COPYRIGHT AND CITATION CONSIDERATIONS FOR THIS THESIS/ DISSERTATION

o Attribution — You must give appropriate credit, provide a link to the license, and indicate if changes were made. You may do so in any reasonable manner, but not in any way that suggests the licensor endorses you or your use.

o NonCommercial — You may not use the material for commercial purposes.

o ShareAlike — If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original.

How to cite this thesis

Surname, Initial(s). (2012). Title of the thesis or dissertation (Doctoral Thesis / Master’s Dissertation). Johannesburg: University of Johannesburg. Available from: http://hdl.handle.net/102000/0002 (Accessed: 22 August 2017). Char extracted from coal ash as a replacement for natural graphite – “Charphite”

By

Charlotte Badenhorst

Thesis

Submitted in fulfilment of the requirements for the degree

of

PHILOSOPHIAE DOCTOR

In

Geology

in the

Faculty of Science

at the

University of Johannesburg, South Africa

Supervisor: Prof. N.J. Wagner

Co-supervisor: Prof. B.R.V. Valentim

Co-supervisor: Prof. K.S. Viljoen

November 2019 Acknowledgements

This work is based on research supported by the National Research Foundation (NRF) of South Africa (Grant number 103466) and the Department of Science and Technology (DST)-NRF ERA-MIN grant for the Charphite project. The author also acknowledges DST-NRF Centre of Excellence for Integrated Mineral and Energy Resource Analysis (CIMERA) for additional financial assistance. K.S. Viljoen acknowledges financial support from the South African Department of Science and Technology through their Research Chairs initiative (Geometallurgy), as administered by the National Research Foundation.

Prof. Nikki Wagner for allowing me to add my own special touches to my work, for minding my commas and semicolons, and for the opportunity to be supervised by one of the greatest.

Prof. Bruno Valentim for encouraging me when the news was bad and for your always insightful inputs (from deciding on a colour scheme for my posters to complex scientific discussions on my papers!).

Prof. Fanus Viljoen for your calm and collective presence, for your patience with all my queries and uncertainties, and for sharing your immense knowledge on everything and anything.

The third ERA-MIN collaboration partners for allowing a rookie to be part of such a brilliant, experienced, and cultural diverse team. It was an honour to be working with you.

UJ’s Geology Department, CIMERA, and Spectrau Laboratory for welcoming a non-geologist into your midst. After three years with you, I feel as old and wise as the world itself! A special thanks to everybody who contributed in some way to my research.

The Coal Research Group for our bi-weekly discussions on all things coal, for our highly entertaining World Coal magazine talks, for our “safe environment” dry run presentations, and for growing from coal students into coal experts.

Eskom, Sasol, and Jonkel Carbons and Grafites for supplying me with much needed samples and for dealing with all the red tape, for supplying me with even more samples when I miscalculated the extend of my research (special thanks to Dr. Kelley Reynolds- Clausen), and for tolerantly answering my sometimes annoying questions.

i | P a g e

South African Coal Ash Association for the opportunity to network and share my ideas and for your financial contribution to our (very expensive!) trip to the 2019 World of Coal Ash Conference in the United States of America (USA). What an experience! Thank you.

Mfesane Tshazi and Dr. Natasia Naudé from the University of Pretoria for usage of your CoronaStat electrostatic separator. Willem Swanepoel and Sean Harbinson from Bureau Veritas Testing and Inspections South Africa for allowing me to spend the day in your labs. Sabine Verryn from XRD Analytical & Consulting for our discussion on XRD structural characterisation. MAK Analytical for thousands of LOI trials. Betachem for kindly providing the reagents used in the froth flotation experiments.

To my family and friends. My mom for raising us on coffee (it came in handy this time). My dad for knowing more about my projects than I do (after 28 years I still think my dad is the smartest person alive). My siblings (four sisters and a not-so-little-anymore brother) for my stories. Jaganmoy Jodder for sharing this small office space with me and for showing me what kindness really means. For all the great people I have met on this journey - You have made me realise how much I still have to learn.

ii | P a g e

Abstract This thesis forms part of the third ERA-MIN collaboration between Portugal, Poland, Romania, Argentina, and South Africa under the project Charphite. The overall aim of the Charphite project was to determine if char found in coal ash can be used as a substitute replacement for natural graphite in green energy applications, including direct use of char in the electrocatalytic oxygen reduction reaction and the hydrogen oxidation reaction.

Ash is a waste product resulting from the combustion of coal. The landfilling or ponding of ash can lead to serious environmental and health concerns, and therefore the utilisation of ash / certain components in the ash is desired. Literature shows that South African coal ash samples (>50 mill ton ash per annum) contain 0.5 to 8 % carbon in ash (chars) and also a significant amount of unburned carbon is associated with carbonaceous shale containing <10 % carbon and >90 % mineral matter. Char in coal ash has a high degree of structural order and can possibly be used as a substitute for natural graphite. The European Union, the United States, and the United Kingdom have all listed natural graphite as one of their raw critical commodities; hence the project has significant merits. Research on producing synthetic graphite from char in coal ash is scarce, with only a limited amount of papers being published as yet.

Due to the extensive nature of the ERA-MIN collaboration, only the results pertaining to the South African involvement are presented in this thesis (see https://www.fc.up.pt/charphite/ for overall results). The characterisation of South African coal ash sources, the separation of char from these sources, and the characterisation / evaluation of the extracted char as a possible precursor for synthetic graphite are presented. A desktop study on natural graphite occurrences in southern Africa, and characterisation of selected samples are also provided.

A combination of size, electrostatic, and magnetic separation steps were used for char extraction. All carbon percentages were determined with the loss on ignition method. The final product grades ranged between 45 and 66 wt. % carbon. A density separation step was added to one of the fly ash samples and a final grade of 83 wt. % of carbon was achieved. Although a significant increase from the initial char in ash percentages was obtained, the carbon recoveries were unfortunately very low (6 to 32 %). Due to the low starting carbon in ash percentages, a significant amount has to be recovered to make economic sense. The reason for the low carbon recoveries might be due to small, unliberated ash minerals that formed part of the char matrix.

The evaluation of the extracted chars (proximate and ultimate analyses, carbon forms, X-ray diffraction mineralogy and structure analyses, petrography and reflectance analyses, and

iii | P a g e

Raman microspectroscopy analysis) showed the presence of strong carbon-carbon bonds, similar to those found in graphite, and limited impurities (oxygen, nitrogen, sulphur, and hydrogen). The anisotropy percentages of the samples ranged between 22 and 49 %; the reference natural graphite sample had an anisotropy of 86 %. The three-dimensional structure of the chars can be described as turbostratic, with randomly orientated carbon layers, small graphite crystallite sizes, and large interlayer spacings. Raman microspectroscopy classified the chars as being “transitional” (being somewhere between amorphous and graphitic carbon) with the possibility as a precursor for synthetic graphite.

iv | P a g e

Contribution to science Journal publications published

Badenhorst, C.J., Wagner, N.J., Valentim, B.R.V., Viljoen, K.S., Santos, A.C., Guedes, A., 2019. Separation of unburned carbon from coal conversion ash: Development and assessment of a dry method. Coal Combustion and Gasification Products 11, 89-96. https://doi.org/10.4177/CCGP-D-19-00002.1

Journal publications in preparation

Badenhorst, C.J., Wagner, N.J., Valentim, B.R.V., Viljoen, K.S., 2019. A review of natural graphite occurrences and mining in southern Africa. Journal of African Earth Sciences.

Badenhorst, C.J., Santos, A.C., Abagiu, T.A., Białecka, B., Całus-Moszko, J., Cruceru, M., Guedes, A., Lázaro-Martínez, J.M., Moreira, K., Popescu, L.G., Predeanu, G., Viljoen, K.S., Valentim, B.R.V., Wagner, N.J., 2019. Recovery of char concentrates for graphite substitution. Minerals.

International conferences

Badenhorst, C., Wagner, N., Valentim, B., Viljoen, F., 2018. Fifty shades of grey: The extraction of char from a variety of coal ash for consideration as synthetic graphite, in: 70th Annual Meeting of the International Committee of Coal and Organic Petrology (ICCP) extended abstracts. ICCP, Brisbane. https://www.iccop.org/documents/70th- iccp-meeting-program-and-abstracts.pdf/

Badenhorst, C., Wagner, N., 2018. Coal char as a substituting material of natural graphite in green energy technologies – Charphite (ERA-MIN meeting feedback presentation, Târgu Jiu).

Badenhorst, C.J., Wagner, N.J., Valentim, B.R.V., Santos, A.C., Guedes, A., Białecka, B., Całus, J., Popescu, L.G., Cruceru, M., Predeanu, G., Viljoen, K.S., Lázaro-Martínez, J.M., Abagiu, T.A., 2019. Char from coal ash as a possible precursor for synthetic graphite – Recent developments of the Charphite project, in: World of Coal Ash (WOCA) Conference proceedings papers. University of Kentucky Center for Applied Energy Research, St. Louis, MO. http://www.flyash.info/

v | P a g e

Local conferences / colloquiums / technical talks

Badenhorst, C., 2017. Char extracted from coal ash as a replacement for natural graphite – Charphite. (Paleoproterozoic Mineralization (PPM) Research Group Technical Talk, Johannesburg).

Badenhorst, C., 2017. Char extracted from coal ash as a replacement for natural graphite – Charphite, in: Coal, carbon, energy and the environment – National student colloquium 2017. University of the Witwatersrand, Johannesburg.

Badenhorst, C., Wagner, N., 2017. Char extracted from coal ash as a replacement for natural graphite – Charphite, in: The Fossil Fuel Foundation Conference on Sustainable Development of South Africa’s Energy Sources extended abstracts. The Fossil Fuel Foundation, Johannesburg.

Badenhorst, C., 2018. Fifty Shades of Grey – Extraction of Remnant Carbon from Coal Ash. (South African Coal Ash Association Technical Talk, Johannesburg).

Badenhorst, C., Wagner, N., Viljoen, F., Valentim, B., 2018. Char extracted from coal ash as a replacement for natural graphite – “Charphite”, in: Geoncongress 2018 extended abstracts. Geological Society of South Africa, Johannesburg.

Badenhorst, C., 2018. Fifty Shades of Grey – Extraction of Remnant Carbon from Coal Ash, in: Annual Postgraduate Research Conference extended abstracts. University of Johannesburg, Johannesburg.

International posters / multimedia

Badenhorst, C., 2018. Fifty shades of ash. (Visualise Your Thesis Competition, Melbourne). https://research.unimelb.edu.au/visualiseyourthesis/showcase.

Local posters / multimedia

Badenhorst, C., Wagner, N., Viljoen, F., Valentim, B., 2018. Char extracted from coal ash as a replacement for natural graphite – “Charphite”, in: Geoncongress 2018 extended abstracts. Geological Society of South Africa, Johannesburg.

