Controlled Synthesis of Boron Carbide Using Solution-Based Techniques

Home , Boron carbides

Controlled Synthesis of Boron Carbide Using

Solution-Based Techniques

JOSHUA WATTS

A thesis submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy

School of Chemistry, Physics and Mechanical Engineering, Institute for Future Environments,

Queensland University of Technology

Brisbane, 2018 ABSTRACT

This body of work details enhanced techniques for the synthesis and characterisation of boron carbide powder. Two different approaches to boron carbide synthesis are developed. Both techniques, although different, demonstrate improvements upon current commercially implemented methods.

These improvements arise from the identification of fundamental parameters that are key to controlling the mechanism of boron carbide formation. Of primary importance is maximising and maintaining the distribution of carbon and boron precursors before phase formation. To this end, solution mixing of precursors is the basis for these synthesis techniques.

In one method, in-situ polymerisation of vinyl acetate monomer in alcohol increases precursor dispersion while simultaneously controlling the quantity of carbon available for reaction. This approach results in lowered calcination times and temperatures as well as control over boron carbide morphology and purity. In comparison, a simplified approach utilising water and sucrose yields exceptional precursor dispersion through the formation of a gel. In this case, dispersion is maintained via the control of precursor handling and storage conditions. These practices drastically affect the homogeneity of carbon and boron constituents in mixed precursor powders. By preventing the exposure of precursor powder mixtures to humidity at key stages in processing, boron carbide yield is increased by up to 20% and residual carbon impurity is minimised. Comparison of the conditions that lead to optimal phase formation within each technique identifies the key parameters to control during synthesis.

i The inherent transparency of boron and carbon to X-rays as well as their similarity

in electronic structure presents challenges in the characterisation of boron

carbides via X-ray diffraction. To address these issues, the diffractometer

conditions critical to precise measurement of X-ray diffraction high transparency

materials are identified. From this analysis, an optimised X-ray diffraction

methodology is developed for data collection that is complementary to crystal

structure modelling. This approach then makes possible the precise determination

of crystallographic properties of importance to boron carbide synthesis such as

phase quantification, purity and crystal structure.

KEYWORDS

Boric acid

Boron carbide

Calcination

Complexation

Dehydration

Fixed incidence parallel beam

Free-radical polymerisation

Homogenous distribution

Inert processing

Low mass absorption

Morphology

Polyvinyl acetate

Powder

Precursor

Processing conditions

Scanning electron microscopy

Solution-based synthesis

Stoichiometry

Sucrose

Vinyl acetate monomer

Water adsorption

X-ray diffraction

iii TABLE OF CONTENTS

ABSTRACT ...... i KEYWORDS ...... iii TABLE OF CONTENTS...... iv LIST OF FIGURES ...... viii LIST OF TABLES ...... xii LIST OF ABBREVIATIONS ...... xiii DECLARATION OF ORIGINAL AUTHORSHIP...... xv ACKNOWLEDGEMENTS ...... xvi LIST OF PUBLICATIONS ...... xvii

CHAPTER 1 INTRODUCTION 1.1 Thesis Outline ...... 2 1.2 Research Topic ...... 3 1.2.1 Boron Carbide ...... 3 1.2.2 Applications of Boron Carbide ...... 4 1.2.3 Synthesis of Boron Carbide ...... 5 1.2.4 Structural and Compositional Characterisation ...... 7 1.3 Research Problems and Rationale ...... 9 1.4 Objective and Aims ...... 12 1.5 References ...... 14

CHAPTER 2 LITERATURE REVIEW 2.1 Structure and Stoichiometry of Boron Carbide ...... 17 2.2 Commercial Synthesis of Boron Carbide ...... 20 2.3 Solution Based Techniques ...... 22 2.3.1 Synthesis from Slurries ...... 22 2.3.2 Sol-gel Synthesis ...... 22 2.3.3 Synthesis from Polymers ...... 24

2.3.4 Synthesis from Polyols ...... 24 2.4 Preferred Synthesis Criteria for Precursor Homogeneity ...... 26 2.5 References ...... 27

CHAPTER 3 SYNTHESIS AND CHARACTERISATION METHODOLOGY 3.1 Choice of Precursor ...... 30 3.1.1 Boron Precursor ...... 30 3.1.2 Carbon Precursor ...... 35 3.2 Process Methodology ...... 40 3.2.1 Precursor Preparation, Handling and Treatment ...... 40 3.2.2 Precursor Calcination ...... 44 3.3 Characterisation Methodology ...... 46 3.3.1 Precursor Characterisation ...... 46 3.3.2 Boron Carbide Characterisation ...... 48 3.4 References ...... 50

CHAPTER 4 SYNTHESIS FROM POLYVINYL ACETATE 4.1 Chapter Foreword ...... 54 4.2 Article 1: In-Situ Carbon Control in the Preparation of Precursors to Boron Carbide by a Non-Aqueous Solution Technique ...... 56 4.3 Abstract ...... 57 4.4 Introduction ...... 58 4.5 Experimental Methods ...... 61 4.5.1 Starting Materials ...... 62 4.5.2 Polymers ...... 62 4.5.3 PVAcB Powders ...... 63 4.6 Results ...... 65 4.6.1 Polymer Analysis ...... 65 4.6.2 PVAcB Powders Analysis ...... 67 4.6.3 Processing Conditions: Formation of Boron Carbide ...... 69

v 4.6.4 Boron Carbide Product ...... 74 4.7 Discussion ...... 78 4.7.1 Polymer Characteristics ...... 78 4.7.2 PVAcB Powder Composition and Morphology ...... 79 4.7.3 Treatment Atmosphere for PVAcB Powder ...... 82 4.7.4 Residual Carbon ...... 85 4.8 Conclusion ...... 88 4.9 Acknowledgements ...... 89 4.10 References ...... 90

CHAPTER 5 SYNTHESIS FROM SUCROSE 5.1 Chapter Foreword ...... 93 5.2 Article 2: Yield Maximisation in Solution Based Synthesis of Boron Carbide . 95 5.3 Abstract ...... 96 5.4 Introduction ...... 97 5.5 Experimental Methods ...... 99 5.5.1 Precursor Processing...... 99 5.5.2 Characterisation ...... 101 5.6 Results and Discussion ...... 104 5.6.1 Precursor Hydration ...... 105 5.6.2 Precursor Morphology ...... 111 5.6.3 Boron Carbide Product ...... 115 5.7 Conclusion ...... 121 5.8 Acknowledgments ...... 122 5.9 References ...... 123

CHAPTER 6 X-RAY DIFFRACTION FOR LOW MASS MATERIALS 6.1 Chapter Foreword ...... 126 6.2 Article 3: Structural Analysis and Phase Quantification of Boron Carbide Powders Using an Optimised X-Ray Diffraction Technique ...... 128

6.3 Abstract ...... 129 6.4 Introduction ...... 130 6.5 Experimental Methods ...... 133 6.5.1 Materials ...... 133 6.5.2 Solution Synthesis of Boron Carbide Powders ...... 133 6.5.3 Powder X-ray Diffraction ...... 134 6.6 Results ...... 139 6.6.1 Aberration Analysis ...... 139 6.6.2 Measurement of Standards ...... 140 6.6.3 FIPB Analysis of Boron Carbide ...... 141 6.6.4 Structure Refinement ...... 142 6.6.5 Quantification ...... 144 6.7 Discussion ...... 148 6.7.1 Diffraction Geometry Selection ...... 148 6.7.2 FIPB Structural Analysis ...... 151 6.7.3 Quantification ...... 160 6.8 Conclusion ...... 165 6.9 Acknowledgements ...... 167 6.10 References ...... 168

CHAPTER 7 CONCLUSION AND RECOMMENDATIONS 7.1 Summary of Findings ...... 171 7.1.1 Comparison of Synthesis Techniques ...... 171 7.1.2 Optimum Synthesis and Processing Conditions ...... 174 7.1.3 Advancement of Compositional Understanding ...... 175 7.2 Future Work ...... 175 7.2.1 Commercial Application ...... 178 7.2.2 Effect of Temperature of Stoichiometry ...... 180 7.2.3 Sinterability and Physical Properties ...... 181 7.3 References ...... 184

vii LIST OF FIGURES

Fig. 1.1. A structural unit of boron carbide (B6.5C) the rhombohedral unit cell is constructed by joining the centres of the icosahedra ...... 4 Fig. 1.2. Micron sized boron carbide powder purchased from Goodfellow ...... 6

Fig. 2.1. The rhombohedral unit cell of boron carbide (B6.5C) highlighting the

polar (B3 = Bp) and equatorial (B2 = Be) icosahedra sites...... 18 Fig. 2.2. Zoom in on a cut out of the boron carbide crystal structure highlighting and icosahedral unit and how it bonds to the central CBC chain...... 19

Fig. 3.1. The different structural units of H3BO3 dehydration products upon heating ...... 34 Fig. 3.2. Borate anion complexation with 1,3-type diol compounds. Both the monodiol and didiol interaction can occur ...... 36 Fig. 3.3. Free-radical polymerisation of PVAc from VA monomer and subsequent hydrolysis to PVAl ...... 38 Fig. 3.4. Molecular structure of sucrose illustrating the excess of OH groups present...... 39 Fig. 4.1. Processing stages with accompanying characterisation techniques ...... 61

Fig. 4.2. DSC data for H3BO3, 1 hour polymerised PVAc and 1 hour polymerised

PVAc in the presence of H3BO3 ...... 66

Fig. 4.3. TGA data for H3BO3, 1 hour polymerised PVAc and 1 hour polymerised

PVAc in the presence of H3BO3 ...... 66 Fig. 4.4. SEM images of typical pre-treated PVAcB powder before (A, B) and after (C) washing. The red arrows highlight a position where a crack has recombined due to the removal of the boron component and the white

arrows highlight the presence and subsequent removal of the H3BO3 ‘strings’ ...... 67 Fig. 4.5. XRD trace comparison of a typical PVAcB powder after pre-treatment,

H3BO3 (ICSD ref. 98-006-1354) and B2O3 (ICSD ref. 98-001-6021) ...... 68

Fig. 4.6. ATR-FTIR scan comparison of unwashed A1-2-550 powder and H3BO3 . 69 Fig. 4.7. PVAcB powders of varying polymerisation times pre-treated under 4 x 10-1 partial vacuum and Ar flow calcined at 1400 °C for 1 hour ...... 70 Fig. 4.8. PVAcB powders of varying polymerisation times pre-treated under 4 x 10-1 partial vacuum and Ar flow calcined at 1400 °C for 2 hours ...... 70

viii

Fig. 4.9. XRD pattern of typical PVAcB powder calcined at 1400 °C for 1 hour without pre-treatment ...... 71 Fig. 4.10. V6-1-550, PV6-1-550 and A6-1-550 PVAcB powders taken to phase formation completion (A6-1-550 = 1400°C for 1 hour, PV6-1-550 and V6- 1-550 = 1400°C for 2 hours) illustrating the effect of vacuum ...... 72 Fig. 4.11. PV6-1-550 PVAcB powder calcined at 1250 °C for 1, 2, 3 and 4 hours under 140 L/m Ar flow...... 73 Fig. 4.12. A1-2-550 PVAcB powder calcined at 1400 °C for 1 and 2 hours under 140 L/m Ar flow ...... 74 Fig. 4.13. SEM images of boron carbide powder calcined at 1400 °C for 2 hours from A1-2-550 (A) and PV18-1-550 (B) PVAcB powders at the same magnification. The smaller particle size of the sample in B is apparent. The inset of Fig. 4.13B shows the presence of residual carbon (scale bar 1 µm) ...... 76 Fig. 4.14. Raman spectra of boron carbide powders calcined from PV18-1-550, PV6-1-550 and PV1-1-550 PVAcB powders ...... 77 Fig. 5.1. Process diagram for boron carbide synthesis ...... 101 Fig. 5.2. Water adsorption rates from powders after the 150 °C vacuum (SB) and 550 °C 6 L/m argon flow (P-SB-0) processing stages ...... 105 Fig. 5.3. XRD patterns of dry SB powder (a=SB-0) and SB powder exposed to atmospheric conditions for 3 days (b=SB-15). The solid lines mark the

peak positions for orthorhombic (BOH)3O3, and the dashed lines mark

the positions for triclinic H3BO3 ...... 106

Fig. 5.4. XRD pattern of dry P-SB-X powder. No crystalline H3BO3 or (BOH)3O3 phases are visible after pre-treatment at 550 °C ...... 108 Fig. 5.5. XRD patterns from ten sequential measurements over time demonstrate

loss of amorphous phase and increase in crystalline H3BO3 phase after exposure of P-SB-X powder to atmospheric conditions. Within 10 min of

exposure crystalline H3BO3 peaks appear within the originally

amorphous pattern (Fig. 5.4). The crystalline H3BO3 peaks gradually increase in intensity as the amorphous signal drops until after 3600 min, where no amorphous signal is detectable ...... 109 Fig. 5.6. SEM images of P-SB-X powder when dry and after 10 and 60 min exposure to atmospheric conditions. All images are taken from the same sample and each image set focuses on the same position in the sample. Insets in Figs. 5.6e and 5.6f are higher magnification of the identified areas (scale bar =1 μm) ...... 111

ix Fig. 5.7. High magnification SEM image of the (BOH)3O3/H3BO3 string-like features that appear after hydration of P-SB-0 powders ...... 113 Fig. 5.8. DSC and TGA data for P-SB-0-0 and P-SB-0-Full powders. The dashed and solid lines represent TGA and DSC, respectively ...... 114 Fig. 5.9. XRD patterns of boron carbide product calcined from pre-treated powders at different hydration stages. The residual carbon content increases as the degree of hydration increases (see insert) ...... 115 Fig. 5.10. SEM images of boron carbide powders formed from precursors with varying degrees of hydration with accompanying process schematic describing the correlating morphologies of precursor and product. The numeral in the bottom right corner of each SEM image labels the corresponding process pathway in the schematic. Low and high magnification SEM images of BC-0-0 (a, b and c) are shown and highlight the uniform and highly faceted boron carbide particles resulting from inert processing. BC-0-Full powder calcined from fully hydrated precursor powder is shown in d for comparison. Boron carbide calcined from precursor powder hydrated directly after solvent removal (BC-30- 0) is shown in e. Process pathway (iv) in the schematic describes a comparative growth mechanism for boron carbide synthesised by Cheng et al...... 117

Fig. 5.11. DLS particle sizing of BC-0-0 and BC-0-Full powder ...... 118 Fig. 6.1. Comparison of focusing FIPB (a) geometry with BB (b) for low MAC samples. The obtained peak shape from both geometries is shown, with the shaded section highlighting the forward peak slope aberration that occurs due to sample transparency ...... 131 Fig. 6.2. The (0 2 2) reflection of the diamond standard measured using BB and FIPB ...... 139 Fig. 6.3. Carbon concentrations for the commercial and High Purity samples calculated using the linear model of Eq. 6.1 from their refined a lattice parameters ...... 142 Fig. 6.4. Structure fit comparison of a commercial sample (Sigma) and a solution- based sample (High Purity) both containing diamond used for spiked wt% analysis. The (0 2 1) reflection is consistently underestimated and the (0 1 2) reflection is consistently overestimated when modelling solution-based samples as demonstrated by the difference plots ...... 144 Fig. 6.5. FIPB traces of the solution-based sample set overlaid and intensity normalised against the internal diamond standard. All reflections are allocated to the boron carbide phase unless otherwise marked. The

boron carbide phase intensity can be seen to clearly increase with reduced water adsorption in the precursor (refer Table 6.1) ...... 145 Fig. 6.6. FIPB traces of the commercial boron carbide sample set overlaid and intensity normalised against the internal diamond standard. All reflections are allocated to the boron carbide phase unless otherwise marked ...... 145 Fig. 6.7. Effect of site thermals on refined phase wt% in solution-based boron carbide powders ...... 147 Fig. 6.8. Effect of C2 occupancy on refined phase weight % in solution-based boron carbide powders ...... 147 Fig. 6.9. The (0 2 1) reflection of boron carbide purchased from Sigma collected using both BB and FIPB. These peaks are superimposed over the (0 2 1)

reflection of a pure B4C sample collected using FIPB to highlight the peak shoulder generated by the carbon deficient boron carbide phase in commercial samples ...... 153

xi LIST OF TABLES

Table 4.1. Treatment conditions for condensed PVAcB powders at different polymerisation times where; (PV) = Ar flow, 4 x 10-1 bar, (A) = 3 L/m Ar flow and (V) = 10-3 bar ...... 64 Table 4.2. Summary of data collected by GPC analysis for as-synthesised PVAc . 65 Table 5.1. Summary of processing pathways with precursor phase content and subsequent boron carbide phase weight % in calcined powders. The suffix ‘Full’ refers to a sample that has been hydration to saturation point ...... 104 Table 6.1. Processing conditions for solution-based boron carbide powders analysed in this study ...... 134 Table 6.2. Peak asymmetry of measured samples in both BB and FIPB geometries. A value of 0 describes perfect symmetry with asymmetry increasing with larger values ...... 139 Table 6.3. Refined wt% and associated relative bias as well as unit cell dimensions for silicon/corundum 1:1 powder mixtures ...... 140 Table 6.4. Refined wt% and associated relative bias as well unit cell dimensions for diamond/cBN 1:1 powder mixtures ...... 140 Table 6.5. Refined hexagonal unit cell sizes of commercial and solution- synthesised boron carbide powders ...... 141 Table 6.6. Calculated carbon concentrations (at%) of analysed boron carbide samples based on the unit cell trend of Aselage et al...... 142 Table 6.7. Refined atomic positions, site occupancies and thermal parameters for the High Purity boron carbide sample based on B3/C2 equated site positions and thermal parameter ...... 142 Table 6.8. Refined atomic positions, site occupancies and thermal parameters for the High Purity boron carbide sample with independent refinement of the C2 substitution site and thermal parameter ...... 143 Table 6.9. Refined phase wt% of solution-synthesised and commercial boron

carbide samples. The reported Rwp values are calculated with the impurity content excluded. Bracketed numbers are refined through independent refinement of the C2 site substitution position and thermal parameter ...... 146

xii

LIST OF ABBREVIATIONS

AR Analytical reagent at% Atomic percent

ATR-FTIR Attenuated total reflectance Fourier transform infrared spectroscopy

B1=Bc Chain boron

B2=Be Equatorial boron of boron carbide

B2O3 Boron trioxide

B3=Bp Polar boron of boron carbide

(B3O4)(OH)(OH2) Polymeric metaboric acid

B3O6 Boroxol ring

BB Bragg-Brentano

(BOH)3O3 Orthorhombic metaboric acid

BxC Boron carbide

BzO Benzoyl peroxide

C1=Cc Chain boron of boron carbide

C1m1 Monoclinic space group #8

CO Carbon monoxide

CO2 Carbon dioxide

DFT Density functional theory

DI Deionised

DLS Dynamic light scattering

DSC Differential scanning calorimetry

EMPA Electron microprobe analysis

Eq. Equation

Fig. Figure

FIPB Fixed incidence parallel beam

GPC Gel permeation chromatography

xiii H3BO3 Boric acid

HBO2 Metaboric acid structural unit

IR Infrared

ICP-OES Inductively coupled plasma optical emission spectroscopy

LaB6 Lanthanum hexaboride

MAC Mass absorption coefficient

MeOH Methanol

Mp Peak molecular weight

Mw Average molecular weight

NaBH4 Sodium Borohydride

Na2B4O7 Sodium borate

NIST National Institute for Standards and Technology

NMR Nuclear magnetic resonance

PB Parallel beam

PDI Polydispersity index

PVAc Polyvinyl acetate

PVAcB Polyvinyl acetate/boron condensate

PVAl Polyvinyl alcohol

R3m Rhombohedral space group #166

SB� Sucrose/boron condensate

SEM Scanning electron microscopy

SPS Spark plasma sintering

STP Standard temperature and pressure

TGA Thermogravimetric analysis

UHP Ultra-high purity

VA Vinyl acetate monomer

wt% Phase weight percent

XRD X-ray diffraction

xiv

QUT Verified Signature ACKNOWLEDGEMENTS

The undertaking of the grand adventure that has been this PhD would not have

been even remotely possible without the tutelage, support and love provided by

those who have accompanied me throughout the journey. I would like to take this

opportunity to acknowledge their wonderful support;

To my principal supervisor Professor Ian Mackinnon, I present my humblest

gratitude and appreciation. Without your instruction, support and genuine

interest in this research it would not have been possible. Thank you very much for

everything you have done for me.

To my associate supervisors Professor Peter Talbot and Professor Jose Alarco, I

cannot even begin to adequately describe how grateful I am for your presence in

my life. Without your kindness, passion for science, willingness to teach and

genuine friendship I honestly have no idea where I would be. I have learnt a great

deal from the both of you not just in science, but in life as well, thank you.

I would also like to thank and acknowledge all of my colleagues and the support

staff in the Institute for Future Environments, Central Analytical Research Facility

and the Banyo Pilot Plant Precinct who provided assistance throughout my

studies, it is greatly appreciated.

Finally, I would like to thank my family and friends, without your unconditional

love and support I would not have been able to make it this far. I love you all very

much, thank you for everything.

xvi

LIST OF PUBLICATIONS

PEER-REVIEWED JOURNAL ARTICLES • Watts, J. L., Talbot, P. C., Alarco, J. A. & Mackinnon, I. D. R., In-situ carbon control in the preparation of precursors to boron carbide by a non-aqueous solution technique. Journal of Materials Science and Engineering A. 2015, 5: p. 8-20.

• Watts, J. L., Talbot, P. C., Alarco, J. A. & Mackinnon, I. D. R., Morphology control in high yield boron carbide. Ceramics International. 2017, 43: p. 2650-2657.

• Watts, J. L., Spratt, H. J., Talbot, P. C., Alarco, J. A., Raftery, N. A. & Mackinnon, I. D. R., Structural analysis and phase quantification of boron carbide powders using an optimised X-ray diffraction technique. Journal of Applied Crystallography. 2018, (in preparation).

• Raftery, N. A., Halstead, B. W. & Watts, J. L., The measurement of mass attenuation coefficient at Cu Kα wavelength of powder XRD samples using parallel beam geometry. Advances in X-ray Analysis. 2015, 58: p. 146-152.

• Sauerschnig, P., Watts, J. L., Vaney, J. B., Talbot, P.C., Alarco, J. A., Mackinnon, I. D. R. and Mori, T., Thermoelectric properties of pure B4C prepared by a solution-based method. Advances in Applied Ceramics. 2018, (in preparation).

POSTER PRESENTATIONS • Watts, J. L., Talbot, P. C., Alarco, J. A. & Mackinnon, I. D. R., A Non-Aqueous Solution Synthesis of Boron Carbide by Control of In-Situ Carbon. 18th International Symposium on Boron, Borides and Related Materials. 2014, August 31st, Honolulu, Hawaii.

• Watts, J. L., Talbot, P. C., Alarco, J. A., Spratt, H. J., Mackinnon, I. D. R. & Raftery, N. A., Fixed-Incidence Parallel beam versus Bragg-Brentano for X- ray diffraction analysis of boron carbide. Australian X-ray and Analytical Association Conference. 2017, February 5th, Melbourne, Australia.

xvii • Watts, J. L., Talbot, P. C., Alarco, J. A. & Mackinnon, I. D. R., Morphology control in high yield boron carbide. 19th International Symposium on Boron, Borides and Related Materials. 2017, September 4th, Freiburg, Germany.

xviii

CHAPTER 1

INTRODUCTION

1.1 THESIS OUTLINE

The work presented here in the mode of ‘Thesis by Publication’ for the fulfilment

of the requirements for the degree of ‘Doctor of Philosophy’ is structured over

seven chapters. Three of these chapters (4, 5 and 6), are each dedicated to

individual published (or submitted), peer-reviewed manuscripts. A foreword to

each of these chapters is included for the purpose of positioning the manuscripts

within the overarching research goals. Chapters 4 and 5 focus on differing

synthesis techniques for high quality boron carbide powders, while Chapter 6

details a technique developed for enhanced characterisation of these powders.

Collectively, these works form a novel and efficient pathway to the formation of

high quality boron carbide powders with wide spread application and benefit to

commercial and technological applications. Chapter 1 serves to introduce the

research topic as well as the associated research goals while Chapter 2 presents a

review of the available peer-reviewed literature relevant to the presented works.

Synthesis techniques detailed in Chapters 4 and 5 follow a basic process

architecture developed to facilitate homogenous precursor dispersion, while

Chapter 3 presents an overview of these processes as well as the characterisation

techniques implemented. A description of the precursor chemicals chosen for

these techniques and the reasoning governing their selection is also detailed.

Finally, Chapter 7 presents a culmination of the findings generated by this research

and details some of the future work that will be/is being undertaken based on

these outcomes.

1.2 RESEARCH TOPIC

1.2.1 BORON CARBIDE

Boron carbide is a synthetic, ultra-hard, non-oxide ceramic material with many uses in specialty technology industries. Boron carbide is the third hardest material known, approaching the hardness of cubic boron nitride and diamond[1]. The exceptional hardness of boron carbide is due to the ability of the boron and carbon atoms to form multiple covalent bonds. This capability allows boron to form many diverse atomic configurations and thus, is a highly versatile element for functional materials synthesis. For boron carbide, the crystal structure consists of covalently linked 12-atom icosahedra that are connected by 3-atom chains (see Fig. 1.1).

Boron carbide is also thought to exist as a solid solution of multiple boron/carbon stoichiometries varying at the unit cell level[2] [3]. The overall stoichiometry of a boron carbide sample is then given as the average of these unit cell compositions and is highly dependent on the synthesis technique and processing conditions used. This solubility in boron carbide is due to the capacity of carbon to easily substitute for boron atoms throughout the structure, although limits to the atomic position(s) and amount(s) of substitution do exist[1] [4].

Fig. 1.1. A structural unit of boron carbide (B6.5C) the rhombohedral unit cell is constructed by

joining the centres of the icosahedra.

The upper limit of the carbon concentration in possible atomic arrangements is

generally identified as B4C. However, there are varying opinions in the literature

on the precise degree of maximum carbon substitution possible in the boron

carbide structure[4]. Nevertheless, with B4C stoichiometry, the hardness of boron

carbide is at a maximum[5].

1.2.2 APPLICATIONS OF BORON CARBIDE

Boron carbide is used in a wide range of engineering applications due to a

combination of useful properties including high hardness, a high resistance to

chemical corrosion, a high melting point and a low specific weight. Boron carbide

in many forms – as powders, sintered billets and coatings − is used as wear resistant refractories, as a coating for cutting tools and for ballistic applications such as armour plating. As boron naturally exists in an isotopic ratio of approximately 20% 10B and 80% 11B[6], boron carbide also finds use as a neutron absorber due to the high neutron absorption cross section of the 10B isotope.

Boron carbide is also a p-type semiconductor with thermoelectric properties with applications specifically in high temperature environments[2]. The recent discovery of high temperature n-type thermoelectric counterparts to boron carbide further increases its capability as a functional high temperature thermoelectric material [7] [8] [9].

1.2.3 SYNTHESIS OF BORON CARBIDE

The carbothermal method is the most common synthetic route to produce boron carbide. Using this method, boron carbide formation is achieved by the reduction of boron oxide (B2O3) in the presence of carbon monoxide (CO) resulting in reduced elemental boron which then reacts with elemental carbon to form boron carbide[1] [4] [10]. Boric acid is most commonly used as the boron source whilst the carbon source varies greatly dependant on the processing method. The reaction can be summarised by the following three equations:

4H3BO3 → 2B2O3 + 6H2O [Eq. 1.1]

B2O3 + 3CO → 2B + 3CO2 [Eq. 1.2]

4B + C → B4C [Eq. 1.3]

Bulk industrial synthesis methods are commonly undertaken at high temperature

using this method, followed by milling and hot pressing and/or sintering to shape

a product[10] [11]. However, the formation of excess or free carbon during these

synthesis methods is problematic, requiring further processing for carbon

removal[11] [12]. The formed boron carbide is also of a particle size too large for

optimal sintering, requiring milling to the desired particle size leading to further

contamination from the grinding media[10]. Fig. 1.2 shows an X-ray diffraction

(XRD) pattern of commercially produced boron carbide powder purchased from

Goodfellow (product ID B 506010). The presence of substantial carbon impurity is

highlighted by the # identifier at approximately 30 °2θ.

Fig. 1.2. Micron sized boron carbide powder purchased from Goodfellow.

Because of these issues, recent research has been focused on alternative lower temperature synthesis methods that result in a fine powder with less residual carbon[12] [13] [14] [15]. To date the best attempts at achieving these outcomes have been through solution-based preparation of precursors to boron carbide.

Solution-based synthesis offers a facile method for the production of boron carbide as it affords a high degree of homogeneity in subsequently condensed precursor powders allowing for enhanced reaction kinetics and greater control over residual-free phase formation. However, despite the ability of solution processing to form highly homogeneous precursors to boron carbide, many inconsistencies still occur within currently reported techniques.

1.2.4 STRUCTURAL AND COMPOSITIONAL CHARACTERISATION

Morphological characterisation of boron carbide can be accomplished via standard techniques such as particle sizing and scanning electron microscopy (SEM).

However, precise boron carbide phase identification and quantification still presents many challenges. These challenges are primarily due to the similarity in electronic and nuclear structure of the boron (11B) and carbon (12C) atoms resulting in very similar radiation scattering cross-sections[4]. This similarity of cross- sections means that differentiation between the two atoms using most forms of characterisation is very difficult. However, some degree of success has been achieved using complementary techniques utilising different forms of radiation[4].

For example, XRD is able to provide basic crystal structure information for boron carbide, but to date it has not been able to conclusively determine precise carbon substitution site occupancies. However, using XRD In concert with other

techniques such as nuclear magnetic resonance (NMR) spectroscopy[16] [17],

neutron diffraction[18] [19] infrared (IR)[6] and Raman spectroscopy[20] [21], a

more precise determination of atomic arrangement and site occupancy can be

achieved. As noted previously, boron carbide generally exists as a solid solution of

multiple unit cell stoichiometries. This feature, in combination with the

characterisation difficulties presented by boron carbide, significantly complicates

correct stoichiometric identification. Thus, the exact boron/carbon ratio within a

particular structure for an unknown sample is often difficult to determine or is not

specified. This characterisation difficulty has resulted in a generalised use of ‘B4C’

in the literature to describe boron carbide, referring to a notional structure,

despite uncertain characterisation techniques. Throughout this thesis, samples will

be identified as ‘boron carbide’, unless the stoichiometry of the sample has been

determined.

