1 GQD-1 on a RDE in a N2- and an O2-saturated 0.1 M KOH solution. (3). LSV curves (10 mV s )

for N-GQD-1, N-GQD-2, GQD-3 and Pt/C on a RDE (1600 rpm) in an O2-saturated 0.1 M KOH solution. Also shown is the LSV curve for 11, a much smaller N-substituted heterocycle with structure shown in the inset. Reprinted with permission from Ref. 193 Copyright 2012 American

Chemical Society. 193

Except for the multiwalled VA-NCNTs, N-doped single-walled carbon nanotubes (N-

SWCNTs) were also synthesized by chemical vapor deposition (CVD) method without involving any metal catalysts. A simple plasma etching technology was applied to generate SiO2

nanoparticles for catalyzing the growth of singlewalled CNTs, which did not involve to any metal

catalysts. The metal-free N-SWCNTs (N ~3.6 at %) are promising ORR catalysts in acidic medium.

N-SWCNTs have showed high electrocatalytic performance with good stability. The electron transfer number per oxygen molecule (n) is derived to be 3.52-3.92, suggesting a 4-electron

process for ORR on the N-SWCNT electrode.194 Beside 1D carbon nanotubes, 2D materials, such

as N-doped graphene thin films were prepared by CVD method in the presence of NH3 on Ni

(~300nm) coated Si substrate. The resulting film is freestanding, flexible and transparent with a

smooth surface. Electrochemical performance evaluation has indicated that the pure graphene

electrode without doping demonstrates a two-step, two-electron process for oxygen reduction,

while N-graphene electrode exhibits a desired one-step, four-electron pathway for the ORR.

Similar to N-CNTs, N-doped graphene had excellent long-term operational stability, CO tolerance,

and insensitivity to methanol.37 All these nitrogen doped carbons exhibit greatly enhanced oxygen

reduction performance due to the doping-induced intramolecular charge redistribution, and the mechanism will be further discussed in the following section 1.3.2.

Except doping electron-rich nitrogen into carbon lattice, boron (B) which is electron deficient, is also introduced into the carbon nanomaterials. Figure 1.6 (c) presents B/N-codoped

graphene (BCN) as the metal-free electrocatalyst for ORR. The BCN graphene with tunable

doping levels were simply synthesized by thermal annealing of graphene oxide in presence of boric

acid under ammonia atmosphere. The proposed chemical structure of the BCN graphene is shown

in Figure 1.6(c)-1, depicting rich B-C, C-N, and B-N domains in the carbon lattice plane. ORR electrocatalytic performance of optimized BCN graphene are better than the commercial Pt/C electrocatalyst in alkaline medium. According to the DFT calculations, BCN graphene with a

modest N- and B-doping level exhibiting the lowest energy gap and highest spin & charge density,

shows the best ORR performance.195 Co-doping strategy is also performed in the carbon nanotubes.

Vertically aligned carbon nanotubes co-doped with B and N (VA-BCN) was obtained by pyrolysis of melamine diborate and the morphologies were presented in Figure 1.6 (d). XPS results indicates the contents of C (85.5%), B (4.2%) and N (10.3%) for VA-BCNs. The electronegativity difference of B, C and N introduces charge redistribution and enhances the oxygen reduction catalysis. As compared to B-doped or N-doped VACNTs, VA-BCN exhibited better ORR electrocatalytic activity with more positive onset, peak potentials and larger reduction current densities as can be seen in Figure 1.6 (d)-3. This suggests the synergetic effect arising from the co-doping of carbon nanomaterials with different heteroatoms.196

Among the above-mentioned heteroatom-doped carbon nanomaterials, none of the doping strategies is able to control the exact locations of heteroatoms in the carbon nanostructures.

However, the location control of the dopant heteroatoms should be very important and meaningful

to tailor the structure-property relationship for heteroatom doped carbon nanomaterials.

Considering this, N-doped holely graphitic carbons were synthesized from 2D covalent organic

polymers (COP) as presented in Figure 1.6 (e).192 In this work, a series of novel COPs with

precisely-controlled locations of N atoms and hole sizes, were first synthesized as precursors for

further carbonization to obtain N-doped holey graphitic carbon materials. The optimized sample

C-COP-4 exhibits comparable ORR activity to the Pt/C catalyst in alkaline medium. Different with previous heteroatom-doped carbon nanomaterials, the N-doping active sites in the C-COP-4 are able to be located due to the well-defined structure of the precursors. Another example of controllable heteroatom doping is the bottom up solution chemistry synthesis of N-doped graphene quantum dots (N-GQD) as presented in Figure 1.6 (f)-1.193 The N-doping is closely correlated to

the properties of the quantum dots. The electrochemical results indicate that the N-doped GQDs

have size-dependent electrocatalytic activity towards ORR.

Previously, the ball milling technique has been demonstrated to be an efficient approach to

exfoliate graphite to obtain graphene materials in large scale. Actually, the most important

characteristic of ball milling is the realization of edge selective functionalization due to the nature

of mechanochemical reactions of the pristine graphite and reagents. Therefore, ball milling can be

considered as a new, facile and cost-effective approach to obtain controllable doping technique.

By selecting different reagents, different edge functionalized graphene materials were prepared as

metal-free electrocatalysts for oxygen reduction in alkaline medium. Several efficient ORR

electrocatalysts via ball milling are summarized in Figure 1.7.

Figure 1.7 Metal-free edge-functionalized graphene via ball milling for oxygen reduction. (a).

Edge-halogenated graphene: (1). A schematic representation for the edge expansions of XGnPs caused by the edge-halogens. (2). CV curves of pristine graphite, ClGnP, BrGnP, IGnP, Pt/C

electrodes in N2- and O2-saturated 0.1 M KOH solution with a scan rate of 0.1 V/s. (3). LSV curves of pristine graphite, ClGnP, BrGnP, IGnP, Pt/C electrodes in an O2-saturated 0.1 M KOH solution with a rotation rate of 1600 rpm. Reprinted from Ref. 58 with permission from the Nature

Publishing Group.58 (b). Edge-sulfurized graphene: (1). Proposed chemical structure of SGnP. (2).

LSV curves of graphite, SGnP and Pt/C electrodes in an O2-saturated 0.1 M KOH solution at a scan rate of 10 mV s −1 with a rotation rate of 1600 rpm. Reprinted with permission from

GnP): (1). A schematic representation for the edge-delamination of SO3-GnP in KOH electrolyte.

For example, halogens were introduced as the reagents with graphite in the ball milling

system.58 A series of edge-halogenated graphene nanoplatelets (XGnPs) were successfully synthesized. High BET surface areas of 471, 579 and 662 m2/g are observed for ClGnP, BrGnP and IGnP, indicating a significant extent of delamination in a good agreement with the proposed model in Figure 1.7 (a)-1. The XGnPs show remarkable electrocatalytic activities toward ORR.

The edges of halogenated graphene have favourable binding affinity with O2 molecule, and the O-

O bond strengths are weakened as a result of the halogenation-induced charge transfer. In another

work as shown in Figure 1.7 (b), edge-sulfurized graphene (SGnP), prepared by ball milling of graphite with sulphur powders, exhibits promising oxygen reduction activities.197 Similarly,

another edge sulfonic acid functionalized graphene (SO3-GnP) was produced by using sulfur trioxide as reagent for ball milling with graphite.57 The overall electrocatalytic activities are closely

related to the edge functionalized induced charge redistribution and edge polarity nature. Apart

from the charge redistribution mechanism, the strong charge repulsions between the graphitic

layers in SO3-GnP and in alkaline electrolyte allow sufficient oxygen diffusion into the graphitic

layers for a more effective ORR process, as shown in Figure 1.7 (c).