Badenhorst, C., 2018. Fifty shades of ash. (Visualise Your Thesis Competition, Johannesburg). (Showcase winner).

vi | P a g e

Table of contents

Acknowledgements ...... i Abstract ...... iii Contribution to science ...... v Table of contents ...... vii List of Figures ...... xi List of Tables ...... xiv Abbreviations and acronyms ...... xvi List of symbols ...... xvii List of amorphous materials and minerals...... xviii Sample nomenclature ...... xix Chapter 1: Introduction ...... 1 1.1 What did we do? ...... 1 1.2 Why did we do this? ...... 4 1.3 How did we approach this thesis? ...... 6 1.3.1 Scope ...... 6 1.3.2 Aim and objectives ...... 7 1.3.3 Thesis outline ...... 7 Chapter 2: Coal conversion ash in South Africa ...... 8 2.1 Introduction ...... 8 2.2 Coal and its usage in South Africa ...... 8 2.3 Coal conversion / ash producing utilities in South Africa ...... 12 2.3.1 Electricity generation...... 12 2.3.2 Synfuels ...... 17 2.4 Ash formation and management in South Africa ...... 18 2.4.1 Minerals in coal ...... 18 2.4.2 Ash formation mechanisms ...... 21 2.4.3 Ash management in South Africa ...... 24 2.4.4 Comparison with the global ash market ...... 28 2.5 Characterisation of South African coal and ash samples ...... 29 2.5.1 Methodology ...... 29 2.5.2 Coal characterisation results ...... 33 2.5.3 Ash characterisation results ...... 39 2.6 Summary ...... 50 Chapter 3: Char-ash separation and char characterisation ...... 51

vii | P a g e

3.1 Introduction ...... 51 3.2 Case studies ...... 51 3.2.1 Hwang et al. (2002) ...... 51 3.2.2 Cabielles et al. (2008) ...... 55 3.2.3 Maroto-Valer et al. (1999a) ...... 58 3.2.4 Other...... 62 3.3 Char-ash separation of South African ash samples ...... 63 3.3.1 Methodology ...... 63 3.3.2 Char-ash separation results ...... 69 3.4 Characterisation of char samples...... 80 3.4.1 The structural progression of carbon with temperature ...... 80 3.4.2 Methodology ...... 83 3.4.3 Char characterisation results ...... 89 3.5 Summary ...... 98 Chapter 4: Char concentrate potential to graphitize ...... 101 4.1 Introduction ...... 101 4.2 A brief summary on ERA-MIN coal and ash characterisation results ...... 101 4.2.1 Methodology ...... 101 4.2.2 Coal and ash characterisation results for ERA-MIN samples ...... 102 4.3 Char concentrate potential to graphitize ...... 107 4.3.1 Methodology ...... 107 4.3.2 Char characterisation results for ERA-MIN concentrates ...... 108 4.4 Summary ...... 114 Chapter 5: Natural graphite in southern Africa ...... 116 5.1 Introduction ...... 116 5.2 An introduction to natural graphite ...... 116 5.3 Natural graphite occurrences in southern Africa ...... 120 5.4 South Africa ...... 121 5.4.1 Northern Cape Province ...... 121 5.4.2 Western Cape Province ...... 122 5.4.3 KwaZulu-Natal Province ...... 122 5.4.4 Mpumalanga Province ...... 123 5.4.5 Free State Province ...... 124 5.4.6 North West Province ...... 124 5.4.7 Limpopo Province ...... 125 5.4.8 Summary of mining potential in South Africa ...... 127

viii | P a g e

5.5 Swaziland ...... 127 5.6 Lesotho ...... 128 5.7 Namibia ...... 129 5.7.1 !Karas region ...... 129 5.7.2 Erongo region ...... 130 5.7.3 Khomas region ...... 131 5.7.4 Hardap region ...... 131 5.7.5 Otjozondjupa region ...... 131 5.7.6 Summary of mining potential in Namibia ...... 132 5.8 Botswana ...... 132 5.8.1 Central District ...... 132 5.8.2 Southern District ...... 134 5.8.3 Summary of mining potential in Botswana ...... 134 5.9 Zimbabwe ...... 134 5.9.1 Matabeleland North Province ...... 135 5.9.2 Mashonaland West Province ...... 135 5.9.3 Midlands Province ...... 136 5.9.4 Harare Province ...... 138 5.9.5 Manicaland Province ...... 138 5.9.6 Masvingo Province ...... 138 5.9.7 Bulawayo Province ...... 138 5.9.8 Summary of mining potential in Zimbabwe ...... 138 5.10 Mozambique ...... 138 5.10.1 Tete Province ...... 139 5.10.2 Manica Province ...... 139 5.10.3 Sofala Province ...... 140 5.10.4 Zambezia Province ...... 140 5.10.5 Nampula Province ...... 140 5.10.6 Niassa Province ...... 140 5.10.7 Cabo Delgado Province ...... 140 5.10.8 Summary of mining potential in Mozambique ...... 142 5.11 Madagascar ...... 142 5.11.1 Summary of mining potential in Madagascar...... 144 5.12 Characterisation of natural graphite from two localities in southern Africa ...... 145 5.12.1 Methodology ...... 145 5.12.2 Natural graphite characterisation results ...... 146

ix | P a g e

5.13 Summary...... 157 Chapter 6: Summary, conclusions and recommendations ...... 159 6.1 Summary ...... 159 6.2 Conclusions ...... 159 6.3 Recommendations ...... 161 References ...... 163 Appendix A – Automated SEM methodology and results ...... 205 Appendix B – Raman microspectroscopy curve fitting procedure ...... 214 B.1 Char samples ...... 214 B.2 Graphite samples ...... 217

x | P a g e

List of Figures

Figure 1.1: Partners that were involved in the third ERA-MIN collaboration...... 2 Figure 1.2: Methodology used in the third ERA-MIN collaboration, with components included in this thesis indicated (adapted from Valentim, 2015)...... 3 Figure 1.3: Critical raw materials identified by the European Union (adapted from European Commission, 2017)...... 5 Figure 2.1: Coalfields of South Africa (adapted from Jeffrey, 2005)...... 9 Figure 2.2: Map indicating Eskom coal-fired power station locations (adapted from eNCA, 2014)...... 12 Figure 2.3: Schematic of a pulverised coal combustion process (adapted from Tishmack and Burns, 2004)...... 15 Figure 2.4: Schematic of a Sasol-Lurgi fixed bed dry bottom gasification process (Van Dyk et al., 2006)...... 17 Figure 2.5: Inorganic modes of occurrence in a coal conversion feed stream (Benson et al., 1993)...... 19 Figure 2.6: Behaviour of included pyrite upon heating (McLennan et al., 2000)...... 20 Figure 2.7: Excluded minerals’ transformation mechanisms in coal (adapted from Van Alphen, 2005)...... 21 Figure 2.8: Ash formation mechanisms of included, excluded, and organically associated inorganic minerals (adapted from Krishnamoorthy and Pisupati, 2015)...... 22 Figure 2.9: Ash tonnages produced, saleable, and sold by individual Eskom power stations (adapted from Reynolds-Clausen and Singh, 2019)...... 26 Figure 2.10: Inertodetrinite found in the coal samples. A) Vitrinite stringers embedded within an inertodetrinite-rich particle, B and C) Variable grain sizes in an aggregated inertodetrinite- rich particle, D) Silicate minerals (quartz and clay) included in a inertodetrinite-rich particle (Reflected-light, oil immersion, ×500)...... 39 Figure 2.11: Quartz found in the coal samples. A) Massive quartz particle found in C PS4, B) Fine quartz fragments embedded in a massive clay particle (Reflected-light, oil immersion, ×500)...... 39 Figure 2.12: Visual examination of ash samples. A) Typical fly ash sample, B) FA PS4 with large, dark char particles clearly visible, C) Typical bottom ash sample, D) Typical gasification ash sample, E) Unreacted coal particles in the gasification ash, F) Char-ash particles in the gasification ash samples burned according to the “shrinking core model”. .... 41 Figure 2.13: Amorphous particles in the ash samples. A) Typical glass particle found in South African ash samples, angular and “cloudy;, B) Baked clay particle with a metakaolinite centre and a glassy rim; C) Typical glass particle for Portuguese ash sample (ERA-MIN consortium example), plerosphere; D) Typical glass particle for Portuguese ash sample (ERA-MIN consortium example), cenosphere (Reflected-light, oil immersion, ×500)...... 45 Figure 2.14: A) Needle-like mullite crystals embedded in a alumino-silicate glass matrix, B) Elongated anorthite crystals observed in the bottom and gasification ash samples (Reflected-light, oil immersion, ×500)...... 46 Figure 2.15: A) and B) Dense ferrospheres in ash samples; C) and D) Dendritic ferrospheres in ash samples; E) Dense ferrosphere found in GA East; F) Dense iron fragments mantling a glass particle (Reflected-light, oil immersion, ×500)...... 47 Figure 2.16: Particle size distributions (PSDs) for A) Fly ash samples, B) Gasification ash samples, C) Bottom ash samples 1, 3 and 4, D) Bottom ash samples 2 and 5...... 49 Figure 3.1: Char-ash separation schematic followed by Hwang et al. (2002)...... 52 Figure 3.2: Char-ash separation schematic followed by Cabielles et al. (2008)...... 57

xi | P a g e

Figure 3.3: Char-ash separation schematic by Maroto-Valer et al. (1999a)...... 60 Figure 3.4: Separation process flow diagram followed in this thesis (excluding froth flotation as this step was not efficient and therefore discarded in end method)...... 66 Figure 3.5: OreKinetics CoronaStat used in the electrostatic separation trials...... 67 Figure 3.6: A) Magnetic material in ash residue after LOI tests on electrostatic separated samples and B) Magnetic material collected with a hand held magnet...... 68 Figure 3.7: Typical char particle found in the ash samples. Included minerals finely disseminated and interwoven into the char matrix can clearly be seen (BA PS4 illustrated) (Reflected-light, oil immersion, ×500)...... 70 Figure 3.8: Mineral deportment categories (Kelly and Spottiswood, 1982)...... 70 Figure 3.9: Milling of char particles will only lead to the “mini-me” effect pertaining. Even milling to particle sizes as small as A) 100 µm and B) 50 µm will only result in a replicate of the original particle forming (BA PS4 illustrated) (Reflected-light, oil immersion, ×500)...... 71 Figure 3.10: Typical char particle found in the Portuguese ash sample. The chars were porous and infilled with small glassy inclusions (Reflected-light, oil immersion, ×500)...... 71 Figure 3.11: Visual confirmation of char enrichment based on particle size (FA PS4 illustrated)...... 72 Figure 3.12: Particle size versus LOI (wt. %) for the fly ash samples...... 72 Figure 3.13: Cumulative carbon grade, carbon recovery, and sample yield curves for A) FA PS4 and B) FA PS2 at different particle sizes...... 73 Figure 3.14: Conductivity versus LOI (wt. %) for the ash samples...... 74 Figure 3.15: Cumulative carbon grade, carbon recovery, and sample yield curves for A) FA PS4; B) FA PS2; C) BA PS4; and D) GA East at different conductivity bins...... 75 Figure 3.16: Structural progression of carbon with temperature (adapted from Bourrat, 2000)...... 81 Figure 3.17: The crystallography of graphite (Pierson, 1993)...... 83 Figure 3.18: Carbon form analysis diagram (ACT-TPM-028)...... 85 Figure 3.19: ICCP char classification (adapted from Wagner et al., 2018)...... 88 Figure 3.20: Diffractograms of FA PS4 CC and FA PS2 CC...... 91 Figure 3.21: Typical Raman microspectroscopy spectrum obtained in the char concentrates: A) 1st and 2nd order spectra and B) 1st order spectrum close-up...... 92 Figure 3.22: Raman 1st order spectrum curve fitting example for char concentrates...... 93 Figure 3.23: A) Anisotropic char and B) Isotropic char. Anisotropic char is distinguished from isotropic char by a spot colour change with a 90° turn of the analyser (reflected-light, ×500, oil objective)...... 97 Figure 3.24: A) Inertinitic (top) and mixed porous (bottom) char particle and B) Mixed porous char particle with a glassy (rare) inclusion (reflected light, ×500 oil objective)...... 98 Figure 4.1: A) A high percentage of crassispheres was observed in the Portuguese char, indicating possible application in fluid adsorption, B) Large network and mixed porous particles were seen in the Polish char, C) Mixed porous particles with rare glassy inclusions were seen in the South African char, D) The Romanian char sample had a higher percentage of fusinoid / solid and was also fragmented due to pulverisation during beneficiation (Reflected-light, oil immersion, ×500)...... 112 Figure 4.2: Typical 1st order Raman microspectra obtained from the anisotropic char particles from Portugal, Poland, and South Africa (A), the isotropic char particles from Romania (B), the Carbon I particles from Romania (C), and the Carbon II particles from Romania (D)...... 113 Figure 5.1: African countries with known natural graphite occurrences as well as African countries with known natural graphite occurrences to be discussed in text...... 120

xii | P a g e

Figure 5.2: Geological map of South Africa with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008)...... 121 Figure 5.3: Geological map of Swaziland with major graphite occurrence and associated lithologies indicated (map modified from Schlüter, 2008)...... 128 Figure 5.4: Geological map of Lesotho with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008)...... 128 Figure 5.5: Geological map of Namibia with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008)...... 129 Figure 5.6: Geological map of Botswana with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008)...... 133 Figure 5.7: Geological map of Zimbabwe with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008)...... 134 Figure 5.8: Geological map of Mozambique with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008)...... 139 Figure 5.9: Geological map of Madagascar with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008)...... 143 Figure 5.10: A collection of minerals found in the Zim Graphite sample: A) and B) Intense reddening can be ascribed to amorphous Fe-oxide (Geuna et al., 2015); C) Graphite embedded in an amorphous Fe-oxide phase; D) Honeycomb mineral (left), graphite (middle), and quartz (right)(Reflected-light, oil immersion, ×500)...... 148 Figure 5.11: Diffractogram for demineralised SA Graphite...... 149 Figure 5.12: Diffractogram for demineralised SA Graphite highlighting the occurrence of rhombohedral graphite...... 150