1.3 RESEARCH PROBLEMS AND RATIONALE

Current approaches to boron carbide synthesis require high calcination temperatures, lack stoichiometric control and result in an impure product with undesirable form. Large amounts of waste are also generated in the form of unreacted precursors and unwanted reaction products[10]. The high calcination temperatures, lack of phase control and large particle size can be attributed to two main factors: (i) poor contact area between precursor units; and (ii) non-optimal processing conditions. For example, in the case of the most widely implemented commercial process, electricity is passed between three graphite rods submerged in a powder mixture of H3BO3 and carbon black. Because the precursors are in a powder form when mixed, the contact area between them is low requiring high calcination temperatures for phase formation. The required heat is generated via electricity passed between the rods and thus, there is a large heat gradient present in the reactor. Consequently, only material close to the centre of the rods gives the desired boron carbide phase[10]. Large amounts of residual carbon are also present in the boron carbide product due to poor phase control, and a large particle size is also common, complicating further processing. Given that the most optimal boron carbide form for solid sintering purposes, one of its largest uses, is with a small particle size[10] [22], this method of synthesis would seem highly inappropriate for widespread use. However, these types of processes remain as the primary commercial route for boron carbide production due to the low cost of the precursors and relative ease of processing[1].

The first attempts to reconcile these issues used slurries in an attempt to improve

precursor mixing and processing but high calcination temperatures are still

required[23]. With complete dissolution of boron and carbon precursors, reaction

kinetics improved and calcination temperatures were reduced. However, even

though the calcination temperature can be lowered significantly through

increasing the homogeneity of the precursor mixture, many problems are still

present in these processes. For example, a primary problem is still residual carbon

in the boron carbide product. Presence of residual carbon is due to the fact that as

the concentration of carbon is increased in the precursor mixture, the boron

components are more successfully isolated from each other, preventing

coalescence into larger agglomerates through melting during processing. This

problem of balancing the stoichiometric requirements for carbothermal reaction

whilst maintaining the highest degree of homogeneity possible in the precursor is

key to optimising phase pure synthesis of boron carbide.

Despite the many different approaches that have been undertaken utilising

solution based processing of precursors to boron carbide, little is understood about

precursor states and their interaction with each other and their environment

during processing. Furthermore, how these interactions then affect precursor

morphology, distribution and subsequent reaction kinetics is not well understood

either. An accurate understanding of these factors is currently not detailed in the

scientific literature. Nevertheless, these factors are important for optimising a

given synthesis technique toward phase purity, reaction control and lowered

calcination temperatures. Correctly identifying key processing conditions for

optimisation is also of importance to development of an improved synthesis technique.

Although significant work has been carried out on advancing synthesis methods for boron carbide production in recent years, little work has been done to improve the characterisation methodologies required for accurate identification of a boron carbide phase and its stoichiometry, despite vast advances in analytical technologies. As a result, research on the effects of these improved synthesis techniques on the stoichiometry of boron carbide is not present in the literature.

Because the high performance properties of boron carbide are enhanced at particular carbon concentrations (e.g. maximum hardness at B4C, highest thermoelectric conversion at B6.5C)[1] [24], an advanced understanding of these effects is of utmost importance to developing an optimised synthesis technique that affords stoichiometric control in the product, especially for commercial applications.

1.4 OBJECTIVE AND AIMS

The objective of this research is to develop synthesis techniques for boron carbide

materials that incorporate a high level understanding of the processing conditions

and precursor interactions to achieve a phase pure product of controllable

morphology and stoichiometry at lowered calcination temperatures. The

techniques that have been developed use organic and fully soluble material as the

carbon source of the reaction and also as an agent for homogenously distributing

the boron precursor at the nano-scale. To achieve these goals, the main aims of

the project are:

• To review the literature available on current and emerging boron carbide

synthesis techniques and identify the parameters required to achieve the

objectives whilst also addressing any additional new information identified

in the data.

• The development of processing procedures that support the key concepts

identified, i.e. to create highly homogeneous precursor powders and retain

that homogeneity throughout processing.

• The development and optimisation of a synthesis technique for boron

carbide that incorporates these concepts giving tuneable and scalable

control of material composition and properties.

• Detailed characterisation of the synthesised boron carbide as well as the

development of new boron carbide characterisation methodologies that

allow for the resolution required to further enhance the synthesis technique.

1.5 REFERENCES

1. Suri, A., Subramanian, C., Sonber, J. & Murthy, T., Synthesis and consolidation of boron carbide: a review. International Materials Reviews, 2012. 55: p. 4-40. 2. Werheit, H. & Shalamberidze, S., Advanced microstructure of boron carbide. Journal of Physics: Condensed Matter, 2012. 24: p. 1-12. 3. Tallant, D. R., Aselage, T. L., Campbell, A. N. & Emin, D., Boron Carbides: Evidence for Molecular Level Disorder. Journals of Non-Crystalline Solids, 1988. 106: p. 370- 373. 4. Domnich, V., Reynaud, S., Haber, R. & Chhowalla, M., Boron Carbide: Structure, Properties, and Stability under Stress. Journal of the American Chemical Society, 2011. 94: p. 3605-3628. 5. Niihara, K., Nakahira, A. & Hirai, T., The Effect of Stoichiometry on Mechanical Properties of Boron Carbide. Journal of the American Ceramic Society, 1984. 67: p. C13-14. 6. Werheit, H., Leithe-Jasper, A., Tanaka, T., Rotter, H. W. & Schwetz, K. A., Isotopic effects on the phonon modes in boron carbide. Journal of Physics: Condensed Matter, 2010. 22: p. 575-579. 7. Mori, T. & Nishimura, T., Thermoelectric properties of homologous p- and n-type boron-rich borides. Journal of Solid State Chemistry, 2006. 179: p. 2908-2915. 8. Mori, T., Nishimura, T., Yamaura, K. & Takayama-Muromachi, E., High temperature thermoelectric properties of a homologous series of -type boron icosahedra compounds: A possible counterpart to -type boron carbide. Journal of Applied Physics, 2007. 101: p. 93714. 9. Mori, T., Nishimura, T., Schnelle, W., Burkhardt, U. & Grin, Y., The origin of the n- type behavior in rare earth borocarbide Y1-xB28.5C4. Dalton Transactions, 2014. 43: p. 15048-15054. 10. Thevenot, F., Boron Carbide - A Comprehensive Review. Journal of the European Ceramic Society, 1990. 6: p. 205-225. 11. Alizadeh, A., Taheri-Nassaj, E. & Ehsani, N., Synthesis of boron carbide powder by a carbothermic reduction method. Journal of the European Ceramic Society, 2004. 24: p. 3227-3234. 12. Kakiage, M., Tahara, N., Yanagidani, S., Yanase, I. & Kobayashi, H., Effect of boron oxide/carbon arrangement of precursor derived from condensed polymer-boric acid product on low-temperature synthesis of boron carbide powder. Journal of the Ceramic Society of Japan, 2011. 119: p. 422-425. 13. Kakiage, M., Tahara, N., Yanase, I. & Kobayashi, H., Low-temperature synthesis of boron carbide powder from condensed boric acid–glycerin product. Materials Letters, 2011. 65: p. 1839-1841. 14. Kakiage, M., Tominaga, Y., Yanase, I. & Kobayashi, H., Synthesis of boron carbide powder in relation to composition and structural homogeneity of precursor using condensed boric acid–polyol product. Powder Technology, 2012. 221: p. 257-263. 15. Tahara, N., Kakiage, M., Yanase, I. & Kobayashi, H., Effect of addition of tartaric acid on synthesis of boron carbide powder from condensed boric acid-glycerin product. Journal of Alloys and Compounds, 2013. 573: p. 58-64. 16. Hynes, T. V. & Alexander, M. N., Nuclear Magnetic Resonance Study of b- Rhombohedral Boron and Boron Carbide. The Journal of Chemical Physics, 1971. 54: p. 5296-5310. 17. Lee, D., Bray, P.J. & Aselage, T. L., The NQR and NMR studies of icosahedral borides. Journal of Physics: Condensed Matter, 1999. 11: p. 4435-4450.

18. Kwei, G. H. & Morosin, B., Structures of the Boron-Rich Boron Carbides from Neutron Powder Diffraction: Implications for the Nature of the Inter-Icosahedral Chains. Journal of Physical Chemistry, 1996. 100: p. 8031-8039. 19. Morosin, B., Kwei, G. H., Lawson, A. C., Aselage, T. L. & Emin, D., Neutron powder diffraction refinement of boron carbides nature of intericosahedral chains. Journal of Alloys and Compounds, 1995. 226: p. 121-125. 20. Shirai, K. & Emura, S., Lattice Vibrations of Boron Carbide. Journal of Solid State Chemistry, 1997. 133: p. 93-96. 21. Tallant, D. R., Aselage, T. L., Campbell, A. N. & Emin, D., Boron carbide structure by Raman spectroscopy Physical Review B, 1989. 40: p. 5649-5655. 22. Cho, N., Silver, K., Berta, Y. & Speyera, R., Densification of carbon-rich boron carbide nanopowder compacts. Journal of Materials Research, 2007. 22: p. 1354- 1359. 23. Weimer, A., Roach, R. & Haney, C., Rapid Carbothermal a Graphite Reduction of Boron Oxide in Transport Reactor. American Institute of Chemical Engineers, 1991. 37: p. 759-768. 24. Thevenot, F & Bouchacourt, M., The correlation between the thermoelectric properties and stoichiometry in the boron carbide phase B4C-B10.5C. Jounal of Materials Science, 1985. 20: p. 1237-1247.

CHAPTER 2

LITERATURE REVIEW

2.1 STRUCTURE AND STOICHIOMETRY OF BORON CARBIDE

As mentioned previously, the boron carbide crystal structure consists of covalently linked 12-atom icosahedra that are connected by 3-atom chains, and has trigonal symmetry with the R 3 m space group[1]. The icosahedra are located at the vertices of the 15-atom rhombohedral unit cell and the 3-atom chain lies along the

(111) axis corresponding to the longest diagonal of the rhombohedra[2]. The boron carbide lattice is more commonly described by its non-primitive hexagonal unit cell, the c value of which is also the longest diagonal of the rhombohedral unit cell[2].

The boron carbide phase range is commonly agreed to span from ~8% (B10C) to

~18.8% (B3.4C) atomic % (at%) carbon[3] [4]. However, evidence for carbon concentrations exceeding 18.8% at% carbon and even surpassing the idealised 20 at% (B4C) stoichiometry does exist[5] [6] [7]. Furthermore, the work reported within this thesis lends further evidence towards the existence of a stable B4C phase at the carbon rich limit. As discussed in Chapter 6 herein, this conflict with the commonly accepted B4.3C carbon rich limit can reconciled by the effect of synthesis conditions such as precursor homogeneity and formation temperature.

As such, controversy over the stoichiometric limits and the precise atomic occupancies throughout the entire boron carbide crystal structure phase range is still present. This ambiguity is due to the inability of most characterisation techniques to differentiate between the electronic and nuclear scattering cross sections of boron and carbon. Through a combination of multiple characterisation techniques, it is generally accepted that B4C, at the carbon rich end of the phase

17 range, consists of B11C icosahedra and C-B-C chains, or (B11C)CBC structural units,

yielding the 15-atom unit cell with the stoichiometry B12C3[5]. Theoretical and

experimental NMR data show that the best solution to the idealised B4C structure

is with a mixture of 2.5% (B12)CBC, 95% (B11C)CBC and 2.5% (B10C2)CBC suggesting

that small amounts of carbon rich and carbon deficient icosahedra are present

that, on average, give the overall B4C stoichiometry[1]. As the carbon

concentration is decreased within the boron carbide phase range, greater

ambiguity and disagreement regarding atomic positioning becomes apparent

within the literature. Some work indicates that boron replaces carbon in the CBC

chain first yielding (B11C)CBB structural units[8] while other research postulates

that the icosahedra first accommodate the loss of carbon giving (B12)CBC units[9].

While in opposition to each other, both interpretations are in agreement with

experimental results from differing characterisation techniques[1], perhaps

indicating that both substitution routes occur to some degree. A recent study

combining computational work with precise single crystal X-ray diffraction data

and high resolution transmission electron microscopy imaging has been able to

further provide more detailed information regarding the complex nature of the

potential icosahedral and chain combinations possible with the boron carbide

structure [10].

Throughout the entire phase range, the carbon atoms within different boron

icosahedra are not ordered relative to one another. However, the substituted

carbon atoms are believed to be positioned at the polar sites of the icosahedra

(the sites bound to neighbouring icosahedra)[2]. The remaining icosahedral

positions, known as the equatorial positions, join the icosahedra to the central chains within the unit cell (see Figs. 2.1 and 2.2).

Fig. 2.1.[5] The rhombohedral unit cell of boron carbide (B6.5C) highlighting the polar (B3 = Bp) and equatorial (B2 = Be) icosahedra sites.

Fig. 2.2. Zoom in on a cut out of the boron carbide crystal structure highlighting an icosahedral unit and how it bonds to the central CBC chain.

Due to the smaller atomic radius of carbon compared to boron, the unit cell experiences a loss in volume as the carbon concentration increases[5]. Aselage et al. documented this trend experimentally, effectively providing a method to determine the carbon concentration of an unknown boron carbide if precise lattice parameters are known[11].

19 2.2 COMMERCIAL SYNTHESIS OF BORON CARBIDE

Many synthetic routes to boron carbide have been developed since it was first

identified in 1858[5]. However, only a few routes have been implemented for

commercial production. Although boron carbide can be synthesised by directly

reacting elemental boron and carbon[12], this technique is not commercially

viable and is only used for research applications due to the high costs of the

purified elements. Boron carbide can also be synthesised by magnesiothermal

reaction wherein magnesium metal is used to reduce boron oxide for subsequent

reaction with a carbon source. However, this method does not see large scale

application in commercial production due to the formation of magnesium

contaminants that require acid washing to remove as well as due to the high cost

of magnesium[2] [5]. The most widely used commercial technique for producing

boron carbide is the reduction of boric acid (H3BO3) with carbon black via the

carbothermal method at >1750 °C in electric arc or Acheson type furnaces[13]. The

overall reaction mechanism for this process can be summarised as follows:

4H3BO3 + 7C → B4C + 6CO + 6H2O [Eq. 2.1]

Despite its wide commercial use, the electric arc furnace and other similar

methods of carbothermal reduction produce boron carbide that is not conducive

to further material processing. Very little stoichiometric control is afforded, a very

large particle size is obtained and residual unreacted carbon is left dispersed

throughout the product. The coarse-grained boron carbide then requires intensive

milling to generate a fine powder suitable for sintering; the intense milling also

results in contamination from grinding media[5]. The use of catalysts to remove residual carbon has been explored[14], but such methods result in contamination by the catalysts or require expensive precursor materials. Micron sized boron carbide powder with no residual carbon has been successfully formed by mixing excess boron oxide (B2O3) powder and carbon under inert conditions[15].

However, the temperature required for complete reaction is still high at 1800 °C.

From the aforementioned synthesis techniques, it is clear that dry powder mixing of precursors is far from an optimal synthesis route to produce boron carbide. Dry powder mixing results in poor contact between the precursor constituents due to a relatively large particle size in the source powders. By increasing the homogeneity of the precursors and hence increasing the contact area between the boron and carbon, the reaction kinetics can be substantially improved.

Increased homogeneity and lower sintering temperatures may also result in a smaller, and possibly even controllable, particle size with greater control over phase and purity.

One of the simplest methods for achieving homogenous mixing is through solution processing of precursor chemicals. By dissolving the relevant precursors in solution, homogeneous mixing is achieved at the molecular/atomic scale, provided full dissolution is achieved. It is via this approach that further advances in boron carbide synthesis occurred, and subsequent further review will focus on this topic.

21 2.3 SOLUTION BASED TECHNIQUES

2.3.1 SYNTHESIS FROM SLURRIES

The first solution based approach to boron carbide formation was undertaken by

A. Weimer et al[16] [17], who utilised dissolved aqueous H3BO3 mixed with a

starch slurry as the carbon source. A precursor powder was formed by subsequent

water evaporation and calcination at 800 °C giving a mixture of B2O3 and carbon.

The powder then required firing at temperatures in excess of 1850°C to achieve

full reaction of the precursors to form boron carbide[16]. The minimisation of

residual carbon was achieved by using approximately 10% excess B2O3 to

compensate for boron loss by gaseous B2O2[17]. Although this approach did make

use of a fully dissolved boron precursor, the use of a slurry instead of a completely

dissolved carbon source allows for large B2O3 particles to form during the

evaporation step, resulting in a less intimate mix of precursors. The poor

dispersion of the precursors is reflected in the high temperatures required for

formation of the final boron carbide phase. As B2O3 melts at ~450 °C, the use of

800 °C temperatures at the precursor powder calcination step would also result in

increased B2O3 agglomeration causing further loss of homogeneity.

2.3.2 SOL-GEL SYNTHESIS

In an attempt to increase the degree of precursor homogeneity further, the sol-

gel method was then applied to boron carbide synthesis. A. Sinha et al. formed

B2O3/carbon powders by using citric acid and H3BO3 as precursors[18]. Citric acid

was added to hot (85 °C) solutions of H3BO3 dissolved in distilled water. The water

was then evaporated by heating and the condensate dried on a hot plate and pyrolysed at 700 °C. The resultant mass was then crushed to form the precursor

B2O3/carbon powder which was compressed in a graphite die and calcined under vacuum. This particular sol-gel approach was successful at lowering the required calcination temperature to 1450 °C. However, a significant amount of residual carbon was still present in the final product. Residual carbon was attributed again to boron loss by volatile boron species. A group using a very similar approach was able to improve upon the amount of residual carbon by using excess H3BO3 in the initial solution. However, they were unable to completely remove the residual carbon[19]. A. Khanra was successful in forming boron carbide via sol-gel methods without pressurising the precursor powder during calcination. However, higher temperatures were required to complete the reaction[20]. From these outcomes it can be seen that although the principles of the citric acid sol-gel method appeared promising, the technique was not entirely successful. The inadequacy of this approach is due to a requirement to apply pressure to achieve lower temperature phase formation resulting in excessive residual carbon. This outcome is most likely attributed to the inability of citric acid to maintain separated B2O3 particles formed from the dehydration of H3BO3, thus enhancing agglomeration at the pyrolysis step and loss of finer-scale homogeneity. By using ammonium polycarbonate as a dispersant, A. Najafi et al. were able to form very fine (~30 nm) boron carbide particles using a sol-gel method[21] [22]. This approach was successful at lowering the calcination temperature to 1270 °C at 3 hours, suggesting a very intimate nano-scale mixing of precursor elements which was reflected in the particle size obtained. However, the overall procedure was

23 complex and required specialised boron and carbon precursors. Residual carbon

was also still present in the final product.

2.3.3 SYNTHESIS FROM POLYMERS

Moderate success at boron carbide formation has been achieved using mixed

aqueous polyvinyl alcohol (PVAl) and H3BO3 solutions[23] [24]. Specifically, Yanase

et al. were able to form boron carbide at 1300°C via this method. However, a long

residence time was required to complete the reaction and some residual carbon

was still present in the final product[23]. The success of this method was

attributed to the ability of H3BO3 to complex with PVAl via B-O-C borate ester

bonds in solution, helping to lock the boron into the carbon matrix locally; thus

preventing agglomeration of the boron source. Subsequent further work

improved upon the removal of residual carbon but with a very long residence time

of 20 hours[25].

2.3.4 SYNTHESIS FROM POLYOLS

The most promising published works in solution based approaches to boron

carbide formation have come from Kakiage et al.[26] [27]over the past few years.

Similar to the work of Yanase et al., relatively simple water soluble polyols

(glycerin, mannitol) were mixed with aqueous H3BO3 solutions. In the case of

glycerin as a precursor, nearly carbon-free boron carbide was formed at 1250 °C

with a residence time of 5 hours[26]. The increased homogeneity of the precursor

powder was again attributed to borate ester bonds formed in solution. However,

as an excess of carbon was utilised (compared to what is required by the

carbothermal method) to obtain sufficient homogeneity to facilitate increased reaction kinetics, a multistage pyrolysis step in air was implemented to remove the excess carbon before calcination. By using mannitol as the precursor and excess H3BO3 nearly carbon-free boron carbide could be obtained at 1300 °C with a 5 hour residence time[27]. The reaction kinetics were again substantially improved by the use of excess mannitol, lowering the calcination temperature to

1250 °C. Again, the presence of excess carbon required a pyrolysis step in air to remove it before calcination.

25 2.4 PREFERRED CRITERIA FOR PRECURSOR HOMOGENEITY

Despite the marked improvements to boron carbide synthesis achieved by solution

mixing of precursors, problems still exist; the most predominant of which is the

presence of residual carbon. This presence is due to the fundamental trade off that

appears consistently in solution based techniques: increasing the carbon content

in the boron carbide precursor results in improved processing conditions, but in

turn results in undesirable unreacted carbon in the final product. In other words,

in solution-based synthesis the presence of residual carbon is simply a result of the

degree to which the carbon is able to segregate the boron precursor. The more

carbon there is present, the easier it is to achieve and maintain a high degree of

homogeneity during the dissolved liquid state. Of course, with too much carbon

present, the stoichiometric requirement for the carbothermal method is exceeded

and unreacted residual carbon becomes an unwanted impurity. Thus, other

methods must be used to maintain the highest degree of homogeneity in mixed

precursor powders once solvent removal has occurred whilst still not exceeding

the calorific requirements of the carbothermal method.

Examination of the available literature on solution based techniques suggests that

the key variables for achieving enhanced homogeneity (and retaining it until phase

formation occurs) are: (i) precursor interactions and their capacity to complex or

form covalent bonds; and (ii) the processing conditions used throughout the

synthesis technique, i.e. temperature and environmental conditions. With careful

consideration and investigation of these two factors, notable advances in boron

carbide synthesis can be achieved.

2.5 REFERENCES

1. Domnich, V., Reynaud, S., Haber, R. & Chhowalla, M., Boron Carbide: Structure, Properties, and Stability under Stress. Journal of the American Chemical Society, 2011. 94: p. 3605-3628. 2. Suri, A., Subramanian, C., Sonber, J. & Murthy, T., Synthesis and consolidation of boron carbide: a review. International Materials Reviews, 2012. 55: p. 4-40. 3. Schwetz, K. A. & Karduck, P, Investigations in the boron-carbon system with the aid of electron probe microanalysis. Journal of the Less-Common Metals, 1991. 175: p. 1-11. 4. Rogl, P. F., Vrestal, J., Tanaka, T. & Takenouchi, S, The B-rich side of the B–C phase diagram. CALPHAD: Computer Coupling of Phase Diagrams and Thermochemistry, 2014. 44: p. 3-9. 5. Thevenot, F., Boron Carbide - A Comprehensive Review. Journal of the European Ceramic Society, 1990. 6: p. 205-225. 6. Gosset, D. & Colin, M., Boron carbides of various compositions: An improved method for X-rays characterisation. Journal of Nuclear Materials, 1991. 183: p. 161-173. 7. Werheit, H., Boron carbide: Consistency of components, lattice parameters, fine structure and chemical composition makes the complex structure reasonable. Solid State Sciences, 2016. 60: p. 45-54. 8. Emin, D., Structure and single-phase regime of boron carbides. Physical Review B, 1988. 38: p. 6041-6055. 9. Kirfel, A., Gupta, A. & Will, G., The Nature of the Chemical Bonding in Boron Carbide, B13C2. I. Structure Refinement. Acta Crystallographica Section B, 1979. 35: p. 1052-1059. 10. Rasim, K., Ramlau, R., Leithe-Jasper, A., Mori, T., Burkhardt, U., Borrmann, H., Schnelle, W., Carbogno, C., Scheffler, M. & Grin, Y., Local atomic arrangements and band structure of boron carbide. Angewandte Chemie International Edition, 2018. 57: p. 6130-6135. 11. Aselage, T. L. & Tissot, R. G., Lattice Constants of Boron Carbide. Journal of the American Ceramic Society, 1992. 75: p. 2207-2212. 12. Benton, S & David, R., Methods for preparing boron carbide articles. US patent no. 3 914 371, 1975. 13. Bigdeloo, J. & Hadian, A., Synthesis of High Purity Micron Size Boron Carbide Powder from B2O3/C Precursor. International Journal of Recent Trends in Engineering, 2009. 1: p. 176-180. 14. Krishnarao, R., Subrahmanyam, J., Kumar, T. & Ramakrishna, V., Effect of K2CO3, and FeCl3 on the formation of B4C through carbothermal reduction of B2O3. Journal of Alloys and Compounds, 2010. 496: p. 572-576 15. Junga, C., Leeb, M & Kim, C., Preparation of carbon-free B4C powder from B2O3 oxide by carbothermal reduction process. Materials Letters, 2004. 58: p. 609-614. 16. Weimer, A., Roach, R. & Haney, C., Rapid Carbothermal a Graphite Reduction of Boron Oxide in Transport Reactor. American Institute of Chemical Engineers, 1991. 37: p. 759-768. 17. Weimer, A., Moore, W., Roach, R., Hitt, J. & Dixit, R., Kinetics of Carbothermal Reduction Synthesis of Boron Carbide. Journal of the American Ceramic Society, 1992. 75: p. 2509-2514.

27 18. Sinha, A., Mahata, T. & Sharma, B., Carbothermal route for preparation of boron carbide powder from boric acid–citric acid gel precursor. Journal of Nuclear Materials, 2002. 301: p. 165-169. 19. Hadian, A. & Bigdeloo, J., The Effect of Time, Temperature and Composition on Boron Carbide Synthesis by Sol-gel Method. Journal of Materials Engineering and Performance, 2008. 17: p. 44-49. 20. Khanra, A., Production of boron carbide powder by carbothermal synthesis of gel material. Bulletin of Material Science, 2007. 30: p. 93-96. 21. Najafi, A., Golestani-Fard, F., Rezaie, H. & Ehsani, N., Effect of APC addition on precursors properties during synthesis of B4C nano powder by a sol–gel process. Journal of Alloys and Compounds, 2011. 509: p. 9164-9170. 22. Najafi, A., Golestani-Fard, F., Rezaie, H. & Ehsani, N., A novel route to obtain B4C nano powder via sol–gel method. Ceramics International, 2012. 38: p. 3583-3589. 23. Yanase, I., Ogawara, R. & Kobayashi, H., Synthesis of boron carbide powder from polyvinyl borate precursor. Materials Letters, 2009. 63: p. 91-93. 24. Fathi, A., Ehsani, N., Rashidzadeh, M., Baharvandi, H. & Rahimnejad, A., Synthesis of boron carbide nano particles using polyvinyl alco hol and boric acid. Ceramics- Silikáty, 2012. 56: p. 32-35. 25. Kakiage, M., Tahara, N., Yanagidani, S., Yanase, I. & Kobayashi, H., Effect of boron oxide/carbon arrangement of precursor derived from condensed polymer-boric acid product on low-temperature synthesis of boron carbide powder. Journal of the Ceramic Society of Japan, 2011. 119: p. 422-425. 26. Kakiage, M., Tahara, N., Yanase, I. and Kobayashi, H., Low-temperature synthesis of boron carbide powder from condensed boric acid–glycerin product. Materials Letters, 2011. 65: p. 1839-1841. 27. Kakiage, M., Tominaga, Y., Yanase, I. & Kobayashi, H., Synthesis of boron carbide powder in relation to composition and structural homogeneity of precursor using condensed boric acid–polyol product. Powder Technology, 2012. 221: p. 257-263.

CHAPTER 3

SYNTHESIS AND CHARACTERISATION

METHODOLOGY

29 3.1 CHOICE OF PRECURSOR

Precursor choice for both boron and carbon sources is of the utmost importance

to achieve an enhanced synthesis technique for scalable, controllable, cost

efficient boron carbide production. When assessing possible precursor choices

four main factors were considered; (i) cost/availability; (ii) the inherent properties

of the chemical (solubility, stability etc.); (iii) the interaction dynamics of the

chosen boron/carbon precursor combination; and (iv) the elemental composition

of the chemical. Avoiding the introduction of elements other than those required

for the carbothermal method (boron, carbon and oxygen) is preferred, mitigating

undesirable phase formation and also minimising the processing steps required to

achieve a phase pure product. Using these factors, the viability of multiple boron

and carbon sources were investigated for use in an enhanced boron carbide

synthesis technique.

3.1.1 BORON PRECURSOR

There is a wide range of boron compounds feasible for use in boron carbide

synthesis, of which H3BO3 most closely meets the requirements detailed above.

Due to its many industrial uses H3BO3 is produced commercially in large quantities

making it readily available at low cost. It can be obtained in high purity, and the

compound contains only elements required for carbothermal synthesis of boron

carbide. Unfortunately, it has poor solubility in water at room temperature (~5%

by weight at 25 °C). However, water solubility does substantially increase with

temperature (~25% by weight at 80 °C). H3BO3 also has increased solubility in

alcohols when compared to water, especially methanol (~20% by weight at 25 °C).

As discussed in a later chapter, high solubility of the boron source is important so that precursor interaction is maximised while dissolved in solution. The increased solubility of H3BO3 at high temperature as well as the option of alternate solvents provides avenues for increasing solution concentration, making it viable for use as a precursor under specific conditions. Other boron precursors considered include elemental boron, sodium borate (Na2B4O7), sodium borohydride (NaBH4) as well as organoboron compounds and boron containing polymers. Although elemental boron is useful for synthesis of targeted boron carbide stoichiometries for analytical studies[1] [2], it is expensive in purified form and therefore not conducive to large-scale synthesis. Elemental boron is also not readily water soluble, with synthesis usually carried out via powder mixing and/or grinding. This technique does not afford the degree of precursor dispersion required for optimised processing and synthesis conditions, with excessive (>1750 °C) temperatures required for phase formation. Similar to H3BO3, Na2B4O7 is produced in large quantities at low cost with comparable solubility. However, the introduction of Na creates unnecessary complexity in synthesis and purification.

Hence Na2B4O7 is not considered a viable precursor to boron carbide. NaBH4, although having a higher cost, is more attractive as a precursor due to its increased water solubility (~50% by weight at 25 °C) and high reducing capability. NaBH4 has been used as a precursor in the synthesis of metal borides wherein drastically reduced formation temperatures are realised through solution-based[3] and molten salt[4] synthesis techniques. In this case, NaBH4 could possibly be considered viable due to the cost advantages associated with reduced synthesis

31 temperatures at large-scale. However, for the purposes of the work presented

here, the presence of unwanted Na in the reactant composition excludes the use

of NaBH4 as a viable boron source. Organoboron compounds/polymers have also

shown promise as precursors to boron carbide[5] [6]. This option is attractive due

to the ability to rigidly fix a desired constituent precursor ratio through covalent

bonds. Depending on the compound’s properties and processing conditions, use

of organoborons may inhibit precursor aggregation. However, the high cost of

these types of compounds limit their viability for large-scale commercial

production. Because no other compound meets the requirements specified, H3BO3

is the only boron precursor used for the synthesis of boron carbide in this work.