1.3.2 Role of Heteroatom Doping

As discussed above, a series of metal-free carbon nanomaterials with different heteroatom

doping, prepared by different doping strategies, have demonstrated to be efficient electrocatalysts

towards oxygen reduction reaction. In this section, the mechanism of oxygen reduction arising

from the heteroatom doping will be further discussed based on the nitrogen-doped carbon

25 nanotubes. The presence of nitrogen in CNTs leads to more chemically active sites, high density of defects, high surface areas, high conductivity, and high electrochemical activity. Figure 1.8 (a) proposes the structure models for the main bonding configurations of nitrogen doping in CNTs.

There are three different types of doped nitrogen species, such as pyridine-like, pyrrole-like, and graphitic/quaternary N, depending on their bonding environments with carbon atoms. The C-N bonding configuration can be characterized by X-ray photoelectron spectroscopy (XPS) analysis.

Figure 1.8 (a). Main bonding configurations for N in CNTs: pyridinic N, pyrrolic N, and graphitic

RDE voltammograms for oxygen reduction in air saturated 0.1 M KOH at the Pt-C/GC (curve 1),

VA-CNT/GC (curve 2), and VA-NCNT (curve 3) electrodes. (c). Calculated charge density distribution for the NCNTs (left) and schematic representations of possible adsorption modes of an oxygen molecule at the CNTs (right, top) and NCNTs (right, bottom). The C atoms around the pyrrolic nitrogen could possess much higher positive charges than do the C atoms around the pyridinic nitrogen. Reproduced with permission from Ref.91 Copyright 2009 AAAS.91 (d).

The experimental results have demonstrated that N-doped carbon nanotubes are efficient

ORR catalysts. The mechanism is related to doping-induced charge redistribution caused by the different electronegativity of carbon atom and dopant atom.91 The electronegativity of an atom is

a measure of its affinity for electrons. In the case of N-doped CNTs, the electronegativity

difference between nitrogen and carbon atoms leads to electron transfer from the carbon atom to

the nitrogen atom. As a result, the carbon atom adjacent to the N becomes positively charged. This

intramolecular charge transfer changes the chemisorption mode of O2 from the usual end-on

adsorption (Pauling model) at the nitrogen-free CNT surface (right top, Figure 1.8 (c) ) to a side-

on adsorption (Yeager model) onto the NCNTs electrode (right bottom, Figure 1.8 (c) ).

Consequently, the ORR onset and peak potentials positively shifts in Figure 1.8 (b) because the

parallel diatomic adsorption could effectively weaken the O-O bonding, facilitating oxygen

reduction at the VA-NCNT electrode.91

Subsequent studies have demonstrated that nitrogen doping could reduce the band gap

between the highest-occupied molecular orbital (HOMO) and the lowest-unoccupied molecular

orbital (LUMO).198 As presented in Figure 1.8 (d), the nitrogen doping lifts the energy level of

HOMO by 0.75 eV and reduces the band gap, facilitating the electron to transfer from the CNTs

to the O2. Concurrently, the electron-rich N dopant atoms become negatively charged, and the

adjacent carbon atoms get positively charged, generating charge redistribution as shown in Figure

1.8 (e). As is well known, the doping-induced charge redistribution also enhances the O2

adsorption,91 enhancing the oxygen reduction process. Based on these studies, it is understandable

that both N-doped, B-doped 199 and B, N-codoped carbon nanomaterials are efficient ORR catalysts. The ORR enhancements are related to the doping-induced charge redistribution caused by the different electronegativity of carbon atoms and dopant atoms. No matter whether the

electronegativity of the dopant atoms is higher or lower than that of carbon atoms, the

electronegativity difference between the heteroatom dopant and carbon atom could create

intramolecular charge redistribution, enhancing the O2 adsorption to facilitate then ORR process.

1.3.2 Other Strategies for ORR Electrocatalysis

Figure 1.9 Metal-free electrocatalysis via intermolecular charge transfer. (a). PDDA

functionalized graphene: (1). Schematic illustration of the electron-withdrawing from graphene by

PDDA to facilitate the ORR process. (2). CV curves of oxygen reduction on the graphene and

PDDA-graphene electrodes in an O2-saturated 0.1 M KOH solution. (3). LSV curves for oxygen

reduction on the graphene, PDDA-graphene, and Pt/C electrodes in an O2-saturated 0.1 M KOH

Society.200 (b). PDDA functionalized ACNT: (1).Illustration of charge transfer process and oxygen reduction reaction on PDDA-CNT. (2).CV curves of ORR on PDDA-ACNT in N2 (black curve)

-1 and O2 (red curve)-saturated 0.1 M KOH solutions at a scan rate of 50 mV s . (3). Linear sweep

-1 voltammetry curves of ORR in an O2-saturated 0.1 M KOH solution at a scan rate of 10 mV s .

As discussed above, heteroatom doping can introduce intromolecular charge transfer to facilitate the oxygen reduction process. Based on this mechanism, other metal-free carbon material synthesis techniques, such as intermolecualr charge transfer is also developed to enhance the oxygen reduction. In particular, poly(diallyldimethylammonium chloride) (PDDA), a typical polyelectrolyete, was used to functionalize graphene200 and carbon nanotubes93 as shown in Figure

1.9 (a) and (b), respectively. The PDDA functionalized carbon nanomaterials (PDDA-G and

PDDA-CNT) show enhanced electrocatalytic activity towards ORR. The PDDA can create a net positive charge for the carbon plane via intermolecular charge transfer, which facilitates the O2

adsoprtion and thus enhances the oxygen reduction process.

More recently, researchers have developed graphitic carbon nitride (g-C3N4) based materials

201 as metal-free electrocatalysts for oxygen reduction. There are two structural isomers for g-C3N4:

one comprises condensed melamine units with a periodic array of single carbon vacancies; the

other is made up of condensed melem (2,5,8,-triamino-tri-s-triazine) subunits, and contains larger

202 periodic vacancies in the lattice. Considering the high nitrogen content, g-C3N4 based materials

may be considered as suitable candidates for oxygen reduction catalysis. However, the

202 electrocatalytic performance of g-C3N4 is very poor due to the low conductivity. Thus,

202,203 researchers have combined g-C3N4 with carbon black or immobilized g-C3N4 onto graphene sheets204-206, and achieved improved activity for ORR. Of particular interest, Qiao et al. have

207,208 developed different carbon nitride based materials for ORR. They have incorporated g-C3N4

into a mesoporous carbon (CMK-3) to enhance the electron transfer efficiency of carbon nitride in

order to achieve better ORR activity.209 They have also synthesized 3D ordered macroporous g-

210 C3N4/C by using SiO2 as hard template for oxygen reduction. These carbon nitrides based

materials exhibit great potential for catalyzing oxygen reduction for fuel cells.

Apart from polyelectrolyte functionalized carbons and carbon nitrides based materials,

conducting polymers (CPs) based materials are also studied and applied as metal-free carbon-

based catalysts for oxygen reduction. Conducting polymers with unique 1D delocalized conjugated structures exhibit excellent electrical, electrochemical properties, and have a variety of applications, such as dye sensitized solar cells (DSSCs),211,212 fuel cells,213 biosensors, and

others.214,215,216 In particular, the high conductivity and electroactive properties make them

promising candidates for catalyzing reactions in fuel cells. Conduction polymers based metal-free

ORR electrocatalysts include inherent CPs, CPs/carbon composites, and CPs-derived heteroatom-

doped carbons. The typical CPs are used for electrocatalysis include polythiophene (PTh) and

poly(3,4-ethylenedioxythiophene) (PEDOT),217 polyaniline (PANi),218 and polypyrrole (PPy).219

PANi and PPy are excellent precursors for further pyrolysis to synthesize N-doped carbon

materials.213 Conducting polymers incorporated into carbon materials like carbon black are also

used for fuel cell applications.220 The CPs/carbon composites exhibit enhanced conductivity and

catalytic activity. Besides, a conductive PEDOT electrode is fabricated via a vapor-phase

polymerization method using porous Goretex film as substrate. This PEDOT electrode displays

high catalytic performance for ORR in the electrolytes with different pH values. Moreover, the

PEDOT electrode with a good long-term stability is free from CO poisoning.221 In this particular case, the mechanism of the oxygen reduction electrocatalysis involves a redox cycling process of

the PEDOT. The PEDOT is oxidized to a higher oxidation degree when the O2 molecules are

absorbed onto its surface and then reduced by itself.