Figure 5.13: Carbonaceous material classification based on d002 and Lc (adapted from Tagiri and Oba, 1986)...... 152 Figure 5.14: Raman microspectroscopy spectra for the natural graphite samples. A) Very ordered graphite; B) Moderately ordered graphite, and C) disordered graphite...... 153 Figure 5.15: A collection of flake graphite particles found in the SA Graphite sample: A) Coarse flake graphite (left) and quartz (right) particles; B) Coarse flake graphite particle; C) Basal plane view of a flake graphite particle; D) Agglomerated graphite flakes with a quartz inclusion (left); E) Agglomerated graphite flakes (basal plane); F) Variety of coarse flake graphite particles (Reflected-light, oil immersion, ×500)...... 155 Figure 5.16: A collection of flake graphite particles found in the Zim Graphite sample: A) Small graphite flakes; B) Small graphite flakes embedded within an amorphous Fe-oxide (Reflected-light, oil immersion, ×500)...... 156 Figure A.1: Phase legend for automated SEM images...... 208 Figure A.2: Massive quartz particle with mullite, kaolinite, glass, calcite, and carbon inclusions...... 208 Figure A.3: Glass particle with large, elongated anorthite needles...... 209 Figure A.4: Glass particle with anorthite needle inclusions...... 209 Figure A.5: Quartz/kaolinite/glass/mullite particle...... 210 Figure A.6: Carbon particle with mineral inclusions...... 212 Figure A.7: Fractional distribution for different liberation classes...... 212 Figure A.8: Fractional distribution for different free surface classes...... 213 Figure B.1: A) Baseline subtraction; B) Curve fitting using the method from Sadezky et al. (2005); C) Curve fitting method followed in this study; D) Curve fitting method followed in this study for disordered carbon. (Actual curve=black, fitted curve=red, G band=light blue, D1 band=green, D2 band=orange, D3 band=dark blue, D4 band=purple)...... 216 Figure B.2: Raman peak fitting example (SA Graphite illustrated)...... 217

xiii | P a g e

List of Tables

Table 2.1: Quality, usage, reserve, and major collieries of South Africa’s 19 coalfields...... 10 Table 2.2: General information on Eskom coal-fired power stations...... 13 Table 2.3: Technical information on South African coal-fired power stations...... 16 Table 2.4: Percentage of saleable ash being sold by Eskom power stations...... 27 Table 2.5: Information on ash quality and disposal methods adopted by Eskom power stations...... 27 Table 2.6: CCP volumes produced and utilised globally (Harris et al., 2019)...... 29 Table 2.7: Sample identification, type, origin, mass, sampling time, and unit for the ash and coal samples...... 30 Table 2.8: Coal and ash characterisation techniques used in the current study...... 31 Table 2.9: Proximate, CV, ultimate, and X-ray fluorescence standards used for the analyses...... 31 Table 2.10: Modified Hower (2012) ash petrographic classification scheme...... 33 Table 2.11: Proximate, CV, and ultimate analyses results for the coals (dry ash free basis abbreviates to d.a.f)...... 34 Table 2.12: XRF analysis results for the ash samples of coal samples and XRD analysis results for the coal samples...... 36 Table 2.13: Petrographic analyses results for the coal samples...... 38 Table 2.14: Proximate, XRF, LOI, XRD, and petrographic analyses results for the ash samples...... 42 Table 3.1: Char-ash separation results produced by Hwang et al. (2002)...... 55 Table 3.2: Char-ash separation results produced by Cabielles et al. (2008)...... 58 Table 3.3: Char-ash separation results produced by Maroto-Valer et al. (1999a) (using petrography)...... 61 Table 3.4: Properties of samples selected for the char-ash separation trials...... 63 Table 3.5: Summary of carbon in ash measurement techniques...... 65 Table 3.6: Carbon grades, carbon recoveries, and sample yields at selected size cut-points for fly ash samples...... 74 Table 3.7: Carbon grades, carbon recoveries, and sample yields at selected conductivity cut- points...... 76 Table 3.8: Carbon grades, carbon recoveries, and sample yields at selected conductivity cut- points, 2nd stage electrostatic separation...... 76 Table 3.9: Carbon grades, carbon recoveries, and sample yields for the non-magnetic fractions...... 78 Table 3.10: Carbon grades, carbon recoveries, and sample yields for the density cut-point (FA PS4)...... 78 Table 3.11: Summary on the char-ash separation: Carbon grades, carbon recoveries, and sample yields...... 80 Table 3.12: Sample information of char concentrates (CC) for further characterisation...... 83 Table 3.13: Char concentrate characterisation techniques...... 84 Table 3.14: Proximate, ultimate, and carbon form standards used on the char concentrates...... 84 Table 3.15: Proximate, ultimate (including atomic H/C and O/H ratios), carbon form, and XRD mineralogy analyses results for FA PS4 CC...... 90 Table 3.16: X-ray diffraction structural results for FA PS4 CC and FA PS2 CC...... 92

xiv | P a g e

Table 3.17: Summary on 1st Order Raman microspectroscopy bands fitted in this study for the char samples...... 94 Table 3.18: Quantitative Raman microspectroscopy curve fitting results for the char concentrates...... 95 Table 3.19: Petrographic analyses results for the char concentrates...... 96 Table 4.1: Information of ERA-MIN ash and coal samples...... 101 Table 4.2: Proximate, XRF, XRD, and PSD analyses results for the ERA-MIN ash samples, proximate and ultimate analyses results for the coal samples, XRF analysis results for the high temperature ashes of the coal samples, and XRD analysis results for the LTA ashes of the coal samples...... 103 Table 4.3: Petrographic results for the ERA-MIN coal samples...... 106 Table 4.4: Petrographic results for the ERA-MIN ash samples...... 107 Table 4.5: Sample information of ERA-MIN char concentrates for characterisation...... 107 Table 4.6: Proximate, ultimate, carbon form, and XRD mineralogy results for the ERA-MIN char concentrates...... 109 Table 4.7: Petrographic analysis results for the ERA-MIN char concentrates...... 111 Table 4.8: Raman microspectroscopy curve fitting results for the ERA-MIN char samples. 115 Table 5.1: Properties of flake, amorphous, and vein natural graphite (compiled from Feytis, 2010; Fogg and Boyle, 1987; Krauss et al., 1988; Luque et al., 2014; Mitchell, 1993; Otto, 2011; Simandl et al., 2015)...... 118 Table 5.2: Natural graphite price (U.S. $ per tonne) for different flake sizes with a 95 % purity (2016-2025) (Spencer and Hill, 2016)...... 119 Table 5.3: Natural graphite producing countries and production volumes in 2018 (U.S. Geological Survey, 2019)...... 119 Table 5.4: Graphite occurrences north of the Soutpansberg Mountain Range, Limpopo Province, South Africa (Wilke, 1969)...... 126 Table 5.5: Summary of graphite occurrences in the Mashonaland West Province, Zimbabwe (Muchemwa, 1987)...... 137 Table 5.6: Natural graphite sample nomenclature, descriptions, and other information...... 145 Table 5.7: Natural graphite characterisation techniques used in the current study...... 146 Table 5.8: Proximate, XRF, and carbon form analyses results for the natural graphite samples...... 147 Table 5.9: XRD mineral identification for natural graphite samples...... 149 Table 5.10: X-ray diffraction structural results for SA Graphite...... 152 Table 5.11: Quantitative Raman microspectroscopy curve fitting results for the natural graphite samples...... 154 Table 5.12: Petrography reflectance analysis results for SA Graphite sample...... 157 Table A.1: Automated SEM instrument specifications...... 205 Table A.2: Ash phase identification via automated SEM...... 206 Table A.3: Elemental assay and elemental distribution, automated SEM...... 207 Table A.4: Phase association / locking data...... 211 Table B.1: Restrictive bounds added to the char Raman microspectopy curve fitting procedure...... 215

xv | P a g e

Abbreviations and acronyms

Abbreviation / Acronym

2H Hexagonal graphite 3R Rhombohedral graphite a.d.b Air dried basis AEA Air-entraining agents ASTM American Society for Testing and Materials BSU Basic structural units C Conductive fraction CCP Coal conversion products CTL Coal to liquid CV Calorific Value d.a.f Dry ash free basis DGC Density Gradient Centrifugation DMS Dense Medium Separation Eskom Electricity Supply Commission ESP Electrostatic precipitator EV Electric vehicles EU European Union FBDB Fixed bed dry bottom FFP Fabric Filter Precipitators FWHM Full width at half maximum GD Graphitization degree ICCP International Committee for Coal and Organic Petrology INCAR El Instituto Nacional del Carbón IRP Integrated Resource Plan ISO International Organization for Standardization HER Hydrogen evolution reaction Li-ion Lithium-ion LOI Loss on ignition LST Lithium heteropolytungstate LTA Low-temperature ash M1/M2/M3 Middlings fractions NC Non-conductive fraction OER Oxygen evolution reaction ORR Oxygen reduction reaction PCC Pulverised coal combustion PSD Particle size distribution Sasol South African Synthetic Oil Liquid SEM Scanning Electron Microscopy ss-NMR solid-state Nuclear Magnetic Resonance U.S. United States USA United States of America XRD X-ray diffraction XRF X-ray fluorescence

xvi | P a g e

List of symbols

Symbol

d002 Interlayer spacing K Constant = 1

La Crystallite width

Lc Crystallite height RA1 Raman area ratio 1 disordered carbonaceous material RA2 Raman area ratio 2 disordered carbonaceous material

R2 Raman area ratio for ordered carbonaceous material

Rmin Minimum reflectance

Rmean Mean reflectance

Rmax Maximum reflectance 휃 Bragg’s angle 훽 Full width at half maximum 휆 X-ray wavelength

xvii | P a g e

List of amorphous materials and minerals

Mineral

Amorphous material C (amorphous carbon) and Al2O3.2SiO2 (metakaolinite)

Anorthite CaAl2Si2O8

Bassanite CaSO4∙0.5H2O/2CaSO4∙H2O

Calcite CaCO3

Cristoballite SiO2

Dolomite CaMg(CO3)2

Gypsum CaSO4∙2H2O

Hematite Fe2O3

Kaolinite Al2Si2O5(OH)4 Lime CaO

Magnetite Fe3O4

Mullite 3Al2O32SiO2 / 2Al2O3SiO2

Muscovite KAl2(AlSi3O10)(FOH)2 Periclase MgO

Pyrite FeS2

Quartz SiO2

Siderite FeCO3

Sillimanite Al2SiO5

xviii | P a g e

Sample nomenclature

Nomenclature

FA Fly ash BA Bottom ash GA Gasification ash C Coal PS Power station CC Char concentrate

xix | P a g e

Chapter 1: Introduction

1.1 What did we do? This thesis forms part of the third ERA-MIN collaboration (2015) on Sustainable Supply of Raw Materials in Europe: Coal char as a substituting material of natural graphite in green energy technologies – Charphite. Partners from Portugal, Poland, Romania, Argentina, and South Africa were involved in this collaboration (Figure 1.1). The aim was to suggest a possible substitute for natural graphite (a global critical raw commodity) in green energy technologies, by forming synthetic graphite from char found in coal ash. Seeing that coal ash is a waste product, this collaborative project, therefore, had both economic and environmental merit.

The Charphite project was divided into four Work Packages (Figure 1.2):

 Work Package 1 concerned the obtainment and characterisation of coal ash and natural graphite samples from the individual consortium countries;  Work Package 2 concerned the extraction of char from the obtained ash samples. As part of this Work Package, the ferrous fraction was simultaneously beneficiated with the char, and the discarded ash was assessed for usage in cement, concrete and ceramic applications. The extracted char was graphitized to form Charphite. Graphitization trials were conducted by the Spanish consultant: El Instituto Nacional del Carbón (INCAR). The extracted char, ferrous fraction and formed Charphite were characterised;  In Work Package 3, a comprehensive electrochemical characterisation on Charphite, natural graphite, exfoliated graphite (exfoliated Charphite and natural graphite), and intercalated graphite (ferrous fraction intercalated with Charphite and natural graphite) took place, and  In Work Package 4, the graphite products mentioned above were immobilised on electrode surfaces and tested in energy-related reactions such as the hydrogen evolution reaction (HER), oxygen evolution reaction (OER), oxygen reduction

reaction (ORR), and water splitting (simultaneous H2 and O2) reactions.