Properties of H3BO3

H3BO3 is produced commercially for use as a fiberglass and heat-resistant glass

additive. To a lesser degree, H3BO3 is also used in cleaning products, fire

retardants, alloys and adhesives as well as other specialty chemicals[7]. To be

more precise, the dehydrated form of H3BO3, boron trioxide (B2O3), is most

commonly used. However, B2O3 itself is not stable in humid conditions and is most

easily obtained from dehydration of H3BO3. The high functionality of H3BO3/B2O3

is derived from the ability of the boron atom to form tetrahedral and trigonal

bonding environments. For example, depending on solution pH, dilute aqueous

- H3BO3 can exist as either B(OH)3 (low pH) or as B(OH)4 (high pH)[8]. This pH

dependent coordination allows for a simple method of manipulation of the boron

atom through pH control in aqueous solution.

H3BO3 undergoes many different temperature dependent phase transformations primarily due to dehydration. H3BO3 itself is stable at standard temperature and pressure (STP) and has trigonal coordination (see Fig. 3.1a). Upon heating past 80

°C, it loses the equivalent of one water molecule per H3BO3 molecule and converts to metaboric acid (HBO2). This description, although useful for quantitatively illustrating water loss, is not entirely accurate. Metaboric acid is not stable as HBO2 and forms (BOH)3O3 moieties comprised of three HBO2 segments (see Fig. 3.1b).

These (BOH)3O3 structural units then form an orthorhombic plate-like crystal structure through hydrogen bonding[9]. Full dehydration of H3BO3 to metaboric acid can be achieved at 130 °C[10]. Metaboric acid is not stable in the presence of humidity and will rapidly reabsorb water to reform H3BO3. Metaboric acid has a second monoclinic form when heated above 140 °C which prevents further dehydration from occurring. The monoclinic form is polymeric in nature, consisting of linked (B3O4)(OH)(OH2) units (see Fig. 3.1c). Although identical in composition to the orthorhombic structural unit, the bonding environment is dissimilar with one of the boron atoms converting to tetrahedral coordination with addition of an

OH2 group. Again, as with the orthorhombic form, the polymeric monoclinic form of metaboric acid is not stable in the presence of humidity and will absorb water to form H3BO3. A third, cubic form of metaboric acid also exists at higher temperatures[11]. However, this phase is only relevant to pressurised/sealed environments as further dehydration will occur well before the cubic form is crystallised.

Fig. 3.1. The different structural units of H3BO3 dehydration products upon heating.

Heating above 130 °C in an unsealed environment results in further dehydration

and B2O3 begins to form. Complete dehydration to B2O3 can be achieved at ~330

°C[10]. B2O3 has drastically different properties compared to its hydrated

precursors; though as with metaboric acid it is highly unstable in the presence of

moisture and will rapidly absorb water converting to metaboric acid and then to

H3BO3. B2O3 is completely amorphous at STP, only crystallising at high

pressures[12]. B2O3 is widely considered to comprise ~75% boroxol ring (B3O6)

units linked in a disordered structure by the remaining mass to total a B2O3

stoichiometry (see Fig. 3.1d); although some ambiguity still exists due to its

vitreous nature[13]. When fully dehydrated, B2O3 exists as a white glassy solid with

a melting point at ~450 °C. This relatively low melting point becomes an issue in

homogenous precursor preparation as the melt can lead to segregation of the

precursors unless properly managed. Indeed, a clear understanding of H3BO3 and

its many phase transitions along with their individual properties is key to the

optimisation and control of boron carbide synthesis. Up until the publication of

Watts et al.[14] (as presented as part of this thesis; refer Chapter 5) this crucial

dependence was not recognised in the literature on boron carbide synthesis.

Indeed, most publications reporting XRD patterns of precursors to boron carbide completely misidentify the diffraction data for H3BO3 as B2O3[15] [16] [17] [18]

[19] [20] [21] [22] [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33]. The misidentification of this phase most likely originated from an earlier inaccurate phase identification that has since been perpetuated through the literature.

However, the distinction between the different phases of H3BO3 is very important as detailed in Chapter 5.

3.1.2 CARBON PRECURSOR

Although the choice of boron precursors is limited, the choices for carbon precursors in boron carbide synthesis are vast. Again, the requirements of low cost, high solubility and elemental composition apply. However, given that only one viable option for the boron precursor could be identified, of greater importance in the carbon source is its interaction dynamics with H3BO3. One strategy for maximising and maintaining homogeneous dispersion of precursors is to facilitate complexing and/or chelation between them. This strategy can effectively ‘lock in’ a desired atomic arrangement, preventing agglomeration or segregation before final product phase formation. In the case of H3BO3, it is known to form complexes with OH group containing compounds[34] [35]. This property has been exploited since the early 1900s, specifically for the titration of H3BO3 using mannitol[36]. The exact complexing mechanism was not completely understood until the 1980s when it was conclusively shown through NMR studies that the dissolved borate ion complexes most effectively with diol type compounds, wherein the OH functional groups are separated by an interstitial

35 atom (1,3-type diols)[37]. At high pH, the tetrahedral coordinated borate anion

can form both monodiol and didiol complexes with these 1,3-type diol compounds

via dehydration (see Fig. 3.2).

Fig. 3.2. Borate anion complexation with 1,3-type diol compounds. Both the monodiol and didiol

interaction can occur.

This didiol complexation with the borate anion is commonly used for the

crosslinking of polymers that contain OH functional groups. In particular, polyvinyl

alcohol forms a very strong complex with the borate anion due its 1,3-diol type

backbone structure[38]. A similar process can also occur with carboxylic acid

containing compounds, wherein the borate anion dehydrates to complex with the

carboxylic OH group[39].

Using this complexation process as the targeted method for precursor interaction,

the main selection criteria then narrows to compounds that contain a large

amount of OH groups, have low cost, high solubility and contain only carbon,

oxygen and hydrogen. High solubility is especially important in this case, as a high

solution concentration is required to maximise precursor interaction to facilitate

complexation. The rationale for this caveat is that for the highest number of OH groups available to complexation with the borate anion, the higher the likelihood that the boron atom will be bound and thus, prevent agglomeration or segregation. A similar rationale is employed in recent literature detailing boron carbide synthesis from solution-based methods as described in sections 2.3.3 and

2.3.4 here. However, a common parallel can be observed in these works, and indeed in most reported solution-based syntheses to boron carbide. As previously identified in section 2.4, this is the relationship between enhanced reaction kinetics, the amount of carbon precursor utilised and the purity of the boron carbide product. Reaction kinetics can be greatly improved by the use of excess carbon to facilitate enhanced boron dispersion in the precursor. This occurs as excess carbon is capable of further separating the boron component with mixed precursors, mitigating coalescence during heat treatment[29]. However, this approach has the drawback of residual carbon impurity in the boron carbide product. Indeed, even when stoichiometric amounts of carbon are used as per the carbothermal method, residual carbon impurity still occurs. Although still related to precursor homogeneity, this effect is attributed to loss of reactant via volatile boron species due to poor encapsulation by the carbon source[40] [20]. This effect can be compensated for by either extra H3BO3 at the expense of reaction kinetics, or by extra carbon at the expense of residual carbon in the final product. However, mitigation of this trade off should be possible by facilitating increased precursor dispersion through careful choice of precursor and processing conditions. Based on the above discussion and selection criteria, two different approaches were

37 developed for the synthesis of boron carbide in this work, with each utilising a

different carbon source.

Polyvinyl acetate as precursor

The first approach employs methanol as solvent to facilitate increased H3BO3

concentration in solution. Unfortunately, polyols are relatively insoluble in

methanol, so a different strategy was developed for dissolution of the carbon

precursor. To this end, polyvinyl acetate (PVAc) generated from vinyl acetate

monomer (see Fig. 3.3) was chosen as the carbon source due its high methanol

solubility and its ability to partially hydrolyse[41] [42] [43]. Partial hydrolysis

substitutes OH groups onto the polymer backbone so that borate anion

complexation can occur, while still maintaining the methanol solubility of the

polymer. The PVAc is also synthesised in-situ while in the presence of dissolved

H3BO3 to further increase the chances of boron covalently incorporating into the

polymer structure. Polymerisation is also carried out in-situ for a second purpose;

to facilitate steric hindrance and further increase precursor dispersion. This aspect

will be discussed further in Chapter 4 along with a detailed description and analysis

of the devised technique.

Fig. 3.3. Free-radical polymerisation of PVAc from VA monomer and subsequent hydrolysis to PVAl.

Sucrose as precursor

The second approach utilises sucrose (Fig. 3.4) as the carbon source and water as the solvent, with a particular focus on precursor handling and processing conditions for maintaining precursor homogeneity. The synthesis of boron carbide utilising sucrose as precursor has been conducted prior to[28] [31] [44] [45] and after[32] the publication of the work conducted here. However, in all cases, the effect of the handling conditions in respect to boron component dispersion is not considered, resulting in impure, inconsistent boron carbide product. Sucrose is chosen due to the large amount of OH groups present in its structure, its extremely low cost and its high water solubility. To increase the poor solubility of H3BO3 in water, dissolution is carried out at 80 °C. To further facilitate precursor interaction and maximise complexation, a precursor gel is formed by slow evaporation of the solvent. A detailed description of this approach is presented in Chapter 5.

Fig. 3.4. Molecular structure of sucrose illustrating the excess of OH groups present.

39 3.2 PROCESS METHODOLOGY

As two different approaches are chosen for investigation, slightly different

processing methodologies are implemented for each developed technique.

Although overall processing largely remains similar, differences arise from the way

in which the condensed precursor is formed and in turn, the way in which the

condensed precursor is handled after condensation. Here, the overall processing

methodology is described, with focus on the differences between the techniques

and why they are required.

3.2.1 PRECURSOR PREPARATION, HANDLING AND TREATMENT

Homogeneously mixed precursor generation and subsequent preparation for final

calcination proceeds fundamentally in three basic steps: (i) full dissolution of

precursor compounds; (ii) solvent removal to form a condensate; and (iii)

condensate preparation for calcination.

Precursor dissolution

In this initial step, care is taken to ensure that full precursor dissolution is achieved,

and that it has occurred in the smallest amount of solvent possible. Full dissolution

is required to ensure that no aggregates or undissolved crystallites are present in

solution, maximising the available molecular surface area for precursor

interaction. A high solution concentration is also required to maximise the

proximity of the precursor constituents (the carbon and boron sources) to

promote complexation. Solution concentration is also maximised for a second

purpose which will be discussed in the next section. Depending on the individual

precursor properties, concentration increases can be achieved through different means (solvent temperature, choice of solvent etc.). In the case of PVAc as precursor, methanol has been implemented as solvent due to the increased solubility of H3BO3 compared with water. This choice also allows for processing of the precursor solution in liquid form at STP. When using water as solvent, as is the case for sucrose as the carbon source, the solution temperature must be elevated to facilitate increased H3BO3 concentration while maintaining full dissolution.

However, this approach may cause issues in processing, as precipitation of the boron source will occur if a high solution temperature is not maintained, as noted in the next section.

Solvent removal

As mentioned above, a second motivation exists in maximising the dissolved mixed precursor solution concentration which is imbedded in the solvent removal process. As identified previously, the primary goal of this work is to maximise precursor homogeneity and dispersion to facilitate increased reaction kinetics, control of morphology and stoichiometry as well as the purity of the product.

Complexation of precursors plays a large role in achieving this. However, processing conditions are also important. In the case of solvent removal, homogenous and fine dispersion of precursors can be promoted through rapid evaporation. When fully dissolved in the liquid state, precursors are at their most intimate level of dispersion as they exist as individual molecular species. By implementing rapid solvent removal, this perfectly distributed liquid form is quickly converted to a solid wherein the high degree of fine dispersion can be

41 maintained. Many different processing architectures exist for achieving this

outcome depending on the properties of the precursor solution. In the case of

PVAc as carbon source, high temperatures and reduced pressures are

implemented to directly convert from the liquid state to the solid state. In this

process lies another motivation for the use of methanol as solvent; because

methanol has a much lower evaporation temperature it can be removed more

easily in comparison to water. However, as an increased temperature must be

maintained for precursor dissolution in the case of sucrose as the carbon source,

direct conversion from the liquid to solid state becomes problematic. Transfer of

the dissolved solution between vessels leads to cooling which in turn causes H3BO3

precipitation to occur. This cooling can be mitigated by the use of heated feedlines

to transfer the precursor solution. Although manageable with the proper

equipment, this option is costly and prone to inconsistencies, and as such was not

a viable option in this work.

For this reason, an alternate process was developed for solvent removal when

using sucrose as the carbon source. A two-step process was developed wherein a

sucrose/H3BO3 gel is formed by slow evaporation of the solvent, followed by a

second solvent removal step where the remaining water in the gel is rapidly

removed. Slow evaporation of the solvent gradually increases precursor contact

as the water evaporates facilitating complexation of the precursors. The resulting

gel can then be easily managed and transferred for rapid solvent removal.

Of great importance in both cases is the temperature implemented for solvent

removal. As described in section 3.1.1, H3BO3 has many different phase

transformations that are temperature dependent. In the case of solvent removal, one of the main goals is to achieve precursor complexation, which is facilitated through interaction of the OH groups in both the boron and carbon sources. If excessive temperatures are implemented in solvent removal, this can lead to dehydration of the boron source as well as structural changes in the carbon source, inhibiting the opportunity to complex. If full conversion to B2O3 occurs

(which begins at temperatures as low as 130 °C), complexation will not occur. This creates a trade off in solvent removal temperature, wherein greater temperatures are able to increase the solvent removal rate. However, damage to the complexation mechanism can occur. For this reason, a temperature of 150 °C is consistently used at the rapid solvent removal stage. This temperature represents the optimum trade off wherein solvent removal rate is maximised and conversion to B2O3 is minimised. Temperatures less than 150 °C do not produce a sufficiently dry condensate product conducive to further processing.

Condensate preparation

Once the dry condensate has been formed, multiple processing stages are required to prepare the sample for final calcination. The dry condensate is generally in the form of a solidified mass. Ring milling is therefore implemented to break the solid condensate into a fine powder. In cases where exposure to atmospheric conditions are to be avoided, the condensate is instead broken down into a powder by hand via mortar and pestle under inert conditions. Next, a ‘pre- treatment’ step is required to compositionally prepare the precursor powder for final calcination. Boron carbide formation proceeds according to Eq. 1.1, 1.2 and

43 1.3 as detailed in section 1.2.3, and any excess of the required elements can lead

to deleterious effects in final phase formation. Of particular importance given the

choice of precursors, is the excess of structurally bound oxygen in the form of

functional groups present in the carbon source. Given that boron carbide

formation is dependent upon the reduction of B2O3 by generated CO, if excess

oxygen is present in the condensed carbon matrix, reduction will preferentially

occur with the carbon source and the B2O3 will remain unreduced. This effect not

only hinders boron carbide formation, but also leads to volatilisation of the boron

source as it is not in a reduced form at high temperature. This mechanism is further

described in section 4.7.2. For this reason, a pre-treatment step is required to

remove any excess oxygen present in the carbon matrix through generation of CO

and CO2. Without the implementation of this step, the final product contains large

amounts of residual carbon with very little boron carbide formed due to loss of

the boron source through volitisation. As this step is dependent upon the amount

of oxygen bound within the condensed carbon matrix, different processing

conditions are required based on the composition of the precursor. The specific

processing conditions required for PVAc and sucrose as carbon precursors are

detailed in Chapters 4 and 5, respectively.

3.2.2 PRECURSOR CALCINATION

Pre-treatment of condensate powder again generates a semi-rigid mass which

requires grinding before calcination. After ring milling/grinding a very fine black

powder is obtained which can then be calcined. Due to the elevated temperatures

required for final phase formation, a specialised graphite insulated furnace is

implemented. A graphite crucible is also used to hold the precursor powder as standard crucible materials (stainless steel, alumina, quartz etc.) will react/decompose at the required temperatures. Due to the static nature of the furnace reaction chamber, only small volumes of powder can be calcined per furnace run (~1 g). If larger volumes of precursor powder are calcined the reaction product is inconsistent. Specifically, the centermost region of the powder mass will contain the targeted, carbon-free, boron carbide product, while the outermost regions of the powder will contain boron carbide with small amounts of residual carbon impurity. This outcome is attributed to an increased reducing environment due to a higher concentration of trapped CO gas in the center of the powder mass.

This effect more efficiently reduces B2O3 to elemental boron in this region, preventing volatilisation of the boron source to a greater degree than in the outermost regions. When using smaller volumes (~1 g or less), this effect is less severe and a higher consistency in the boron carbide product is achieved.

Calcination is always carried out in inert conditions, specifically, under flowing ultra-high purity (UHP) Ar gas. Vacuum calcination is not feasible when using the carbothermal method as it further promotes B2O3 volitisation through the removal of generated CO gas required for local reducing conditions. The actual temperatures implemented for boron carbide phase formation vary depending on the precursors used, the calcination time as well as the prior pre-treatment and handling conditions. Using the techniques developed in this work, complete phase formation can be achieved at temperatures as low as 1250 °C. The specifics of the calcination procedures used for both PVAc and sucrose based synthesis techniques are detailed in Chapters 4 and 5, respectively.

45 3.3 CHARACTERISATION METHODOLOGY

A wide range of characterisation techniques are applicable to the analysis of boron

carbide and precursors. However, as the work presented here does not involve

sintered/solid bodies, physical properties testing of boron carbide is not

conducted. This aspect, although highly pertinent to boron carbide research, is not

investigated primarily because this work focuses on synthesis techniques.

Therefore, the majority of characterisation in this work is directed towards the

enhancement of powder characterisation techniques.

Boron carbide is also a highly refractory material, making the formation of

sintered/solid bodies non-trivial, requiring specialised equipment and processes.

For these reasons, the preparation and characterisation of boron carbide solid

bodies was not investigated. Although to some degree, based on stoichiometric

determination, the physical properties of boron carbide (as a solid body) can be

inferred. As discussed in the final section of this thesis, physical properties testing

of boron carbide synthesised by the sucrose-based technique developed as part

of this thesis has commenced in collaboration with the National Institute for

Materials Science, Japan. This work focuses on forming solid bodies from boron

carbide powders synthesised by different techniques (solution-based and

commercially produced) for subsequent analysis and comparison of their physical

properties[46]. Of particular interest in this collaborative work is the

measurement of thermoelectric properties.

3.3.1 PRECURSOR CHARACTERISATION

Precursors to boron carbide are almost always in powdered form, therefore, techniques that are conducive to powder analysis are primarily used. This type of analysis does not apply to the precursor dissolution stage. As the main requirement at this stage of processing is the full dissolution of the precursor compounds, any precipitation that occurs during optimisation can be identified with the naked eye. However, polymer analysis is carried out from solution as is described in Chapter 4.

The main characterisation technique implemented for the analysis of precursor powders is laboratory XRD due to its ease of use and ability to provide large amounts of information relevant to synthesis optimisation. At the condensation stage, XRD can determine the degree of crystallinity of the material as well as the phase content. Due to the amorphisation of the boron component and the many forms it can take with increasing temperature (see Fig. 3.1), the ability to detect and differentiate these phases is crucial. Thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) also find use in precursor analysis for identifying deleterious melts as well as dehydration and phase formation. As this work also focuses on morphological control, scanning electron microscopy (SEM) plays a large role. Precursor powders after pre-treatment are highly conductive due to the de-oxygenised carbon matrix and hence do not require conductive coating for analysis. However, analysis of condensed powder before pre- treatment is prone to the release of volatile species and degradation and is therefore not viable for analysis under typical SEM conditions. Attenuated total

47 reflectance Fourier transform infrared spectroscopy (ATR-FTIR) is also useful for

phase identification of the boron source in precursor powders.

A large part of the synthesis optimisation procedure developed for the sucrose-

based technique requires precise measurement of water adsorption in the

precursor powders after pre-treatment. This requirement presented technical

challenges as complete inert handling of the precursor powders was required prior

to commencing the measurement of water adsorption. To this end, a specialised

methodology for measuring water adsorption from inert powders was developed,

and is detailed in Chapter 5.

3.3.2 BORON CARBIDE CHARACTERISATION

Similar to precursor powders, the primary characterisation method implemented

for boron carbide analysis is XRD. Unfortunately, due to the high X-ray

transparency of the material, substantial complications in data collection and

modelling are encountered when using standard techniques. XRD is a standard

analysis technique for the collection of structural data and phase identification in

powdered materials. For this reason, the development of an XRD technique

tailored towards improved data collection in high X-ray transparency materials

was deemed of high importance. Through the identification of the physical

processes responsible for poor data quality presented by standard collection

techniques, a data collection methodology was developed by optimisation of the

diffractometer hardware parameters. This in turn allows for highly accurate

modelling of the boron carbide powders synthesised in this work (and indeed by

other means), leading to improved structure and phase analysis. The improved

resolution in structural data and phase differentiation made possible by this newly developed technique also facilitates further progress in synthesis optimisation.

Chapter 6 of this thesis reports the details of this technique as well as the overarching positive implications it has for boron carbide analysis and development. It should be noted that although the developed XRD technique has exceptional capability in boron carbide structure and phase analysis, XRD on its own is not sufficient for the precise determination of free carbon within a given sample. Although the presence of amorphous/graphitic carbon can be detected by XRD, the absence of this signal does not necessarily negate the presence of trace amounts of carbon in the sample. As such, the use of the term ‘residual carbon-free’ in this thesis refers to the absence of a detectable carbon signal in associated XRD data. Accurate determination of the carbon content with a given sample requires the use of complementary analysis techniques such as chemical dissolution methods or Raman analysis.

SEM and dynamic light scattering (DLS) are also routinely implemented for morphological characterisation of synthesised boron carbide. As with pre-treated precursor powders, boron carbide does not require a conductive coating for quality high magnification images to be obtained.

49 3.4 REFERENCES

1. Aselage, T. L. & Tissot, R. G., Lattice Constants of Boron Carbide. Journal of the American Ceramic Society, 1992. 75: p. 2207-2212. 2. Bouchacourt, M. & Thevenot, F., Analytical investigation in the B-C system. Journal of the Less-Common Metals, 1981. 82: p. 219-226. 3. Kapfenberger, C., Hofmann, K. & Albert, B., Room-temperature synthesis of metal borides. Solid State Sciences, 2011. 5: p. 925-930. 4. Portehault, D., Devi, S., Beaunier, P., Gervais, C., Giordano, C., Sanchez, C. & Antonietti, M., A general solution route toward metal boride nanocrystals. Angewandte Chemie International Edition 2011. 50: p. 3262-3265. 5. Yanase, I., Ogawara, R. & Kobayashi, H., Synthesis of boron carbide powder from polyvinyl borate precursor. Materials Letters, 2009. 63: p. 91-93. 6. Mirabelli, M. G. L. & Sneddon, L. G., Synthesis of boron carbide via poly( vinylpentaborane) precursors. Journal of the American Chemical Society, 1988. 110: p. 3305-3307. 7. Woods, W.G., An Introduction to Boron: History, Sources, Uses, and Chemistry. Environmental Health Perspectives, 1994. 102: p. 5-11. 8. Anderson, J.L., Eyring, E. M. & Whittaker, M. P., Temperature jump rate studies of polyborate formation in aqueous boric acid. Jounal of Physical Chemistry 1964. 68: p. 1128-1132. 9. Peters, C. R. & Milberg, M. E., The Refined Srtucture of Orthorhombic Metaboric Acid. Acta Crystallographica, 1963. 17: p. 229-234. 10. Balci, S., Sezgi, N. A. & Eren, E., Boron Oxide Production Kinetics Using Boric Acid as Raw Material. Industrial and Engineering Chimstry Research, 2012. 51: p. 11091-11096. 11. Zachariasen, W. H., The Crystal Structure of Cubic Metaboric Acid. Acta Crystallographica, 1963. 16: p. 380-384. 12. Aziz, M., Nygren, E., Hays, J. F. & Turnbull, D., Crystal growth kinetics of boron oxide under pressure. Journal of Applied Physics, 1985. 57: p. 2233-2242. 13. Soper, A. K., Boroxol Rings from Diffractoin Data on Vitreous Boron Trioxide. Journal of Physics: Condensed Matter, 2011. 23: p. 365402. 14. Watts, J. L., Talbot, P. C., Alarco, J. A. & Mackinnon, I. D. R., Morphology control in high yield boron carbide. Ceramics International, 2017. 43: p. 2650-2657. 15. Kakiage, M., Tahara, N., Yanagidani, S., Yanase, I. & Kobayashi, H., Effect of boron oxide/carbon arrangement of precursor derived from condensed polymer-boric acid product on low-temperature synthesis of boron carbide powder. Journal of the Ceramic Society of Japan, 2011. 119: p. 422-425. 16. Kakiage, M., Tahara, N., Yanase, I. and Kobayashi, H., Low-temperature synthesis of boron carbide powder from condensed boric acid–glycerin product. Materials Letters, 2011. 65: p. 1839-1841. 17. Tahara, N., Kakiage, M., Yanase, I. & Kobayashi, H., Effect of addition of tartaric acid on synthesis of boron carbide powder from condensed boric acid-glycerin product. Journal of Alloys and Compounds, 2013. 573: p. 58-64. 18. Hadian, A. & Bigdeloo, J., The Effect of Time, Temperature and Composition on Boron Carbide Synthesis by Sol-gel Method. Journal of Materials Engineering and Performance, 2008. 17: p. 44-49. 19. Khanra, A., Production of boron carbide powder by carbothermal synthesis of gel material. Bulletin of Material Science, 2007. 30: p. 93-96.

20. Sinha, A., Mahata, T. & Sharma, B., Carbothermal route for preparation of boron carbide powder from boric acid–citric acid gel precursor. Journal of Nuclear Materials, 2002. 301: p. 165-169. 21. Maqbool, M., Rafi-ud-din, Zahid, G. H., Ahmad, E., Asghar, Z., Subhani, T., Shahzad, M. & Zaleem, I., Effect of saccharides as carbon source on the synthesis and morphology of B4C fine particles from carbothermal synthesis precursors. Materials Express, 2015. 5: p. 390-400. 22. Kobayashi, T., Yoshida, K. & Yano, T., Effects of addition of seed grains on morphology and yield of boron carbide powder synthesized by carbothermal reduction. Ceramics International, 2013. 39: p. 3849–3856. 23. Rafi-ud-din, Zahid, G. H., Asghar, Z., Maqbool, M., Ahmad, E., Azhar, T., Subhani, T. & Shahzad, M., Ethylene Glycol Assisted Low-temperature Synthesis of Boron Carbide Powder from Borate Citrate Precursors Journal of Asian Ceramic Societies 2014. 2: p. 268-274. 24. Konno, H., Erata, T., Fujita, K., Aoki, Y., Shiba, K. & Inoue, N., Formation of B/C composites containing B4C from sugar– organoborane complexes. Carbon, 2001. 39: p. 771-785. 25. Kakiage, M., Shoji, T. & Kobayashi, H, Low-temperature carbothermal nitridation of boron oxide induced by networked carbon structure. Journal of the Ceramic Society of Japan, 2016. 124: p. 13-17. 26. Sharifi, E. M., Karimzadeh, F. & Enayati, M. H., Mechanochemical assisted synthesis of B4C nanoparticles. Advanced Powder Technology, 2011. 22: p. 354- 358. 27. Chen, S., Wang, D., Huang, J. & Ren, Z., Synthesis and characterization of boron carbide nanoparticles. Applied Physics A, 2004. 79: p. 1757-1759. 28. Pilladi, T. R., Ananthasivan, K. & Anthonysamy, S., Synthesis of boron carbide from boric ox ide-sucrose gel precursor. Powder Technology, 2012. 246: p. 247-251. 29. Kakiage, M., Tominaga, Y., Yanase, I. & Kobayashi, H., Synthesis of boron carbide powder in relation to composition and structural homogeneity of precursor using condensed boric acid–polyol product. Powder Technology, 2012. 221: p. 257-263. 30. Bigdeloo, J. & Hadian, A., Synthesis of High Purity Micron Size Boron Carbide Powder from B2O3/C Precursor. International Journal of Recent Trends in Engineering, 2009. 1: p. 176-180. 31. Pilladi, T. R., Ananthasivan, K., Anthonysamy, S. & Ganesan, V., Synthesis of nanocrystalline boron carbide from boric acid–sucrose gel precursor. Jounal of Materials Science, 2012. 47: p. 1710-1718. 32. Vijay, S.K., Krishnaprabhu, R., Chandramouli, V. & Anthonysamy, S., Synthesis of nanocrystalline boron carbide by sucrose precursor methodoptimization of process conditions. Ceramics International, 2018. 44: p. 4676-4684. 33. Sheng, Y., Li, G., Meng, H., Han, Y., Xu, Y., Wu, J., Xu, J., Sun, Z., Liu, Y. & Zhang, X., An improved carbothermal process for the synthesis of fine-grained boron carbide microparticles and their photoelectrocatalytic activity. Ceramics International, 2018. 44: p. 1052-1058. 34. Kurokawa, H., Shibayama, M., Ishimaru, T. & Nomura, S., Phase behaviour and sol- gel transition of poly (vinyl alcohol )-borate complex in aqueous solution. Polymer Communications, 1991. 33: p. 2182-2188. 35. Hansen, T. S., Mielby, J. & Riisager, A., Synergy of boric acid and added salts in the catalytic dehydration of hexoses to 5-hydroxymethylfurfural in water. Green Chemistry, 2010. 13: p. 109-113. 36. Hollander, M. & Rieman, W., Titration of Boric Acid in Presence of Mannitol. Ind. Eng. Chem. Anal. Ed., 1945. 17: p. 602-603.