In this chapter, different metal-free materials for oxygen reduction have been introduced, including heteroatom-doped carbons, edge-functionalized carbons, polyelectrolyte functionalized carbons, carbon-nitride based materials, and conducting polymers. The ORR enhancement mechanism from heteroatom doping has also been discussed. Table 1.2 summarizes the preparation routes and electrochemical performance of many important metal-free catalysts. The synthesis methods include solution chemistry, template-based method, CVD, pyrolysis, post treatment, and many other combined approaches. These metal-free materials exhibit excellent electrocatalytic performance, good operation stability, and provides alternatives for the commercial Pt/C catalysts in fuel cells.

Table 1.1 Electrocatalytic performance and preparation methods of metal-free carbon nanomaterials.

1.4 Research Objectives

Researchers have developed precious metals, transition metal oxides, carbon-based

transition metal hybrids and metal-free materials for oxygen reduction. In particular, metal-free

carbon based nanomaterials with low cost, large surface area, high conductivity, rich electroactive

sites, and corrosion resistance, are greatly desired for the electrochemical reduction of oxygen.

Heteroatom-doped carbon nanomaterials have been developed by different strategies, and are

considered as the most important metal-free catalysts for oxygen reduction. Although a variety of

heteroatom-doped carbon materials have been developed, the electrochemical performance of

ORR can still be improved. Besides, most ORR studies of the metal-free electrocatalysts were

conducted on a three-electrode system in alkaline medium. The electrochemical behavior of the

metal-free carbon-based nanomaterials in the acidic medium and single fuel cell still remains

largely unclear.

In this work, metal-free carbon-based nanomaterials for oxygen reduction electrocatalysis were developed by heteroatom doping and rational structure design. Different metal-free materials including CNTs, graphene, graphitic carbon nitrides, and biomass, were selected as the building blocks for constructing novel carbon nanomaterials with highly porous structures and abundant electrochemical active sites. The oxygen reduction performance of these materials were investigated on the three-electrode system in alkaline and acidic medium. PEMFCs based on metal-free carbon catalysts at cathode were also fabricated and studied. The rationally designed carbon-based materials exhibit excellent oxygen reduction catalytic activities in both the three- electrode electrochemical cell and the single PEMFC cell.

CHAPTER II: Heteroatoms-codoped Carbon Nanotubes as High Performance Metal-free

Electrocatalysts for Energy Conversion

Abstract

Previously, researchers have demonstrated that nitrogen-doped carbon nanomaterials are

effective metal-free ORR catalysts due to the heteroatom induced charge redistribution. Based on

this understanding, we developed a new class of carbon nanotubes with heteroatoms codoping by

different heteroatoms as metal-free electrocatalysts for oxygen reduction and hydrogen evolution

reactions. Boron and nitrogen codoped carbon nanotubes (BN-CNT) were synthesized by thermal annealing of o-MWCNTs/boric acid precursors and exhibit excellent ORR and HER

electrocatalytic activity. Besides, nitrogen and fluorine codoped carbon nanotubes (N, F-CNTs) were prepared by simple ball milling technique in large quantity. The metal-free N, F-CNTs, show

outstanding catalytic activity towards ORR with a high power density of ~200.2 w/g in a PEMFC.

The BN-CNT and N, F-CNTs with excellent electrochemical activities exhibit synergetic effect of

codoping, providing an effective strategy in the development of metal-free electrocatalysts.

2.1 Introduction

As can be seen from above discussions, it has been demonstrated that nitrogen doped carbon

materials are promising alternative catalysts to Pt/C.15 Nitrogen-doped graphene and nitrogen-

doped carbon nanotubes exhibit an extreme low ORR overpotential, larger reduction current, excellent long-term stability, and good tolerance against methanol cross-over and CO poisoning.15

According to the theoretical calculations,91 the N-doping induced charge transfer in the carbon materials has been identified to be responsible for catalyzing oxygen reduction by changing the adsorption mode of O2. Apart from the electron-rich N, researchers have also extended the study

of doping atoms to include electron-deficient B.199 Metal-free boron-doped CNTs have also been

successfully synthesized by chemical vapor deposition with tunable boron doping concentration.

The B-CNT shows good ORR performance, excellent stability, and are free from methanol cross-

over and CO poisoning.199 Theoretical calculation has suggested that the positively charged B

199 atoms enhances the O2 chemisorption related to ORR. Moreover, the π-electrons accumulate on the B dopant sites could be easily transferred to the chemisorbed O2 molecules for ORR. Thus, the

ORR enhancement mechanism of N doped carbon materials and B-CNT are quite similar, which

is originated from the heteroatom doping induced charge redistribution. The heteroatom doping

process through substituting carbon atoms with heteroatoms like N, B, P and S atoms can

effectively tailor the electron-donating properties, and hence enhanced catalytic activities. The

electronegativity difference between carbon atom and heteroatoms results in the localized charge

redistribution, generating large quantities of electroactive sites for catalysis.

Based on this understanding, we have developed heteroatom-codoped carbon materials with

more than one type of heteroatoms as metal-free electrocatalysts for energy conversion

applications. The heteroatoms were selected according to the electronegativity (χ) and the atomic

size. In particular, codoping with two elements, one with higher and one with lower

electronegativity than that of C (χ =2.55), for example, B (χ =2.04) and N (χ =3.04) may create a unique electronic structure with a synergistic coupling effect between heteroatoms. In the first part

of the present work, boron and nitrogen codoped carbon nanotubes (BN-CNT) were synthesized by thermal annealing of o-MWCNTs/boric acid precursors to demonstrate a bifunctional metal- free catalysts for ORR and HER. The BN-CNT shows better ORR performance than Pt/C and

excellent HER activity in electrolytes at different PH values ranging from 0 to 13.

We further synthesized and explored nitrogen and fluorine codoped carbon nanotubes (N, F-

CNTs), which were prepared by simple ball milling technique in large quantity at low cost. The

metal-free N, F-CNTs shows outstanding catalytic activity towards ORR on the three-electrode system and then applied in the fuel cell as cathode catalyst. The N, F-CNTs based single cell show a high power density of ~200.2 W/g.

The BN-CNT and N, F-CNTs with excellent electrochemical activities exhibit synergetic effect upon codoping which provides an effective pathway in the development of metal-free electrocatalysts for energy conversion and storage.

2.2 Experiment Section

Synthesis of oxidized carbon nanotubes (o-MWCNTs)

The commercially available multiwall carbon nanotubes were first purified by 0.1M HCl at

80o C for 24 hours to remove the residual catalysts. The HCl treated MWCNTs were then washed

by DI-water and dried in vacuum. 200 mg of HCl treated MWCNTs was dispersed in 200 mL of

a 3:1 mixture of concentrated H2SO4 (98%) and HNO3 (70%) and heated in a water bath at 60ºC

for 3 hours with vigorous stirring.223 After oxidization, the o-MWCNTs were recovered by

filtration and purified by DI-water before doing dialysis for 3 days followed by freeze drying.

Synthesis of boron, nitrogen-codoped carbon nanotubes (BN-CNT) and boron-doped carbon nanotubes (B-CNT)

Typically, 100 mg of o-MWCNTs and 10 mg boric acid were dispersed in 30 mL DI-water.