Complete collaboration details and results can be accessed on the Charphite website (https://www.fc.up.pt/charphite/).

This thesis covers only part of the overall project (Figure 1.2) and focusses specifically on the South African output.

1 | P a g e

Poland • Główny Instytut Górnictwa • CARBO-GRAF Sp.o.o

Portugal (lead) Romania • University of Porto • University POLITEHNICA of Bucharest • Rede de Química e Tecnologia • University “Constantin Brâncuşi” of Târgu Jiu • Pegop power plant

Argentina • University of Buenos Aires South Africa • University of Johannesburg

Figure 1.1: Partners that were involved in the third ERA-MIN collaboration.

2 | P a g e

Sampling (2016)

Fly ash and bottom ash char Natural graphite Work Package 1

Residue re-use (2018) Beneficiation (2017) Characterisation (2016) Work Package 2

Metallic concentrates Char concentrates Work Package 3

Characterisation (2018 / 2019) Graphitization (2019) Work Package 4

This thesis Charphite

Technological assessment (2019)

Charphite / natural graphite, exfoliated products and ferrous composites

Immobilisation on electrode surfaces (2019)

Electrode materials

Figure 1.2: Methodology used in the third ERA-MIN collaboration, with components included in this thesis indicated (adapted from Valentim, 2015).

3 | P a g e

1.2 Why did we do this? Ash is described as a waste product resulting from the burning of coal. In South Africa alone, more than 50 million tonnes of coal ash are produced annually (Department of Environmental Affairs, 2018a). Globally, this figure amounts up to 780 million tonnes (Heidrich et al., 2013). The landfilling and ponding of ash lead to numerous problems. For example, in 2008 one of the largest ash spillage disasters occurred when a retaining pond wall at Kingston Fossil Plant (Tennessee) gave way, spilling gallons of wet coal ash into a nearby river (CBS News, 2008a; CBS News, 2008b), and harmful toxic elements such as arsenic, mercury, lead, and chromium were released into the river, consequently leading to serious health issues in exposed humans and animals (however, in South Africa the trace elements are encapsulated or captured by alumino-silicate glasses and are therefore not harmful) (CBS News, 2008a; CBS News, 2008b). Said toxic elements are also detrimental to ecosystems and the environment when leached into soils and groundwater (Carlson and Adriano, 1993; Fulekar and Dave, 1986). Therefore, it comes as no surprise that ash producers are constantly seeking new and innovative ideas to re-use ash.

Currently the main practice is to use ash in concrete and cement applications, but utilisation in soil amelioration, waste immobilisation, clay bricks, road stabilisation, fillers for polymers and rubbers, acid mine drainage, and mine backfilling are also feasible options (Ahmaruzzaman, 2010; Blissett and Rowson, 2012; Kruger and Krueger, 2005; Yao et al., 2015). Another approach is to “pick out” certain components in the ash for advanced utilisation or the manufacturing of value-added products (Heidrich et al., 2013). Examples include rare earth elements (which is a current trending topic amongst researchers and government parastatals), cenospheres, zeolites, ferrospheres, and char (Franus et al., 2015; Heidrich et al., 2013). Coal char is of specific interest in the current research.

Depending on the furnace temperature, residence time, parent coal, and utility technology, char in the ash can vary from 0 – 45 %, and it is essentially an indication of the efficiency of the coal-burning process (Bartoňová, 2015). South African coal ash samples contain 0.5 to 8 % carbon in ash and also a significant amount of unburned carbon is associated with carbonaceous shale containing <10 % carbon. The char fraction is known to have a high porosity and specific surface area, making it ideal for fluid adsorption (mercury in particular) and activated carbon applications (Hwang and Li 2000; Hwang et al., 2002; Li et al., 2002; Li and Maroto-Valer 2012; Luo et al., 2004; Zhang et al., 2003). However, the porous nature of the char also causes difficulties when utilising the ash in concrete and cement applications - the char adsorbs added air-entraining agents (AEA) and surfactants (American Society for Testing and Materials (ASTM) C618a, 2012; Pedersen et al., 2008). Therefore, by first

4 | P a g e removing and subsequently using the char, the remaining ash can securely be used for cement and concrete production.

Another possible value-added product that can be formed from char is synthetic graphite (Bartoňová, 2015, Hower et al., 2017). Char particles have a high degree of structural order and a lamellar microtexture, making it ideal for this type of application (Baltrus et al., 2001; Bartoňová, 2015; Hurt et al., 1995; Maroto-Valer et al., 2002).

The European Union (EU), United States (U.S.), and British Geological Survey have all recently listed natural graphite as one of the raw critical commodities, exhibiting it as both a high supply risk and of great economic importance (Figure 1.3) (British Geological Survey, 2015; European Commission, 2017; Fortier et al., 2018).

Figure 1.3: Critical raw materials identified by the European Union (adapted from European Commission, 2017).

The critical status has partially come from the increase in graphite demand, especially concerning usage in economic valuable applications such as lithium-ion batteries (Li-ion), fuel cells, and pebble-bed nuclear reactors (Desjardins, 2012; Luque et al., 2014; Pierson, 1993). A decrease in graphite supply also influenced the critical status. China - the world’s leading natural graphite supplier (~65 %) - is drastically reducing its output due to environmental concerns (Du Venage, 2014; U.S. Geological Survey, 2019). In 2013, more than 55 graphite mines and producers in the province of Shandong were forced to shut down due to their wastewater, dust, and gas emissions not meeting regulation requirements (Du Venage, 2014). It is estimated that by 2025, China’s production will decrease from a current 780 000 to 300 000 tonnes / annum (Spencer and Hill, 2016).

Three options have been identified to fill the graphite supply / demand gap. The first pertains to the vast amounts of untapped natural graphite reserves that are located in Africa (Feytis,

5 | P a g e

2010). Currently, Zimbabwe, Madagascar, Mozambique, and Namibia are Africa’s only notable graphite producers with a combined ~3.5 % contribution to total world production (U.S. Geological Survey, 2019). Possible economic deposits have also been identified in Tanzania, Kenya, South Africa, Botswana, Ethiopia, Malawi, Swaziland, Uganda, Zambia, and Angola. Political and economic instabilities in Africa were, however, identified as being problematic, with investors reluctant to place their trust in uncertainties (Feytis, 2010).

The second option concerns the production of synthetic graphite. Although synthetic graphite can be used as a graphitic source, the high costs associated with manufacturing (~U.S. $10 000 – 20 000 / tonne selling price) can lead to its natural counterpart (~U.S. $800 – 900 / ton selling price) being preferred (Spencer and Hill, 2016). However, the high quality (high carbon content) of synthetic graphite is an advantage. Currently, synthetic graphite is manufactured from petroleum coke, but anthracite coal has also been identified as a favourable precursor, and in this thesis, the usage of char in coal ash as a possible precursor was evaluated.

The third option will be the recycling of graphite from used batteries (Vanderbruggen et al., 2018).

Research on producing synthetic graphite from char in coal ash is scarce, with only (to the author’s knowledge) papers by Cabielles et al. (2008; 2009) and Cameán and Garcia (2011) being published as yet. Yeh et al. (2011) published work on the graphitization of char in oil- fired fly ash. A definite gap or niche in research can thus be identified regarding this topic.

1.3 How did we approach this thesis? 1.3.1 Scope Due to the extensive nature of the ERA-MIN collaboration, only the results pertaining to the South African involvement are presented in this thesis (Figure 1.2). Consequently, this thesis deals with the obtainment and characterisation of coal ash sources from South Africa (part of Work Package 1), the development of a char-ash separation process, and the characterisation of the extracted chars (part of Work Package 2). Natural graphite sources from southern Africa are also identified, obtained, and characterised (part of Work Package 1). The South African partner (University of Johannesburg) also assisted with the characterisation of material from all participating countries; these results will be briefly discussed in this thesis. The graphitization of the concentrated chars to form Charphite, the characterisation of the Charphite, the electrochemical characterisation, the electrode immobilisation, and the testing of the graphitic material in energy-related reactions were conducted by other partners, and therefore not considered in this thesis.

6 | P a g e

1.3.2 Aim and objectives The aim of this thesis is to test the applicability of char in South African ash to act as a possible precursor for synthetic graphite.

The main objectives are as follows:

1. Desktop study on coal conversion ash in South Africa - aids in determining the availability (utilities, volumes, current usage, accessibility) of ash in South Africa; 2. Obtainment and characterisation of ash and corresponding feed coal from power plants and a gasification unit in South Africa – informs on ash quality in South Africa; 3. Desktop study on the separation of char from coal ash – articulates on how to produce and recover high-grade char concentrates; 4. Development of a process to separate char from the coal ash studied – the aim is to achieve high carbon grades (~90 %) and carbon recoveries, and 5. Characterisation of extracted char – supplies information on the possible usage of char concentrates (pre-graphitization) in graphitic applications.

As secondary objectives, the following pertains:

6. Predicts / comments on the graphitization ability of the concentrated char – South African sample compared to other ERA-MIN samples. 7. Desktop study on natural graphite occurrences in southern Africa – informs on possible usage / sustainability for future generations; 8. Characterisation of natural graphite in southern Africa – quality observations.

1.3.3 Thesis outline The thesis does not follow a conventional structure; rather each chapter addresses a particular aspect of the project. The intention is that each chapter should be able to stand alone, but also with cohesion throughout the document. The chapters are divided as follows:

Chapter 2: Coal conversion ash in South Africa (Objectives 1 and 2).

Chapter 3: Char-ash separation and char characterisation (Objectives 3 to 5).

Chapter 4: Char concentrate potential to graphitize (Objective 6).

Chapter 5: Natural graphite in southern Africa (Objectives 7 and 8).

Chapter 6: Summary, conclusions and recommendations.

7 | P a g e

Chapter 2: Coal conversion ash in South Africa

2.1 Introduction More than 150 million tonnes of coal are burned annually in South Africa, producing more than 50 million tonnes of ash. Although coal conversion ash is still classified as waste by many, increasing interest in its unique set of properties has led to the realisation that it can be effectively exploited in various applications. The expression “one man’s trash is another man’s treasure” readily comes to mind. Examples include utilising this ash in concrete and cement applications, or utilising certain components in the ash (e.g. the char fraction) in value-added applications. Before utilisation can take place, however, it is first necessary to understand the ash in question. For this reason, subsequent sections will focus on coal conversion ash in South Africa, its availability and management (desktop study - Sections 2.2 to 2.4) and quality (characterisation of a selection of ash samples used for the current study - Section 2.5).

2.2 Coal and its usage in South Africa South Africa has 19 coalfields, all situated in the geological Karoo Supergroup (Figure 2.1), and collectively hosting 40 billion tonnes of recoverable ore reserve (Barker, 1986; Cairncross, 2001; Hancox and Götz, 2014; Hancox, 2016; Jeffrey, 2005; Prévost, 2003; South African Coal Roadmap, 2011; Wagner et al., 2018). An inventory of coal quality, utilisation, reserve amount, and major collieries for each coalfield is given (Table 2.1).

The Witbank Coalfield is South Africa’s major producing coalfield (55 % of all coal being mined) but is fast approaching depletion with only an estimated 9 billion tonnes of recoverable reserve left (Hancox and Götz, 2014; Jeffrey, 2005; South African Coal Roadmap, 2011). The coal is typically high ash yield, inertinite rich, bituminous coal, and is mainly used for electricity generation (Pretorius et al., 2002; Wagner et al., 2018). The Waterberg Coalfield is seen as a possible substitute for the Witbank Coalfield, but its low ash yield and semi-soft coking nature makes it more ideal for metallurgical and exporting purposes (Jeffrey, 2005; Pretorius et al., 2002; Wagner et al., 2018). The Springbok Flats Coalfield is also seen as a viable replacement, but due to the high uranium content, utilisation is currently not feasible (Ndhlalose et al., 2015).

The Highveld Coalfield is exploited mainly for the production of synfuels and has comparable quality to the Witbank Coalfield (Pretorius et al., 2002). Although it also has an estimated 9 billion tonnes of coal reserve left, the utilisation rate is much lower than for the Witbank Coalfield, and it is predicted that it can still serve the synfuels industry for a considerable number of years (Jeffrey, 2005; South African Coal Roadmap, 2011). In the Free State

8 | P a g e

Coalfield, a new synfuels project is on the way which focusses on underground coal gasification (50-MW e Theunissen Underground Coal Gasification Project) (Arnoldi, 2017; Van Dyk et al., 2015).