51 37. Shibayama, M., Sato, M., Kimura, Y., Fujiwara, H. & Nomura, S., 11B N.M.R study on the reaction of poly(vinyl alcohol) with boric acid. Polymer Communications, 1988. 29: p. 336-340. 38. Ochiai, H., Shimizu, S., Tadokoro, Y. & Murakami, I., Complex formation between poly(vinyl alcohol) and borate ion. Polymer Communications 1981. 22: p. 1456- 1458. 39. Kustin, K. & Pizer, R., Temperature-jump study of the rate and mechanism of the boric acid-tartaric acid complexation. Journal of the American Chemical Society, 1969. 91: p. 317-322. 40. Kakiage, M., Tominaga, Y., Yanase, I. & Kobayashi, H., Synthesis of boron carbide powder in relation to composition and structural homogeneity of precursor using condensed boric acid–polyol product. Powder Technology, 2012. 221: p. 257-263. 41. Semsarzadeh, M. A., Karimi, A. & Eshtad, M., Polymerizations of Vinyl Acetate in Solution. Iranian Polymer Journal, 1997. 6: p. 261-268. 42. Joshi, D. P. & Pritchard, J. G., Partly alcoholized poly(vinyl acetate) polymers: kinetics of formation and reaction with iodine. Polymer 1977. 19: p. 427-430. 43. Minsk, L. M., Priest, W. J. & Kenyon, W. O. , The Alcoholysis of Polyvinyl Acetate. Journal of the American Chemical Society, 1941. 63: p. 2715-2721. 44. Konno, H., Sudoh, A., Aoki, Y. & Habazaki, H., Synthesis of C/B4C composites from sugar-boric acid mixed solutions. Molecular Crystals and Liquid Crystals, 2002. 386: p. 15-20. 45. Foroughi, P. & Cheng, Z., Understanding the morphological variation in the formation of B4C via carbothermal reduction reaction. Ceramics International, 2016. 42: p. 15189-15198. 46. Sauerschnig, P., Watts, J. L., Vaney, J. B., Talbot, P. C., Alarco, J. A., Mackinnon, I. D. R. and Mori, T., Thermoelectric properties of pure B4C prepared by a solution- based method. Advances in Applied Ceramics, 2018. (to be subbmitted).

CHAPTER 4

SYNTHESIS FROM POLYVINYL ACETATE

53 4.1 CHAPTER FOREWORD

Based on a review of the literature detailing the synthesis of boron carbide from

solution-based techniques, the key parameter identified for enhancing control of

the product (in stoichiometry, morphology and purity) is to form and maintain a

consistent, fine dispersion of precursors until phase formation occurs. To this end,

the first technique developed in this work implemented multiple strategies for

achieving this goal. As described in Chapter 3 for the case of PVAc as the carbon

source, these strategies include: (i) use of solvent (methanol) with high H3BO3

solubility for increased solution concentration; (ii) choice of carbon precursor with

the ability to complex with H3BO3 (hydrolysis of PVAc and/or incorporation

through in-situ polymerisation); (iii) rapid solvent removal; (iv) control of

processing temperatures to avoid failure of the complexation mechanism and

melting of B2O3; and (v) steric hindrance. The use of steric hindrance is the primary

reason for the use of in-situ polymerisation of PVAc and stems from the reaction

kinetics/excess carbon trade off described in section 2.4. The implementation of

in-situ polymerisation allows for control of the amount of carbon available for

reaction in the amounts required for the carbothermal method (through

polymerisation time) while simultaneously creating excess carbon in the form of

unreacted vinyl acetate (VA) monomer. VA is the starting material used in the

polymerisation of PVAc, and in its monomeric form is highly volatile. In this

property lies the strategy for enhancing steric hindrance through excess carbon

while still maintaining the stoichiometric requirements of the carbothermal

method. Excess VA is used than required for the carbothermal method, then

subsequent partial polymerisation of the VA monomer generates the required amount of carbon in the form of PVAc polymer. This approach leaves residual monomer and smaller oligomers present in solution and in condensed precursors to facilitate steric dispersion of the boron source. As these monomers/oligomers have low boiling points compared to the PVAc polymer, they remain present in the initial stages of processing to enhance dispersion of the boron source. At later stages of processing, these monomers/oligomers are then removed through volatilisation. In this way, excess carbon is present to enhance dispersion early in processing, but is removed at elevated temperatures and hence is not present as residual carbon impurity in the final product. Through implementation of these multiple strategies for creating and maintaining enhanced precursor dispersion, this method was successful in reducing calcination times and temperatures.

However, some drawbacks in the technique do exist. Although not initially obvious, these failings only become apparent after comparison with the second, sucrose-based, synthesis technique. Indeed, through the comparison of these two different techniques, a great deal of information can be extracted with regard to synthesis techniques and synthesis mechanisms, which is highly valuable for process optimisation. These details will be further elaborated in Chapter 7 after both synthesis techniques and their characterisation have been described.

55 4.2 ARTICLE 1

IN-SITU CARBON CONTROL IN THE PREPARATION OF PRECURSORS TO

BORON CARBIDE BY A NON-AQUEOUS SOLUTION TECHNIQUE

Joshua L. Watts1,2, Peter C.Talbot1,2, Jose A. Alarco1,2 and Ian D. R. Mackinnon1

1Institute for Future Environments 2Science and Engineering Faculty, School of Chemistry, Physics and Mechanical Engineering

Queensland University of Technology, Brisbane, QLD Australia 4001

Journal of Materials Science and Engineering A

Volume 5: Pages 8-20.

Published May, 2015

4.3 ABSTRACT

Synthesis of high quality boron carbide powders is achieved by carbothermal reduction of B2O3 from a condensed H3BO3/PVAc product. Precursor solutions are prepared via free radical polymerisation of VA monomer in methanol in the presence of dissolved H3BO3. A condensed product is then formed by flash evaporation under vacuum. As excess VA monomer is removed at the evaporation step, the polymerisation time is used to manage availability of carbon for reaction.

This control of carbon facilitates dispersion of H3BO3 in solution due to the presence of residual VA monomer. Boron carbide powders with very low residual carbon are formed at temperatures as low as 1250 °C with a 4 hour residence time.

57 4.4 INTRODUCTION

Boron carbide is used in a wide range of engineering applications due to a

combination of properties including high hardness, a high resistance to chemical

corrosion, a high melting point and a low specific weight. Boron carbide in many

forms – as powders, sintered billets and coatings − is used as wear resistant

refractories, as a coating for cutting tools, for ballistic applications such as armour

plating and as a neutron absorber. Bulk industrial synthesis methods are commonly

undertaken at high temperature, followed by milling and hot pressing and/or

sintering to shape a product[1] [2]. However, the formation of excess free carbon

during these synthesis methods is problematic[2] [3].

Although boron carbide can be synthesised by directly reacting elemental boron

and carbon[4], this technique is rarely used due to the high costs of the purified

elements. Boron carbide is also made by a magnesiothermal reaction wherein

magnesium metal is used to reduce boron oxide for subsequent reaction with a

carbon source. However, acid washing is required to remove magnesium

contaminants[1]. The most widely used commercial technique for producing boron

carbide is the reduction of H3BO3 with carbon black (referred to as the

carbothermal method) at ~1750 °C in electric arc furnaces[5].

The overall reaction mechanism for the carbothermal process is:

4H3BO3 + 7C → B4C + 6CO + 6H2O

The commercial method for carbothermal reduction results in high amounts of

residual carbon that adversely affect the properties of a final or formed product

and thus further processing is required to remove this carbon. For example, the coarse-grained boron carbide powder requires intensive milling to generate a finer powder suitable for sintering; the intense milling also results in contamination from grinding media. Because of these issues, research is focused on alternative lower temperature synthesis methods that result in a fine powder with less residual carbon[3] [6] [7] [8]. The use of catalysts to remove residual carbon has been explored[9], but such methods result in contamination by the catalysts or require expensive precursor materials. Micron sized boron carbide powder with no residual carbon has been successfully formed by mixing excess B2O3 powder and carbon under inert conditions[10]. However, the temperature required for complete reaction is still high at 1800 °C.

Solution based synthesis offers a facile method for the production of boron carbides with potential to form a fine powder at relatively low temperatures.

Complete dissolution of precursor components, if achieved, affords a high degree of homogeneity in subsequently condensed products. This feature may allow enhanced reaction kinetics and greater control over residual-free phase formation.

By dissolving suitable boron compounds and a suitable carbon source in solution, a condensed precursor powder may be achieved by evaporation of the solvent.

Early attempts to synthesise boron carbides, including boron carbide, using solution-based techniques include the work by Weimer and colleagues which required high calcination temperatures[11] [12] [13]. Other methods include the application of pressure during calcination[14] [15] or specialised precursors to form boron carbides[16] [17]. In many cases, the presence of residual carbon is

59 evident[11] [12] [14] [15] [17]. Recently, Yanase et al. report production of boron

carbide at 1300 °C with low residual carbon using mixed aqueous PVAl and H3BO3

solutions[18]. Subsequently, Kakiage et al. improved upon the removal of residual

carbon but with a long residence time of 20 hours at 1300 °C[3].

Kakiage et al.[6] [7] show that water soluble carbon sources (glycerin, mannitol)

when mixed with aqueous H3BO3 solutions are suitable precursors for formation

of boron carbide. With glycerin, nearly carbon-free boron carbide was formed at

1250 °C with a residence time of 5 hours[6]. With mannitol as precursor and H3BO3

in excess, nearly carbon-free boron carbide is obtained at 1300 °C with a 5 hour

residence time[7]. However, in both cases, optimised synthesis with excess carbon

required a pyrolysis step in air to remove this carbon before calcination. We report

here an improved solution-based synthesis for boron carbide that utilises the

polymerisation of a short-chain monomer in the presence of completely dissolved

H3BO3 in a non-aqueous solvent such as methanol.

4.5 EXPERIMENTAL METHODS

Precursor powders are prepared by a four stage process as follows: (i) dissolution of H3BO3 and VA monomer in methanol (MeOH); (ii) a polymerisation stage under

N2 inert atmosphere to generate PVAc; (iii) a flash evaporation stage to remove the solvent/excess monomer to form the dry PVAc/H3BO3 precursor powder, and (iv) a pre-treatment stage where various atmospheric conditions are trialed. Pre- treated powders are then calcined in Ar to form the final boron carbide product.

Precursor powders formed via flash evaporation after the polymerisation stage are referred to as ‘PVAcB’ powders. A summary of the process stages is shown in Fig.

4.1 including the characterisation techniques employed at each stage for reference. Analysis of generated polymers and polymer/H3BO3 products was carried out on powders formed from solvent removal by a rotary evaporator.

Fig. 4.1. Processing stages with accompanying characterisation techniques.

61 4.5.1 STARTING MATERIALS

H3BO3 (99.5%), VA monomer (≥99%), and benzoyl peroxide (75%, remainder

water) from Sigma-Aldrich, AR grade MeOH (99.8%) from Chem-Supply and basic

alumina (90 active) from Merck were used as received.

4.5.2 POLYMERS

Synthesis

50 g of VA was added to 20 g of MeOH and passed over a column of basic alumina

to remove inhibitor. 24.73 g of H3BO3 was then dissolved in 100 g of methanol and

added slowly to the VA solution. 1.33 g of benzoyl peroxide (BzO) initiator

compound was then added and dissolved. A viscous clear colourless solution was

obtained. The solution was then bubbled with N2 for 30 min to remove O2 and then

heated to 65 °C under N2 atmosphere to initiate polymerisation. Separate

polymerisation reactions were then held at 65 °C for a range of times under N2

atmosphere. Polymerisation times of 1, 6 and 18 hours are the focus of this paper

as these times yielded relevant results.

Characterisation

The peak molecular weight (Mp), weight average molecular weight (Mw) and the

polydispersity index (PDI) of synthesised polymer precursors was determined via

gel permeation chromatography (GPC) using a Waters 2487 absorbance detector

in series with a Waters 2414 refractive index detector. Chromatography treatment

used three consecutive phenomenex, phenogel 5 μm columns (300 x 7.8 mm; 104

Å, 103 Å, 50 Å) operating at 30 °C using tetrahydrofuran as eluent at a flow rate of

1 mL/min. These columns were preceded by a Phenomenex 5 μm Linear Mixed

Bed guard column (50 x 7.8 mm). Polymer weights were determined by gravimetric analysis; H3BO3 was not added to the solution before polymerisation due to the azeotrope it forms with methanol which leads to loss of H3BO3 during solvent evaporation and hence errors in the final weight measurement of the polymer.

After polymerisation and solvent removal using a rotary evaporator, the polymer was dried overnight in a vacuum oven at 100 °C and weighed. Differential scanning calorimetry (DSC) and thermogravimetric (TGA) data were collected simultaneously using a Netzsch STA-449F3 instrument.

4.5.3 PVAcB POWDERS

Preparation

PVAcB powders were formed by flash evaporation of solutions prepared as per section 4.5.2 under vacuum within 24 hours of polymerisation. The dry PVAcB powder obtained was then ring milled for 10 sec at 750 rpm. The PVAcB powder was then pre-treated at 450 °C or 550 °C for 1 to 2 hours under three different atmospheric conditions: (i) 3 L/m Ar flow (full Ar flow); (ii) Ar flow with 4 x 10-1 bar absolute pressure (partial vacuum); or (iii) 10-3 bar absolute pressure with no Ar flow (full vacuum), in order to determine optimum phase formation conditions. A summary of these pre-treatment conditions for the PVAcB powders is given in

Table 4.1. The prefixes PV, A and V reflect the atmospheric conditions used (partial vacuum, full Ar flow and full vacuum respectively).

63 Table 4.1. Treatment conditions for condensed PVAcB powders at different polymerisation times where; (PV) = Ar flow, 4 x 10-1 bar, (A) = 3 L/m Ar flow and (V) = 10-3 bar. Pre-treatment → Partial Vacuum (PV) Full Ar Flow (A) Full Vacuum (V) Polymerisation time → 1 hr 6 hr 18 hr 1 hr 6 hr 6 hr 1 hr 450 °C - - - - A6-1-450 - 1 hr 550 °C PV1-1-550 PV6-1-550 PV18-1-550 - A6-1-550 V6-1-550 2 hr 550 °C - - - A1-2-550 - -

Calcination

Pre-treated PVAcB powders were again ring milled and placed in a graphite crucible

and calcined at 1250-1400 °C with a ramp rate of 25 °C/min for 1 to 4 hours under

140 L/m Ar flow.

Characterization

Pre-treated and calcined PVAcB powders were analysed using X-ray diffraction

(XRD), scanning electron microscopy (SEM), attenuated total reflectance Fourier

transform infrared spectroscopy (ATR-FTIR) and Raman spectroscopy. To analyse

powders with the boron component removed, a hot DI water wash was employed.

XRD measurements were collected using a PANalytical X’Pert PRO diffractometer

with Co Kα1 radiation. SEM images were obtained using a JEOL JSM-7001 field

emission scanning electron microscope. ATR-FTIR measurements were taken with

a Nicolet iS50 FT-IR spectrometer while Raman spectra were collected with a

Renishaw System 1000 Raman Microscope with a 40 sec exposure per scan using

a 633 nm excitation wavelength from a HeNe laser. The laser power at the sample

was ~2 mW focused to a spot size of ~1 µm through a x50 objective lens. All spectra

are normalized to the area of the peak at 1088 cm-1 commonly identified as the

breathing mode of the boron icosahedra in boron carbide[19] [20].

4.6 RESULTS

Experiments were undertaken to characterise the starting materials, the formed

PVAc polymer, polymer/H3BO3 mixture, PVAcB powders and the final boron carbide product. Table 4.1 summarizes the pre-treatment conditions for each sample.

4.6.1 POLYMER ANALYSIS

The weights of as-synthesised PVAc for 1, 6 and 18 hour polymerisation times determined via gravimetric analysis are 13.20 g, 34.73 g and 41.05 g, respectively.

The molecular weights, Mp and Mw as well as the PDI of PVAc with polymerisation times of 1, 6 and 18 hours determined by GPC analysis are summarised in Table

4.2. Polydispersity is a measure of polymer molecular mass distribution given by

PDI = Mw/Mn where Mn is the number average molecular weight. These molecular weight determinations reveal a decrease in Mp and Mw values with longer polymerisation time.

Table 4.2. Summary of data collected by GPC analysis for as-synthesised PVAc.

Polymerisation Time Mp Mw PDI 1 hour 36151 82037 2.60 6 hours 26623 28045 2.40 18 hours 23233 22155 2.78

DSC and TGA characterisation was performed on pure H3BO3, 1 hour polymerised

PVAc as well as PVAc polymerised for 1 hour in the presence H3BO3 for comparison.

The results of this characterisation are shown in Fig. 4.2 and Fig. 4.3.

Fig. 4.2. DSC data for H3BO3, 1 hour polymerised PVAc and 1 hour polymerised PVAc in the presence

of H3BO3.

Fig. 4.3. TGA data for H3BO3, 1 hour polymerised PVAc and 1 hour polymerised PVAc in the presence

of H3BO3.

4.6.2 PVAcB POWDER ANALYSIS

SEM images of typical pre-treated PVAcB powders before and after washing with hot DI water are shown in Fig. 4.4. Note the fine scale circular features ranging in size from 100 nm to ~1 µm. Close inspection reveals very fine (~10 nm cross section) string-like features coating the carbon matrix (highlighted by the white arrow in Fig. 4B).

Fig. 4.4. SEM images of typical pre-treated PVAcB powder before (A, B) and after (C) washing. The red arrows highlight a position where a crack has recombined due to the removal of the boron component and the white arrows highlight the presence and subsequent removal of the H3BO3

‘strings’.

67 PVAcB powders were analysed by XRD after pre-treatment. A typical XRD pattern

of this powder is shown in Fig. 4.5 and compared against XRD patterns for H3BO

and B2O3. This comparison suggests that the string-like features observed in the

SEM images of unwashed pre-treated PVAcB powders (Fig. 4.4B) are H3BO3.

Fig. 4.5. XRD trace comparison of a typical PVAcB powder after pre-treatment, H3BO3 (ICSD ref. 98-

006-1354) and B2O3 (ICSD ref. 98-001-6021).

The ATR-FTIR transmittance spectrum of a typical unwashed pre-treated PVAcB

powder compared with pure H3BO3 is shown in Fig. 4.6. The spectra are nearly

identical indicating that the signal from the H3BO3 is far more intense than any

response from the carbon matrix. This result also corroborates the information

gained from XRD analysis. ATR-FTIR of washed PVAcB powders with the boron

component removed is a match for an aromatic hydrocarbon.

Fig. 4.6. ATR-FTIR scan comparison of unwashed A1-2-550 powder and H3BO3.

4.6.3 PROCESSING CONDITIONS: FORMATION OF BORON CARBIDE

Synthesis conditions were systematically evaluated to establish the key determinants for optimised formation of a phase pure, carbon-free boron carbide product. These conditions included polymerisation time, pre-treatment atmospheric conditions and calcination time. For each suite of conditions, the same heating rate has been used at the particular temperature range.

Polymerisation time

PVAcB powders of 1, 6 and 18 hour polymerisation times that were pre-treated for

1 hour at 550 °C with partial vacuum conditions (PV1-1-550, PV6-1-550 and PV18-

1-550 respectively) were analysed by XRD after calcination for 1 hour (Fig. 4.7) and

2 hours (Fig. 4.8) at 1400 °C. Note in Fig. 4.7 the difference in relative proportions of carbon resultant from different polymerisation times. The proportion of carbon is the lowest for PVAcB powder with a 1 hour polymerisation time.

Fig. 4.7. PVAcB powders of varying polymerisation times pre-treated under 4 x 10-1 partial vacuum

and Ar flow calcined at 1400 °C for 1 hour.

Fig. 4.8. PVAcB powders of varying polymerisation times pre-treated under 4 x 10-1 partial vacuum

and Ar flow calcined at 1400 °C for 2 hours.

Pre-treatment atmosphere

A typical XRD pattern of PVAcB powder calcined at 1400 °C for 1 hour without pre- treatment is shown in Fig. 4.9. Note the excess carbon and absence of an H3BO3 phase.

Fig. 4.9. XRD pattern of typical PVAcB powder calcined at 1400 °C for 1 hour without pre-treatment.

PVAcB powders with 6 hour polymerisation time that were pre-treated under full

Ar flow, partial vacuum and full vacuum conditions (A6-1-550, PV6-1-550 and V6-

1-550 respectively) were calcined to reaction completion to evaluate the influence of vacuum on the carbon precursor component. XRD traces of the calcined products are shown in Fig. 4.10; note the substantial difference between traces due to the variation in pressure. The time required to reach complete boron carbide formation is reduced by 1 hour for A6-1-550 PVAcB powder.

Fig. 4.10. V6-1-550, PV6-1-550 and A6-1-550 PVAcB powders taken to phase formation completion

(A6-1-550 = 1400°C for 1 hour, PV6-1-550 and V6-1-550 = 1400°C for 2 hours) illustrating the effect

of vacuum.

Optimisation

PV6-1-550 PVAcB powder was also calcined for 1, 2, 3, and 4 hours under 140 L/m

Ar flow at 1250 °C to demonstrate that a lower calcination temperature is possible

as data shown in Fig. 4.6 and Fig. 4.7 suggest that material polymerised for six

hours has an optimum carbon-boron ratio to form near carbon-free boron carbide.

The XRD patterns of these calcined powders are shown in Fig. 4.11.

Fig. 4.11. PV6-1-550 PVAcB powder calcined at 1250 °C for 1, 2, 3 and 4 hours under 140 L/m Ar flow.

Because residual carbon is observed in calcined product formed from PVAcB powders pre-treated under full Ar flow conditions at atmospheric pressure (see

A6-1-550, Fig. 4.10), a reduction of the polymerisation time to 1 hour is possible.

XRD traces of 1 hour polymerised PVAcB powder pre-treated under full Ar flow

(A1-2-550) calcined at 1400 °C are shown in Fig. 4.12 and demonstrate that a 2 hour calcination is sufficient for the formation of the near carbon-free product.

Fig. 4.12. A1-2-550 PVAcB powder calcined at 1400 °C for 1 and 2 hours under 140 L/m Ar flow.

4.6.4 BORON CARBIDE PRODUCT

SEM

The morphologies of near carbon-free boron carbide (calcined from A1-2-550

PVAcB powder at 1400 °C for 2 hours) and boron carbide with residual carbon

(calcined from PV18-1-550 PVAcB powder at 1400 °C for 2 hours) are shown in Fig.

4.13A and Fig. 4.13B, respectively. In Fig. 4.13A, well-formed euhedral grains range

in size from ~1 µm to 10 µm in size. In Fig. 4.13B, similar euhedral particles, but

with a smaller size range (from < 1 µm to ~5 µm), are shown. The inset in Fig. 4.13B

shows residual carbon as sub-micron particles on the surface of boron carbide

grains.

Raman Spectra

Raman spectra from three different boron carbide samples corresponding to those

identified in Fig. 4.8 (calcined from PV18-1-550, PV6-1-550 and PV1-1-550 PVAcB

powders for 2 hours at 1400 °C) are shown in Fig. 14.4. Each spectrum is from different locations on individual grains in each sample. For each sample, up to 20 different spectra are obtained under the same experimental conditions. Three representative spectra from each sample are shown in Fig. 4.14.

In Fig. 4.14, peaks at 1340 cm-1 and 1589 cm-1 are characteristic signatures of residual carbon[21]. The intensity of these peaks indicates the relative presence or absence of residual carbon within the sample. Peaks at lower wavenumber between 250 cm-1 and 400 cm-1 are known to vary in intensity dependent on the carbon content within the actual boron carbide crystal structure[19]. In Fig. 4.14, spectra for sample PV1-1-550 show very low signatures for residual carbon. This qualitative result is consistent with XRD data shown in Fig. 4.8. Similarly, the majority of grains analysed in sample PV6-1-550 also show a low residual carbon signature using Raman spectroscopy. Sample PV18-1-550 shows a consistently high number of grains with high residual carbon content which is also reflected in the XRD pattern obtained on the bulk sample.

Fig. 4.13. SEM images of boron carbide powder calcined at 1400 °C for 2 hours from A1-2-550 (A)

and PV18-1-550 (B) PVAcB powders at the same magnification. The smaller particle size of the

sample in B is apparent. The inset of Fig. 4.13B shows the presence of residual carbon (scale bar 1

µm).

Fig. 4.14. Raman spectra of boron carbide powders calcined from PV18-1-550, PV6-1-550 and PV1-

1-550 PVAcB powders.

77 4.7 DISCUSSION

4.7.1 POLYMERISATION CHARACTERISTICS

Gravimetric analysis was employed to quantify the amount of polymer formed for

polymerisation times of 1, 6 and 18 hours. A substantial increase in the final PVAc

weight was observed between 1 and 6 hour polymerisations. After 18 hours, only

a small amount of extra PVAc was formed. This outcome suggests that complete

polymerisation occurs soon after 6 hours. The 18 hour polymerisation condition

was chosen to ensure polymerisation had gone to completion to maximise the

amount of carbon formed. The increased carbon content with polymerisation time

is also reflected in the XRD patterns of calcined PVAcB powders formed at different

polymerisation times as shown in Fig. 4.7.

With free radical polymerisation, the formation of high molecular weight polymer

occurs immediately upon initiation, and under ideal conditions the molecular

weight of the polymer remains unchanged throughout the course of

polymerisation[22]. GPC analysis of 1, 6 and 18 hour polymerised PVAc shows

decreased values for Mp and Mw with increased polymerisation time - contrary to

expectation (see Table 4.2). This effect may be due to gradual contamination of the

system by O2 which results in radical pacification through the formation of

peroxides[23]. Contamination from atmospheric oxygen may occur through the

seals of the reaction vessel, especially at longer polymerisation time. GPC

chromatograms of 1, 6 and 18 hour polymerised PVAc also show evidence of some

low molecular weight species which suggests the presence of some residual VA monomer and PVAc oligomers.

DSC and TGA characterisation was carried out on three samples: (a) pure H3BO3;

(b) 1 hour polymerised PVAc; and (c) 1 hour polymerised PVAc in the presence of

H3BO3. DSC data show a peak at 225 °C that is not present in either the pure H3BO3 sample or the 1 hour PVAc sample, indicating that complexation between the

H3BO3 and the polymer has occurred (see Fig. 4.2). This mechanism is also suggested by the small hump observed in the TGA data shown in Fig. 4.3. TGA data were also used to identify the temperature required for pre-treatment of precursor condensates. As shown in Fig. 4.3, no further weight loss is observed above 500

°C. Thus, a temperature of 550 °C is considered a suitable pre-treatment temperature to ensure sufficient decomposition of the sample before calcination.

4.7.2 PVAcB POWDER COMPOSITION AND MORPHOLOGY

PVAcB powders calcined without pre-treatment predominantly contain residual carbon, exhibit minimal boron carbide and contain no H3BO3 (see Fig. 4.9).

Furthermore, PVAcB powder pre-treated at 450 °C exhibits a similar but less dramatic increase in residual carbon compared with PVAcB powder pre-treated at

550 °C. This outcome suggests that the carbon matrix is not available to react optimally with boron under these conditions, and thus, results in loss of boron by volatilisation of gaseous boron-oxide species[24] [25]. This volatilisation is attributed to the presence of excess oxygen in the PVAc which reacts preferentially with the reducing CO gas atmosphere (see Eq. 4.4). After dehydration of H3BO3 in

79 the PVAcB powder has occurred, the carbothermal reaction can be represented

by:

2B2O3 + 7C → B4C + 6CO [Eq. 4.1]

The overall carbothermal process takes place in two stages, the first of which is the

reduction of B2O3 by CO, followed by the reaction of elemental boron with carbon

to form B4C as shown in Eq. 4.2 and Eq. 4.3:

B2O3 + 3CO → 2B + 3CO2 [Eq. 4.2]

4B + C → B4C [Eq. 4.3]

If excess oxygen is present in the precursor carbon matrix, reduction by CO will

prefer reaction with the matrix leaving B2O3 unreduced:

B2O3 + CO + O(matrix) → B2O3 + CO2 [Eq. 4.4]

By comparing Eq. 4.2 and Eq. 4.4 it can be seen that the presence of excess oxygen

within the carbon matrix greatly hinders the carbothermal reaction process and

leads to boron volatilisation as the boron component is insufficiently reduced at

high temperature. For this reason, a pre-treatment stage is utilised to remove

residual oxygen from the polymer matrix before calcination.

ATR-FTIR of unwashed pre-treated powders shows a transmittance spectrum that

matches H3BO3 (Fig.4.6). This spectrum shows no attributes for carbon despite

being present in large quantities. Since the penetration depth of an IR signal is

limited, the spectrum in Fig. 4.6 suggests that H3BO3 forms as a homogeneous

coating on the carbon matrix. This finding is contrary to earlier work on solution

based methods[3] [6] [7] [8] [18] in which it is proposed that boron at this stage of processing is B2O3. Based on the ATR-FTIR data shown in Fig.4.6, and the well- known hygroscopic nature of B2O3[26], the precursor, once exposed to air after pre-treatment, will rapidly reabsorb water to form H3BO3 as per the reaction shown in Eq. 4.5. The PVAcB powder XRD pattern shown in Fig. 4.5 matches the literature data for H3BO3 and supports this interpretation of the ATR-FTIR spectrum shown in Fig. 4.6.

B2O3 + 3H2O → 2H3BO3 [Eq. 4.5]

SEM images of pre-treated PVAcB powder before washing (Fig. 4.4A and Fig. 4.4B) reveal a porous structure with pore sizes ranging from 100 nm up to 1 µm. The pore structure is very similar to that reported by Kakiage et al.[3] for condensed precursors. A typical SEM image of pre-treated PVAcB powder after washing can be seen in Fig. 4.4C. Comparison with the same area without washing (Fig. 4.4B) shows that the pores are of similar dimension. These images, as well as the ATR-

FTIR results detailed above, show that the boron precursor is H3BO3 rather than

B2O3, and appears to create a surface coating inside the pores of the carbon matrix.

Fine strings of H3BO3 with diameters of less than ~10 nm are observed (see white arrow in Fig. 4.4B). This description of the reaction mechanism is different from that proposed by Kakiage et al.[3] who consider that pores in their precursor material are filled with B2O3. Equivalent SEM images of the polyvinyl alcohol/H3BO3 powder prepared by Kakiage et al. or analyses of the precursor powder before washing are not available for comparison.

81 Unwashed pre-treated PVAcB powders also exhibit surface cracking of the carbon

matrix. These surface cracks are attributed to water absorption by B2O3 after

exposure to air which causes swelling as it hydrates to form H3BO3 (see Eq. 4.5).

After washing with hot water, these cracks appear to ‘heal’ as they close on the

surface, as highlighted by the red arrows in Fig.4.4B and Fig. 4.4C.

4.7.3 TREATMENT ATMOSPHERE FOR PVAcB POWDERS

The treatment atmosphere of PVAcB powders is found to have a significant effect

on precursor composition, specifically carbon content, and hence different

atmospheric treatment conditions were trialed and analysed to ascertain the

optimum pre-treatment conditions to form near carbon-free boron carbide.

Partial Vacuum

Fig. 4.7 shows XRD traces of PV1-1-550, PV6-1-550 and PV18-1-550 PVAcB

powders after calcination at 1400 °C for 1 hour under 140 L/m Ar flow. The carbon

peak increases with increased polymerisation time. This trend shows that control

of carbon content in PVAcB powder is achieved by variation of the polymerisation

time. Heat treatment for a further 1 hour at 1400 °C results in reaction completion

for all samples as seen in Fig. 4.8. PV6-1-550 PVAcB powder calcines to nearly

carbon-free boron carbide. This result indicates that 6 hours is the optimum

polymerisation time to generate sufficient PVAc to balance the requirements of

the carbothermal reaction under partial vacuum conditions.