The mixture was sonicated for 0.5 h and stirred for 12 h, followed by freeze drying, and this sample was noted as 100-OCNT-10B. To adjust the doping level, different amounts of boric acid (5 mg

and 25 mg) were mixed with 100mg o-MWCNTs which were noted as 100-OCNT-5B and 100-

OCNT-25B, respectively. B, N-CNT with different doping level were prepared by annealing the precursors in the mixture of NH3 and Ar (NH3: Ar = 200 mL/min: 100 mL/min) at 900ºC for 1

hour. The products are noted as BN-CNT 100/5, BN-CNT 100/10 and BN-CNT 100/25,

respectively. The BN-CNT in this document is corresponding to BN-CNT 100/10. Boron-doped

carbon nanotubes (B-CNT) were prepared by annealing the precursor 100-OCNT-10B in Ar (200 mL/min) at 900º C for 1 hour.

Synthesis of nitrogen-doped carbon nanotubes (N-CNT) and reduced carbon nanotubes (r-CNT)

100 mg o-MWCNTs were dispersed in 30mL DI-water and sonicated for 0.5 hour. The o-

MWCNTs dispersion was then freeze dried. N-CNT was prepared by annealing o-MWCNTs in

the mixture of NH3 and Ar (NH3: Ar = 200 mL/min: 100 mL/min) at 900º C for 1 hour. Reduced

CNTs (r-CNT) were prepared by annealing o-MWCNTs in the mixture of H2 and Ar (H2: Ar =

200mL/min: 100mL/min) at 900ºC for 1 hour.

Preparation of N, F codoped CNTs (N, F-CNTs) and BM-MWCNTs:

Commercial multiwall carbon nanotubes (MWCNTs) were selected as the starting materials and ammonia fluoride was introduced as heteroatom sources of nitrogen and fluoride.

MWCNTs were first purified by 0.1 M HCl for 24 h at 80o C to remove the metallic impurities, if any. The N, F-CNTs were prepared by the ball milling method. Typically, 0.1g MWCNTs and

1.0g NH4F were placed in an 80 mL grind bowl with 100 g balls of 5 mm in diameter, and then 4

mL N-Methyl-2-pyrrolidone (NMP) was added to the container to assist the milling process. The container was fixed in the planetary ball-mill machine and agitated with 500 rpm for 48 h. The

obtained product was washed with acetone and DI-water for several times to remove the NMP.

The product was further washed with 0.1 M HCl to remove metal impurities, if any, followed by freeze drying. N, F-CNTs with different heteroatom doping levels were synthesized by tuning the weight ratio of the reactants and ball milling reaction time. In detail, different amount of NH4F

(2.0 g, 1.0 g, 0.4 g and 0.2 g) were mixed with 0.1 g MWCNTs at 500 rpm for 48 h, and the corresponding product were noted as N, F-CNTs-A, N, F-CNTs-B, N, F-CNTs-C and N, F-CNTs-

D, respectively. The N, F-CNTs mentioned in the paper is corresponding to N, F-CNTs-B. Besides,

1.0 g NH4F and 0.1 g MWCNTs were also milled for different reaction time (24 h, 48 h and 72 h),

the products were noted as N, F-CNTs-24h, N, F-CNTs-48h and N, F-CNTs-72h.

For comparison purpose, 0.1 g MWCNTs and 4 mL NMP were directly ball milled at 500 rpm for 48 h without any NH4F involved. The product after purifying with HCl and freeze dried,

was denoted as BM-MWCNTs.

Structure characterization

SEM images were taken using high resolution field emission scanning electron microscopy

FEI Nova Nanolab200. TEM images were taken by transmission electron microscopy FEI Tecnai

TF20 FEG. XRD was carried out on a Miniflex Desktop X-ray Diffractometer. XPS was conducted on VG Microtech ESCA 2000 using a monochromic Al X-ray source (97.9 W, 93.9 eV). The

Raman spectra were collected by the Raman spectroscopy (Renishaw), using 514 nm laser.

Thermogravimetric Analysis was performed on TGA (TA instrument Q50) with a heating rate of

10o C/min in air condition. Nitrogen adsorption isotherms were measured at -196° C on TriStar II

3020 Version 2.00 volumetric adsorption analyzers manufactured by Micromeritics. Before

adsorption measurements, each sample was degassed at 150° C under vacuum for 24 h. The specific surface area of the samples was calculated using the Brunauer-Emmett-Teller (BET) method within the relative pressure range of 0.02-0.30. Incremental pore size distributions were obtained from the nitrogen adsorption isotherms by the Dollimore-Heal method provided by

Micromeritics.

Electrochemical characterization

Electrochemical measurements were performed on an electrochemical workstation

(CHI760C, CH Instrument, U.S.A.) with a three-electrode electrochemical cell. All of the experiments were conducted at room temperature. A platinum wire was used as the counter electrode, and a silver/silver chloride electrode (Ag/AgCl) and saturated calomel electrode (SCE) were used as reference electrodes in O2 saturated 0.5 M H2SO4 and 0.1 M KOH electrolytes,

respectively. The catalyst was drop casted on the glass carbon, followed by casting with a Nafion solution (0.05 wt % in ethanol) as the binder. The loading amount of the catalyst was about 250

μg/cm2.

The ORR activity of the electrocatalysts was evaluated by cyclic voltammetry (CV) and

linear sweep voltammetry (LSV) techniques on rotating disk electrodes (RDEs) in oxygen

saturated 0.1 M KOH electrolyte. Oxygen evolution activity was also evaluated in 0.1 M O2- saturated KOH. Hydrogen evolution (HER) measurement was conducted in N2-saturated electrolytes with different PH values. These electrolytes included H2SO4 (PH=0, PH=3, PH=5),

Na2SO4 (PH=7) and KOH (PH=13).

Fabrication of membrane electrode assembly (MEA)

At first, 15 mg cathode catalyst powders (N, F-CNTs and BM-MWCNTs) were dispersed in the mixture of 0.5 mL distilled H2O, 1.0 mL isopropanol and 300 mg 5 wt.% Nafion solution by

ultra sonication and stirring. Then the cathode catalyst ink was coated on the 5.0 cm2 cathode

carbon paper (ElectroChem Inc, Carbon Micro-porous Layer (CMPL)) with a brush. Commercial

Hispec 4100 Pt/C catalyst was used in the anode. The anode catalyst ink was prepared by the same method, and then brush coated on the anode carbon paper. The catalyst loadings in the cathode were 3.0 mg/cm2 and Pt loadings in the anodes were 1.0 mg/cm2, respectively. MEAs were fabricated by hot pressing the cathode, DuPont Nafion membrane 211 and the anode together under the pressure of 60lb cm-2 at 130oC for 2 min.

Test of PEM fuel cell

The fuel cell performance was tested at a single cell system. Single fuel cell was assembled

with the as-prepared MEAs. H2 and O2 were used as the fuel and oxidant with 30 psi in the test

process. Fuel cell polarization plots were recorded using fuel cell test stations (Arbin Instruments,

USA). Pure hydrogen and O2, humidified at 80° C, were purged to the anode and cathode,

-1 -1 respectively, at flow rates of 300 mL min (H2) and 500 mL min (O2).

2.3 Results and Discussion

The morphology of the BN-CNT was investigated by SEM and TEM. As can be seen in

Figure 2.1 (a), the BN-CNT prepared by freeze drying shows loosely packed foam-like structure.

The SEM images at higher magnification in Figures 2.1 (b) and (c) present each sheet of the BN-

CNT consisting of abundant nanotube bundles. TEM images in Figure 2.2 provide more detailed information of BN-CNT. As can be seen in Figure 2.2 (a), the BN-CNT does not aggregate too much, providing more potential electroactive sites for catalysis. Figure 2.2 (c) and (d) clearly show

40 that the BN-CNT has a well-defined multiwall structure with a wall number of 8~15. The diameter of the BN-CNT nanotube is between 5 to 10 nm. The multiwall structure ensures the good conductivity of the BN-CNT, which is a key factor related the electrocatalytic activity.

Figure 2.1 Morphology of BN-CNT. (a). SEM image of freeze dried BN-CNT foam, the scale bar is 10 μm. (b), (c) SEM images of BN-CNT sheets at different magnification. The scale bar is

(b). 5 μm and (c). 300nm, respectively.