South Africa also has several low production anthracite / semi-anthracite deposits in the Somkhele, Nongoma, Utrecht, Klip River, and Kangwane Coalfields (Pretorius et al., 2002). However, much of the anthracite grade coal has already been mined out.

Overall, the quality of South African coal can be described as low-grade with high ash yield, low calorific values (CV), and high percentages of unreactive inertinite macerals (South African Coal Roadmap, 2011). However, the sulphur content of South African coal is low when compared to other countries (Kalenga et al., 2011; Snyman and Botha, 1993). All export coal is beneficiated to lower the ash yield and better the quality.

Figure 2.1: Coalfields of South Africa (adapted from Jeffrey, 2005).

9 | P a g e

Table 2.1: Quality, usage, reserve, and major collieries of South Africa’s 19 coalfields.

Coalfield Coal quality1, 2, 4, 6, 11 Coal utilisation2, 3, 4, 5, 6 Recoverable Major collieries2, 3, 4, 6, 7, 8, 9, 10 reserves (Mt)3, 7

Limpopo (Tuli) Presumably higher grade coal Coking and steam coal 107 Makhado, Vele than Witbank and Highveld

Waterberg (Ellisras) 8-14 % ash yield, 27-30 MJ/kg Steelmaking and power generation, 6 744 Grootegeluk CV, 0.5-0.8 % sulphur, Medium exporting to Mozambique Rank C, bituminous

Soutpansberg West Mining not feasible Mining not feasible No current mining (Mopane)

Soutpansberg Central Mining not feasible Mining not feasible 257 Soutpansberg No current mining (Tshipise) combined

Soutpansberg East Coking coal, high ash yield (~60 Hard coking coal at Tshikondeni No current mining - (Venda-Pafuri) %) Tshikondeni previously

Springbok Flats High ash yield (40-60 %), high Not mined - high uranium content 1 700 No current mining uranium

Witbank 13-28 % ash yield, 21-29 MJ/kg Major South African producing 8 509 Goedehoop, Greenside, CV, Medium Rank C, bituminous coalfield (55 % of all collieries), Kleinkopje, Landau, Kriel, power generation, export, Mafube, Optimum, Delmas, metallurgical, synfuels, and other Arnot, Matla, North Block Complex, New Clydesdale, Leeuwpan, Inyanda, Klipspruit, Khutala, Douglas

Kangwane Anthracite and semi-anthracite One operating mine, semi-anthracite 146 Nkomati Anthracite (Komatipoort) and anthracite

Free State Very high ash yield (37.9 %), Underground coal gasification 4 919 New Vaal

10 | P a g e

bituminous project, low-grade steam coal

Vereeniging-Sasolburg Bituminous Steam coal and synfuels 1 708 New Vaal, Wonderwater Strip Mine, Sigma

South Rand Dull bituminous coal Not currently mined 716 No current mining

Highveld 20-35 % ash yield, 18-25 % CV, Major South African producing 9 475 New Denmark, Isibonelo, low percentage inertinite, coalfield (25 %), synfuels, export Forzando, Kriel, Matla, bituminous coal, steam coal Secunda collieries, Twistdraai

Ermelo Bituminous, presumably better Steam and exporting 4 388 Coastal fuels, Spitzkop, quality than Witbank and Highveld Savmore

Klip River Anthracite and coking coal Anthracite and coking coal 529 Springlake, Aviemore, Magdalena

Utrecht Anthracite and coking coal Anthracite and coking coal 541 Small-scale mining

Vryheid Anthracite and coking coal, high Anthracite and coking coal 100 Vaalkrantz sulphur (1.17 %)

Nongoma Anthracite Anthracite Zululand anthracite 6 combined Somkhele High-grade anthracite Anthracite Somkhele

Molteno-Indwe Low-grade bituminous Brick making 47 Small-scale mining

1Malaza (2014); 2Wagner et al. (2018); 3Jeffrey (2005); 4Pretorius et al. (2002); 5Ndhlalose et al. (2015); 6Hancox and Götz (2014); 7South African Coal Roadmap (2011); 8Chamber of Mines of South Africa (2018); 9Barker (1986); 10Jeffrey et al. (2014); 11Kataka et al., 2018

11 | P a g e

2.3 Coal conversion / ash producing utilities in South Africa Approximately 224 million tonnes of coal are mined annually in South Africa, of which 31 % is exported, 42 % used in electricity generation, 21 % used in synthetic fuels, and the remainder in metallurgical and smaller industries (Eberhard, 2011; Prévost, 2003). The electricity generation and synfuels production industries are of importance for the current research, and will subsequently be discussed.

2.3.1 Electricity generation In South Africa, coal’s share in the energy mix is about 72 % (Eskom, 2017a). Eskom is South Africa’s biggest coal-fired electricity utility with 15 baseload power stations in operation or the construction phase (Eskom, 2018). Regional Municipalities and Public- Private Partnerships own an additional 14 coal-fired power stations (Department of Energy, 2018). The output of these 14 power stations is, however, marginal (~4389 MW) (Department of Energy, 2018), and will not be considered further in this thesis.

Eskom power stations are situated in the Mpumalanga, Limpopo, and Free State Provinces, where the majority of the South African coalfields are also located (Figure 2.2).

Figure 2.2: Map indicating Eskom coal-fired power station locations (adapted from eNCA, 2014).

A list of the Eskom power stations is given, with some general information on each (Table 2.2).

12 | P a g e

Table 2.2: General information on Eskom coal-fired power stations.

Power Status1 Commission date1, 2, 7 Decommissioning Coal supplier1, 4, 5, 6 Coal quality1, 8 station date3, 7

Arnot Return to service 1971-1975 & 1997-1998 2021-2029 Optimum, Arnot, North Block Complex, 21.92 MJ/kg, 26.91 % ash yield, 0.78 and Mafube % sulphur Camden Return to service 1967-1969 & 2005-2008 2021-2023 Usutu South, West, and East collieries 24.65 MJ/kg Duvha Baseload 1980-1984 2030-2034 Douglas, and South 32 (Wolvekrans) 23.3 MJ/kg, 28.82 % ash yield Grootvlei Return to service 1969-1977 & 2008-2011 2025-2028 Multiple sources - Hendrina Baseload 1970-1976 2020-2026 Optimum colliery 24.0 MJ/kg, 25.5 % ash yield, 0.9 % sulphur Kendal Baseload 1988-1992 2038-2043 Khutala, Douglas, and Grootegeluk - Komati Return to service 1961-1966 & 2009-2013 2024-2028 Koornfontein and Blinkpan 21.6 MJ/kg, 27.84 % ash yield, 0.84 % sulphur Kriel Baseload 1976-1979 2026-2029 Kriel colliery 23.4 MJ/kg, 23.4 % ash yield, 1 % sulphur Lethabo Baseload 1985-1990 2035-2040 New Vaal colliery, unusually low-grade 16 MJ/kg, 37.8 % ash yield, 0.59 % sulphur Majuba Baseload 1996-2001 2046-2050 No dedicated mine, 15 suppliers of coal - Matimba Baseload 1987-1991 2037-2041 Grootegeluk colliery, 2100 t/h 16-22 MJ/kg, 36 % ash yield Matla Baseload 1979-1983 2029-2033 Matla colliery 24 MJ/kg; 19.1 % ash yield, ~1 % sulphur Tutuka Baseload 1985-1990 2035-2040 New Denmark colliery 24.5 MJ/kg, 22.4 % ash yield Medupi Newly build 2015- 50 year life Grootegeluk colliery - Kusile Newly build 2017- 50 year life Grootegeluk colliery -

1Eskom (2018); 2Eskom (2017b); 3The Star (2017); 4South African Coal Roadmap (2011); 5Reynolds-Clausen (2016a); 6Seriti (2017); 7Department of Energy (2018); 8Reynolds-Clausen (2016b)

13 | P a g e

It can be seen that the decommissioning dates for all stations, except for newly build Medupi and Kusile, are set before 2050. This forms part of the government’s Integrated Resource Plan (IRP), in which coal-fired dependence will be reduced from the current 72 % to 45 % before the end of 2030 (Department of Energy, 2018). This IRP will require the decommissioning of 12 000 MW in the period 2020-2030 (Department of Energy, 2018). By 2050 a further amount (unspecified at this stage) will be decommissioned, with only Medupi and Kusile left as coal-fired utilities. The specifics of the IRP are beyond the scope of the current research but can be viewed at http://www.energy.gov.za/IRP/irp-update-draft-report- 2018.html.

Most of the Eskom power stations are fed with a mix of coal supplies, leading to inconsistencies in output quality. In 2008, the so-called “load shedding” concept was introduced to South Africa in which scheduled rolling black-outs occur across the country due to Eskom’s generation problems. In 2018, Gupta-owned coal mines (e.g. Optimum) went into business rescue due to an alleged corruption scandal (TimesLive, 2018), affecting coal supply to several power stations.

Pulverised coal combustion (PCC) technology is used in the Eskom power stations. Detailed information is given in Table 2.3. A schematic of a PCC process is shown in Figure 2.3.

Coal enters the process via a conveyor belt and is subsequently pulverised in a mill. Milling until 70 % of the entering coal particles are smaller than 75 µm is custom in South Africa (Belaid et al., 2014; Van Alphen, 2017). The pulverised coal is then injected, together with an airstream, into a boiler operating at temperatures in the range of 1400 °C to 1700 °C (Andrews, 2012). The high operating temperatures, compared to northern countries (1200 to 1400 ºC) are due to high inertinite maceral contents in South African coal, which burn at higher temperatures than vitrinite macerals (Andrews, 2012). The fluids within the coal matrix (moisture and methane) evaporate at the elevated temperatures (Harvey and Ruch, 1984). Organic components (and some inorganic matter) in coal (mostly C, O, N, H, S) devolatilise and react with the air to form flue gases such as CO2, NO푥, 퐻2푂and SO푥. The energy released from these reactions is used to produce electricity by heating and converting water to steam, which subsequently drives the generation turbines.

The inorganic matter in coal is often referred to as “mineral matter” and includes discrete mineral crystals as well as any organically associated inorganic elements (Harvey and Ruch, 1984). During heating, the mineral matter can devolatilise, but the majority, however, will melt (Tishmack and Burns, 2004). When cooled, the molten particles form ash. The detailed ash formation mechanisms are discussed in Section 2.4.1. The fine, lightweight ash particles are called fly ash and leave the boiler with the flue gases. Some of the melted ash may

14 | P a g e agglomerate on the furnace walls (causing slagging, fouling, and clinker formation) and, in time, will fall to the bottom of the furnace (Van Alphen, 2017). Hence, the nomenclature bottom ash.

In conventional PCC processes, fly ash yields ~65 % and bottom ash ~35 % of the total ash weight (Helmuth, 1987). Fly ash is separated from the flue gases via particulate cleaning technologies such as electrostatic precipitators (ESP) or Fabric Filter Precipitators (FFP). In Table 2.3 the technologies used in each South African power station are indicated. Although ESPs are mostly used as an ash capturing device, Eskom is planning to retrofit with FFPs by the end of 2024. The bottom ash samples are collected in hoppers.

SO2 is separated in a flue gas desulphurisation scrubber (Glasser, 2004). However, in South

Africa, only the Kusile power station is implementing SO2 removal. The remaining flue gases, enriched in CO2, are vented into the atmosphere.

Figure 2.3: Schematic of a pulverised coal combustion process (adapted from Tishmack and Burns, 2004).

15 | P a g e

Table 2.3: Technical information on South African coal-fired power stations.