In contrast, PV1-1-550 PVAcB powder gives boron carbide with residual H3BO3

impurity because there is not enough carbon present for reaction under these

conditions. In addition, an 18 hour polymerisation time gives boron carbide with a large amount of residual carbon and no residual H3BO3. This outcome is due to the excess carbon present in the 18 hour polymerised PVAcB powder. Fig. 4.11 shows

XRD patterns for PV6-1-550 PVAcB powder calcined at 1250°C for 1, 2, 3 and 4 hours under 140 L/m Ar flow. Boron carbide starts to form after 1 hour, and the reaction has gone to completion after 4 hours. Minimal residual carbon and no

H3BO3 is observed in the final boron carbide product after 4 hours at 1250 °C.

Boron carbide calcinations taken to completion that contain excess carbon (PV18-

1-550 PVAcB powder calcined at 1400 °C for 2 hours in this example) exhibit regions of hollow spherical carbon shells attached to the surface of the boron carbide particles (Fig. 4.13B insert). The average particle size of these boron carbide powders (Fig. 4.13B) is also noticeably smaller compared to near carbon- free boron carbide powders owing to the increased H3BO3 dispersion that results from the extra carbon presence from the 18 hour polymerisation time. Free carbon is also observed as independent agglomerates throughout the sample.

Full Ar Flow

The presence of residual carbon in the XRD pattern of A6-1-550 in Fig. 4.10 is due to the absence of vacuum in the pre-treatment stage (see Full Vacuum section below). Because of this effect, the polymerisation time can be reduced significantly to achieve the required amount of carbon for optimum phase formation without residual H3BO3 and minimal residual carbon under atmospheric pressure pre- treatment conditions. The reduced calcination time of A6-1-550 (1 hour compared to 2 hours for PV6-1-550 and V6-1-550) to reach complete boron carbide phase

83 formation can be attributed to excess carbon that is present in the PVAcB powder.

As reported previously[6] [7], homogeneity of the boron component within the

precursor powder is increased by the presence of excess carbon. Although this

effect is desirable, it is not practical as it results in excess carbon impurity in the

final product.

Fig. 4.12 shows XRD patterns of 1 hour polymerisation PVAcB powder pre-treated

at 550 °C for 2 hours in full Ar flow (A1-2-550) after final calcination at 1400 °C for

1 and 2 hours under 140 L/m Ar flow. After 1 hour, a significant amount of boron

carbide formation has occurred with carbon and H3BO3 phases still present. After

2 hours, boron carbide phase formation has gone to completion. At 2 hours

calcination time, almost carbon-free boron carbide with no H3BO3 component is

achieved indicating that a 1 hour polymerisation time is optimum for PVAcB

powders pre-treated under 3 L/min full Ar flow conditions. The increased

calcination time required for A1-2-550 PVAcB powder compared to A6-1-550

PVAcB powder (an extra 1 hour at 1400 °C) is due to the reduced amount of carbon

in the PVAcB powder from a shorter polymerisation time (1 hour).

SEM images of near carbon-free boron carbide calcined from A1-2-550 PVAcB

powder at 1400 °C for 2 hours show interconnected particles with sizes ranging

from sub-micrometer to ~10 µm (Fig. 4.13A). Some rod-like structures are also

dispersed throughout the agglomerates. Free carbon is not observed on the

surface of the boron carbide grains nor as separate particles within the sample.

These same observations are true for SEM images of near carbon-free boron

carbide calcined from PV6-1-550 PVAcB powder

Full Vacuum

Full vacuum pre-treatment (1 x 10-3 bar absolute pressure, no Ar flow) was carried out at 550 °C for 1 hour on 6 hour polymerised PVAcB powder (V6-1-550) for comparison with 6 hour polymerised PVAcB powder pre-treated with partial vacuum and full Ar flow conditions (PV6-1-550 and A6-1-550 respectively). Fig.

4.10 shows a comparison of the XRD patterns collected for these three precursor materials taken to boron carbide phase formation completion. Product calcined from V6-1-550 PVAcB powder shows no residual carbon and a residual H3BO3 component. Product formed from PV6-1-550 PVAcB powder shows no residual components. Product from A6-1-550 PVAcB powder shows a residual carbon component.

From these results, it can be concluded that applying vacuum to the pre-treatment stage removes extra carbon from the PVAcB powder and the amount removed is dependent on vacuum conditions. This effect can be accounted for by the low vapour pressure of residual monomer at higher temperature resulting in removal of the monomer and other short chain polymer moieties that are not removed during the evaporation step. The presence of these short chain polymer fragments is confirmed via GPC analysis (see section 4.7.1).

4.7.4 RESIDUAL CARBON

As shown earlier, the polymerisation time is optimized to minimize the residual carbon content via appropriate balance of reactants for subsequent carbothermal reaction. Raman spectra shown in Fig. 4.14 provide useful insight on the influence

85 of polymerisation time and the form of carbon in the final boron carbide product.

As mentioned previously, independent carbon agglomerates, attached to the

boron carbide grains or as independent particles, are not observed in SEM images

of near carbon-free boron carbide samples (calcined from PV6-1-550 and A1-2-550

PVAcB powder), yet a small amount of carbon is detectable from XRD analysis.

Raman investigation reveals that this small amount of residual carbon is present

on the surface of the boron carbide grains. Furthermore, the amount of carbon

detected varies from grain to grain, but the typical amounts are within the ranges

shown by the peak intensities in Fig. 4.14.

In the case of boron carbide calcined from PV1-1-550 PVAcB powder (1 hour

polymerisation), low amounts of boron carbide and no residual carbon or H3BO3 is

observed after calcination (compare XRD intensities in Fig. 4.8) due to poor

homogeneity of the boron and carbon components in the precursor as well as the

proclivity of B2O3 to readily volatilize at temperature. Despite this, the Raman

spectra show that small amounts of residual carbon are still present on the surface

of the boron carbide grains. Boron carbide calcined from PV18-1-550 PVAcB

powder shows increased carbon content which is consistent with XRD and other

data shown previously and is due to an increased polymerisation time.

As noted in section 4.6.4, Raman spectra can provide useful qualitative indications

of sample stoichiometry[19]. Near carbon-free boron carbide calcined from PV6-

1-550 PVAcB powder shows a consistent intensity of the peaks at 270 cm-1 and 320

cm-1 over all Raman spectra obtained from all grains. This outcome, examples of

which are shown in Fig. 4.14, indicates that the quality − or stoichiometry − of the

boron carbide structure formed by this synthesis is consistent across all individual grains. However, the Raman spectra for boron carbide samples calcined from both

PV18-1-550 and PV1-1-550 PVAcB powder show significant variations in the intensities of peaks at 270 cm-1 and 320 cm-1 for different grains. Thus, the stoichiometry of boron carbide powder containing residual impurities is variable within the sample, while near carbon-free boron carbide powder shows consistency in structural carbon content and hence overall stoichiometry throughout the entire sample.

87 4.8 CONCLUSION

By polymerising VA monomer in the presence of dissolved H3BO3 in methanol, the

amount of carbon available for reaction in PVAcB powders can be controlled via

the polymerisation time. Increased precursor dispersion results from the presence

of excess carbon in the form of unreacted VA monomer in solution. This feature

affords excellent homogeneity of the reactants without requirement for excess

carbon in the PVAcB powder. Using this technique, near carbon-free boron carbide

powders are formed after two hours at 1400 °C as well as after four hours at 1250

°C without the need for carbon removal from the PVAcB powder via pyrolysis in

air.

4.9 ACKNOWLEDGMENTS

The author would like to gratefully acknowledge Mitchell De Bruyn, Tony Raftery,

Alison Chou, Llew Rintoul, John Colwell and James Blinco for useful discussions and assistance with characterisation as well as the Institute for Future Environments for providing funding.

89 4.10 REFERENCES

1. Thevenot, F., Boron Carbide - A Comprehensive Review. Journal of the European Ceramic Society, 1990. 6: p. 205-225. 2. Alizadeh, A., Taheri-Nassaj, E. & Ehsani, N., Synthesis of boron carbide powder by a carbothermic reduction method. Journal of the European Ceramic Society, 2004. 24: p. 3227-3234. 3. Kakiage, M., Tahara, N., Yanagidani, S., Yanase, I. & Kobayashi, H., Effect of boron oxide/carbon arrangement of precursor derived from condensed polymer-boric acid product on low-temperature synthesis of boron carbide powder. Journal of the Ceramic Society of Japan, 2011. 119: p. 422-425. 4. Benton, S. & David, R., Methods for preparing boron carbide articles. US patent no. 3 914 371, 1975. 5. Bigdeloo, J. & Hadian, A., Synthesis of High Purity Micron Size Boron Carbide Powder from B2O3/C Precursor. International Journal of Recent Trends in Engineering, 2009. 1: p. 176-180. 6. Kakiage, M., Tahara, N., Yanase, I. and Kobayashi, H., Low-temperature synthesis of boron carbide powder from condensed boric acid–glycerin product. Materials Letters, 2011. 65: p. 1839-1841. 7. Kakiage, M., Tominaga, Y., Yanase, I. & Kobayashi, H., Synthesis of boron carbide powder in relation to composition and structural homogeneity of precursor using condensed boric acid–polyol product. Powder Technology, 2012. 221: p. 257-263. 8. Tahara, N., Kakiage, M., Yanase, I. & Kobayashi, H., Effect of addition of tartaric acid on synthesis of boron carbide powder from condensed boric acid-glycerin product. Journal of Alloys and Compounds, 2013. 573: p. 58-64. 9. Krishnarao, R., Subrahmanyam, J., Kumar, T. & Ramakrishna, V., Effect of K2CO3, and FeCl3 on the formation of B4C through carbothermal reduction of B2O3. Journal of Alloys and Compounds, 2010. 496: p. 572-576. 10. Junga, C., Leeb, M & Kim, C., Preparation of carbon-free B4C powder from B2O3 oxide by carbothermal reduction process. Materials Letters, 2004. 58: p. 609-614. 11. Weimer, A., Roach, R. & Haney, C., Rapid Carbothermal a Graphite Reduction of Boron Oxide in Transport Reactor. American Institute of Chemical Engineers, 1991. 37: p. 759-768. 12. Weimer, A., Moore, W., Roach, R., Hitt, J. & Dixit, R., Kinetics of Carbothermal Reduction Synthesis of Boron Carbide. Journal of the American Ceramic Society, 1992. 75: p. 2509-2514. 13. Khanra, A., Production of boron carbide powder by carbothermal synthesis of gel material. Bulletin of Material Science, 2007. 30: p. 93-96. 14. Sinha, A., Mahata, T. & Sharma, B., Carbothermal route for preparation of boron carbide powder from boric acid–citric acid gel precursor. Journal of Nuclear Materials, 2002. 301: p. 165-169. 15. Hadian, A. & Bigdeloo, J., The Effect of Time, Temperature and Composition on Boron Carbide Synthesis by Sol-gel Method. Journal of Materials Engineering and Performance, 2008. 17: p. 44-49. 16. Najafi, A., Golestani-Farda, F., Rezaiea, H. R. & Ehsania, N., Effect of APC addition on precursors properties during synthesis of B4C nano powder by a sol–gel process. Journal of Alloys and Compounds, 2011. 509: p. 9164-9170. 17. Najafi, A., Golestani-Fard, F., Rezaie, H. & Ehsani, N., A novel route to obtain B4C nano powder via sol–gel method. Ceramics International, 2012. 38: p. 3583-3589.

18. Yanase, I., Ogawara, R. & Kobayashi, H., Synthesis of boron carbide powder from polyvinyl borate precursor. Materials Letters, 2009. 63: p. 91-93. 19. Tallant, D. R., Aselage, T. L., Campbell, A. N. & Emin, D., Boron Carbides: Evidence for Molecular Level Disorder Journals of Non-Crystalline Solids 1988. 106: p. 370- 373. 20. Shirai, K. & Emura, S., Lattice Vibrations of Boron Carbide. Journal of Solid State Chemistry, 1997. 133: p. 93-96. 21. Ferrari, A. C. & Robertson, J., Interpretation of Raman spectra of disordered and amorphous carbon. Physical Review B, 2000. 61: p. 14095. 22. Odian, G., Principles of Polymerisation, Fourth Edition. John Wiley & Sons, Inc., 2004. 23. Allen, P. W., Inhibition Periods in the Polymerization of Vinyl Acetate. Journal of Polymer Science, 1955. 17: p. 156-158. 24. Rentzepis, P., White, D. & Walsh, P., The Reaction Between B2O3 and C: Heat of Formation of B2O2. Journal of Physical Chemistry, 1960. 64: p. 1784-1787. 25. Lamoreaux, R. & Hildenbrand, D., High-Temperature Vaporization Behavior of Oxides II. Journal of Physical and Chemical Reference Data, 1986. 16: p. 419-443. 26. Perez-Enciso, E., Ramos, M. & Vieira, S., Low-temperature specific heat of different B2O3 glasses. Physical Review B, 1997. 56: p. 32-35.

CHAPTER 5

SYNTHESIS FROM SUCROSE

5.1 CHAPTER FOREWORD

The development of a second synthesis technique was initially driven by the findings generated in the final optimisation stages of the PVAc-based synthesis technique described in Chapter 4. Specifically, this was the realisation that pre- treated precursor powder aging had a significant effect on calcined boron carbide purity. Pre-treated PVAcB powders that were exposed to atmospheric conditions for longer periods of time contained less residual carbon impurity in subsequently calcined boron carbide powders, while pre-treated PVAcB powders that were kept under inert conditions before calcination contained large amounts of residual carbon. The mechanism of this effect was initially not well understood, so further investigation was carried out in a simplified environment to minimise process variables. To this end, sucrose was chosen as the carbon source and water as the solvent for reasons previously detailed (section 3.1.2). Preliminary investigations with PVAcB pre-treated precursors identified morphological changes due to B2O3 conversion to H3BO3 upon exposure to water adsorption from atmospheric humidity as the reason for final product variability. Because of this, a methodology for complete processing of mixed precursor powders under inert conditions was devised. The effects of atmospheric water adsorption in precursor powders on phase purity could then be conclusively identified. As will be detailed later in this chapter, the effect of water adsorption in condensed and pre-treated precursor powders has drastic effects on calcined boron carbide yield, morphology and purity. The effects of water adsorption are also found to be very different for precursors generated through the PVAc-based technique and the sucrose-based

93 technique. As mentioned previously, these differences will be further elaborated

in Chapter 7 after the details of the sucrose-based technique and the

characterisation of powders generated from this method have been fully

described.

5.2 ARTICLE 2

MORPHOLOGY CONTROL IN HIGH YIELD BORON CARBIDE

Joshua L. Watts1,2, Peter C.Talbot1,2, Jose A. Alarco1,2 and Ian D. R. Mackinnon1

1Institute for Future Environments 2Science and Engineering Faculty, School of Chemistry, Physics and Mechanical Engineering

Queensland University of Technology, Brisbane, QLD Australia 4001

Ceramics International

Volume 43, pages 2650-2657

Published November, 2016

95 5.3 ABSTRACT

The production of boron carbide powder with uniform particle size in high yield is

demonstrated via precise control of precursor processing and handling conditions.

A gel is formed by complete dissolution of boric acid (H3BO3) and sucrose in water

which is dried, pre-treated at 550 °C and then calcined at 1400 °C to form the

boron carbide product. Optimised synthesis conditions are obtained by managing

exposure of precursor powders to atmospheric conditions. Exposure of precursor

powders to humid environments causes significant changes in morphology due to

water adsorption by dehydrated H3BO3 variants. This hydration drives a loss of

contact surfaces between boron and carbon components, decreasing boron

carbide yield and increasing residual carbon impurity. Reactant dispersion in

precursor powders is also shown to have a direct effect on formed boron carbide

morphology.

5.4 INTRODUCTION

Boron carbide is a non-oxide ceramic with many specialised applications owing to low weight, extreme hardness and a high melting point (~2300 °C)[1]. Boron carbide also has a high neutron capture cross-section, a high resistance to chemical attack and is a p-type semiconductor that exhibits thermoelectric properties[2]. Boron carbide is produced commercially by reacting powdered carbon and H3BO3 in an electric-arc furnace at high temperature (1750 °C and above)[1] [3]. The mechanism of formation by this means is commonly referred to as the carbothermal method and is summarised by the following reaction:

4H3BO3 + 7C → B4C + 6CO + 6H2O[4] [Eq. 5.1]

The reaction proceeds in three steps: (i) dehydration of H3BO3 to boron oxide

(B2O3); (ii) reduction of B2O3 to elemental boron by carbon monoxide (CO); and (iii) reaction of elemental boron and carbon to form boron carbide[5]. Electric-arc synthesis yields large solid bodies of boron carbide which suffer from poor stoichiometric control and residual carbon impurities[1]. The solid product also needs to be crushed to a powder of the desired particle size resulting in the introduction of further impurities[4]. Commercial interest in the production of a fine-grained, high purity boron carbide powder has led to significant effort in identifying alternative synthesis methods. An alternative synthesis technique uses a solution-based approach to prepare precursors to boron carbide. This process enhances homogeneity of the reactants through complete dissolution of the base compounds[6] [7]. In general, a solution-based approach yields a powdered boron

97 carbide product (1–10 μm average particle size) at lowered calcination

temperatures (1250–1350 °C) with reduced residual carbon impurity[8] [9] [10].

Despite these advances, literature describing solution-based synthesis of boron

carbide commonly misidentifies the crystalline H3BO3 component in mixed

carbon/boron precursor powders as B2O3[3] [6] [8] [9] [11] [12] [13] [14] [15] [16]

[17]. We show here that the chemical form of the boron constituent can

significantly affect the distribution of the reactants in precursor powders. Correct

identification and manipulation of the boric phase in precursor powders through

process control is therefore important for optimisation of boron carbide yield and

purity. Recently, Cheng et al. have briefly described the conversion of B2O3 to

H3BO3 through moisture adsorption in precursors to boron carbide[18]. However,

the consequences of this effect on calcined boron carbide powder have yet to be

reported. Here we examine the effect of water adsorption in prepared precursor

powders with respect to subsequent boron carbide yield, phase purity and

morphology. Our investigation covers a wide range of humidity exposure times

from a few minutes to several hours and days at multiple processing stages.

Shorter exposure times are potentially more representative of standard

processing conditions, while longer timeframes are relevant to the storage of

precursor mixtures before final calcination. This range of exposure times serves to

maximize the extent of hydration to better understand overall effects.

5.5 EXPERIMENTAL METHODS

Boron carbide powders are prepared by a four stage process: (i) dissolution of

H3BO3 and sucrose in DI water to create a sucrose/H3BO3 gel; (ii) an evaporation stage under vacuum to remove residual solvent from the gel to form a powder;

(iii) a pre-treatment stage at 550 °C under argon (Ar) gas flow; and (iv) calcination of pre-treated powders under Ar flow at 1400 °C for 4 hours to form the final boron carbide product. Calcination at 1400 °C for 4 hours is used to ensure all reactions are complete. This condition enables direct comparison between fully calcined samples for a consistent measure of yield and purity, leaving precursor water adsorption as the only processing variable in these experiments.

5.5.1 PRECURSOR PROCESSING

H3BO3 (99.5%) and sucrose (≥99%) from Sigma-Aldrich were used as received. In a typical gel preparation, molar quantities required by the carbothermal reaction of sucrose and H3BO3 (0.145:1 mol respectively) are completely dissolved in deionized (DI) water at 80 °C. The solution is then held at this temperature while retaining any evaporate and stirred until the solution changes from colourless to dark orange. The solvent is then allowed to evaporate with continued stirring until a viscous sucrose/H3BO3 (SB) gel forms.

Precursor SB powders are formed by evaporation of residual water from the SB gel under vacuum at 150 °C in a sealed vessel. After 30 min, the vessel is transferred into a glovebox and opened under inert conditions. A dry SB product is collected and ground to a powder via mortar and pestle. Prior to pre-treatment

99 at 550 °C, SB powders are exposed to atmospheric conditions for variable times

(in this work, ‘atmospheric conditions’ refer to conditions at 40% relative

humidity). Sample weight gain is denoted by ‘X′, where X=the percentage weight

(wt%) gain of the dry powder after controlled exposure. The SB-X powder is then

pre-treated at 550 °C for 1 hour under 6 L/m Ar flow. Pretreated SB-X (P-SB-X)

product is transferred under inert conditions to a glovebox for storage where it is

again ground to a powder via mortar and pestle. P-SB-X powders are then exposed

to atmospheric conditions for variable times before calcination. Sample weight

gain is denoted by ‘Y′, where Y=the wt% gain of the dry powder after controlled

exposure. P-SB-X-Y powders are then placed in a graphite crucible and calcined at

1400 °C with a ramp rate of 25 °C/min for 4 hours under 3 L/m Ar flow to produce

the boron carbide product (BC-X-Y). A summary of the complete process is shown

in Fig. 5.1 for reference.

100

Fig. 5.1. Process diagram for boron carbide synthesis.

5.5.2 CHARACTERISATION

Powders are analysed using X-ray diffraction (XRD), scanning electron microscopy

(SEM), thermogravimetric analysis (TGA), differential scanning calorimetry (DSC) and dynamic light scattering (DLS). XRD measurements are collected using a

PANalytical X′Pert PRO diffractometer with Co Kα1 radiation. SEM images are obtained using a JEOL JSM-7001 field emission SEM. TGA/DSC data are collected with a Netzsch STA 449 F3 Jupiter and DLS measurements are obtained with a

Malvern Mastersizer 3000.

101 XRD of precursor powders that do not require inert conditions are scanned in

Bragg-Brentano geometry with a 15 mm mask, 2 degree divergence slit and 0.04

degree Soller slits on the incident beam arm, and a 5 mm anti-scatter slit with 0.04

degree Soller slits on the diffracted beam arm. XRD of humidity-sensitive dry

powders is accomplished with a gas-tight sample stage. X-rays probe the sample

through a Mylar film window in the sample chamber wall. To ensure the powders

remain dry, they are loaded onto the stage under inert conditions in a glove box

and the sample chamber is sealed. The stage is then mounted onto the

diffractometer and the sample scanned under the conditions as mentioned above,

but with a 10 mm mask to ensure the beam does not fall outside the Mylar

window. Phase quantification of calcined powder is achieved by Rietveld

refinement using a diamond internal standard (25 wt%).

SEM of dry powders is achieved by mounting the powder on a conductive adhesive

under inert conditions. The mounted dry powder is then transferred to the

microscope under inert conditions and quickly loaded. The samples experience

approximately 5 s exposure to the atmosphere during loading. SEM of hydrated

powders is achieved by removing the mounted dry sample from the microscope

after examination and exposing it to atmospheric conditions for the desired length

of time. The sample is then placed back under inert conditions in the microscope

for re-examination. This approach allows the same position on the sample to be

analysed for direct comparison after hydration has occurred.

The measurement of water adsorption rates in SB and P-SB-X powders is achieved

by transferring a known weight of dry precursor powder under inert conditions

into a custom fabricated aluminum weigh boat enclosed by an air tight aluminum

102

seal. The weigh boat is then placed on a 0.001 mg sensitive balance (0.0025 mg standard deviation) under atmospheric conditions. The seal is then broken and the weight increase due to water adsorption from the atmosphere is recorded.

103 5.6 RESULTS AND DISCUSSION

Exposure of precursor powders to humid conditions is found to have a significant

effect on the formation of boron carbide. Specifically, the impurity content, yield

and morphology of synthesised boron carbide are all directly affected by water

adsorption in precursor powders post solvent removal. These factors can be

understood, and by extension manipulated, by careful examination and

identification of the boron component under the different processing conditions

implemented during synthesis. Table 5.1 summarises the processing pathways

evaluated and the sample set utilised to demonstrate the effects of precursor

powder hydration on boron carbide formation.

Table 5.1

Summary of processing pathways with precursor phase content and subsequent boron carbide

phase weight % in calcined powders. The suffix ‘Full’ refers to a sample that has been hydration to

saturation point.

1400 °C 4 hours 3 L/m 150 °C 30 min vacuum 550 °C 1 hour 6 L/m Ar Ar

Hydration Sample Boric phase Hydration Sample Boric phase Sample Yield (%) (wt %) ID content (wt %) ID content ID - Full P-SB-0- - BC-0-Full 74 Full H3BO3

B2O3

50 P-SB-0- (BOH)3O3 BC-0-50 75 B2O3 H3BO3 0 SB-0 (BOH)3O3 50 H3BO3 B2O3 30 P-SB-0- (BOH)3O3 BC-0-30 77 30 H3BO3

B2O3 0 P-SB-0- - BC-0-0 89 0 -

B2O3 B2O3 15 SB-15 (BOH)3O3 0 P-SB- - BC-15-0 90 H3BO3 15-0 -

B2O3 B2O3 30 SB-30 (BOH)3O3 0 P-SB- - BC-30-0 90 H3BO3 30-0 -

104

The nomenclature in Table 5.1 is used for sample referral, with the suffix ‘Full’ referring to a sample that has been hydrated to the saturation point, where no further weight gain is observed upon exposure to atmospheric conditions.

5.6.1 PRECURSOR HYDRATION

Powders formed by residual water removal from SB gels as well as powders post pre-treatment are hygroscopic. However, despite the common hygroscopic nature of both SB and P-SB-X powders, the water adsorption characteristics of these powders are different. Typical measured water adsorption rates of dry SB and dry

P-SB-X powder are shown in Fig. 5.2.

Fig. 5.2. Water adsorption rates from powders after the 150 °C vacuum (SB) and 550 °C 6 L/m argon flow (P-SB-0) processing stages.

Powder at the SB and P-SB-X processing stages also exhibit significant changes in phase content and morphology upon exposure to atmospheric conditions.

Obvious changes are visually apparent; for example dry SB powders become noticeably wet upon atmospheric exposure while P-SB-X powder grains bind

105 together forming a rigid mass. P-SB-X powders gain over half their original dry

weight within the first 8 hours of exposure to atmospheric conditions, and after

approximately 36 hours very little further weight gain is measured (Fig. 5.2). On

the other hand, SB powders adsorb water at a much slower, nearly linear rate.

Adsorption in SB powders is still apparent after 2 weeks exposure to atmospheric

conditions.

XRD traces of dry SB powder (SB-0) and SB powder exposed to atmospheric

conditions for 3 days (SB-15) are shown in Fig. 5.3.

Fig. 5.3. XRD patterns of dry SB powder (a=SB-0) and SB powder exposed to atmospheric conditions

for 3 days (b=SB-15). The solid lines mark the peak positions for orthorhombic (BOH)3O3, and the

dashed lines mark the positions for triclinic H3BO3.

106

XRD traces of dry SB powder (SB-0) and SB powder exposed to atmospheric conditions for 3 days (SB-15) are shown in Fig. 5.3. SB-0 powder contains three phases; H3BO3, metaboric acid ((BOH)3O3=3HBO2) and an amorphous phase (Fig.

5.3a). In comparison, XRD analysis of SB-15 powder (Fig. 5.3b) reveals a decrease in (BOH)3O3 and amorphous content and an increase in the H3BO3 content. The presence of crystalline sucrose is not detected in SB powders. As exposure to atmospheric conditions increases, the amorphous content gradually decreases while the H3BO3 content increases. The (BOH)3O3 phase also decreases with increased exposure time but does not completely disappear. All diffraction peaks are assigned to H3BO3[19] [20] and (BOH)3O3[21] [22] except for a single unidentified peak (marked by ‘*’) at 18.9 °2θ. This peak is not present in dry SB powders, but gradually forms with increasing exposure time. (BOH)3O3 is known to have multiple crystal structures depending on the dehydration conditions used[23]. However, this unassigned peak does not match either the monoclinic or the cubic form of metaboric acid[24] [25]. Because the unknown peak forms as a result of hydration, we suggest that a H3BO3 analogue may be present that deviates from the typical triclinic crystal structure. As mentioned previously P-SB-

X powders are also hygroscopic, however, dry P-SB-X powders are almost amorphous as shown in Fig. 5.4.

107

Fig. 5.4. XRD pattern of dry P-SB-X powder. No crystalline H3BO3 or (BOH)3O3 phases are visible

after pre-treatment at 550 °C.

Crystalline H3BO3 or (BOH)3O3 phases are not detectable in the XRD pattern of

these samples. Notably, dry P-SB-X powder contains very broad peaks at high

angle (approximately 74 and 85 °2θ) that cannot be identified. B2O3 displays low

to medium-range order[26] [27] which may account for these peaks. Alternatively,

the peaks may be due to a nanocrystalline boron oxide analogue resulting from

dehydration. These XRD peaks disappear as the dry powder adsorbs water and

thus, suggests that they are related to the dry boron component.

Similar to dry SB powders, dry P-SB-X powders experience a loss in the amorphous

signature centered at approximately 25.5 °2θ and a proportionate increase in the

crystalline phase with increased atmospheric exposure time (Fig. 5.5).

108

Fig. 5.5. XRD patterns from ten sequential measurements over time demonstrate loss of amorphous phase and increase in crystalline H3BO3 phase after exposure of P-SB-X powder to atmospheric conditions. Within 10 min of exposure crystalline H3BO3 peaks appear within the originally amorphous pattern (Fig. 5.4). The crystalline H3BO3 peaks gradually increase in intensity as the amorphous signal drops until after 3600 min, where no amorphous signal is detectable.

By taking an XRD trace of P-SB-X powder every 10 min after initial exposure to atmospheric conditions for 60 hours, formation of the crystalline phase and loss of the amorphous phase is clearly observed. For clarity, Fig. 5.5 shows only ten of these XRD patterns collected over 60 hours. After 60 hours of atmospheric exposure, the amorphous phase is completely converted to H3BO3.

As seen in Fig. 5.2, SB powder absorbs much less water and does so at a slower rate compared to P-SB-X powders. This difference in adsorption can be attributed to the processing temperatures used. H3BO3 is known to have multiple stages of dehydration when subjected to heat, the first stage to (BOH)3O3 occurring at temperatures between 80 and 130 °C[28]. The second dehydration stage occurs

109 at temperatures above 130 °C and results in dehydration to B2O3[23]. The

combined balanced equation for these two dehydration stages is shown in Eq. 5.2.

6H3BO3 →(80–130°C) 2(BOH)3O3 + 6H2O →( > 130°C) 3B2O3 + 9H2O [Eq. 5.2]

As B2O3 is known to be amorphous and highly hygroscopic,[26] [29] this

mechanism accurately accounts for the water adsorption observed in dry

precursor powders post heat treatment at either 150 °C (SB) or 550 °C (P-SB-X).