Figure 2.2 Morphology of BN-CNT. (a), (b). TEM images of BN-CNT, the scale bar is 200 nm

and 20 nm for (a) and (b), respectively. (c), (d) TEM images of individual nanotube structure of

BN-CNT. The scale bar is 10 nm.

Raman and X-ray diffraction measurements were carried out to analyze the structure of the

heteroatom doped carbon nanotubes. Figure 2.3 (a) presents the Raman spectra of BN-CNT, B-

CNT, N-CNT and r-CNT. All of them possess two characteristic peaks, D band (a feature of

disordered carbon or defective graphitic structures) at 1349 cm−1 and G band (a feature of graphitic

layers) at 1580 cm−1. The D’ band at around 1607 cm−1 are also observed for B-CNT and BN-

CNT, corresponding to the structure defects related to boron doping. The peak at around 2690

cm−1 is considered as 2D band existing in the heteroatoms doped CNT, but not in r-CNT. The

intensity ratio of D band to G band (ID/IG) is also calculated to evaluate the defect degree of the

doped carbon nanotubes. As expected, BN-CNT shows the highest ID/IG among all these samples

due to the codoping induced defects. The introduction of boron and nitrogen atoms into the carbon

nanotubes results in a great damage in the actual graphitic C-C bond structure. The ratio of ID/IG

for N-CNT and B-CNT are 1.25 and 1.16, respectively. Both of them has a higher ID/IG ratio than

r-CNT without heteroatoms. Figure 2.3 (b) presents the XRD patterns of all these samples. As can

be seen, all the XRD patterns show a broad peak at around 25.3°, corresponding to the C (002)

plane. The presence of the graphitic carbon peak at 25.3° is in good agreement with the Raman

results. To further study the elemental composition of heteroatom doped carbon nanotubes, X-ray

photoelectron spectroscopy (XPS) analysis was carried out, and the survey spectra of different

samples are presented in Figure 2.3 (c). As can be seen, all of XPS spectra process a strong C 1s

peak at ~280 eV and a weak O 1s at ~533 eV. The N-CNT has N 1s peak at ~ 400 eV, while the

B-CNT shows B 1s peak at ~191 eV. As expected, the BN-CNT presents both N 1s (7.6%) and B

1s (5.3%) peaks. It should be noted that the doping level of N and B in the BN-CNT is higher than that of N-CNT (N: 2.3%) and B-CNT (B: 2.3%), indicating a synergetic effect in the doping process of the heteroatoms B and N. The B-CNT presents a relatively high O content of 12.9%, and this is related to the oxygen rich precursor, boric acid. As the ammonia is a reductive gas, the

oxygen content is greatly reduced after high temperature treatment in the NH3. To further

investigate the state of C, O, B and N in the codoped BN-CNT, high resolution XPS spectra were

carried out.

Figure 2.3 Structure analysis of BN-CNT. (a).Raman spectra and (b).XRD patterns of B-CNT,

N-CNT, BN-CNT and r-CNT. (c). XPS survey spectra of B-CNT, N-CNT and BN-CNT. (d). BET of BN-CNT, insert: pore size distribution of BN-CNT.

Figure 2.4 XPS analysis of BN-CNT. (a) High resolution XPS C1s deconvoluted spectrum of

BN-CNT; (b) High resolution XPS B1s deconvoluted spectrum of BN-CNT; (c). High resolution

XPS N1s deconvoluted spectrum of BN-CNT and (d). High resolution XPS O1s deconvoluted spectrum of BN-CNT.

As shown in Figure 2.4 (a), the high resolution C 1s spectrum of BN-CNT can be curve fitted into C-C (284.5 eV), C-B (285.1 eV), C-N (285.7 eV) and O=C-OH (289.5 eV), respectively. For the high resolution B 1s spectrum of BN-CNT in Figure 2.4 (b), the B 1s can be deconvoluted into

two main component peaks located at 190.3, and 192.0 eV, corresponding to B-C and B-O bonding,

respectively. An extreme weak B-N bonding can be observed at 191.2 eV, indicating negligible

B-N domains in the carbon structures. The weak B-N bonding at 398.1 eV is also shown in the

high resolution N 1s spectrum of BN-CNT, which is in good agreement with the B1s spectrum.

Besides, a distinct peak at 398.4 eV in the N 1s spectrum is corresponding to the pyridinic N,

which is considered as highly electroactive towards different electrochemical reactions such as

oxygen reduction. The high resolution O 1s spectrum of the BN-CNT is presented in Figure 2.4

(d), the O 1s can be fitted into two peaks located at 531.6 eV and 532.9 eV, corresponding to C=O

and B-O & C-OH, respectively. This is consistent with the results from C 1s and B 1s spectra.

Based on the XPS analysis, it can be considered that BN-CNT consisting of most pyridinic N

species without obvious B-N domains.

To better understand the codoping effect on different electrochemical behavior, the high

resolution XPS spectra of N-CNT and B-CNT were also carried out and presented in Figure 2.5.

In the N-CNT, the high resolution C1s spectrum can be fitted into C-C (284.6 eV), C-N (286.1 eV)

and C-O (288.9 eV), respectively. The C-N bonding is furthered investigated by the high resolution

N 1s spectrum. As can be seen in Figure 2.5 (c), there are two main N binding configurations in the ammonia treated carbon nanotubes, which are pyridinic-N at 398.4 eV and pyrrolic-N at 399.8 eV. Thus, the C-N configuration in the N-CNT is different from the BN-CNT. Figure 2.5 (b) shows the C 1s spectrum of the B-CNT, which can be deconvoluted into C-C (284.5 eV), C-B (285.4 eV) and C-O (288.9 eV), respectively. The high resolution B 1s spectrum can be fitted into two peaks, a strong B-O peak at 192.6 eV and a weaker B-C peak at 191.7 eV. The strong B-O peak is in good agreement with the high oxygen content from XPS analysis. The oxygen rich precursor boric acid may dope the carbon structure with oxygen groups.

Figure 2.5 XPS analysis of N-CNT and B-CNT. High resolution XPS C1s deconvoluted

spectrum of (a) N-CNT and (b). B-CNT; (c) High resolution XPS N1s deconvoluted spectrum of

N-CNT; (d) High resolution XPS B1s deconvoluted spectrum of B-CNT. High resolution XPS

C1s deconvoluted spectrum of (e) N-CNT and (f). B-CNT.

Figure 2.6 presents the proposed chemical structure of boron, nitrogen coped BN-CNT based on the XPS analysis. There are mainly C-B, C-N and very few B-N bonds in the carbon structure according to the XPS results. As we shall see later, the BN-CNT are demonstrated to be a bifunctional electrocatalyst for oxygen reduction (ORR) and hydrogen evolution reactions (HER).

As proposed in Figure 2.6, ORR and HER mainly occur on the electroactive sites, which are located at the heteroatoms and adjacent carbons. According to previous study, the heteroatom doping will introduce charge redistribution (blue dots in Figure 2.6) due to the different electronegativity between dopants and carbon atoms.

Figure 2.6 Proposed chemical structure and electrocatalytic process of oxygen reduction and hydrogen evolution on BN-CNT.

Figure 2.7 Catalytic activity towards electrochemical reduction of oxygen in basic electrolyte

at room temperature. (a) Cyclic voltammetry (CVs) of BN-CNT in O2-saturated and N2-saturated

0.1M KOH obtained at a sweep rate of 50 mV s −1; (b) LSVs of BN-CNT on the RRDE at 1600rpm

in O2-saturated 0.1M KOH, the insert: electron transfer number estimated from the ring and disk

currents; (c). LSVs of BN-CNT at different rotating speeds (400rpm, 600rpm, 900rpm, 1200rpm

−1 and 1600rpm) at a sweep rate of 5 mV s in 0.1M O2-saturated KOH; (d). Koutecky-Levich plots

of BN-CNT derived from Figure 2.7 (c); (e). Durability curves (i-t) of BN-CNT and Pt/C obtained in at -0.3 V versus SCE at a rotation rate of 1000 rpm.