Power Total Total Boiler Coal feed per Boiler final steam Boiler Turbine Ash capturing method1,2,3,4,5 station installed nominal units2 boiler (max pressure volume efficiency capacity capacity tonne/hour)2 (MPa)/Temperature (°C)2 (m3)2 (%)2 (MW)1 (MW)1

Arnot 2 352 2 232 6 4167 16.7/516 (367×96) 34.48 FFP Camden 1 561 1 481 8 90.7 11.0/543 (352×65) 32.00 FFP Duvha 3 600 3 450 6 1500-1800 17.1/540 13 080 37.60 Units 1-3 FFP, Scheduled retrofit units 4-6 to FFP 2021-2023 Grootvlei 1 180 1 120 6 - - - 32.90 Units 1,5,6 FFP, Scheduled retrofit units 2,3,4 to FFP 2016-2017 Hendrina 1 893 1 793 10 1000 11/540 4100 34.20 FFP Kendal 4 116 3 840 6 1161 17.24/540 - 35.30 ESP Komati 990 904 9 1000 - - 30.00 ESP Kriel 3 000 2 850 6 1400 17.2/516 10 856 34.99 ESP, Retrofit units to FFP 2019-2024 Lethabo 3 780 3 558 6 2100-2400 17.32/540 23 049 - ESP Majuba 4 110 3 843 6 - - - - ESP, Retrofit units to FFP 2019-2024 Matimba 3 990 3 690 6 1750 16.1/540 17 223 35.60 ESP Matla 3 600 3 450 6 1800 17.2/540 12 193 37.60 ESP, Retrofit units to FFP 2020-2024 Tutuka 3 654 3 510 6 1500 17.1/540 13 080 38.00 ESP, Retrofit units to FFP 2018-2024 Medupi 1 588 1 437 6 - - - - FFP Kusile 799 720 6 - - - - FFP 1Department of Energy (2018); 2Eskom (2018); 3The Star (2017); 4Pretorius (2015); 5Reynolds-Clausen (2016a)

16 | P a g e

2.3.2 Synfuels Sasol is the only bituminous coal to liquid (CTL) utility globally and is partially responsible for the synfuels industry in South Africa. The CTL plant is located in Secunda (Mpumalanga Province) where patented Sasol-Lurgi fixed bed dry bottom (FBDB) gasifiers are used for coal conversion (Figure 2.4).

Coal is crushed to a typical top size of 65 mm before entering the gasifiers (Bunt and Waanders, 2008). The crushed coal is fed into the gasifiers, which operate at temperatures below the ash melting point (1300-1350 °C) (Sasol, 2018; Van Dyk et al., 2006). Steam and limited oxygen are used as reagent gases, with the result being syngas formation (퐶푂, 퐻2,

푁2, 퐶푂2, and 퐶퐻4) (Sha, 2009). The syngas is subsequently converted to liquid petroleum and chemicals via the downstream Fischer-Tropsch process (Van Dyk et al., 2006). The ash is removed by an ash lock and transported to ash dumping sites (Bunt and Waanders, 2008). The ash distribution is coarse and rock-like, known as gasification ash.

Figure 2.4: Schematic of a Sasol-Lurgi fixed bed dry bottom gasification process (Van Dyk et al., 2006).

The coal used in Sasol operations is solely supplied by Sasol Mining (Pty) Ltd. An average of 40 million tonnes of coal is mined annually at the Bosjesspruit, Brandspruit, Middelbult, Syferfontein, and Twistdraai collieries in the Highveld Coalfield (Eberhard, 2011; Sasol, 2018; South African Coal Roadmap, 2011). Approximately 7 % of the coal is exported to international consumers, 70 % used in Sasol Synfuels, and the rest for private electricity

17 | P a g e generation at the Sasol plant. Sasol is looking at replacing some of their mines, notably Twistdraai with Thubelisha, Brandspruit with Impumelelo, and Middelbult with Shondoni (Sasol, 2018; South African Coal Roadmap, 2011). Coal fed to Sasol Synfuels is usually a blend of coal from the mines mentioned above (Sasol, 2018). The quality of the coal can be described as low-grade with a CV below 21 MJ/kg and an ash yield ranging between 20 and 35 air dried basis (a.d.b) wt. % (South African Coal Roadmap, 2011). The sulphur content ranges between 1 and 2 a.d.b wt. % (South African Coal Roadmap, 2011).

Sasol Synfuels was commissioned in 1950-1955 and although recent retrenchment of employees is causing concern, Sasol has denied withdrawal of operations from South Africa.

2.4 Ash formation and management in South Africa 2.4.1 Minerals in coal Mineral matter in coal can be classified as syngenetic or epigenetic. Syngenetic minerals were deposited in the original plant or peat swamp through fluvial action, wind, and precipitation (Bryers, 1986; Speight, 2013; Van Alphen, 2005). Epigenetic minerals were incorporated after coal formation, when percolating waters deposited the minerals into the cavities of the coal (Bryers, 1986; Speight, 2013; Van Alphen, 2005).

Although more than 100 minerals have previously been identified in coal, the major minerals are well-known and can be classified into four groups (Speight, 2013):

 Clay minerals - kaolinite, illite, muscovite, and smectite,  sulphides and sulphates – gypsum, anhydrite, pyrite, marcosite, and sphalerite,  carbonates – calcite, dolomite, and siderite, and  oxides – quartz and rutile

After beneficiation in a processing plant, the minerals remaining in the coal feed to a boiler / gasifier can be classified as included minerals, excluded minerals, and organically associated inorganic elements (mainly salts) (Figure 2.5).

18 | P a g e

Organically associated inorganic elements

Figure 2.5: Inorganic modes of occurrence in a coal conversion feed stream (Benson et al., 1993).

The included minerals can be defined as mineral grains closely associated with the organic matrix, and are syngenetic (Tomeczek and Palugniok, 2002). Wigley et al. (1997) found that included minerals were more frequent in South African and Australian pulverised coals than in British and North American pulverised coals. Russell et al. (2002) reported high percentages of included pyrite, calcite, and dolomite minerals in a high ash anthracite from Spain. Wells et al. (2004) also indicated that pyrite is often included in coals from the United Kingdom. Take note that other minerals such as clay and quartz can also be included. In an oxidising environement, included pyrite will undergo the following transformation mechanism (Figure 2.6):

At temperatures between 500 °C and 700 °C, solid pyrite (FeS2) decomposes into solid pyrrhotite (FeS) and gaseous sulphur (Van Alphen, 2005). The gaseous sulphur reacts with excess oxygen in the atmosphere to form sulphur dioxide (Bryers, 1986). At 1100 °C, the pyrrhotite melts before it is oxidised to a wusite melt (FeO) (Van Alphen, 2005). The oxidation step includes a molten Fe-O-S stage followed by removal of S to form a complete molten wusite phase (Mitchell and Akanetuk, 2002). When expelled from the high temperature environment wusite will rapidly cool to form magnetite (Fe3O4) and then (only at high oxygen levels) oxidises slowly into hematite (Fe2O3) (Mitchell and Akanetuk, 2002, Van Alphen, 2005). The pyrrhotite and wusite melts will also react with silicates to form an iron rich glass.

19 | P a g e

Figure 2.6: Behaviour of included pyrite upon heating (McLennan et al., 2000).

Excluded minerals consist of distinct mineral grains with limited organic constituents being present, and might be more easily separated than included minerals (Tomeczek and Palugniok, 2002). Excluded minerals are largely removed before conversion. Russell et al. (2002) reported that clays, quartz, and pyrite are the major excluded minerals in high ash anthracite from Spain. Wells et al. (2004) also reported that quartz is the main excluded mineral in coals from the United Kingdom. The transformation mechanisms for excluded minerals in coal are provided (Figure 2.7).

Upon heating kaolinite will go through a series of irreversible dehydration and decomposition reactions. At 400 – 600 °C dehydration of kaolinite takes place to form metakaolinite (Insley and Ewell, 1935). Metakaolinite will then decompose into spinel and amorphous oxide (McConnell and Fleet, 1970). This decomposition takes place rapidly at 925 – 985 °C (Insley and Ewell, 1935). Temperatures above 1250 °C will lead to spinel decomposing into platelet mullite and cristoballite (crystalline 푆푖푂2 but with a different crystal structure than quartz) (McConnell and Fleet, 1970). Finally, at >1400 °C needle mullite will form (Bryers, 1986). Due to the high melting point of mullite it may not be completely liquid even at temperatures as high as 1800 °C (Bryers, 1986).

Carbonates decompose to their oxide forms and release carbon dioxide upon heating e.g. calcite forms lime and carbon dioxide (Speight, 2013). The lime will liquefy at 2570 °C (Bryers, 1986). When contacted with sulphur dioxide and oxygen the lime will further react to form calcium sulphate or anhydrite (Van Alphen, 2005).

The excluded pyrite follows the same transformation path as the included pyrite except that the additional iron glass phase does not form.

Quartz is the only mineral in coal that is inert to high temperatures (Speight, 2005). However, at 1750 °C cristoballite can form (Bryers, 1986). Quartz can melt at 1713 °C (Bryers, 1986).

20 | P a g e

Figure 2.7: Excluded minerals’ transformation mechanisms in coal (adapted from Van Alphen, 2005).

The organic associated inorganic elements are, as opposed to the other two modes, not distinct granules but rather chemically bonded to the organic components (Harvey and Ruch, 1984). Organic associated inorganic elements include K, Na, and Cl.

2.4.2 Ash formation mechanisms The schematic in Figure 2.8 shows the ash formation pathways followed by the different mineral modes of occurrence.

21 | P a g e

Figure 2.8: Ash formation mechanisms of included, excluded, and organically associated inorganic minerals (adapted from Krishnamoorthy and Pisupati, 2015).

When exposed to high temperatures in the conversion furnace, the moisture in coal particles evaporate and the organic volatile matter undergoes pyrolysis. The volatile gas consequently combusts into a flame and forms a barrier through which the oxygen cannot further penetrate; thus leaving the coal particles in a reducing atmosphere.

At this stage, the coal particles may experience primary thermal fragmentation due to either a thermal shock or a pressure build-up of volatile gases in the coal matrix (Dacombe et al., 1999; Dakič et al., 1989). A fraction of the included minerals may subsequently be liberated (Blissett and Rowson, 2012). Primary fragmentation is usually present in large particle conversion processes, such as fixed bed gasification (Zhang et al., 2002). Bunt (2006) has noticed significant primary thermal fragmentation (and mechanical fragmentation due to particle collision with the gasifier walls) in Sasol gasifiers. After pyrolysis and primary fragmentation, a carbonaceous residue known as char is left. As the flame enveloping these newly formed char particles subsides as the char moves away from the volatiles-rich flame, the oxygen will come into contact with the particles and burn-out starts. Everson et al. (2005) proposed that for large particles in Sasol gasifiers, burning will take place from the outside in (shrinking core model). The temperature of the char particles at this point greatly exceeds the furnace temperature, with the result being the melting of residual included minerals and the decomposition of the carbonates (e.g. calcite, ankerite) and the sulphides (e.g. pyrite) (Bryers, 1986; Flagan and Seinfeld, 1988). The melted minerals may coalesce / agglomerate on the surface of the char particles to form ash (Flagan and Seinfeld, 1988). As burn-out

22 | P a g e continues, the porous char structure becomes fragile and collapses - termed secondary thermal fragmentation (Cui and Stubington, 2001; Shah et al., 2015). This phenomenon is present in small combustion as well as large gasification char particles and partially liberates the included coalesced ash grains. Take also note that the porous char forms from low-to- medium vitrinite and reactive semifusinite coals. For high inertinite coals this behaviour will not occur. The liberated and coalesced ash particles will react with each other, as well as the oxygen environment to shape the final ash composition (Krishnamoorthy and Pisupati, 2015). When expelled from the furnace, the molten ash will rapidly cool to form solid particles. Due to this rapid cooling, crystallisation is hindered and the majority of the ash particles are amorphous, glassy solids (Malhotra, 2008). The size distribution of the resultant ash particles ranges between 0.2-10 μm (Tomeczek and Palugniok, 2002) (but this is affected by the feed particle size and composition). Any partially or unburned char will also exit the furnace with the ash. These unburned char particles might still encompass a fraction of the ash components that were not liberated during fragmentation.

The excluded minerals are not included in the flame barrier described earlier, and therefore they do not reach the same temperature as the included and organically bounded minerals (Tomeczek and Palugniok, 2002). The transformation can hence be significantly different than for the included minerals (Shah et al., 2015). According to Shah et al. (2015), the excluded minerals can melt and fuse, exit the furnace without any alteration, or it can fragment. Typically, an excluded mineral can lose its water of hydration, evolves gas, become oxidised or reduced, and melt all in the course of the furnace temperature history (Bryers, 1986).

In the feed to a gasifier, stones are also present which will only heat during gasification (Krishnamoorthy and Pisupati, 2015).

Organically associated inorganic elements (e.g. K, Na, and Cl) react with each other to form artefacts and gases. Although unlikely, some thermal refractory minerals, such as quartz

(SiO2), lime (CaO), and periclase (MgO), can vaporise as particular matter containing Si, Mg, and Ca elements reported as oxides (Tomeczek and Palugniok, 2002). The vaporised species will homogeneously nucleate, coagulate, agglomerate, and condense to form an ultrafine ash fraction (<0.1 μm) (Flagan and Seinfeld, 1988). Some of the vaporised species will also heterogeneously condense onto the surface of the larger ash particles, enriching these particles with trace elements (Flagan and Seinfeld, 1988). Although the vaporised species only contribute ~1 % to the final ash distribution, it concerns environmentalists due to its contribution to air pollution (Flagan and Seinfeld, 1988).