According to Eq. 5.2, heating of precursor powders beyond 80 °C results in the loss

of water and conversion to (BOH)3O3 followed by further dehydration to B2O3 at >

130 °C. Dry SB powder, with exposure to 150 °C for 30 min, contains less converted

amorphous B2O3 with some residual (BOH)3O3 and H3BO3. Comparatively dry PSB-

X powder experiences a higher temperature for a longer time and hence contains

only amorphous B2O3. This interpretation is in agreement with XRD analysis of dry

SB and P-SB-X powders (Figs. 5.3 and 5.4 respectively). Therefore, P-SB-X powder

experiences a greater wt% gain compared to dry SB powder during atmospheric

exposure due to the increased B2O3 content as shown in Fig. 5.2.

As the amorphous content in both SB and P-SB-X powders completely disappears

with hydration, and is accompanied by increasing amounts of crystalline

H3BO3/(BOH)3O3, we propose that the amorphous signature observed in XRD

traces of precursor powders emanates only from the B2O3 phase. Therefore, it is

likely that B2O3 does not produce any crystalline signal observed in XRD analysis of

precursor powders. This is further supported by the fact that B2O3 is typically

difficult to crystallise and does not exhibit long-range structural order at ambient

pressure[30] [31] [32].

110

5.6.2 PRECURSOR MORPHOLOGY

Water adsorption in P-SB-X powder causes dramatic morphological change. SEM images of P-SB-X powder after 0, 10 and 60 mins exposure to atmospheric conditions are shown in Fig. 5.6.

Fig. 5.6. SEM images of P-SB-X powder when dry and after 10 and 60 min exposure to atmospheric conditions. All images are taken from the same sample and each image set focuses on the same position in the sample. Insets in Figs. 5.6e and 5.6f are higher magnification of the identified areas

(scale bar =1 μm).

Because XRD of dry P-SB-X powder shows significant generation of H3BO3 within

10 min of exposure to atmospheric conditions (Fig. 5.5) this time period was chosen for the initial exposure. Each row of images, i.e. Fig. 5.6a-c, d-f and g-i,

111 highlight specific features at the same position in the sample during the hydration

process.

SEM of P-SB-0 powder in Fig. 5.6d shows a large dense carbon/B2O3 particle with

a rough fracture surface resultant from the grinding process. These particles make

up the bulk of P-SB-0 powders and range in size from ~10–50 μm. B2O3 spheres of

approximately 1–5 μm in size occasionally adhere to or embed in the surface of

these much larger particles (Fig. 5.6g). These spheres likely result from localised

inhomogeneities in otherwise homogeneously distributed precursor constituents.

SEM analysis after 1 min of exposure to atmospheric conditions does not yield any

significant morphological change. After 10 and 60 min exposure to atmospheric

conditions, very drastic changes in the powder are observed. Cracks/fissures

visible on the surface of the smooth particles (highlighted by white arrows in Fig.

5.6a and b) are forced open by the generation of very fine ‘strings’ (beginning in

Fig. 5.6b after only 10 min exposure and becoming very pronounced after 60 min

in Fig. 5.6c). The same effect is observed at fracture surfaces of the carbon/B2O3

particles (Fig. 5.6d-f). If a crack or some form of breach is not present in the surface

of a particle these strings are not observed, suggesting that in P-SB-0 powders the

B2O3 phase is well encapsulated and distributed within the dense carbon/B2O3

particles. An exception to this observation is the spheres of B2O3 found on the

surface of these particles (Fig. 5.6g). The spheres break down into similar string-

like structures and branch out across the surface of the particles (Fig. 5.6h and i).

A high magnification SEM image of the string-like features that appear in hydrated

P-SB-X powder is shown in Fig. 5.7.

112

Fig. 5.7. High magnification SEM image of the (BOH)3O3/H3BO3 string-like features that appear after hydration of P-SB-0 powders.

This confirms that the hydration effects on phase morphology and distribution become noticeable within timeframes that are comparable to typical processing times. Such timeframes may be expected to become shorter in less controlled atmospheres with higher humidity.

The generation of this string-like morphology is consistent with the mechanism defined in Eq. 5.2. Metaboric acid is known to polymerise via repeating (BOH)3O3 monomer units and extensive hydrogen bonding[24] [33]. As the sample absorbs water, the reverse of Eq. 5.2 occurs through hydration and B2O3 converts to polymeric (BOH)3O3 according to Eq. 5.3. The metaboric acid polymeric strings then absorb further water to form H3BO3.

B2O3 + H2O → (BOH)3O3 + H2O → [(BOH)3O3]n + H2O → H3BO3 [Eq. 5.3]

The behavior of the boron component in P-SB-X-Y powders upon heating under inert conditions is analysed using TGA and DSC under Ar flow and is consistent with both XRD and SEM observations. TGA and DSC data for P-SB-0-0 and completely hydrated P-SB-0-Full powder is shown in Fig. 5.8.

113

Fig. 5.8. DSC and TGA data for P-SB-0-0 and P-SB-0-Full powders. The dashed and solid lines

represent TGA and DSC, respectively.

P-SB-0-0 powder, which experiences no exposure to atmospheric conditions

throughout the entirety of processing, shows no sign of dehydration and

experiences minimal weight loss. This is due to the retention of the B2O3 phase

resulting from inert processing conditions. In comparison, P-SB-0-Full powder

which is hydrated to saturation, exhibits multiple endothermic processes and a

large loss in weight in the TGA analysis due to the dehydration of the H3BO3 and

(BOH)3O3 phases generated from exposure to atmospheric conditions. The

formation of these phases causes segregation as observed in the SEM images of

Fig. 5.6 leading to a loss of contact area between reactants.

114

5.6.3 BORON CARBIDE PRODUCT

The separation of the carbon and boron reactants due to hydration has a very significant effect on boron carbide yield and residual carbon contents in calcined powders. XRD analyses of boron carbide calcined from dry (BC-0-0), partially hydrated (BC-0-30 and BC-0-50) and fully hydrated (BC-0-Full) precursors show an increase in the residual carbon content with increasing hydration (Fig. 5.9).

Fig. 5.9. XRD patterns of boron carbide product calcined from pre-treated powders at different hydration stages. The residual carbon content increases as the degree of hydration increases (see insert).

A similar trend is also observed for boron carbide phase weight % such that as hydration increases, boron carbide yield decreases (Table 5.1). The increase in residual carbon and drop of boron carbide yield in powders calcined from hydrated precursors correlates well with the observations from SEM and XRD analysis. The formation of (BOH)3O3 and H3BO3 results in an overall decrease in the contact surface area of the boron and carbon components. As more water is absorbed, additional B2O3 is converted and further homogeneity between

115 precursors is lost. The minimisation of residual carbon and maximisation of boron

carbide yield is therefore optimised in precursor powders kept dry post

pretreatment. Hydration of SB powders after the 150 °C vacuum stage (SB-15, SB-

30) does not have any significant effect on the yield of boron carbide powders that

are then kept dry post pre-treatment before calcination (BC-15-0, BC-30-0). As

dehydration to B2O3 has only partially occurred at this stage due to the low

processing temperature and time implemented (150 °C, 30 mins), hydration back

to H3BO3 does not significantly affect reactant dispersion and hence boron carbide

yield is unchanged provided inert handling is employed post pre-treatment. This

finding indicates that inert handling of precursor powders is more critical to

maximizing boron carbide yield only after the 550 °C pre-treatment stage.

The effects of hydration are also evident in the morphologies of boron carbide

calcined from P-SB-0-0 (BC-0-0) and P-SB-0-Full (BC-0- Full) powder. Again, this can

be related to the loss of precursor homogeneity due to exposure to atmospheric

conditions. BC-0-Full (Fig. 5.10d) is calcined from fully hydrated pre-treated

powder, and consequently exhibits large rod-like particles and an overall larger

particle size due to the loss of fine homogeneity in the reactant dispersion (see

process pathway (ii) in the Fig. 5.10 schematic).

116

Fig. 5.10. SEM images of boron carbide powders formed from precursors with varying degrees of hydration with accompanying process schematic describing the correlating morphologies of precursor and product. The numeral in the bottom right corner of each SEM image labels the corresponding process pathway in the schematic. Low and high magnification SEM images of BC-

0-0 (a, b and c) are shown and highlight the uniform and highly faceted boron carbide particles resulting from inert processing. BC-0-Full powder calcined from fully hydrated precursor powder is shown in d for comparison. Boron carbide calcined from precursor powder hydrated directly after solvent removal (BC-30-0) is shown in e. Process pathway (iv) in the schematic describes a comparative growth mechanism for boron carbide synthesised by Cheng et al.[18].

However, boron carbide calcined from precursor powder that is kept under inert conditions throughout the entire process (BC-0-0) exhibits a very uniform and highly faceted euhedral particle size and shape as seen in Fig. 5.10a, b and c (see process pathway (i) in the Fig. 5.10 schematic). This is also reflected in DLS measurements taken of both BC-0-0 and BC-0-Full powders (Fig. 5.11).

117

Fig. 5.11. DLS particle sizing of BC-0-0 and BC-0-Full powder.

A smaller overall particle size is observed for boron carbide calcined from dry

precursor powder (BC- 0-0). DLS sizing also shows two distinct particle sizes in

produced boron carbide powders; approximately 1 μm (primary) and 5 μm

(secondary) for BC-0-0 powder.

This dual-uniformity in particle size is most likely due to the slight homogeneity

loss ascribed to the B2O3 spheres scattered throughout the precursor. The primary

particles as seen in BC-0-0 powder are shown in Fig. 5.10a, with the secondary

particle size shown embedded in a primary particle aggregate for comparison in

Fig. 5.10b. Interestingly, although boron carbide yield is not affected by hydration

post solvent removal at the SB stage as mentioned previously (BC-15-0, BC-30-0),

the boron carbide particle size is significantly changed as a result of hydration at

this stage. Fig. 5.10e shows an SEM image of BC-30-0 powder which contains a

much larger overall particle size (5–20 μm) when compared to BC-0-0 powder (Fig.

5.10b for comparison at the same magnification). This effect is most likely due to

small growth in the H3BO3/(BOH)3O3 regions within the condensed carbon matrix

118

due to hydration of partially dehydrated boric phases at the SB stage (see process pathway (iii) in the Fig. 5.10 schematic).

These observations are contrary to the findings of Cheng et al.[18] regarding boron carbide particle growth and nucleation using identical precursors. However, the findings of Cheng et al. and the results presented here can both be reconciled provided the effect of reactant dispersion on calcined boron carbide morphology is taken into account. In addition, the effects of calcination temperature proposed by Cheng et al. are also important[18]. Cheng et al. conclude on the basis of nucleation and particle growth mechanisms[18], that a uniform and fine boron carbide particle distribution is not possible from a low temperature calcination

(1250–1450 °C). However, our results clearly demonstrate that a uniform and fine boron carbide particle size is possible within this temperature range, provided that intimate reactant dispersion is maintained throughout processing.

Two key processing conditions contribute to a lack of uniform and fine boron carbide particle size at 1450 °C as noted in the work of Cheng et al[18]. Firstly, initial solvent removal to form the mixed precursor powder is performed in air (no vacuum), and secondly, precursor powders are not isolated from atmospheric moisture after pre-treatment. Boron carbide in Fig. 5.7b of Cheng et al.[18] displays identical morphological characteristics to boron carbide synthesised under similar conditions in this work (BC-0-Full) by exposure to atmospheric conditions before calcination (see Fig. 5.10d). This implies reactant segregation in the precursor powder of Cheng et al. due to hydration and/or initial inhomogeneous precursor distribution. Poor reactant dispersion in precursors is further evident in the precursor hydration results presented[18]. For example, dry

119 precursors prepared by Cheng et al. gain weight by only 11% over a period of 4

days while precursors prepared in this work gain 60% of their weight in 8 hours.

This discrepancy can be attributed to large B2O3 agglomerates resulting from the

poor reactant dispersion. As these large agglomerates convert to H3BO3 from

moisture present in the air, a capping layer is formed preventing further

conversion of the B2O3 and resulting in a much smaller weight gain (see process

pathway (iv) in the Fig. 5.10 schematic).

This mechanism is also evident in the XRD pattern presented by Cheng et al. for a

hydrated precursor powder[18]. After 15 days, this powder retains an amorphous

B2O3 signal despite undergoing complete hydration. In comparison, for hydrated

precursors synthesised in this work, the amorphous B2O3 signature is completely

converted to H3BO3 within 2 days. Conversion proceeds to completion in

precursors due to a smaller B2O3 particle size that allows complete conversion to

H3BO3 (see Fig. 5.10g-i). As noted by Cheng et al.[18], higher calcination

temperatures (>1700 °C) and rapid heating rates will decrease boron carbide

particle sizes calcined from mixed precursors.

From these results it is clear that the key factor influencing boron carbide

morphology, yield and purity is the distribution of the reactants in precursor

powders. Post solvent removal, homogeneity is easily lost due to the hygroscopic

nature and low melting point of B2O3. It is therefore critical to maximize the

contact surface area between boron and carbon constituents during solvent

removal, and furthermore, to maintain this intimate distribution for the entirety

of processing.

120

5.7 CONCLUSION

The distinction between H3BO3 and B2O3 in precursor powders is critical to the precise control of boron carbide synthesis. Recognition of this fundamental precept is important for a clear understanding of reaction kinetics, morphology and the reduction of residual carbon in calcined powders. Retention of the amorphous boron oxide phase is desirable in precursor powders post pre- treatment to avoid the loss of contact area between boron and carbon due to atmospheric water adsorption. This segregation of reactants results in calcined powders that contain increased residual carbon with poor boron carbide yield and variable morphology. By minimising the conversion of B2O3 to H3BO3 in precursor powders through inert handling, a euhedral and uniform boron carbide particle size is achieved with increased yield and minimal residual carbon. This approach increases boron carbide yield by 15% compared to boron carbide calcined from fully hydrated precursor powder. As the morphology of boron carbide powder is closely related to the reactant distribution in mixed precursors, this provides an avenue for tailoring boron carbide morphology through hydrated phase manipulation.

121 5.8 ACKNOWLEDGEMENTS

The author would like to gratefully acknowledge the Institute for Future

Environments for research support as well as Tony Raftery, Henry Spratt,

Mahboobeh Shahbazi, Elizabeth Graham and the Central Analytical Research

Facility for assistance with characterisation.

122

5.9 REFERENCES

1. Thevenot, F., Boron Carbide - A Comprehensive Review. Journal of the European Ceramic Society, 1990. 6: p. 205-225. 2. Domnich, V., Reynaud, S., Haber, R. & Chhowalla, M., Boron Carbide: Structure, Properties, and Stability under Stress. Journal of the American Chemical Society, 2011. 94: p. 3605-3628. 3. igdeloo, J. & Hadian, A., Synthesis of High Purity Micron Size Boron Carbide Powder from B2O3/C Precursor. International Journal of Recent Trends in Engineering, 2009. 1: p. 176-180. 4. Suri, A., Subramanian, C., Sonber, J. & Murthy, T., Synthesis and consolidation of boron carbide: a review. International Materials Reviews, 2012. 55: p. 4-40. 5. Alizadeh, A., Taheri-Nassaj, E. & Ehsani, N., Synthesis of boron carbide powder by a carbothermic reduction method. Journal of the European Ceramic Society, 2004. 24: p. 3227-3234. 6. Sinha, A., Mahata, T. and& Sharma, B., Carbothermal route for preparation of boron carbide powder from boric acid–citric acid gel precursor. Journal of Nuclear Materials, 2002. 301: p. 165-169. 7. Yanase, I., Ogawara, R. & Kobayashi, H., Synthesis of boron carbide powder from polyvinyl borate precursor. Materials Letters, 2009. 63: p. 91-93. 8. Kakiage, M., Tominaga, Y., Yanase, I. & Kobayashi, H., Synthesis of boron carbide powder in relation to composition and structural homogeneity of precursor using condensed boric acid–polyol product. Powder Technology, 2012. 221: p. 257-263. 9. Rafi-ud-din, Zahid, G. H., Asghar, Z., Maqbool, M., Ahmad, E., Azhar, T., Subhani, T. & Shahzad, M., Ethylene Glycol Assisted Low-temperature Synthesis of Boron Carbide Powder from Borate Citrate Precursors Journal of Asian Ceramic Societies 2014. 2: p. 268-274. 10. Chen, X. W., Dong, S. M., Kan, Y. M., Zhou, H. J., Hu, J. B. & Ding, Y. S, Effect of Glycerine Addition on the Synthesis of Boron Carbide from Condensed Boric Acid- polyvinyl Alcohol Precursor. RSC Advances, 2016. 6: p. 9338-9343. 11. Kakiage, M., Tahara, N., Yanase, I. and Kobayashi, H., Low-temperature synthesis of boron carbide powder from condensed boric acid–glycerin product. Materials Letters, 2011. 65: p. 1839-1841. 12. Tahara, N., Kakiage, M., Yanase, I. & Kobayashi, H., Effect of addition of tartaric acid on synthesis of boron carbide powder from condensed boric acid-glycerin product. Journal of Alloys and Compounds, 2013. 573: p. 58-64 13. Kakiage, M., Tahara, N., Yanagidani, S., Yanase, I. & Kobayashi, H., Effect of boron oxide/carbon arrangement of precursor derived from condensed polymer-boric acid product on low-temperature synthesis of boron carbide powder. Journal of the Ceramic Society of Japan, 2011. 119: p. 422-425. 14. Maqbool, M., Rafi-ud-din, Zahid, G. H., Ahmad, E., Asghar, Z., Subhani, T., Shahzad, M. & Zaleem, I., Effect of saccharides as carbon source on the synthesis and morphology of B4C fine particles from carbothermal synthesis precursors. Materials Express, 2015. 5: p. 390-400. 15. Pilladi, T. R., Ananthasivan, K. & Anthonysamy, S., Synthesis of boron carbide from boric oxide-sucrose gel precursor. Powder Technology, 2012. 246: p. 247-251. 16. Pilladi, T. R., Ananthasivan, K., Anthonysamy, S. & Ganesan, V., Synthesis of nanocrystalline boron carbide from boric acid–sucrose gel precursor. Jounal of Materials Science, 2012. 47: p. 1710-1718.

123 17. Hadian, A. & Bigdeloo, J., The Effect of Time, Temperature and Composition on Boron Carbide Synthesis by Sol-gel Method. Journal of Materials Engineering and Performance, 2008. 17: p. 44-49. 18. Foroughi, P. & Cheng, Z., Understanding the morphological variation in the formation of B4C via carbothermal reduction reaction. Ceramics International, 2016. 42: p. 15189-15198. 19. Kuznetsov, A. Y., Pereira, A. S., Shiryaev, A. A., Haines, J., Dubrovinsky, L., Dmitriev, V., Pattison, P. & Guignot, N., Pressure-Induced Chemical Decomposition and Structural Changes of Boric Acid. Journal of Physical Chemistry B, 2006. 110: p. 13858-13865. 20. Zachariasen, W. H., The Precise Structure of Orthoboric Acid. Acta Crystallographica, 1954. 7: p. 305-310. 21. Peters, C. R. & Milberg, M. E., The Refined Srtucture of Orthorhombic Metaboric Acid. Acta Crystallographica, 1963. 1`7: p. 229-234. 22. Tazaki, H., The Structure of Orthorhombic Metaboric Acid, HBO2. Journal of Science of the Hiroshima University, Series A: Mathematics, Physics, Chemistry 1940. 10: p.55-61. 23. Kocakusak, S., Akcay, K., Ayok, T., Koroglu, H. J., Koral, M., Savasci O. T. & Tolun, R., Production of Anhydrous, Crystalline Boron Oxide in Fluidized Bed Reactor. Chemical Engineering and Processing, 1996. 35: p. 311-317. 24. Zachariasen, W. H., The Crystal Structure of Monoclinic Metaboric Acid. Acta Crystallographica, 1962. 16: p. 385-389. 25. Zachariasen, W. H., The Crystal Structure of Cubic Metaboric Acid. Acta Crystallographica, 1963. 16: p. 380-384. 26. Perez-Enciso, E., Ramos, M. & Vieira, S., Low-temperature specific heat of different B2O3 glasses. Physical Review B, 1997. 56: p. 32-35. 27. Soper, A. K., Boroxol Rings from Diffractoin Data on Vitreous Boron Trioxide. Journal of Physics: Condensed Matter, 2011. 23: p. 365402. 28. Balci, S., Sezgi, N. A. & Eren, E., Boron Oxide Production Kinetics Using Boric Acid as Raw Material. Industrial and Engineering Chimstry Research, 2012. 51: p. 11091-11096. 29. Broadhead, P. & Newman, G. A., The Vibrational Spectra of Orthoboric Acid and its Thermal Decomposition Products. Journal of Molecular Structure, 1971. 10: p. 157. 30. Aziz, M., Nygren, E., Hays, J. F. & Turnbull, D., Crystal growth kinetics of boron oxide under pressure. Journal of Applied Physics, 1985. 57: p. 2233-2242. 31. Gurr, G. E., Montgomery, P. W., Knutson, C. D. & Gorres, B. T., The Crystal Structure of Trigonal Diboron Trioxide. Acta Crystallographica B, 1969. 26: p. 906- 915. 32. Youngman, R. E., Haubrich, S.T., Zwanziger, J. W., Janicke, M. T. & Chmelka, B. F., Short- and Intermediate-Range Structural Ordering in Glassy Boron Oxide. Science, 1995. 269: p. 1416-1420. 33. Freyhardt, C. C., Wiebcke, M. & Felshe, J., The Monoclinic and Cubic Phases of Metaboric Acid (Precise Redeterminations). Acta Crystallographica C, 1999. 56: p. 276-278.

124

CHAPTER 6

X-RAY DIFFRACTION FOR LOW MASS

MATERIALS

125 6.1 CHAPTER FOREWORD

The development of an enhanced X-ray characterisation technique for boron

carbide powder analysis was driven by multiple factors: (i) the absence of

structural information in peer-reviewed literature pertaining to solution-based

boron carbide synthesis; (ii) the lack in structural resolution attributed to difficulty

in differentiation between boron and carbon due to similar X-ray scattering cross

sections; and (iii) the inability of standard XRD techniques to accurately and

reproducibly determine the phase wt% of boron carbide in calcined powders.

Attempts to accurately and reproducibly measure the phase wt% in synthesised

boron carbide powder were initially unsuccessful due to the high X-ray

transparency of the material as well as the small sample volumes generated.

Variations in boron carbide X-ray reflection intensities across aged precursor

sample sets indicated that water adsorption had a direct effect on boron carbide

yield. Because of these variations, the ability to accurately measure phase wt%

was deemed key to quantifying and understanding the effects of water adsorption

on boron carbide powders.

A key objective of this research was to elicit stoichiometric control in the synthesis

of boron carbide powders. Without accurate and reproducible measurement of

structural parameters, precise determination of stoichiometry would not be

possible. Solution-based techniques are also likely to enhance dispersion of

reactants in precursor powders. Hence, measurement of detectable variations in

boron carbide stoichiometry from solution-based samples was important for

comparison with commercial production methods. A high throughput analytical

126

tool with the required resolution to accurately investigate structural details of low mass absorption coefficient (MAC) materials was of paramount importance. A number of alternative analytical techniques were investigated for this purpose including electron microprobe analysis (EMPA) and chemical dissolution followed by inductively coupled plasma optical emission spectroscopy (ICP-OES) analysis.

However, these methods were considered unsuitable due to technical issues. For example, EPMA required provision of suitable standards with reliable atomic concentrations in the material which could not be obtained. Furthermore, boron carbide is difficult to completely dissolve for solution-based analysis using standard techniques. The failings of standard X-ray diffraction methodologies in the analysis of low volume, low MAC materials were then identified for subsequent optimisation at the hardware level. These improvements provided accurate and reproducible stoichiometric and phase determination in synthesised boron carbide powders. In addition, these improvements allow differentiation of different boron carbide phases within the same sample. From this data, novel inferences were made regarding boron carbide formation and structure, as detailed in this chapter.

127 6.2 ARTICLE 3

STRUCTURAL ANALYSIS AND PHASE QUANTIFICATION OF BORON

CARBIDE POWDERS USING AN OPTIMISED X-RAY DIFFRACTION

TECHNIQUE

Joshua L. Watts1,2, Henry Spratt1,2, Peter C.Talbot1,2, Jose A. Alarco1,2, N.A.

Raftery1,2 and Ian D. R. Mackinnon1

1Institute for Future Environments 2Science and Engineering Faculty, School of Chemistry, Physics and Mechanical Engineering

Queensland University of Technology, Brisbane, QLD Australia 4001

Journal of Applied Crystallography

Manuscript in preparation

To be submitted

128

6.3 ABSTRACT

Structural analysis and phase quantification of solution-synthesised and commercially available boron carbide powders is conducted via Rietveld refinement of X-ray diffraction data collected with a Co source. A fixed incidence parallel beam geometry is applied and compared against the more commonly used

Bragg-Brentano geometry. Fixed incidence parallel beam is found to have a significant advantage in phase quantification through the minimisation of peak aberrations due to the high X-ray transparency of boron carbide. Analysed commercial boron carbide powders are shown to contain varying contents of both

B4C and B~6C phases, while solution-synthesised powders are comprised of B4C only. The refined site occupancy value for carbon substitution on the polar position in the boron icosahedra is also found to correlate well with the structural carbon content based on unit cell measurements.

129 6.4 INTRODUCTION

Boron carbide is of practical interest due to its potential in a wide range of

applications owing to many functional properties. These properties include

extreme hardness, high melting point, low specific weight, chemical inertness,

high neutron capture cross-section as well as semiconducting and p-type

thermoelectric properties. Commercially, boron carbide is most commonly

produced in electric-arc or Acheson type furnaces at high temperatures[1].

Consequently, the product is very coarse and contains impurity in the form of free

carbon[2]. Boron carbide formed in this way can also be structurally carbon

deficient[3], deviating from the desired B4C stoichiometry responsible for

maximum hardness[4]. Solution-based boron carbide synthesis is then of interest

due to the potential to form a fine grained, high purity powder. Substantial recent

research effort has been focused on this area[5] [6] [7] [8] [9].

Despite the significant advances in solution-based synthesis of boron carbide over

the past few decades, detailed reliable information regarding the crystal structure

and stoichiometry of powders produced via this technique is not present in the

literature. This absence is due to difficulties encountered in both chemical and

physical analytical techniques. For example, the differentiation between free and

structural carbon is ambiguous in chemical analytical techniques[1]. Boron carbide

can also exist as a solid-solution with different B/C ratios (~8 to ~20 at% C)[10]

leading to complications in accurate stoichiometric phase identification.

130

Further complications arise due to the similar X-ray scattering cross sections of boron and carbon, which are difficult to differentiate for precise structural analysis. Boron and carbon are also highly transparent to X-rays. Thus, boron carbide has a low mass absorption coefficient (MAC). A low MAC presents practical problems in X-ray diffraction (XRD) analysis of boron carbides because the extended X-ray penetration depth generates severe peak asymmetry in the collected data[11]. The use of a non-focusing geometry such as fixed incidence parallel beam (FIPB) allows for the mitigation of this peak asymmetry as only diffracted X-rays parallel with the collimator reach the detector as shown in Fig.

6.1a. This effectively removes any diffraction signal below the sample surface, providing improved peak symmetry when compared with Bragg-Brentano (BB) geometry[12].

Fig. 6.1. Comparison of focusing FIPB (a) geometry with BB (b) for low MAC samples. The obtained peak shape from both geometries is shown, with the shaded section highlighting the forward peak slope aberration that occurs due to sample transparency.

In this work, the crystallographic properties of commercially available and solution-synthesised boron carbides are investigated using both BB and FIPB geometries. These experiments are carried out for three specific purposes: (i) to

131 ascertain whether any advantage exists in the structural analysis of boron carbides

by minimising/removing aberrations at the hardware level; (ii) to investigate and

compare the structural characteristics of boron carbides synthesised via

commercial and by solution-based techniques using a FIPB geometry; and (iii) to

develop an optimised technique for phase quantification in low-volume boron

carbide powders. The developed technique is not only limited to boron carbide

and may find use in the analysis and quantification of other low MAC powders (e.g.

organics/pharmaceuticals) wherein only small sample volumes are available.

132

6.5 EXPERIMENTAL METHODS

6.5.1 MATERIALS

Boron carbide powders obtained from Goodfellow (product number: B506010),

Sigma (product number: 101637639) and American Elements (product number:

BO-C-02M-P.40UM) were characterised as received. Diamond and cubic boron nitride (cBN) powder (Micron+) were obtained from Element Six. Silicon (SRM

640d) and corundum powders were obtained from the National Institute of

Standards and Technology and Baikowski International respectively. Boric acid

(H3BO3) (99.5%) and sucrose (≥99%) used for solution-synthesis of boron carbide were purchased from Sigma-Aldrich and used as received.

6.5.2 SOLUTION SYNTHESIS OF BORON CARBIDE POWDERS

Boron carbide powders synthesised in-house are prepared by a five-stage process as follows: (i) dissolution of H3BO3 and sucrose in DI water; (ii) slow heating to evaporate the solvent and create a gel; (iii) a high temperature evaporation stage to remove the residual solvent to form the dry sucrose/H3BO3 precursor powder;

(iv) a pre-treatment stage at 550 °C under argon (Ar) flow; and (v) a calcination step at 1400 °C under Ar to form the final boron carbide product. Through control of precursor handling and processing conditions the boron carbide yield is able to be varied[13]. For this study, five such samples are chosen with different processing and handling conditions (summarised in Table 6.1). Detailed synthesis and characterisation of these powders has been reported previously[13].

133 Table. 6.1.

Processing conditions for solution-based boron carbide powders analysed in this study.

Sample Name Processing Conditions

BC-Full Precursor exposure to atmospheric conditions (40% relative humidity) until saturation point

BC-50 Precursor exposure to atmospheric conditions (40% relative humidity) until 50% weight gain

BC-30 Precursor exposure to atmospheric conditions (40% relative humidity) until 30% weight gain

BC-0 Full processing of precursor under inert conditions

High Purity Full processing of precursor under inert conditions with 14% extra H3BO3 added

6.5.3 POWDER X-RAY DIFFRACTION

Collection Methodology

XRD measurements are collected using a PANalytical X’Pert PRO diffractometer

with Co Kα1 radiation (40 kV, 40 mA). Samples for XRD analysis are mounted in a

circular depression 1 mm deep and 18 mm in diameter inlaid into single crystal

quartz disks. This method of mounting is used for all samples in both FIPB and BB

geometries unless otherwise specified. All measurements are conducted from 18

to 120 °2θ. As the optics of the diffractometer are interchanged on a daily basis

due to routine instrument use by other researchers, a line position alignment is

conducted for both BB and FIPB geometries before any sample data is collected.