The ORR electrocatalytic activity of BN-CNT was investigated on a conventional three- electrode system. Figure 2.7 (a) presents cyclic voltammetry (CV) curves measured in O2 and N2

saturated 0.1M KOH solutions. As can be seen, a voltammogram without any obvious peak was

observed in the absence of O2, however, when oxygen was introduced, a characteristic ORR peak

starting at about -0.20 V was observed, showing the electrochemical reduction of oxygen. To better investigate the kinetics of ORR process, rotation ring-disk electrode (RRDE) was carried out to evaluate the ORR performance of the BN-CNT. As shown in 2.7 (b), the ORR onset potential

(Eonset) and half-wave potential (E1/2) of the BN-CNT is about -0.03 V and -0.12 V, respectively.

According to the current densities collected on the ring and disk electrodes, the electron transfer

number is calculated to be about 4.0, as shown in the insert of Figure 2.7 (b). A series of linear

sweep voltammograms (LSVs) on a rotating disk electrode (RDE) of BN-CNT electrode were also measured from 400 to 1600 rpm, respectively. As can be seen in Figure 2.7 (c), LSVs for BN-CNT show typical increasing current with higher rotational speeds due to shortened diffusion distance at higher speeds. Eonset and E1/2 observed in Figure 2.7 (c) are in consistent with LSV curve on

RRDE. LSV curves at RDE under different rotation speed were also used to estimate the electron

transfer number in the ORR process. The corresponding Koutechy-Levich (K-L) plots at different

potentials are shown in Figure 2.7 (d), and the electron transfer number was calculated to be 4.

The durability of the BN-CNT was evaluated by the chronoamperometric response under a

constant cathodic voltage of -0.3 V. As shown in Figure 2.7 (e), the metal-free catalyst shows

excellent stability with a negligible attenuation after 30000 s, and a high relative current of 97.5%

yet persists, which is much better than the commercial Pt/C catalyst.

To investigate the heteroatom doping effect, the CV and LSVs of B-CNT, N-CNT and r-

CNT were also measured for comparison. As can be seen in Figure 2.8 (a), BN-CNT, B-CNT and

N-CNT have distinct reduction peaks of oxygen at around -0.2 V, while r-CNT without any doping shows negligible ORR activity. It should be noted that the BN-CNT exhibited a more positive O2

reduction onset and half-wave potentials than that of single heteroatom doped B-CNT and N-CNT,

as depicted in Figure 2.8 (e). More detailed electrochemical behavior can be seen in the LSVs

obtained on RDE of B-CNT, N-CNT and r-CNT electrodes. As shown in Figure 2.8 (b) and (c),

the N-CNT shows a more positive onset potential than B-CNT, while B-CNT shows a better kinetic

process of oxygen reduction. The boron, nitrogen codoped BN-CNT takes advantage of both boron

doping and nitrogen doping, exhibiting an extreme low ORR overpotential and a positive half-

wave potential, which can be attributed to the high density of electroactive catalytic sites.

According to the XPS analysis, there are mainly three type of electroactive sites in the BN-CNT,

which are B-C sites, N-C sites and very small B-N domains. As mentioned before, B (χ =2.04) has

a lower electronegativity than that of C (χ =2.55), while N (χ =3.04) has a higher electronegativity.

In the B-C and B-N domains, the electrons are transferred to the atoms with a higher

electronegativity, consequently the boron dopant atoms get positively charged.

Figure 2.8 Catalytic activity towards electrochemical reduction of oxygen in basic electrolyte

at room temperature. (a) CVs of B-CNT, N-CNT, BN-CNT and r-CNT in O2-saturated 0.1M

KOH obtained at a sweep rate of 50 mV s −1; (b) LSVs of BN-CNT on the RRDE at 1600rpm in

O2-saturated 0.1M KOH, the insert: electron transfer number estimated from the ring and disk

currents; LSVs of (b). B-CNT (c). N-CNT and (d). r-CNT at different rotating speeds (400rpm,

−1 600rpm, 900rpm, 1200rpm and 1600rpm) at a sweep rate of 5 mV s in 0.1M O2-saturated KOH;

(e). LSVs of B-CNT, N-CNT, BN-CNT, r-CNT and Pt/C electrodes obtained at a rotation speed

−1 of 1600rpm at a sweep rate of 5 mV s in 0.1M O2-saturated KOH.

On the contrary, in the N-C sites, the C atoms adjacent to the N atoms get positively

charged due to its lower electronegativity than that of the N atom. The doping-induced charge

redistribution further changes the adsorption modes of oxygen and facilitate the oxygen reduction

process.91 The better electrocatalytic ORR performance of codoped CNT can be understood from

two aspects. First, the doping level of boron and nitrogen in the BN-CNT is higher than that in B-

CNT or N-CNT, indicating the boron and nitrogen atoms can facilitate synergistically the other

atom’s incorporation. The higher doping level provides more potential catalytic active sites,

resulting in better ORR performance than B-CNT and N-CNT. On the other hand, the codoping effectively combines the features of both B and N, generating a synergistic effect in catalyzing oxygen reduction. Without incorporation of heteroatoms, the r-CNT show negligible catalytic activity and sluggish kinetics towards oxygen reduction, as shown in Figure 2.8 (d). According to the LSVs obtained at 1600rpm in Figure 2.8 (e), the metal-free BN-CNT even showed better ORR performance compared to the commercial Pt/C catalyst.

Figure 2.9 LSVs of (a). BN-CNT 100/5 and (b). BN-CNT 100/25 at different rotating speeds

−1 (400rpm, 600rpm, 900rpm, 1200rpm and 1600rpm) at a sweep rate of 5 mV s in 0.1M O2-

saturated KOH; (c). LSVs of BN-CNT 100/5, BN-CNT 100/10 and BN-CNT 100/25 electrodes

−1 obtained at a rotation speed of 1600rpm at a sweep rate of 5 mV s in 0.1M O2-saturated KOH.

(d). XPS survey spectra of BN-CNT 100/5, BN-CNT 100/10 and BN-CNT 100/25.

After confirming the enhancement of codoping effect, BN-CNT with different doping levels were synthesized by tuning the weight ratio of precursors. BN-CNT 100/25 with a higher percentage of boric acid and BN-CNT 100/5 were prepared by the same route. The B and N

contents were investigated by XPS and the survey spectra are shown in Figure 2.9 (d). The element

compositions of B, N, C and O of B-CNT, N-CNT, BN-CNT with different doping levels are summarized in Table 2.1. As can be seen in Figure 2.9 (d) and Table 2.1, the BN-CNT 100/25 exhibits the highest B (8.8%) and N (9.3%) content and lowest C content (76.3%).

Table 2.1 XPS analysis of heteroatoms doped carbon nanotubes.

Samples C (at.%) N (at.%) B (at.%) O (at.%)

B-CNT 84.8 / 2.3 12.9

BN-CNT 100/5 86.5 5.7 4.4 3.4

BN-CNT 100/10 84.3 7.6 5.3 2.8

BN-CNT 100/25 76.3 9.3 8.8 5.6

N-CNT 93.9 2.3 / 3.8

Although BN-CNT 100/25 and BN-CNT 100/10 have been through the same ammonia treatment (temperature and reaction time), the higher N content is observed due to the higher B doping level. The B dopants facilitate the formation of B-N bonding due to electrostatic attraction towards N ions, and this is in good agreement with previously discussed synergistic doping effect.