23 | P a g e

2.4.3 Ash management in South Africa Power station ash In South Africa, more than 50 million tonnes of coal ash are produced annually from power stations (Department of Environmental Affairs 2018a). Approximately 88 % consists of fly ash, and 12 % of bottom ash (Department of Environmental Affairs, 2018a). Eskom alone produces more than 34 million tonnes of ash annually (excluding newly built Medupi and Kusile), with only 7 % being re-used (Reynolds-Clausen and Singh, 2019). The low utilisation percentage in South Africa is due to ash being classified as a waste, and subsequently due to the strict regulations associated with this classification (Department of Environmental Affairs, 2018a; Department of Environmental Affairs, 2018b). An ash producer may apply to the Minister of Environmental Affairs for the exclusion of a waste stream / portion of a waste stream from the definition of waste, if the producer can demonstrate that the ash will be used beneficially, undertake a risk assessment, and submit a risk management plan (Department of Environmental Affairs, 2018b). Failure to comply with these specifications can lead to 15-years imprisonment, a fine, or both imprisonment and a fine (Department of Environmental Affairs, 2018b).

Another hurdle to the utilisation of ash in South Africa is something as simple as distance. The majority of South African power stations are situated in the province of Mpumalanga, and any off-takers outside this region are burden with high transportation costs (Kruger, 2017). As a result, only industries within a radius of 50 km are served by Eskom ash facilities (Department of Environmental Affairs, 2018a).

Eskom power stations maintain a Zero Liquid Effluent Discharge policy (Reynolds-Clausen and Singh, 2019). Saline effluents are being disposed of at the ash-handling facility, with the ash acting as a salt sink for the effluents (Reynolds-Clausen and Singh, 2019). More or less 74 % of the ash produced is used for this purpose (Reynolds-Clausen and Singh, 2019). The elevated salt levels in the ash render it unusable, also contributing to the low utilisation and reclamation percentages. Desalinised water is recycled.

A breakdown of ash production and utilisation from individual Eskom power stations is provided in Figure 2.9 and Table 2.4. Fly and bottom ash from the Medupi and Kusile power stations are excuded, seeing that they are not in full operation yet. Saleable can be defined as any ash not used for an effluent sink. It is unclear if the data concerns both fly and bottom ash.

Lethabo power station produces the largest amount of ash (6.48 million tonnes per annum). Lethabo burns very low-grade coal from the Free State Coalfield. Kendal, Kriel, Lethabo, Majuba, Matimba, and Matla are the only power stations currently selling their ash. Of these

24 | P a g e six, Matla sells 100 % of saleable ash, Lethabo more than 94 %, and Matimba only 0.6 %. Kendal has the largest amount of ash available to sell but is currently only selling 18.1 %. In total, only 27 % of saleable ash is sold.

The ash sold from Eskom power stations is mainly used in paint, rubber, geopolymer, soil, amelioration, mine backfilling, mine drainage treatment, road construction, brick making, and concrete applications (Eskom Rotek Industries, 2019). Eskom ash was also used in the building of the Burj Kalifa, the Gautrain railway, the Maputo-Katembe bridge, and in the building of the Medupi power station (Eskom Rotek Industries, 2019, Hunter, 2018).

Information on ash quality and ash disposal methods is given for the individual power stations (Table 2.5). The loss on ignition (LOI) percentages for fly ash range from very low (0.4 % Lethabo and Majuba) to moderately low (7.8 % Komati). Bartoňová (2015) reported on LOIs as high as 45 % from international power stations. The low percentages from Eskom indicate that they are burning coal efficiently as most of the input carbon is converted. The LOI percentages for the bottom ash are slightly higher.

Eskom utilises both wet and dry ash disposal methods. Ash is transported as a slurry to wet impoundments, where it acts as a salt sink for saline effluents (Reynolds-Clausen and Singh, 2019). Ash is transported to dry ash landfills, where it is dampened with 10 % water to reduce dust formation (Reynolds-Clausen and Singh, 2019). The fly and bottom ash streams are mixed before being transported to the appropriate disposal facility (Reynolds-Clausen, 2016a).

25 | P a g e

40.0

35.0

30.0

25.0

20.0

15.0

10.0

5.0

0.0 Arnot Camden Duvha Grootvlei Hendrina Kendal Komati Kriel Lethabo Majuba Matimba Matla Tutuka Total Produced (Mt) 1.6503 1.068 2.506 0.996 1.824 4.906 0.324 2.302 6.48 3.428 4.752 3.04 2.926 36.2023 Saleable (Mt) 0.4 0 0.2 0 0.2 3 0 1 1.3 1 2 0.4 0 9.5 Sold (Mt) 0 0 0 0 0 0.542 0 0.312 1.227 0.042 0.012 0.4 0 2.535

Figure 2.9: Ash tonnages produced, saleable, and sold by individual Eskom power stations (adapted from Reynolds-Clausen and Singh, 2019).

26 | P a g e

Table 2.4: Percentage of saleable ash being sold by Eskom power stations.

Kendal Kriel Lethabo Majuba Matimba Matla Total

푆표푙푑 18.1 31.2 94.4 4.2 0.6 100.0 29.7 % 푆푎푙푒푎푏푙푒

Table 2.5: Information on ash quality and disposal methods adopted by Eskom power stations.

ID Loss on ignition (%)1 Disposal method1,2

Arnot 6.1 Wet, new dam required by 2028 Camden 3.2 Wet Duvha 5.8 Wet, 31 years remnant life Grootvlei 6.0 Wet Hendrina 4.5 Wet, new dam required 2022 Kendal 2.0 Dry Komati 7.8 Wet, 14 years remnant life Kriel 1.5 Wet Lethabo 0.4 Dry Majuba 0.4 Dry Matimba 1.0 Dry Matla 1.1 Wet Tutuka 2.9 Dry Medupi 1.0 Dry Kusile 0.6 Dry 1Reynolds-Clausen (2016a); 2Eskom (2018)

27 | P a g e

Gasification ash At Sasol, coal is fed to a gasification unit as well as a steam generation unit (Mahlaba, 2011). Coarse ash, known as gasification ash, is discarded via a hopper from the gasification unit, and fine ash (fly ash) is mainly obtained from the steam generation unit (Mahlaba, 2011).

Sasol Synfuels produces 7 million tonnes of gasification ash annually but is currently only re- using a small percentage (Ginster and Matjie, 2005; Mathews, 2016). Previously, gasification ash was utilised by a brick manufacturer in the Vaal Triangle (Ginster and Matjie, 2005). Transportation costs are currently an issue and the majority of the ash is therefore landfilled (Ginster and Matjie, 2005). Other utilisation possibilities for the gasification ash include acting as lightweight aggregates (not economically feasible), cement extenders, or adding it to old underground mine workings (Ginster and Matjie, 2005).

According to Mahlaba et al. (2011) 4 million tonnes of fine ash is generated annually from Sasol Synfuels, where it is landfilled.

2.4.4 Comparison with the global ash market According to Harris et al. (2019), approximately 1221.9 million tonnes of coal conversion products (CCPs) are produced globally per annum (includes fly ash, bottom ash, boiler slag, fluidized-bed combustion ash, and flue gas desulfurization material). A country breakdown is given in Table 2.6. China produces the most ash (50.3 %), followed by India (17.6 %), and the United States of America (9.6 %). It is, however, predicted that ash produced from India will increase exponentially in the coming years, due to coal combustion continuing to be their major source of electricity generation (Harris et al., 2019). Globally, 3 billion tonnes of ash have been stockpiled (Harris et al., 2019). It is estimated that since 1949 more than 2.2 billion tonnes of ash have been deposited on landfills in China, covering 300 square kilometres of land (Barnes and Sear, 2006, Harris et al., 2019).

Due to the large variety of ash properties, re-utilisation opportunities are endless. More than 670 million tonnes of CCPs are utilised globally per year (Harris et al., 2019). A country breakdown is provided (Table 2.6). Countries such as Australia, the Middle East and Africa, and Russia are known to have very low ash utilisation rates (<50 %) (Harris et al., 2019). The Middle East and Africa, for instance, are only re-using ~10.6 % of their ash and Russia only ~27.2 % (Harris et al., 2019). Environmental, health, and legislation problems are associated with low utilisation rates (Openshaw, 1992). Transportation costs are another utilisation problem to consider. Fly ash purchase costs can range between U.S. $25 and $75 per tonne (Boral, 2018). However, the costs associated with ash transportation can double or even triple this price (Kruger, 2017; The Aberdeen Group, 1985).

28 | P a g e

Other countries such as Japan, Europe (EU15), and Israel have very high utilisation rates (Harris et al., 2019). Japan is the forerunner with 100 %, then Europe with 94.3 %, and then Israel with 90.9 % (Harris et al., 2019). The global average utilisation rate is 56 %.

Table 2.6: CCP volumes produced and utilised globally (Harris et al., 2019).

Country CCP Percentage of CCP utilisation Utilisation production total global (2011) (tonnes) percentage per (2011) (tonnes) production (%) country (%)

China 565 000 000 50.3 396 000 000 70.1 India 197 000 000 17.6 132 000 000 67.0 United States of America 107 400 000 9.6 60 100 000 56.0 Europe 100 000 000 8.9 Europe (EU15) 40 300 000 3.6 38 000 000 94.3 Middle East and Africa 32 200 000 2.9 3 400 000 10.6 Russia 21 300 000 1.9 5 800 000 27.2 Australia 12 300 000 1.1 5 350 000 43.5 Japan 12 300 000 1.1 12 300 000 100.0 Korea 10 300 000 0.9 8 800 000 85.4 Canada 4 800 000 0.4 2 600 000 54.2 Israel 1 100 000 0.1 1 000 000 90.9 Asia (other) 18 200 000 1.6 12 300 000 67.6 Total 1221 900 000 100.00 677 650 000 55.5

The majority of ash is utilised in cement applications, but other applications such as soil amelioration, waste immobilisation, clay bricks, road stabilisation, fillers for polymers and rubbers, acid mine drainage, and mine backfilling should not be excluded (Ahmaruzzaman, 2010; Blissett and Rowson, 2012; Kruger and Krueger, 2005; Yao et al., 2015).

2.5 Characterisation of South African coal and ash samples Having set the scene regarding ash production in South Africa, attention now moves to address ash utilised in the current research.

2.5.1 Methodology Fly, bottom, and gasification ash were supplied by South African combustion and gasification utilities (Table 2.7). The combustion utilities were chosen based on historical LOI data (provided by Reynolds-Clausen, 2016a). The older, less efficient power stations with higher LOI percentages were selected. The feed coal sources to these power stations were also obtained. Gasification ash samples from the East and West Sasol gasification units were selected. Due to confidential reasons, the coal feed to the gasifiers was not supplied.

It is assumed that all samples were obtained in a representative manner following in-house sample protocols based on International Organization for Standardization (ISO) 13909-2:

29 | P a g e

2016. The samples are considered to represent a snap-shot in time as the coal feed to some processes is known to vary due to blending from several mines.

The bulk samples were air-dried overnight and divided into smaller, representative fractions using the cone and quartering technique as a first means, followed by rotary splitting. All samples were stored in air-tight containers or bags to prevent oxidation and moisture adsorption. A representative sample from each coal and ash was characterised, and the remainder of the ash sample was utilised in char-ash separation trials (Chapter 3). Due to the coarser sizes of the bottom and gasification ash samples, the samples were pulverised for 3 minutes to a -212 μm particle size before characterisation commenced.

Table 2.7: Sample identification, type, origin, mass, sampling time, and unit for the ash and coal samples.