Alignment involves first measuring a powdered lanthanum hexaboride (LaB6)

standard (NIST 660a). After the initial sample mounting the LaB6 powder is not

removed from, nor disturbed, in the holder over the course of all reported

measurements. The scan data is then modelled by restraining the unit cell size to

the certified value and an instrument description generated. The instrument

description is then used to refine the measured line positions of the (1 1 1), (0 0

134

2), (0 2 1), (2 1 1), (0 2 2), (0 0 3), (0 3 1), (3 1 1) and (2 2 2) reflections of the LaB6 standard. The average difference between the measured and the certified line positions is then calculated and applied to the diffractometer as a sample offset.

The LaB6 standard is then measured with the applied sample offset and the refined line positions compared against the NIST certified line positions to confirm successful alignment. This methodology realises a consistent average difference between measured and NIST certified line position of less than the calculated error in line position measurement for both BB and FIPB geometries.

BB measurements are acquired with a 10 mm mask, 2° anti-scatter slit, 0.5° divergence slit and 0.04 radian Soller slits on the incident beam arm, and a 5 mm anti-scatter slit with 0.04 radian Soller slits and an X’Celerator detector on the diffracted beam arm. FIPB measurements are collected with a 10 mm mask, 1.4 mm anti-scatter slit, 0.5° divergence slit and 0.04 radian Soller slits with an incidence angle (ω) of 9° on the incident beam arm, and a 0.09° parallel plate collimator with 0.04 radian Soller slits and a proportional detector on the diffracted beam arm. All samples for both geometries are spun at 2 revolutions per second during measurement.

Standard preparation

Two high MAC and two low MAC standard materials are chosen so that 1:1 by weight mixes of comparable MAC materials can be prepared and analysed by alternate geometries. Silicon (MAC = 94 cm2 g-1) and corundum (MAC = 47.86 cm2 g-1) are used for the high MAC standards whilst diamond (MAC = 6.8 cm2 g-1) and cBN (MAC = 7.85 cm2 g-1) are used as the low MAC standards. Reported MACs are

135 calculated using X’Pert Highscore Plus (v4.5, PANalytical). Three separate mixtures

of the 1:1 standards are prepared to analyse any variation that might be present

due to sample preparation. One specific mixture from each set is remounted and

rescanned three times to analyse any variation due to sample mounting.

Silicon/corundum and diamond/cBN powder mixtures are prepared by thorough

dispersion via mortar and pestle.

Rietveld Refinement

TOPAS (v5, Bruker) is used for all Rietveld refinement and data modelling. BB

instrument functions used the linear position sensitive detector (lpsd) macros,

along with the full axial divergence macros (axial_conv) and two additional user

defined convolutions. For all data sets the included CoKα7 and CoKβ6 emission

profiles are employed. However, the area and width of the Kβ profile is refined but

set to the ratios of the starting values during instrument function determination

for both BB and FIPB geometries. For FIPB instrument functions, the peak shape is

modelled with numerous user defined convolutions, the equatorial divergence

correction, and the FIPB corrections provided by Rowles et al.[14] and references

therein. For sample data, all of the aforementioned parameters are fixed to the

instrument function values. Thus, the refined terms for sample data includes

specimen zero error, background (same number of terms for every data set),

Gaussian and Lorentzian crystallite size or strain terms as required, unit cell

parameters, structural parameters as required, and a scale factor. Numbers in

parentheses are the estimated standard deviations in the least significant digit of

the refined parameter calculated by TOPAS.

136

Quantification

Quantification of boron carbide samples is achieved through the internal standard method. Diamond (25% by weight) is used as the spiked phase in all quantifications due to its similar MAC to boron carbide which minimises the deleterious effects of microabsorption. After each diffractometer alignment, a new pure diamond structure refinement is carried out for use in subsequent phase quantifications.

The refined thermal parameter from the pure diamond sample is then restrained for refinement as the internal standard. For the solution-based samples, structure solution S1 (see Section 6.6.4) is used as a starting point for quantification.

Rhombohedral (R 3 m, no. #166) symmetry is used for all boron carbide quantification refinements.�

A typical refinement methodology for quantification is as follows: (i) structure and thermal parameters from structure solution S1 (see Table 6.7) imported and restrained and the unit cell, Lorentzian and Gaussian size terms and Lorentzian strain terms refined; (ii) site positions released and refined; (iii) the calculated occupancy from the a parameter of the unit cell added and restrained with refinement of the site positions. The effects of allowing the carbon substitution site position and thermal to refine independently on quantification is also analysed. A typical quantification refinement is as follows: (i) structure and thermal parameters from structure solution S2 (see Table 6.8) are imported and restrained and the unit cell, Lorentzian and Gaussian size terms and Lorentzian strain terms refined; (ii) site positions released and refined; (iii) sites restrained and thermals refined; (iv) thermals restrained and the calculated occupancy from

137 the a parameter of the unit cell added and restrained with refinement of the site

positions; (v) all parameters refined (except occupancy calculated from the a unit

cell parameter).

For all boron carbide phase weight quantification measurements 25% by weight

diamond powder internal standard is used. In a typical sample preparation, 60 mg

diamond powder is thoroughly mixed with 180 mg boron carbide powder via

mortar and pestle.

138

6.6 RESULTS

6.6.1 ABERRATION ANALYSIS

Table 6.2 gives a measure of the peak asymmetry for all materials analysed in this study for both BB and FIPB geometries. The closer the peak asymmetry value is to

0 the more symmetric the average measured peak shape for a sample.

Table. 6.2.

Peak asymmetry of measured samples in both BB and FIPB geometries. A value of 0 describes perfect symmetry with asymmetry increasing with larger values.

High MAC Low MAC

Sample LaB6 Silicon Corundum Diamond cBN Boron carbide

Asymmetry BB 14.53 9.73 9.99 47.08 33.35 44.85

FIPB 8.80 6.62 5.31 7.69 3.84 11.97

Fig. 6.2 gives a visual example of the resultant peak asymmetry in the second most intense reflection (0 2 2) of the diamond standard when using a focusing (BB) geometry compared against FIPB. This reflection is chosen as an example as any contribution from axial divergence is minimised.

Fig. 6.2. The (0 2 2) reflection of the diamond standard measured using BB and FIPB.

139 6.6.2 MEASUREMENT OF STANDARDS

Rietveld refinement is applied to calculate the phase abundances (wt%) of the

standard mixtures for comparison against the weighed values. The unit cell

parameters for both the high MAC and low MAC standard mixtures are also

measured for comparison across the two different geometries. The averaged

refined values are shown in Tables 6.3 and 6.4.

Table. 6.3.

Refined wt% and associated relative bias as well as unit cell dimensions for silicon/corundum 1:1

powder mixtures.

Remount x3 (same mix) Remix x3 (different mix) Sample FIPB BB FIPB BB Si 640D wt% 48.7(3) 48.0(3) 48.1(3) 48.1(1) R. Bias (%) 2.7(6) 4.0(7) 3.8(6) 3.7(3) Unit cell (Å) 5.43136(3) 5.43136(3) 5.43137(3) 5.43137(1)

Al2O3 wt% 51.3(3) 52.0(3) 51.9(3) 51.9(3) R. Bias (%) 2.7(6) 4.0(7) 3.8(6) 3.7(3) Unit cell (Å) - a 4.75967(4) 4.75966(5) 4.75970(5) 4.75967(1) b 12.99350(12) 12.99317(7) 12.99346(3) 12.99326(16)

Table. 6.4.

Refined wt% and associated relative bias as well unit cell dimensions for diamond/cBN 1:1 powder

mixtures.

Remount x3 (same mix) Remix x3 (different mix) Sample FIPB BB FIPB BB Diamond wt% 52.0(2) 52.3(2) 52.2(1) 52.2(2) R. Bias (%) 4.1(4) 4.6(3) 4.4(3) 4.5(4) Unit cell (Å) 3.567271(16) 3.567019(13) 3.567255(26) 3.567016(18) cBN wt% 48.0(2) 47.8(2) 47.8(1) 47.8(2) R. Bias (%) 4.1(4) 4.6(3) 4.4(3) 4.5(4) Unit cell (Å) 3.616186(5) 3.615768(18) 3.616186(17) 3.615781(18)

140

6.6.3 FIPB ANALYSIS OF BORON CARBIDE

Table 6.5 lists the unit cell parameters measured using FIPB for all boron carbide samples analysed in this study. Note the absence of a secondary phase in the solution-based samples.

Table. 6.5.

Refined hexagonal unit cell sizes of commercial and solution-synthesised boron carbide powders.

Primary Secondary

Lattice parameter a (Å) c (Å) a (Å) c (Å)

Commercial

Goodfellow 5.6021(2) 12.0785(6) 5.6195(10) 12.1188(29)

Sigma 5.6019(1) 12.0753(3) 5.6200(6) 12.1156(15)

American Elements 5.6017(1) 12.0739(2) 5.6227(3) 12.1275(8)

Synthesised

BC-Full 5.5991(1) 12.0673(5) - -

BC-50 5.5992(1) 12.0675(5) - -

BC-30 5.5991(1) 12.0671(5) - -

BC-0 5.5989(1) 12.0677(3) - -

High Purity 5.5990(1) 12.0680(2) - -

By implementation of the unit cell size correlation with stoichiometry reported by Aselage et al.[15], the carbon concentration of all detected phases can be calculated. The following linear model (Eq. 6.1) is derived by fitting the a lattice parameter of the unit cell data presented by Aselage et al. and is used to calculate the carbon concentrations reported in Table 6.6. Fig 6.3 shows the unit cell (a parameter) of all detectable phases within the commercial samples as well as the High Purity sample superimposed on the derived liner model of Eq. 6.1.

. at% carbon = . [Eq. 6.1] 𝑎𝑎−5 6768 −0 0039

141

Fig. 6.3. Carbon concentrations for the commercial and High Purity samples calculated using the

linear model of Eq. 6.1 from their refined a lattice parameters.

Table. 6.6.

Calculated carbon concentrations (at%) of analysed boron carbide samples based on the unit cell

trend of Aselage et al.

Carbon American BC- High Goodfellow Sigma BC-50 BC-30 BC-0 concentration Elements Full Purity

Primary (at%) 19.15 19.21 19.26 19.92 19.90 19.92 19.97 19.95

Secondary (at%) 14.69 14.56 13.87 - - - -

6.6.4 STRUCTURE REFINEMENT

Two different structures are realised through refinement of long scan time (32

hours) variable count rate data collected from the High Purity solution-synthesised

sample based on two different approaches. The first approach assumes the

inability of the FIPB technique to resolve any difference between the site position

and thermal parameter of the substitution site when occupied by carbon. By

142

restraining the C2 site position and thermal parameter to equal that of B3 the refined structure (S1) of Table 6.7 is obtained.

Table. 6.7.

Structure solution S1. Refined atomic positions, site occupancies and thermal parameters for the

High Purity boron carbide sample based on B3/C2 equated site positions and thermal parameter.

Site No. Positions Occupancy beq x y z Rwp=6.09

C1 6 1 0.98(3) 0 0 0.1185(1) B1 3 1 1.80 (5) 0 0 0 B2 18 1 1.26(3) 0.5041(1) 0.4959(1) 0.1918(1) B3 18 0.780(13) 1.52(3) 0.4415(1) 0.5585(1) 0.0524(1) C2 18 0.220(13) 1.52(3) 0.4415(1) 0.5585(1) 0.0524(1)

The second approach assumes the ability of the FIPB technique to resolve a change in the site position once occupied with carbon. By allowing the C2 site position and thermal parameter to refine independently of B3 the structure (S2) refinement of

Table 6.8 is obtained.

Table. 6.8.

Structure solution S2. Refined atomic positions, site occupancies and thermal parameters for the

High Purity boron carbide sample with independent refinement of the C2 substitution site and thermal parameter.

Site No. Positions Occupancy beq x y z Rwp=5.88

C1 6 1 1.04(3) 0 0 0.1186(1) B1 3 1 2.09(5) 0 0 0 B2 18 1 1.40(3) 0.5033(1) 0.4967(1) 0.1914(1) B3 18 0.851(12) 1.13(6) 0.4406(2) 0.5594(2) 0.0563(3) C2 18 0.149(12) 0.88(25) 0.4498(7) 0.5502(7) 0.0304(15)

143 Fig. 6.4 compares the fits and difference plots of a commercial sample (Sigma) vs.

a solution-based sample (High Purity). Both fits are characteristic of a typical

refinement for each respective sample set.

Fig. 6.4. Structure fit comparison of a commercial sample (Sigma) and a solution-based sample

(High Purity) both containing diamond used for spiked wt% analysis.

6.6.5 QUANTIFICATION

FIPB traces of the solution-based sample set can be seen in Fig. 6.5, intensity

normalised against the internal diamond standard.

144

Fig. 6.5. FIPB traces of the solution-based sample set overlaid and intensity normalised against the internal diamond standard. All reflections are allocated to the boron carbide phase unless otherwise marked. The boron carbide phase intensity can be seen to clearly increase with reduced water adsorption in the precursor (refer Table 6.1).

FIPB traces of the commercial boron carbide sample set can be seen in Fig. 6.6, intensity normalised against the internal diamond standard.

Fig. 6.6. FIPB traces of the commercial boron carbide sample set overlaid and intensity normalised against the internal diamond standard. All reflections are allocated to the boron carbide phase unless otherwise marked.

145 The refined phase wt% values of these traces obtained by the internal standard

method can be seen in Table 6.9.

Table. 6.9.

Refined phase wt% of solution-synthesised and commercial boron carbide samples. The reported

Rwp values are calculated with the impurity content excluded. Two values for the solution-based

samples are reported. The first values are refined using structure solution S1 (see Table 6.7), while

bracketed numbers are refined using structure solution S2 (see Table 6.8)

Phase abundance (wt%) Primary Secondary Residual Rwp

Synthesised

BC-Full 75.0 [81.0] - 25.0 [19.0] 12.3 [11.9]

BC-50 75.9 [81.8] - 24.1 [18.2] 11.7 [11.3]

BC-30 77.7 [83.6] - 22.3 [16.4] 11.8 [11.5]

BC-0 90.0 [94.6] - 10.0 [5.4] 10.3 [10.0]

High Purity 94.3 [98.8] - 5.7 [1.2] 8.8 [8.5]

Commercial

Goodfellow 78.1 15.5 6.4 6.4

Sigma 77.7 19.6 2.7 5.6

American Elements 70.1 25.7 4.2 6.6

146

Fig. 6.7 shows the effect on refined boron carbide wt% versus thermal value within the solution-based boron carbide sample set.

Fig. 6.7. Effect of site thermals on refined phase wt% in solution-based boron carbide powders.

Fig 6.8 shows the refined wt% values based on the C2 site occupancy within the solution-based boron carbide sample set.

Fig. 6.8. Effect of C2 occupancy on refined phase weight % in solution-based boron carbide powders.

147 6.7 DISCUSSION

6.7.1 DIFFRACTION GEOMETRY SELECTION

FIPB provides a distinct advantage in the X-ray analysis of low MAC materials

through utilisation of a non-focusing geometry. This geometry in conjunction with

a parallel plate collimator prevents any diffracted X-rays off-axis to the detection

position from reaching the detector. This is important for low MAC materials as

the X-ray penetration depth becomes significant and results in substantial

diffraction occurring below the sample surface. In a focusing geometry such as BB

this sub-surface diffraction manifests as a severe forward peak slope aberration

(see Fig. 6.2).

Another advantage of FIPB is the ability to restrain the incidence angle (ω) so that

the beam penetration depth and irradiation length are constant. In symmetric BB

geometry, the irradiated length and penetration depth varies with incidence angle

such that the incident beam penetrates further into the sample at higher angle

(see Fig. 6.1b). This becomes important for thin bed heights due to small sample

volumes (especially for low MAC materials) which are often produced in lab-scale

synthesis. In the case of reduced sample thickness in a low MAC material,

complete sample penetration can occur with a symmetric geometry, resulting in

an intensity loss that increases with incidence angle. In other words, it cannot be

guaranteed in BB geometry, for thin samples, that the sample is infinitely thick

with respect to the beam. However, this requirement of powder XRD can be

realised by utilising a FIPB geometry for data collection as seen in Fig. 6.1a. The

148

removal of peak shift associated with sample height displacement that occurs in a focusing geometry is also possible with the FIPB technique[12]. In addition, the irradiated length at low incidence angles in BB geometry may be longer than the sample, requiring the use of small divergence slits and a corresponding poor diffracted intensity. This is more of a problem for samples with peaks at low diffraction angles (<10 °2θ) such as organic compounds.

A comparison of BB and FIPB geometries by way of collected peak asymmetry for all samples analysed in this study can be seen in Table 6.2. Even with high MAC materials such as LaB6 and Silicon, an appreciable increase in peak symmetry is achieved with FIPB. However, the greatest improvement is observed in the low

MAC materials due to prevention of sub-surface diffraction reaching the detector.

The forward peak slope aberration that manifests due to a focusing collection geometry (BB) can generally be modelled by additional peak shape convolutions on a phase-by-phase basis if it is greater than the instrument function’s axial divergence correction. However, in the case of boron carbide, there is a very specific reason why it is advantageous to remove this aberration through hardware optimisation as will be discussed in the next section.

The effects of using a FIPB geometry compared to BB for low MAC materials is first investigated practically by measurement and refinement of a set of pure standards

(see Tables 6.3 and 6.4). Co Kα1 radiation is used because it is less penetrating than Cu Kα1. The refined wt% values compared across BB and FIPB geometries show negligible variation in both the high MAC and low MAC sample sets. There is also negligible difference between the refined values for the remixing and

149 remounting data sets across the same geometry. Because the standard material

mixtures employed here have comparable MACs, micro-absorption (a major

problem in quantitative phase analysis that is often insurmountable[16]) is

minimised. This approach means that significant potential sources of error in the

quantifications in Tables 6.3 and 6.4 are sample homogeneity and crystallinity - i.e.

possibility of undetectable amorphous content - and the thermal parameters of

the atomic positions in the reference crystal structures used[17]. The Round Robin

on quantitative phase analysis found an accuracy of 0.4 to 1.5% can be achieved

at the 95 wt% level, and 9 to 26% relative at the 1 wt% level[16]. Another study on

Portland cements found relative errors of 2 to 3% for main phases and 5 to 10%

for low content phases[18]. In simple samples of quartz and corundum, with or

without an additional silica flour amorphous phase, uncertainties within ± 2 wt%

absolute from weighed were observed[19]. Given our relative biases are < 5 wt%

for all data sets, we consider the cBN, diamond, corundum and silicon phases

essentially fully crystalline with negligible amorphous content. Thus, the diamond

phase analysed is a suitable internal standard for boron carbide quantification.

The average refined unit cell values for the high MAC sample sets are also

consistent across both geometries, varying only within the refined error. However,

a deviation between BB and FIPB is observed in the refined unit cell values for the

low MAC sample sets (Table 6.4). Although consistent within the same geometry,

the low MAC sample sets reveal refined unit cell values that vary substantially

outside the refined error when comparing BB to FIPB. This consistent discrepancy

can be attributed to minimisation of aberrations in the collected data by

150

optimisation at the hardware level by using FIPB. Given that the FIPB values are refined from aberration-minimised data, it follows that these values represent a more accurate measure of the unit cell parameters. This becomes increasingly important in the analysis of boron carbide. Not only is boron carbide a low MAC material, but the stoichiometry of a given sample is most conveniently determined by measurement of its unit cell parameters. It has been demonstrated experimentally that as the structural carbon concentration of boron carbide increases from the boron rich limit, a decrease in the a lattice constant of the hexagonal unit cell occurs[20]. This trend has been quantified[15], allowing for stoichiometric determination in single phase boron carbide samples provided the lattice parameters are known. Given that the a lattice constant can vary by up to

0.045 Å within the stable stoichiometric range[15], this corresponds to an approximate shift of 0.00038 Å per 0.1% increase in carbon concentration. This is within the range of the measured unit cell variation observed between BB and

FIPB geometries for the low MAC standards (~0.00041 Å for cBN and ~0.00025 Å for diamond). This observed capability of FIPB to measure precise unit cell parameters in low MAC materials is therefore highly relevant for stoichiometric determination in boron carbides.

6.7.2 FIPB STRUCTURAL ANALYSIS

In this study, a total of eight boron carbide samples are analysed using FIPB. Three of these samples are commercially available powders with the remaining samples synthesised in-house using a solution-based method. The combined list of samples is given in Table 6.5.

151 Boron carbide phase resolution

Removal of the forward peak slope aberration becomes particularly important in

the analysis of these samples due to the stoichiometric effect on unit cell size

observed in boron carbide. This relationship creates an associated peak shift in

collected data as the unit cell size changes according to the Bragg equation. As the

carbon concentration drops from the carbon rich limit the unit cell size (and

interplanar spacing) increases, which results in a corresponding peak shift to lower

values of θ. Given the reported stoichiometric range of boron carbide from ~8 to

~20 at% carbon, this translates to an approximate calculated range of 0.37 °2θ for

the primary (0 2 1) reflection of boron carbide. This range lies well inside the extent

of the forward peak slope aberration that is present when using a focusing

geometry. Consequently, if a secondary, carbon deficient boron carbide phase is

present within a sample that is predominantly carbon rich, any signal from the

former will be embedded in the forward peak slope aberration of the latter. The

carbon deficient phase is then obscured and cannot be detected or modelled

accurately. This situation is precisely the case observed in commercial boron

carbide powders analysed using BB. Analysis of the commercial powders reported

in this study using a conventional BB geometry give no indication about the

presence of a secondary boron carbide phase. However, when analysed with the

FIPB technique, a clear shoulder becomes evident on the lower 2θ side of all

diffraction peaks. This can be clearly seen when comparing the same commercial

sample analysed by both BB and FIPB to the High Purity boron carbide sample (also

collected with FIPB) as seen in Fig. 6.9.

152

Fig. 6.9. The (0 2 1) reflection of boron carbide purchased from Sigma collected using both BB and

FIPB. These peaks are superimposed over the (0 2 1) reflection of a pure B4C sample collected using

FIPB to highlight the peak shoulder generated by the carbon deficient boron carbide phase in commercial samples.

It should be noted that among the samples analysed in this study, the forward peak shoulder is only observed in boron carbide samples obtained from commercial suppliers. All samples synthesised in-house show highly symmetric diffraction peaks, without any detectable forward shoulder, indicating that they comprise a single boron carbide phase only. With the forward peak slope aberration removed through the implementation of FIPB, the secondary phase present in the commercial samples can be detected and modelled. It also becomes possible to measure the unit cell parameters of the secondary, carbon deficient phase (see Table 6.5).

Boron carbide phase stability

From the measured unit cell values, the structural carbon content of the identified boron carbide phases can then be inferred by the unit cell/stoichiometry trend

153 reported by Aselage et al.[15]. Although there may be some error in the absolute

values reported for the carbon concentrations due to their dependence on a linear

fit to data reported by Aselage et al., their relative values are highly accurate due

to the stringent collection methodology implemented.

Given the above, the data then reveals important correlations. Firstly, the carbon

concentrations of the boron carbide powders synthesised via the solution-based

method are consistently in the ~20 at% range. As reported previously[13],

residual-carbon impurity in these samples is minimised and the boron carbide

yield maximised through control of atmospheric conditions, i.e. prevention of

exposure to water in the form of humidity. It is observed that as water adsorption

in the precursor increases, the residual-carbon impurity also increases with a

corresponding decrease to the boron carbide yield (see Table 6.9). These changes

are shown to be a direct result of precursor homogeneity loss that increases with

water adsorption. As this occurs, the boron source has less direct contact with the

carbon source leading to decreased reaction kinetics and evaporation of the boron

source. This result, in turn, leads to residual unreacted carbon and poor yield. It is

interesting then, that despite the boron loss altering the targeted (20 at% carbon)

precursor ratio, a very consistent carbon concentration is observed in the final

solution-based boron carbide products (see Table 6.6). This outcome is further

demonstrated in the High Purity sample, wherein excess boron precursor (14% by

weight, see Table 6.1) has been added to manage the very small amount of

residual-carbon present after the reaction of precursors targeted at a 20 at%

carbon stoichiometry. Even in this sample, where a substantial excess of boron is

154

added, the ~20 at% carbon (B4C) boron carbide phase is still formed. This is suggestive of high thermodynamic stability for the B4C phase. Since residual- carbon impurity is not incorporated into the structure despite being available, the level of carbon certainly does not extend beyond this limit. In addition, the ~20 at% carbon limit is in good agreement with the current literature consensus.

However, there remains some ambiguity as to the most stable boron carbide phase when approaching this carbon limit. Density functional theory (DFT) modelling of boron carbide does appear to favour the B4C phase as the most stable form[21] [22]. However, evidence supporting a higher degree of stability in a slightly carbon deficient B4.3C phase has been reported[23]. Despite the wide range of possible stoichiometries with the potential to form in the boron carbide system, the B4C phase is consistently produced by the solution-based synthesis technique implemented here.

Secondly, there is an observed consistent carbon deficiency in the primary boron carbide phase measured in the commercial samples when compared to the solution-based samples. Varying between 19.15 and 19.26 at% carbon, these samples are closer to the B4.3C (18.87 at% carbon) model proposed by Werheit et al.[23]. This discrepancy in the primary phase carbon concentration may possibly be reconciled by the effect of formation temperature. Due to the enhanced precursor homogeneity afforded by the solution-based synthesis method, relatively low temperatures (1400 °C in this case) can be used to achieve complete boron carbide phase formation. In contrast to this, commercial techniques employing electric arc or Acheson type furnaces require much higher

155 temperatures (>2000 °C) due to poor reactant dispersion and the desire for an

increased reaction rate[1] [3]. The decreased stability of B4C at elevated

temperature has also been predicted in DFT calculations by Widom et al.[22], and

indeed the results of this study do lend experimental evidence to this finding.

However, further experimental investigation is warranted for confirmation.

There is also some degree of consistency in the unit cell parameters of the

secondary boron carbide phase detected in the commercial samples. Although

more variable across the sample set than the primary phase, the secondary phase

range lies between a B6.2C and B5.8C stoichiometry and is suggestive of another

high stability phase region at lower carbon concentration. This suggestion is also

in agreement with the DFT calculations of Widom et al. wherein a stable B6.5C

phase consisting of B12 icosahedra and C-B-C chains is predicted[22]. The presence

of this carbon deficient secondary phase is also encouraged by the commercial

synthesis methods employed wherein poor precursor dispersion and large heat

gradients could play a role in the formation of alternate boron carbide phases.

Carbon substitution

Given the exceptional quality of the solution-synthesised High Purity sample, a

high intensity FIPB scan of the sample is conducted via a 32 hour scan time with

increasing count time at higher angle for use in a full structure refinement. The

site position for carbon substitution beyond the idealised C-B-C chain with B12

icosahedra (13.33 at% carbon) structure is refined at the polar (B3) position within

p p the boron icosahedra only, giving a C-B-C chain B11C type structure where C

denotes the polar position within the boron icosahedra. This is chosen not only

156

because of the current literature consensus based on experimental and theoretical nuclear magnetic resonance (NMR), Raman and infrared spectroscopy investigations[10], but also because of the structure refinement itself. Allowing for carbon substitution on either the equatorial position within the icosahedra or the chain boron position consistently refines with zero substitution. However, refinement on the polar icosahedral position with the relationship shown in Eq.

6.2 yields a reasonable occupancy and improves the Rwp of the fit.

C2occ = 1 – B3occ [Eq. 6.2]

The approach yields a sensible refined value for the carbon site occupancy (Table

6.7). However, compared with the calculated occupancy from the a parameter of the unit cell, the value is overestimated by 2.18 at% carbon. Given the smaller ionic radius of carbon compared to boron, a change in the bonding environment would be expected at the substitution site. By allowing for a change in the carbon position upon substitution within boron icosahedra (see Table 6.8), the carbon substitution occupancy refines to within 0.66 at% carbon of the value based on the carbon concentration calculated from the a parameter of the measured unit cell. The Rwp of the fit is also improved from 6.09 for the position-restricted refinement to 5.88 with no restrictions on the substitution site position and thermal. This finding is counter to the current consensus wherein the carbon substitution site occupancy is deemed beyond the capability of X-ray investigation due to the similarity in form factors between the carbon and boron atom. The apparent ability to not only accurately refine this value, but also exclude the possibility of carbon substitution on the other boron sites, can be attributed to

157 two main factors; the exceptional quality and purity of the boron carbide sample

analysed as well as the meticulous nature of the of XRD technique employed for

analysis.

However, given the data available, it is not possible to definitively conclude the

ability of the technique to resolve the carbon site substitution to this level of detail

despite the accuracy in the refined site occupancy. Repeat synthesis and

subsequent measurement and refinement across a larger sample set is required

for confirmation. Attempts at refining the carbon substitution site occupancies of

the primary boron carbide phase in the commercial and residual-carbon

containing solution-based samples consistently produce occupancies higher than

those expected when compared to the calculated value from the a parameter of

the measured unit cell. This occurs even when allowing for independent

refinement of the C2 substitution position. This result is most likely due to the

impure nature of these samples interfering with accurate modelling as well as the

reduced counting statistics at high angle due to the constant count time data

collection strategy employed for these samples[24]. It should be noted that

attempts to refine for carbon substitution on the polar position of the boron

icosahedra in the secondary carbon deficient phase consistently yield a zero

occupancy. This result is in agreement with the C-B-C chain and B12 icosahedra

model of the B6.5C phase predicted by DFT modelling[22]. Given the possibility that

the highly accurate site occupancy calculated by the second approach to the

structure refinement may be artificial, structure data from the restricted S1

solution is reported.

158

Pattern fitting anomalies

Rietveld refinement of all solution-based samples reveals another recurring variation when compared to commercial samples with regard to peak fitting intensities. Specifically, when modelling solution-based samples, the (0 2 1) reflection is underestimated and the (0 1 2) reflection is overestimated (see Fig.

6.4a). This does not occur in the commercial samples where the intensities for these reflections are matched well by the rhombohedral structure model (see Fig.

6.4b). Attempts to model the intensity variation with preferred orientation are unsuccessful, and given the approximately spherical shape of the solution-based boron carbide particles, preferential stacking is unlikely to be responsible. This intensity mismatch occurring on the (0 1 2) and (0 2 1) reflections is consistent across all solution-based samples. When normalised against the (1 0 4) reflection, the (0 2 1) reflection intensity of solution-based samples is on average ~13% larger compared to commercial samples, and when normalised against the (0 0 3) reflection, the (0 1 2) reflection is ~25% smaller. Given that the normalised peak intensity differences are consistent within each sample set (commercial vs. solution-based) it can be inferred that the intensity variation between them is a direct result of their differing stoichiometries (19.21 at% carbon averaged for the commercial set compared to 19.93 at% carbon for the solution-based sample set).