However, the electrocatalytic activity of the BN-CNT is not only related to the doping effect, and the electrical conductivity is also an important factor in this respect. For the BN-CNT 100/25 with the highest heteroatom doping level, the excessive B-N domains may degrade the conductivity and then decrease the electrocatalytic performance. This claim is demonstrated by the electrochemical performance in Figure 2.9. As shown in Figure 2.9 (a) and (b), BN-CNT 100/5 and BN-CNT

100/25 show worse ORR electrocatalytic activity, as reflected by the more negative half-wave

potential and sluggish kinetics. The BN-CNT 100/5 with lower B and N doping level, does not have enough electroactive catalytic sites. Although BN-CNT has a very positive onset potential,

the kinetics of oxygen reduction is quite reluctant due to the low density of active sites. In the case

of BN-CNT 100/25, as discussed, the excessive heteroatoms doping damages the carbon lattice

structures and decreases the conductivity, resulting in the poor electrocatalytic performance. Thus,

as shown in Figure 2.9 (c), BN-CNT 100/10 with modest doping level, which has good

conductivity and abundant catalytic sites, present the best ORR performance as expected.

Figure 2.10 Catalytic activity towards electrochemical oxygen evolution in basic electrolyte

at room temperature. (a). LSVs of B-CNT, N-CNT, BN-CNT, r-CNT and Pt/C electrodes

−1 obtained at a rotation speed of 1600rpm at a sweep rate of 5 mV s in 0.1M O2-saturated KOH.

(b). LSVs of BN-CNT 100/5, BN-CNT 100/10 and BN-CNT 100/25 electrodes obtained at a

−1 rotation speed of 1600rpm at a sweep rate of 5 mV s in 0.1M O2-saturated KOH. (c).

Summarization of ORR and OER performance of B-CNT, N-CNT, BN-CNT with different doping

level, r-CNT and Pt/C electrodes. All the potentials in the table are vs. RHE.

The electrocatalytic oxygen evolution reaction (OER) activities were also investigated in

O2-saturated 0.1M KOH solution using a standard three-electrode system. Figure 2.10 (a) shows

the LSV curves of the different catalysts, including BN-CNT, B-CNT, N-CNT, r-CNT and Pt/C.

The boron, nitrogen codoped BN-CNT with the largest current density exhibited the earliest onset

potential of 1.52V (Vs. RHE), lower than that of the B-CNT ( 1.65 V), N-CNT ( 1.61 V),

suggesting that∼ codoping of boron and nitrogen can effectively enhance∼ the oxygen ∼evolution.

Besides, the heteroatom doped carbons, including BN-CNT, B-CNT and N-CNT, showed much

better OER catalytic activity than r-CNT with an onset potential of 1.74 V. It should be noted

here that the OER performance of the metal-free BN-CNT is superior∼ to Pt/C catalyst. The OER

catalytic performance of BN-CNT with different doping levels were also measured in the alkaline medium, as shown in Figure 2.10 (b). Similarly for the ORR performance, BN-CNT 100/10 with a modest doping level exhibited the best catalytic activity towards oxygen evolution. The modest doping effectively balances the introduction of heteroatoms active sites and conductivity, achieving the optimized electrochemical performance towards different reactions. The ORR and

OER performance of heteroatoms doped CNT and Pt/C are summarized in Figure 2.10 (c).

Figure 2.11 Electrocatalytic activity towards hydrogen evolution reaction. LSVs of BN-CNT,

−1 N-CNT, B-CNT, r-CNT and Pt/C obtained at a sweep rate of 10 mV s in (a). 0.5M N2 saturated

H2SO4 and (b). 0.1M N2 saturated KOH; Tafel plots of BN-CNT, N-CNT, B-CNT, r-CNT and

Pt/C obtained in (c). 0.5M N2 saturated H2SO4 and (d). 0.1M N2 saturated KOH; Potential value

of BN-CNT, N-CNT, B-CNT, r-CNT and Pt/C when the current density is 10mA/cm2 in (e). 0.5M

N2 saturated H2SO4 and (f). 0.1M N2 saturated KOH.

The boron, nitrogen codoped BN-CNT are also demonstrated to be efficient for hydrogen evolution reaction in both acidic and alkaline medium. Figure 2.11 (a) and (b) show the LSV curves of different electrode materials in 0.5 M N2-saturated H2SO4 and 0.1 M N2-saturated KOH, respectively. Figure 2.11 (c) and (d) present the Tafel plots derived from Figure 2.11 (a) and (b), respectively. According to the LSVs in Figure 2.11 (a) and (b), the boron, nitrogen codoped BN-

CNT presents much higher electrocatalytic activity towards HER than N-CNT, B-CNT in different electrolytes, while r-CNT shows negligible HER performance. In the acidic medium, BN-CNT shows the Tafel slope of 80.6 mV/decade, which is better than those of N-CNT (153.4 mV/decade),

B-CNT (142.3 mV/decade) and undoped r-CNT (182.1 mV/decade), indicating BN-CNT has the highest electrocatalytic kinetics for HER. Figure 2.11 (e) presents the overpotential value when the current density achieve 10 mA/cm2 in acidic medium. As can be seen, the metal-free material

BN-CNT shows an extreme low overpotential of 120 mV to achieve the current density of 10

mA/cm2, which is much lower than those of N-CNT (200 mV), N-CNT (290 mV) and r-CNT

(690 mV). This value is only slightly higher than that of Pt/C catalyst (60 mV). The HER electrocatalytic activity of different heteroatom doped CNT in the alkaline medium shows similar

trend as the acidic medium. The HER performance given in Figure 2.11 (b), (d) and (e) follows

the order: BN-CNT > N-CNT > B-CNT > r-CNT. In 0.1 M N2-saturated KOH, BN-CNT shows

the Tafel slope of 275.8 mV/decade, which is better than those of N-CNT (332.5 mV/decade), B-

CNT (535.9 mV/decade), r-CNT (560.8 mV/decade) and Pt/C (320.6 mV/decade).

Figure 2.12 (c) shows the LSV curves of BN-CNT obtained at different scan rates from 5

mV s-1 to 50 mV s-1 in acidic and alkaline electrolytes. As can be seen, the BN-CNT is quite stable

electrodes under different scanning rates. Besides, the long-term stability of the BN-CNT was

carried out in 0.5 M H2SO4 and 0.1 M KOH, respectively. As shown in Figure 2.12 (d), the BN-

CNT can persist at a typical current density of 10 mA cm-2 for up to 40000 s and keeps generating

H2, indicating the excellent durability of the metal-free BN-CNT. Apart from typical electrolytes like 0.5 M H2SO4 (PH=0) and 0.1 M KOH (PH=13), different PH valued electrolytes including

H2SO4 (PH=3, PH=5) and Na2SO4 (PH=7) were also selected as the medium for HER. Figure 2.12

(a) presents the LSV curves of BN-CNT in electrolytes with different PH values. Figure 2. 12 (b)

shows the overpotential reached to 20 mA cm-2 and gradually increases with increasing the PH value of acidic and alkaline electrolytes. The BN-CNT shows the best and worst HER activities in the 0.5 M H2SO4 (PH=0) and in the neutral Na2SO4 (PH=7) electrolytes.

The mechanism of HER enhancement from heteroatom doping is similar to the ORR. In the

boron, nitrogen codoped BN-CNT, the B-C, N-C and C-C with rich defects can act as electroactive sites to enhance the sorption of H+ ions from the electrolytes. The electron-rich N atoms influences

the charge redistribution, affecting the localized chemical and electronic properties. According to

the previous report, nitrogen doping can lower the C-H bond energy and stabilizes H* on the electrode surface, which is very important since it prevents desorption of protons prior to the conversion reaction.39 Similarly, the electron-deficient heteroatom B creates the positive sites for adsorption and catalysis of HER. Thus, the boron, nitrogen codoped BN-CNT exhibit highly

60 promising HER catalytic performance. The synergistic effect in the HER catalysis can also be considered.

Figure 2.12 Electrocatalytic activity towards hydrogen evolution reaction. (a). LSVs of BN-

CNT in different PH-valued electrolytes; (b). Overpotential at 20mA cm-2 in different PH-valued electrolytes on BN-CNT electrode; (c). LSVs of BN-CNT obtained at different sweep rates (5, 10,

−1 20, 50 mV s ) in 0.5M N2 saturated H2SO4 and 0.1M N2 saturated KOH; (d). Stability

measurements of BN-CNT electrode in in 0.5M N2 saturated H2SO4 and 0.1M N2 saturated KOH.

Figure 2.13 Electrocatalytic activity of BN-CNT towards hydrogen evolution reaction. LSVs

of BN-CNT with different doping levels (BN-CNT 100/5, BN-CNT 100/10 and BN-CNT 100/25)

in (a). 0.5M N2 saturated H2SO4 and (c). 0.1M N2 saturated KOH; Tafel plots of B, N-CNT with

different doping levels in (b). 0.5M N2 saturated H2SO4 and (d). 0.1M N2 saturated KOH.

The HER electrocatalytic activity of BN-CNT with different doping level were measured in

0.5 M N2-saturated H2SO4 and 0.1 M N2-saturated KOH, respectively. Figure 2.13 (a) and (c) show the LSV curves of different BN-CNT in 0.5 M N2-saturated H2SO4 and 0.1 M N2-saturated KOH,

respectively. Figure 2.13 (b) and (d) present the Tafel plots derived from Figure 2.13 (a) and (c),

respectively. The HER electrocatalytic activity follow the same trends in acidic and alkaline

medium: BN-CNT 100/10 > BN-CNT 100/5 > BN-CNT 100/25. As shown in Figure 2.13 (b), BN-

CNT 100/10 shows the Tafel slope of 80.6 mV/decade in 0.5 M H2SO4, which is better than that

for BN-CNT 100/5 (193.2 mV/decade) and BN-CNT 100/25 (336.2 mV/decade). Similarly in 0.1

M KOH, according to Figure 2.13 (d), the BN-CNT 100/10 shows the Tafel slope of 275.8 mV/decade, while BN-CNT 100/5 and BN-CNT 100/25 have the Tafel slope of 455.5 mV/decade and 524.2 mV/decade, respectively. As discussed before, BN-CNT 100/10 with modest doping level balances the advantages of rich heteroatom active sites and the conductivity, achieving the optimized electrochemical performance for ORR, OER and HER.

We successfully used thermal annealing strategy to obtain boron, nitrogen codoped BN-CNT

as metal-free bifunctional electrocatalysts for ORR and HER. In BN-CNT, the two dopant atoms have higher and lower electronegativity than C, respectively. Based on this work, we further extended our study to codoping CNTs with nitrogen and fluorine. It would be interesting study because both N (χ =3.04) and F (χ =3.98) have higher electronegativity than that of C (χ =2.55).

Nitrogen, fluorine codoped carbon nanotubes (N, F-CNTs) were synthesized by simple ball milling

commercial multiwall carbon nanotubes with NH4F, which served as the heteroatom sources of

both N and F. The one-step preparation method is quite simple, cost-effective and the product can

63 be obtained in large quantities, providing a new strategy to synthesize carbon nanomaterials codoped with heteroatoms.

Figure 2.14 (a), (b). SEM images of N, F-CNTs, the scale bar is 20 μm and 1 μm for (a) and (b), respectively. (c), (d). TEM images of N, F-CNTs at different magnification. The scale bar is (c).

20nm and (d). 100nm. (e). Proposed chemical structure of N, F-CNTs.

The morphology of the N, F-CNTs was investigated by SEM and TEM. As shown in Figure

2.14 (a) and (b), the N, F-CNTs prepared by ball milling show bamboo-like nanotube structure.

According to the TEM images in Figure 2.14 (c) and (d), the N, F-CNTs exhibit a typical multiwall structure with a diameter of ~15 nm. Figure 2.14 (e) shows the proposed chemical structure of the

N, F-CNTs. As can be seen, the fluorine and nitrogen substitute the carbon atoms in the nanotube, resulting in the defect sites around the heteroatoms as well as C atoms. This proposed structure is further evidenced by structure characterizations presented in Figure 2.15.

Raman and X-ray diffraction measurements were first carried out to analyze the structure of

the heteroatom doped carbon nanotubes. Figure 2.15 (a) presents the Raman spectra of only

MWCNTs, ball milled MWCNTs, and N, F-CNTs. All of them process G band (a feature of

graphitic layers) at 1583 cm−1 and D band (a feature of disordered carbon or defective graphitic

−1 structures) at 1350 cm . The intensity ratio of D band to G band (ID/IG) was also calculated to

evaluate the defect degree of the doped carbon nanotubes. The ID/IG of ball milled MWCNTs (BM-

MWCNTs) is about 1.35, which is higher than that of only MWCNTs (ID/IG=1.25) due to the

defects induced by mechanical stress inside ball milling system. When introducing reagent NH4F

into the ball milling process, N, F-CNTs shows the highest ID/IG ratio, which can be attributed to

their ball milling process and also codoping induced defects.

Figure 2.15 (b) presents the XRD patterns of only MWCNTs, undoped BM-MWCNTs and

N, F-CNTs. All XRD patterns show a distinct diffraction peak around 25.3° and a weaker

diffraction peak around 43.4°, which correspond to the diffraction from (002) and (100) planes of

carbon, respectively. The presence of the graphitic carbon peak at 25.3° is in good agreement with

the Raman results. Interestingly, N, F-CNTs have the widest full-width at half-maximum (FWHM)

of C (002) peak, indicating a large extent exfoliation of MWCNTs via ball milling technique.

Thermogravimetric Analysis (TGA) was employed in air to study the stability of N, F-CNTs

against thermal agitation/oxidation. As can be seen in Figure 2.15 (c), N, F-CNTs exhibits a decomposition temperature at around 574o C, which is lower than that of BM-MWCNTs (615o C)

and MWCNTs (662o C). This indicates that ball milling process introduces abundant defects,

which consequently decreases the thermal stability of the carbon nanotubes because the defective

sites are more reactive to oxygen and other reactants. This observation is in good agreement with

the Raman results. The TGA analysis also demonstrates that N, F-CNTs are free of any metal impurities after HCl washing. X-ray photoelectron spectroscopy (XPS) analysis was carried out to

study the elemental composition. As shown by the survey spectrum in Figure 2. 15 (d), N, F-CNTs

process a strong C 1s peak at ~280 eV, N 1s peak at ~400eV, F 1s peak at ~684 eV and very weak

O 1s peak at ~533 eV, indicating the successful codoping of N and F into the carbon nanotubes,

which is consistent with the proposed chemical structure in Figure 2.14 (e). The atomic content of

C, N and F are 92.4%, 1.3% and 2.6%, respectively. The high resolution N 1s and F 1s spectra are

shown in Figure 2.15 (e) and (f). As evident in Figure 2.15 (e), the C-N bonding is mainly related

to the quaternary-N and pyrrolic-N obtained via ball milling MWCNTs with NH4F.

The ORR electrocatalytic activity was investigated on a conventional three-electrode

system. Figure 2.16 (a) presents the CV curves of N, F-CNTs and undoped BM-MWCNTs in O2-

saturated 0.1M KOH aqueous solutions, respectively. It is clear that the N, F-CNTs possess a

distinct oxygen reduction peak at about -0.3V, while BM-MWCNTs do not indicate any obvious

66 peak. This implies that the ORR catalyst obtained by codoping CNTs with both N and F atoms performs efficiently.

Figure 2.15 Structure analysis of N, F-CNTs. (a). XRD patterns, (b). Raman spectra and (c).

TGA curves of only MWCNTs, BM-MWCNTs and N, F-CNTs. (d). XPS survey spectrum of N,

F-CNTs; (d). High resolution XPS N1s deconvoluted spectrum of N, F-CNTs; (e) High resolution

XPS F1s spectrum of N, F-CNTs.