This outcome suggests a significant structural difference between the two different stoichiometries, wherein rhombohedral symmetry is insufficient for accurate modelling of the peak intensities of carbon-rich samples approaching a

B4C stoichiometry. Widom et al. have predicted through DFT calculation the

159 existence of a stable low-temperature monoclinic B4C phase in which carbon

substitution is ordered throughout the structure, occurring at the same polar

position in each icosahedra[22]. However, attempts to model the solution-based

samples using monoclinic symmetry (space group C1m1, no. #8) do not yield an

improvement in the fit, even when refined against each of the six different polar

positions in the icosahedra. This result suggests that long-range ordering of the

carbon substitution position is not occurring in these samples. The monoclinic B4C

phase is predicted to be unstable above ~330 °C, converting to rhombohedral

symmetry through rotations of the B11C icosahedra and a slight loss of polar

carbon substitution[22]. Given that these samples were calcined at 1400 °C, a high

degree of substitutional disorder is expected. However, the intensity discrepancies

in the solution-based samples could possibly be due to some remnant of ordering

in the carbon substitution position. The slight loss of carbon substitution that is

predicted to occur with increasing temperature by Widom et al. could also explain

why all of the solution-based samples have a slight carbon deficiency below the

nominal targeted B4C stoichiometry[22].

6.7.3 QUANTIFICATION

Solution-based boron carbide

Based on structure solutions S1 and S2, two different values for boron carbide

phase abundance is refined for each solution-synthesised sample (see Table 6.9).

Assuming the inability of the FIPB technique to discern a change in the carbon

substitution, refinement of phase abundances using solution S1 is relatively

straight forward. This is because the site position and thermal parameter of the

160

carbon substitution are restricted to equal that of the equivalent polar boron atom. However, in the case of refinement using structural solution S2, restrictions are required on certain parameters due to the extra degrees of freedom allowed by releasing the carbon substitution site position and thermal parameter.

Specifically, refinement of thermal parameters is carried out with restrictions on the B1 and C2 sites to ensure they do not stray outside of the relative values obtained from structural solution S2. Without this additional restriction, the thermal values of the B1 and C2 sites refine nonsensical values with large error. As the S2 solution yielded a thermal value for the carbon substitution position close to that of the value obtained for a boron atom at the same site (B3), the C2 thermal value is restricted to be within the same margin (±0.2 of the refined value for the

B3 thermal). This restraint is also implemented as the C2/B3 position is equivalent in having the same atomic environment so similar thermals are expected.

Similarly, the B1 thermal value is restricted to be within ±1 of the C1 thermal to match the margin from the S2 solution. The refined values can be seen in Table

6.9 within brackets next to the restricted refinement results. A noticeable shift in the quantification result is clearly seen. Indeed, a shift in the refined wt% values will always occur depending on the thermal values used because anything which alters the Rietveld refinement scale factors has a large effect on quantification[17]

[19] and is especially true of thermal parameters. For simple two-phase mixtures

(one of which is the standard) with minimal peak overlap, we posit that refinement of thermal parameters during quantification is justified and worthy of investigation.

161 The effect on refined boron carbide wt% versus site thermal parameter within the

solution-based boron carbide sample set can be seen in Fig 6.7. The absolute value

varies by up to 10% depending on what thermals are used for refinement.

However, the relative difference in wt% between the samples remains the same.

As the relative values remain constant, the methodology implementing

independent site and thermal refinement of the carbon substitution position

(structural solution S2) may be more accurate as it provides a better description

of the High Purity sample. Assuming the High Purity sample is indeed without any

residual amorphous impurity as XRD analysis appears to suggest (no non-

diffracting impurity content), it’s use in calibration of the absolute quantification

value is vital. Without the use of a high purity boron carbide sample, it would not

be possible to obtain an absolute value for these quantification results. A similar

drift in the refined wt% values can be seen based on the C2 site occupancy (see

Fig. 6.8), although on a smaller scale (~2% across the stoichiometric range). This

variance in wt% is why the C2 site occupancy is always restrained to the calculated

value based on the a lattice parameter of the refined unit cell for quantification.

Commercial boron carbide

For refinement of the commercial sample set, a high purity sample of similar unit

cell and synthesis technique could not be obtained. Given the impure nature of

the commercial samples (see Fig. 6.6 for XRD traces used in quantification), a high

intensity structure refinement is not reliable enough to obtain a thermal

relationship for the B3/C2 sites. Therefore, the same approach as the S1 structure

solution is applied to commercial sample quantification. A variable count time

162

structure refinement is conducted on the American Elements sample and the refined site positions and thermals used as the starting values. As the commercial samples contain two different boron carbide phases, a further restriction is required for refinement to ensure that reasonable values for the relative wt% of these phases are obtained. Without restriction on the refined Lorentzian and

Gaussian size terms, the secondary phase refines a value for the crystallite size which is too small (sub 45 nm) to match the synthesis conditions. This occurs as there are multiple combinations of peak heights and widths for the primary and secondary phases that are able to produce an accurate description of the combined peak shape. As the commercial samples have experienced temperatures of >2000 °C, it is unlikely that such a small crystallite size would result. This variation in probable models of the overall peak shape not only affects the quantification value, but also affects the refined unit cell value for the secondary phase. This discrepancy is best addressed by restraining the refined crystallite size of the primary and secondary phases to be equal. Given both phases exist in the same sample, they have experienced the same synthesis conditions, and a similar crystallite size could be expected. The refined wt% values for the commercial sample set can be seen in Table 6.9 and a typical fit can be seen in Fig.

6.4b.

A much lower Rwp for the commercial samples is always obtained compared to the solution-based samples (refer Table 6.9). This is because of the peak intensity mismatch detailed earlier that occurs in solution-based samples only (see difference plots in Fig. 6.4). The refined wt% values for the primary, secondary and

163 residual phases reflect the collected traces well. The American Elements sample

refines the greatest amount of secondary boron carbide phase, and compared to

the Sigma and Goodfellow traces, there is noticeably more volume under the

forward shoulder in the American Elements sample. The Goodfellow sample

refines the highest value for the primary phase content, and when normalised

against the internal diamond standard, it is the Goodfellow sample that has the

most intense peaks across the sample set. This result is with the one exception of

the (0 2 1) reflection, which is again the same reflection that showed discrepancies

in the solution-based samples. This observation lends further evidence to the

hypothesis that the intensity of this peak may be directly linked to structural

changes in the sample related to the carbon content. It should be noted that a

significantly larger value for the Lorentzian strain term is consistently refined for

the secondary boron carbide phase when compared to the primary phase. As

there is significant variation in the secondary phase unit cell size across the

commercial sample set, there does appear to be a range of possible

stoichiometries that are stable at lower carbon concentration. It follows that

within the same sample it is most likely not just one carbon-deficient phase but a

range of phases with an average reflected by the refined unit cell parameters.

Significant strain within this phase range is then expected. The variation in the

carbon content of the secondary boron carbide phase is most likely a direct effect

of synthesis temperature.

164

6.8 CONCLUSION

Utilising FIPB XRD with a small incidence angle (ω = 9°) and a low MAC material as internal standard, highly accurate and reproducible unit cell size measurement and phase quantification of low-volume boron carbide powders is achieved.

Increased accuracy and reproducibility of refinement results is attributed to the minimisation of aberrations through hardware optimisation that would otherwise be detrimental to modelling. By removing the peak asymmetry generated by a focusing geometry and restraining the beam penetration depth and irradiated length, an appreciable difference in the refined unit cell values for the low MAC standards is clearly observed. The removal of the forward peak slope aberration also allows for the detection of a secondary, carbon-deficient phase in commercial boron carbide samples. Through unit cell measurement, this secondary phase is shown to vary between a B6.2C and B5.8C stoichiometry, indicative of stability for a low-carbon boron carbide phase range at high temperatures.

Solution-synthesised boron carbide samples are shown to contain only one phase with consistent B4C stoichiometry. In the case of the solution-based samples, the primary phase never exceeds nor falls below the ~20 at% carbon limit despite the fact that in specific cases significant extra boron or carbon is available for reaction.

This result suggests a very high stability for the ~20 at% carbon B4C phase at low synthesis temperature (1400 °C). Comparatively, the primary phase in commercial samples is found to be significantly carbon deficient. This is most likely due to the effect of synthesis temperature on stoichiometry as predicted in DFT calculations by Widom et al.[22]. Indeed, the results presented here indicate a viable pathway

165 for the synthesis of any desired carbon concentration boron carbide phase in high

purity. However, further experimental data is required to confirm the effect of

synthesis temperature on stoichiometry.

The FIPB methodology developed here is shown to have significant merit in the

structural analysis and quantification of low-volume boron carbide powders. The

ability to detect and differentiate between multiple boron carbide phases in a

given boron carbide powder sample provides an avenue for quality assessment

and optimisation that is particularly relevant to technological application of the

material. The developed technique may also find application in XRD analysis of

other low MAC materials such as pharmaceuticals and other organic compounds.

166

6.9 ACKNOWLEDGEMENTS

The author would like to gratefully acknowledge the Institute for Future

Environments for research support as well as the Central Analytical Research

Facility for assistance with characterisation.

167 6.10 REFERENCES

1. Suri, A., Subramanian, C., Sonber, J. & Murthy, T., Synthesis and consolidation of boron carbide: a review. International Materials Reviews, 2012. 55: p. 4-40. 2. Bigdeloo, J. & Hadian, A., Synthesis of High Purity Micron Size Boron Carbide Powder from B2O3/C Precursor. International Journal of Recent Trends in Engineering, 2009. 1: p. 176-180. 3. Thevenot, F., Boron Carbide - A Comprehensive Review. Journal of the European Ceramic Society, 1990. 6: p. 205-225. 4. Niihara, K., Nakahira, A. & Hirai, T., The Effect of Stoichiometry on Mechanical Properties of Boron Carbide Journal of the American Ceramic Society, 1984. 67: p. C-13-C-14. 5. Sinha, A., Mahata, T. and Sharma, B., Carbothermal route for preparation of boron carbide powder from boric acid–citric acid gel precursor. Journal of Nuclear Materials, 2002. 301: p. 165-169. 6. Yanase, I., Ogawara, R. & Kobayashi, H., Synthesis of boron carbide powder from polyvinyl borate precursor. Materials Letters, 2009. 63: p. 91-93. 7. Kakiage, M., Tominaga, Y., Yanase, I. & Kobayashi, H., Synthesis of boron carbide powder in relation to composition and structural homogeneity of precursor using condensed boric acid–polyol product. Powder Technology, 2012. 221: p. 257-263. 8. Watts, J. L., Talbot, P. C., Alarco, J. A. & Mackinnon, I. D. R., In-Situ Carbon Control in the Preparation of Precursors to Boron Carbide by a Non-Aqueous Solution Technique Journal of Materials Science and Engineering A, 2015. 5: p. 8-20. 9. Chen, X. W., Dong, S. M., Kan, Y. M., Zhou, H. J., Hu, J. B. & Ding, Y. S, Effect of Glycerine Addition on the Synthesis of Boron Carbide from Condensed Boric Acid- polyvinyl Alcohol Precursor. RSC Advances, 2016. 6: p. 9338-9343. 10. Domnich, V., Reynaud, S., Haber, R. & Chhowalla, M., Boron Carbide: Structure, Properties, and Stability under Stress. Journal of the American Chemical Society, 2011. 94: p. 3605-3628. 11. Raftery, N.A., Bekessy, L. K. & Bowpitt, J., Analysis of low mass absorption materials using glancing incidence X-ray diffraction Advances in X-ray Analysis 2007. 50: p. 173-176. 12. Raftery, N. A. & Vogel, R., Limitations of asymmetric parallel-beam geometry. Journal of Applied Crystallography, 2004. 37: p. 357-361. 13. Watts, J. L., Talbot, P. C., Alarco, J. A. & Mackinnon, I. D. R., Morphology control in high yield boron carbide. Ceramics International, 2017. 43: p. 2650-2657. 14. Rowles, M. R. & Madsen, I. C., Whole-pattern profile fitting of powder diffraction data collected in parallel-beam flat-plate asymmetric reflection geometry. Journal of Applied Crystallography, 2010. 43: p. 632-634. 15. Aselage, T. L. & Tissot, R. G., Lattice Constants of Boron Carbide. Journal of the American Ceramic Society, 1992. 75: p. 2207-2212. 16. Madsen, I. C., Outcomes of the International Union of Crystallography Commission on Powder Diffraction Round Robin on Quantitative Phase Analysis: samples 1a to 1h. Journal of Applied Crystallography, 2001. 34: p. 409-426. 17. Snellings, R., Machiels, L., Mertens, G. & Elsen, J., Rietveld refinement strategy for quantitative phase analysis of partially amorphous zeolitized tuffaceous rocks. Geologica Belgica, 2010. 13: p. 183-196. 18. Torre, A. G. de la & Aranda, M. A. G., Accuracy in Rietveld quantitative phase analysis of Portland cements. Journal of Applied Crystallography, 2003. 36: p. 11169-11176.

168

19. Madsen, I. C., Description and survey of methodologies for the determination of amorphous content via X-ray powder diffraction. Zeitschrift für Kristallographie – Crystalline Materials, 2011. 226: p. 944-955. 20. Bouchacourt, M. & Thevenot, F., Analytical investigation in the B-C system. Journal of the Less-Common Metals, 1981. 82: p. 219-226. 21. Vast, N., Sjakste, J. & Betranhandy, E., Boron carbides from first principles. Journal of Physics: Conference Series, 2009. 176: p. 1-18. 22. Widom, M. & Huhn, W. P., Prediction of orientational phase transition in boron carbide. Solid State Sciences, 2012. 14: p. 1648-1652. 23. Werheit, H. & Shalamberidze, S., Advanced microstructure of boron carbide. Journal of Physics: Condensed Matter, 2012. 24: p. 385406. 24. David, W. I. F., Powder diffraction: least-squares and beyond. Journal of research of the National Institute of Standards and Technology, 2004. 109: p. 107-123.

169

CHAPTER 7

CONCLUSION AND

RECOMMENDATIONS

170

7.1 SUMMARY OF FINDINGS

7.1.1 COMPARISON OF SYNTHESIS TECHNIQUES

As mentioned in the foreword to Chapter 5, very distinct differences were observed in the effects of water absorption on PVAc and sucrose generated precursor powders. Specifically, the degree of precursor dispersion in sucrose- based pre-treated powders was found to be, at the initial processing stages, much greater. As seen in section 5.6.2, upon exposure to atmospheric conditions, sucrose-based precursor powders begin to break apart as the intimately distributed B2O3 absorbs water. This results in the loss of enhanced precursor contact which is detrimental to the control of phase purity, yield and morphology.

Therefore, processing of sucrose-based precursor powders under inert conditions is of high importance to controlled boron carbide synthesis. In the case of PVAc- based precursors to boron carbide, the complete opposite is observed. Processing of these precursors under complete inert conditions results in poor yield, large amounts of residual carbon impurity and highly variable morphology in the boron carbide product. However, exposure of PVAc-based precursors to atmospheric conditions before final calcination results in increased yield, reduced residual carbon and a more consistent morphology. This discrepancy between the two synthesis techniques is due to differences in the initial reactant dispersion obtained at the pre-treatment stage. Similar to the process identified in sucrose- based precursors (section 5.6.2), PVAc-based precursors will adsorb water as B2O3 converts first to polymeric metaboric acid and then H3BO3. However, for PVAc precursors, this process facilitates an increase in contact between reactants. The

171 condensed carbon matrix formed in PVAc-based precursor powders after pre-

treatment is a highly porous structure. Whereas, before exposure to atmospheric

conditions, the B2O3 constituent exists as partially segregated spheres of the order

of 100-400 nm in diameter embedded within the carbon structure. Indeed, the

porous carbon structure results from the formation of these B2O3 spheres. This

partial segregation of precursors was initially not identified by previous

researchers, as precursor powders were routinely exposed to atmospheric

conditions after pre-treatment. Upon exposure to humidity, the B2O3 spheres

convert to polymeric metaboric acid and begin to spread throughout the porous

carbon matrix, increasing reactant contact and dispersion. These ‘strings’, now in

the form H3BO3 due to conversion to the fully hydrated phase, can be seen in pre-

treated precursor SEM images in section 4.6.2. The success of the PVAc-based

technique in lowering calcination temperatures and minimising residual carbon is

then attributed to the enhanced reactant dispersion that occurs after conversion

of the B2O3 spheres to H3BO3 as it spreads throughout the porous carbon matrix.

After conversion of B2O3 to H3BO3 and subsequent reticulation throughout the

porous carbon scaffold, the degree of dispersion in PVAc-based powders is still

lacking in comparison to sucrose-based precursor powders processed under inert

conditions. This difference is clearly observed in the morphology of boron carbide

formed from each technique (see Fig. 4.13A and Fig. 5.10a and b). Boron carbide

formed from the PVAc-based technique has a larger average particle size as well

as a larger variation in particle size and shape (Fig. 4.13A). This morphology is

consistent with that described in process pathway (ii) of Fig. 5.10, and indeed, the

172

processing conditions described do closely reflect those implemented in the PVAc- based synthesis technique. Conversely, the boron carbide particle size obtained from inert processing of sucrose-based precursors is smaller and of high consistency (Fig 5.10a and b), suggesting a more intimate reactant dispersion in the mixed precursor powder before calcination.

The developed PVAc-based synthesis technique is therefore considered inferior to the sucrose-based technique due to comparatively poor dispersion in precursor powders at the pre-treatment stage. This failing is possibly attributed to insufficient complexation of the borate anion with the PVAc polymer in solution. A second cause may be the formation of a very fine emulsion within the methanol precursor solution. PVAc is highly water insoluble, and when fully dissolved, the presence of water will cause the polymer to crosslink and precipitate out of solution. Because of this mechanism, small amounts of water in the precursor solution may lead to microscopic inhomogeneities due to polymer aggregation, impeding its ability to effectively disperse the boron precursor. Although great lengths were taken to keep the solution environment anhydrous, water contamination may have occurred. The dehydration of the borate anion upon complexation may also have played a role in providing water, facilitating polymer crosslinking. A combination of these effects would create a lack of intimate reactant dispersion allowing for boron oxide melts and subsequent agglomeration at the pre-treatment stage. A similar porous structure to that obtained in PVAc-based precursor powders is also observed consistently in other techniques presented in the literature, suggesting a very similar mechanism is also occurring.

173 Comparatively, precursor differentiation is not possible at the pre-treatment stage

in sucrose-based mixed precursor powders processed under dry conditions,

suggesting fine dispersion at the nanometer scale. This fine dispersion is reflected

in the consistent and sub-micrometer morphology of the final boron carbide

product. Despite the drawbacks of the PVAc-based synthesis technique, it still

warrants further investigation due to the ability to from a well dispersed precursor

powder without the requirement of dry processing conditions.

7.1.2 OPTIMUM SYNTHESIS AND PROCESSING CONDITIONS

Through comprehensive review of available literature pertaining to the solution-

based synthesis of boron carbide, the key factor for enhancing synthesis capability

was identified as the attainment of homogeneous reactant mixtures with fine

dispersion. The failing of previously described techniques in solution–based

synthesis of boron carbide was also identified as an inability to retain a

homogenous and fine dispersion throughout the entire processing activity. The loss

of fine dispersion before calcination is routinely observed in previously reported

synthesis techniques, and is responsible for inconsistent morphology and residual

carbon impurity in the final boron carbide product.

The primary process identified that gives rise to these variabilities is the

temperature and environmental dependent phases of H3BO3, the most commonly

used boron precursor in solution-based syntheses of boron carbide. Through

dehydration and subsequent water absorption, the many phases of H3BO3 undergo

multiple morphological changes. The most notable of which is the swelling and

polymerisation of metaboric acid due to water absorption by B2O3. This mechanism

174

can lead to significant loss of reactant dispersion if not managed correctly.

Therefore, the processing conditions conducive to mitigating the deleterious effects of the dynamic morphology inherent in condensed H3BO3/carbon precursor powders were identified. Through the application of these processing and handling techniques, as well as others described in section 3.2, a highly efficient synthesis technique for the formation of boron carbide in high yield, consistent stoichiometry and purity with controllable morphology at lowered calcination temperatures was realised. Without the development of the enhanced XRD technique detailed in

Chapter 6 this degree of synthesis optimisation would not have been possible.

7.1.3 ADVANCEMENT OF COMPOSITIONAL UNDERSTANDING

Through the identification and removal of aberrations resultant from the low MAC of boron carbide, an XRD technique with improved accuracy conducive to structure modelling was achieved. These improvements facilitated accurate and reproducible structural and phase wt% analysis in low-volume boron carbide powders. This outcome in turn allowed for improved synthesis optimisation.

Implementation of the developed FIPB technique for solution-based and commercially synthesised boron carbide powder analysis also realised novel developments in the understanding of the compositional properties of boron carbide.

Solution-synthesised boron carbide is shown to have a highly consistent B4C stoichiometry despite the initial reactant composition. Conversely, boron carbide synthesised by commercial means is found to be carbon deficient in structure with variable stoichiometry. This result indicates a high stability for the B4C phase when

175 synthesised through solution-based methods. This stability is attributed to two

main conditions: (i) the high degree of precursor dispersion; and (ii) the synthesis

temperatures used. As commercially synthesised boron carbide is formed at much

higher temperatures when compared to solution-synthesised boron carbide

(>1750 °C compared to <1400 °C respectively), a correlation can be made between

the resultant stoichiometry and synthesis temperature. This trend of structural

carbon concentration with temperature has also been observed in DFT calculations

of boron carbide[1], and the results presented here validate this finding.

The enhanced resolution afforded by the optimised XRD technique also allows for

the differentiation between multiple phases of alternate stoichiometry within the

same boron carbide sample. This approach serves as an excellent tool for assessing

the purity of boron carbide samples, with solution-synthesised articles shown to

comprise only a single B4C phase. However, analysis of commercial samples reveals

the presence of multiple phases of varying stoichiometry; a carbon-rich phase as

well a carbon-deficient phase. As mentioned previously, the carbon-rich phase is

found to be slightly carbon deficient in respect to the precise B4C stoichiometry.

This deficiency is most likely due to the synthesis temperature used but may also

be due to poor reactant dispersion. The carbon-deficient phase is shown to have

high variability across different commercial samples and has a stoichiometry

averaging approximately B6C. This phase duality detected in commercial samples

is also predicted in DFT calculations wherein a high thermodynamic stability in

both the B6.5C and B4C phases is reported[1]. This ability to differentiate between

alternate stoichiometry boron carbide phases within the same sample would not

176

be possible without the optimised XRD technique developed here. This capability has wide reaching implications for the technological advancement and commercial application of boron carbide.

177 7.2 FUTURE WORK

7.2.1 COMMERCIAL APPLICATION

As described in the objectives section of this thesis, an important facet that was

considered and integrated into the developed synthesis techniques was scalability.

This objective was included to ensure that the developed techniques were

conducive to large-scale commercial production. Of the two synthesis techniques

reported here, the sucrose-based technique (Chapter 5) exhibits great potential

for commercial application. This potential is not only because of the abundant,

low-cost precursor chemicals used and the simplicity of the processing

architecture developed, but also because the technique provides high purity

stoichiometric B4C in high yield with controlled morphology and lowered

calcination temperature. Each of these targeted improvements in the boron

carbide product, achieved through the synthesis optimisation conducted in this

work, are crucial to commercial viability.

As boron carbide is most commonly used commercially for ballistic and wear-

resistant applications, the ability of the sucrose-based synthesis technique to

consistently calcine the B4C phase (the hardest analogue across the boron carbide

stoichiometric range) is of high technological importance. The degree of purity in

the boron carbide product afforded by the synthesis technique is also critical to

the generation of a high quality boron carbide consumer product. Solid bodies of

boron carbide are formed by powder sintering at high temperatures (and/or

pressures depending on the technique implemented). The capacity for boron

178

carbide powders to efficiently sinter is highly dependent on the purity of the material, i.e. the presence of residual carbon impurity inhibits this process.

Although carbon-based sintering aids can be used, they are in a different form to the inhomogenously distributed graphitic carbon that commonly results from poorly controlled synthesis conditions. The ability to control the primary boron carbide particle size through hydrated precursor phase manipulation also provides another avenue for improved commercial implementation. Sintering efficiency is also particle size dependent (as are other physical properties of boron carbide), and the ability to tune the primary particle size without the need to grind down the boron carbide source reduces processing steps and maintains product integrity.

Finally, the lowered calcination temperatures required for final phase formation as well as the increased product yield provide significant cost savings for large-scale production. The ability to calcine at temperatures less than 1400 °C, when compared against commercially implemented calcination temperatures that range from 1750 °C to 2200 °C and even higher, increases commercial viability through reduction in operational power costs. The increased yield that is achievable also lowers production costs through efficient use of the feed precursors, maximizing production output and minimising waste production (in the form of boron loss through B2O3 volitisation and carbon as residual impurity).

For these enhanced properties to be realised in large-scale production, future work then falls to the scale-up of the synthesis process, particularly in the final calcination stage. All aspects of the synthesis process are designed with scalability

179 in mind as per the initial objective guidelines. However, as described in section

3.2.2, the final calcination step presents processing quantity limitations due to the

static nature of the calcination environment. With the cause of this limitation

identified, future work would then involve the design of a graphite and/or

refractory based furnace with a dynamic calcination environment.

7.2.2 EFFECT OF TEMPERATURE ON STOICHIOMETRY

As reported in Chapter 6, a significant difference in the stoichiometry of solution-

based and commercially synthesised boron carbide powders was detected through

the accurate measurement of unit cell parameters. Given that solution-based

boron carbide samples experience a much lower calcination temperature (1400

°C) compared to that of commercially synthesised samples (>1750 °C), it can be

inferred that calcination temperature may affect the final stoichiometry obtained

in the boron carbide product. This finding is corroborated by the DFT calculations

of Widom et al.[1], wherein carbon substitution within the boron icosahedra is

predicted to decrease with increasing temperature. However, given the small

sample set available in this work, further experiment is required to confirm this

effect. If found to be a legitimate process affecting boron carbide composition, it

would potentially provide a simplistic method to tailor boron carbide

stoichiometry in high purity. This ability would be highly practicable for commercial

application of boron carbide.

Confirmation of this effect would involve the synthesis of high purity boron carbide

through the sucrose-based technique developed here, followed by subsequent

further calcination at stages of gradually increasing temperatures and dwell times.

180

Measurement of the calcined boron carbide unit cell dimensions would then be carried out using the FIPB XRD technique developed here (Chapter 6) at each stage of calcination. Given the exceptional purity and stoichiometric consistency in the starting boron carbide material, the effect of stoichiometry on calcination temperature and time could then be accurately quantified. This process, in conjunction with a second analytical technique for stoichiometric measurement, could also be used to further the work first pioneered by Bouchacourt et al.[2] and

Aselage et al.[3]. The probable calcination temperature effect on stoichiometry in combination with the ability to accurately measure the structural parameters of low MAC materials could lead to increased accuracy in quantification of the unit cell/stoichiometry relationship. Increased accuracy in this trend would be highly relevant to further boron carbide development and technological application.

7.2.3 SINTERABILITY AND PHYSICAL PROPERTIES

As the end-use application of boron carbide generally involves solid bodies, the sinterability of produced powders is of high importance. Future work in this area would involve comprehensive review of available literature pertaining to sintering of boron carbide powders. Based on the optimum identified parameters in both sintering method and requirements (equipment, particle size, sintering aids etc.), appropriate boron carbide powders could be synthesised using a high degree of control in morphology and purity afforded by the sucrose-based technique.

Subsequent sintering to solid bodies and physical properties measurement would then be conducted. Given the enhanced purity and stoichiometric consistency of powders generated by the sucrose-based method, substantial improvements in

181 physical properties associated with the B4C phase would be expected. Through

stoichiometric control of the boron carbide phase and the enhanced structural

characterisation provided by the FIPB XRD technique, a wide range of different

physical properties could be investigated, with their relationship to stoichiometry

more accurately determined.

Preliminary work in this area has already been conducted through spark plasma

sintering (SPS) of B4C synthesised by the sucrose-based technique and commercial

boron carbide samples (the same samples analysed in Chapter 6). This work, along

with physical properties measurement of the sintered bodies, was carried out in

collaboration with P. Sauerschnig, J. B. Vaney and T. Mori at the National Institute

for Materials Science, Japan[4]. This preliminary work has shown good sinterability

of the sucrose-based B4C with a relative density of 99.6% achieved at 1900 °C with

a 10 min holding time[4]. The sintering was conducted without the use of an

optimised particle size distribution or sintering aids. Therefore, improvements in

these values may be possible. The electrical and thermal properties of the sintered

sucrose-based B4C and commercial boron carbide samples were measured for

comparison of their thermoelectric conversion efficiency, of which the sucrose-

based B4C performed the worst.[4] Given the fact that the carbon-deficient B6.5C

phase has been shown to have highest thermoelectric efficiency across the

stoichiometric range[5], it is no surprise that the high purity B4C presents the

lowest figure of merit. This data also further validates the presence of the carbon-

deficient phases detected in the commercial samples by the FIPB XRD technique

182

in Chapter 6. Research collaboration in this area is ongoing, with the results obtained to date to be published in a peer-reviewed journal.

183 7.3 REFERENCES

1. Widom, M. & Huhn, W. P., Prediction of orientational phase transition in boron carbide. Solid State Sciences, 2012. 14: p. 1648-1652. 2. Bouchacourt, M. & Thevenot, F., Analytical investigation in the B-C system. Journal of the Less-Common Metals, 1981. 82: p. 219-226. 3. Aselage, T. L. & Tissot, R. G., Lattice Constants of Boron Carbide. Journal of the American Ceramic Society, 1992. 75: p. 2207-2212. 4. Sauerschnig, P., Watts, J. L., Vaney, J. B., Talbot, P. C., Alarco, J. A., Mackinnon, I. D. R. and Mori, T., Thermoelectric properties of pure B4C prepared by a solution- based method. Advances in Applied Ceramics, 2018. (to be submitted) 5. Bouchacourt, M. & Thevenot, F., The correlation between the thermoelectric properties and stoichiometry in the boron carbide phase B4C-B10.5C. Jounal of Materials Science, 1985. 20: p. 1237-1247.

184

QUT Verified Signature

QUT Verified Signature QUT Verified Signature

QUT Verified Signature