Sample Sample type Sample origin Mass Sampling LOI ID obtained time and (historical (kg) unit data)

FA PS1 Fly ash Samples supplied 7.2 17/10/2016 6.1 BA PS1 Bottom ash by Eskom power 9.7 at 13h20 Power Power

station 1 C PS1 Coal utility 20.2 Unit 5

FA PS2 Fly ash Samples supplied 6.7 13/10/2016 5.8 BA PS2 Bottom ash by Eskom power 13.7 at 13h30 C PS2 Coal utility 20.0 Unit 1 Power Power station 2

FA PS3 Fly ash Samples supplied 21.0 27/10/2016 6.0 BA PS3 Bottom ash by Eskom power 9.9 at 14h45 C PS3 Coal utility 15.9 Unit 5 Power Power station 3

FA PS4 Fly ash Samples supplied 13.2 17/10/2016 4.5 BA PS4 Bottom ash by Eskom power 9.7 at 14h40 C PS4 Coal utility 19.7 Unit 8 Power Power station 4

FA PS5 Fly ash Samples supplied 4.1 14/10/2016 7.8 BA PS5 Bottom ash by Eskom power 12.7 at 11h40 C PS5 Coal utility 17.8 Unit 3 Power Power station 5

GA East Gasification ash Sample supplied 23 28/07/2016 by Sasol Synfuels at 10h00

East Unit

ash GA West Gasification ash Sample supplied 23 28/07/2016 by Sasol Synfuels at 10h00 Gasification West Unit

The characterisation techniques are provided in Table 2.8. Proximate, CV, ultimate, X-ray fluorescence (XRF), X-ray diffraction (XRD), and petrographic analyses were used to characterise the coal samples. This was done to understand the properties of the ash better,

30 | P a g e seeing that ash characteristics are dependent on the coal quality being burned. The ash samples were characterised using visual, proximate, XRF, LOI, XRD, petrographic, and particle size distribution (PSD) analyses.

Table 2.8: Coal and ash characterisation techniques used in the current study.

Analysis C PS1 C PS2 C PS3 C PS4 C PS5 FA PS1 FA PS2 FA PS3 FA PS4 FA PS5 BA PS1 BA PS2 BA PS3 BA PS4 BA PS5 GA East GA West

Visual × × × × × × × × × × × × Proximate × × × × × × × × × × × × × × × × × CV × × × × × Ultimate × × × × × X-ray fluorescence × × × × × × × × × × × × × × × × × Loss on ignition × × × × × × × × × × × × X-ray diffraction × × × × × × × × × × × × × × × × × Petrographic × × × × × × × × × × × × × × × × × Particle size distribution × × × × × × × × × × × ×

Proximate, CV, ultimate, and XRF analyses were outsourced to Bureau Veritas Testing and Inspections South Africa (Centurion). The standards are provided in Table 2.9.

Table 2.9: Proximate, CV, ultimate, and X-ray fluorescence standards used for the analyses.

Analysis Standard

Sample preparation ACT-TPM-001 based on ISO 13909-4: 2001

Moisture ACT-TPM-010 based on ISO 11722: 1999 Volatile matter ACT-TPM-012 based on ISO 562: 2010 Proximate Ash yield ACT-TPM-011 based on ISO 1171: 2010 Fixed carbon By difference

CV CV ACT-TPM-014 based on ISO 1928: 2009

Total sulphur ACT-TPM-013 based on ISO 19579: 2006 Ultimate Carbon, nitrogen, hydrogen ACT-TPM-027 based on ISO 29541: 2010 Oxygen By difference

X-ray fluorescence ASTM D4326: 2013 by fusion bed

The XRF method was based on the ASTM D4326 (2013) by fusion bed standard. A Thermo Scientific sequential XRF instrument was used with an Rh x-ray source at 50kV/50mA. The samples were dried (110 °C) and combusted at 1000 °C. The roasted material was then fused into a glass bead in a 1:9 ratio with a 66:34 Litetraborate Limetaborate fluxing agent at 1050 °C.

31 | P a g e

The LOI analysis was outsourced to MAK Analytical South Africa (Bronkelspruit). The standard used entailed gradual heating of the samples in an oxidising atmosphere to 500 °C, maintaining this temperature for 30 minutes before gradual heating to 815 °C. The samples were kept at 815 °C for 60 minutes before their mass loss was determined. Converting the mass loss to a percentage provides the LOI value. LOI is a rapid, cost-effective method of estimating char in ash percentages, and it is also able to provide an easy comparison between the samples.

XRD analyses were outsourced to Dr. Sabine Verryn from XRD Analytical & Consulting (http://www.xrd.co.za/). The material was prepared via a backloading preparation method. A PANalytical Empyrean diffractometer with PIXcel detector and fixed slits with Fe filtered Co- K∝ radiation was used for detection while X’Pert Highscore plus software was used for mineral identification. The Rietveld method was used to semi-quantify the samples. The amorphous percentages for the ash samples were determined by adding a 20% Si standard and micronizing it in a McCrone micronizing mill. The amorphous percentages can have errors as high as 15 wt. %. To obtain an indication of the amorphous phase composition and mineral association, automated Scanning Electron Microscopy (SEM) was used. However, as this was only conducted on GA East, the results are presented in Appendix A. It was not deemed necessary to progress with the other samples.

A Zeiss Imager M2M reflected-light petrographic microscope with an oil immersion objective and a combined magnification of ×500 was used for morphological composition determination. The samples were pulverized to <1 mm, mounted in epoxy resin, and polished (based on ISO 7404 part 2, 2009). Hilgers Fossil software was used for the mean random vitrinite reflectance determination for the coal samples (based on ISO 7404 part 5, 2009). Hilgers Diskus software was used to conduct the maceral point count for the coal samples (based on ISO 7404 part 3, 2009). Quantitative and qualitative point count analyses were conducted on the parent ash samples, applying the method presented in Table 2.10 (based on Hower, 2012).

A dry sieving technique was used to separate the ash into different size fractions. Although dry sieving of powders may cause agglomeration and blinding problems, availability of equipment and time limitations restricted experimentation to this method. Sieves were stacked in a √2 series, loaded with 150 g of sample and sieved for 20 minutes. Each sample was repeated at least three times. A Rosin-Rammler model was fitted to the data to predict the particle size distributions outside the upper and lower sieve boundaries. For the fly ash samples, 38, 53, 75, 106, 150, and 212 µm sieves were used. For the bottom ash samples,

32 | P a g e

75, 150, 300, 600, 1000, and 2000 µm sieves were used. For the gasification ash, 150, 300, 600, 1000, 2000, and 5000 µm sieves were used.

Table 2.10: Modified Hower (2012) ash petrographic classification scheme.

Organics Inorganics

Anisotropic char Glass Isotropic char Quartz Inertinitic char Mullite Unreacted/partially devolatilised coal Dense iron Dendritic iron Baked clay Anorthite Other 2.5.2 Coal characterisation results The proximate, CV and ultimate results for the coal samples are presented in Table 2.11. The proximate and CV results are typical for South African coals; namely, low-grade with high ash yields and moderate to low CVs (Pretorius et al., 2002). Except for C PS4, the coals are all ranked as medium-volatile bituminous (ASTM D388, 2018). C PS4 is ranked as high-volatile bituminous (ASTM D388, 2018). The ultimate results show low sulphur contents (compared to other coal hosting countries), which is also typical for South African coals (Kalenga et al., 2011; Pretorius et al., 2002; Snyman and Botha, 1993). The atomic H/C and O/C ratios are comparable to the Van Krevelen bituminous coal rank (Van Krevelen, 1993).

33 | P a g e

Table 2.11: Proximate, CV, and ultimate analyses results for the coals (dry ash free basis abbreviates to d.a.f).

C PS1 C PS2 C PS3 C PS4 C PS5

Proximate and CV Moisture (a.d.b wt. %) 4.8 2.8 3.9 3.8 4.6 Ash yield (a.d.b wt. %) 23.6 24.9 28.1 30.2 26.6 Volatile matter (a.d.b wt. %) 20.8 22.9 20.0 22.5 22.6 Fixed carbon (a.d.b wt. %) 50.8 49.4 48.0 43.5 46.2

CV (a.d.b MJ/kg) 22.27 23.19 20.87 20.61 20.85

Volatile matter (d.a.f wt. %) 29.1 31.7 29.4 34.1 32.8 Fixed carbon (d.a.f wt. %) 70.9 68.3 70.6 65.9 67.2

CV (d.a.f MJ/kg) 31.10 32.07 30.69 31.23 30.31

Fuel ratio 2.4 2.2 2.4 1.9 2.0 Rank (ASTM D388) Medium- Medium- Medium- High- Medium- volatile bit. volatile bit. volatile bit. volatile bit. volatile bit.

Ultimate Total sulphur (a.d.b wt. %) 0.51 0.75 1.58 0.73 1.06 Carbon content (a.d.b wt. %) 57.40 59.40 54.10 52.50 54.10 Hydrogen content (a.d.b wt. %) 2.87 3.14 2.77 3.06 2.92 Nitrogen content (a.d.b wt. %) 1.23 1.17 1.07 0.90 1.08 Oxygen content (a.d.b wt. %) 9.59 7.89 8.50 8.79 9.61 Moisture and ash (a.d.b wt. %) 28.4 27.7 32.0 34.0 31.2

Total sulphur (d.a.f wt. %) 0.71 1.04 2.32 1.11 1.54 Carbon content (d.a.f wt. %) 80.17 82.16 79.56 79.55 78.63 Hydrogen content (d.a.f wt. %) 4.01 4.34 4.07 4.64 4.24 Nitrogen content (d.a.f wt. %) 1.72 1.62 1.57 1.36 1.57 Oxygen content (d.a.f wt. %) 13.39 10.91 12.50 13.32 13.97

Atomic H/C 0.60 0.63 0.61 0.69 0.64 Atomic O/C 0.13 0.10 0.12 0.13 0.13

The XRF and XRD results for the coal samples are presented in Table 2.12. Because the fusion method using sodium borate was followed during the sample preparation of samples for the XRF analysis, the concentrations of elements in the samples were calculated and reported as elemental oxides.

Although more than 100 minerals have previously been identified in coal, the major minerals are well-known and can be classified into four groups: clay minerals, sulphides and sulphates, carbonates, and oxides (Speight, 2013).

The clay group consists out of hydrated alumino-silicates, namely kaolinite, illite, muscovite and smectite, and contributes significantly to the mineral make-up of coal (Harvey & Ruch,

34 | P a g e

1984, Speight, 2013). The samples tested in this thesis contained significant amounts of kaolinite (12.53 to 17.00 wt. %). Muscovite was only present in C PS4 while illite and smectite were absent in all the samples. These results are typical for South African coals with Hattingh (2012) reporting kaolinite percentages ranging between 12 and 27 wt. %, limited illite and muscovite, and no smectitie in his samples from the Witbank and Soutpansberg coalfields.

Natural sulphates in fresh, unoxidised coal are uncommon (except for gypsum in some cases) and are rather formed during ignition when sulphur dioxide absorbs onto the constituents of ash to form e.g. calcium sulphate/anhydrite (Speight, 2005, Speight, 2013). The most well-known sulphide minerals in coal are pyrite, marcosite and sphalerite (Harvey & Ruch, 1984). Small amounts of gypsum were present in the coals from this study (0.48 to 0.91 wt. %), with the gypsum content in C PS4 being slightly higher than for the other samples (1.4 wt. %). The pyrite percentages were minor (0.08 to 0.9 wt. %) with macrosite and sphalerite being absent. These results are typical for South African coals with Hattingh (2012) reporting pyrite percentages ranging between 0.5 and 5 wt. % and no macrosite and sphalerite in his samples from the Witbank and Soutpansberg coalfields.

Carbonates are recognised by their –CO3 bond and include calcite, dolomite and siderite (Harvey & Ruch, 1984). In the South African coals tested in this thesis, all three these minerals are found in minor and trace amounts. Calcite ranges from 0.18 – 0.56 wt.%, dolomite ranges from 0.19 – 1.12 wt.% and siderite ranges from 0 – 0.47 wt.%. The dolomite content in C PS5 is higher than for the other samples. This corresponds to the higher CaO and MgO contents observed in the XRF data for this sample. Hattingh (2012) found calcite percentages ranging from 0.6 – 6.3 wt. %, dolomite from 0.5 – 4.4 wt. % and siderite from 0 – 0.5 wt. % for his South African samples.

Quartz and rutile are examples of oxides that can be found in coal. The coals form this thesis contained significant amounts of quartz (3.5 to 8.5 wt. %). C PS4 has a higher quartz content, while C PS5 has a lower quartz content compared to the other samples. This corresponds to the slightly higher and lower SiO2 percentages (XRF) for C PS4 and C PS5 respectively. Hattingh (2012) reported trace quantities of rutile (0.1 – 0.16 wt. %) and minor quantities of quartz (0.19 – 11.51 wt. %) in South African coals originating from the Witbank and Soutpansberg coalfields.

Take note that low-temperature ashing did not form part of the XRD sample preparation, therefore amorphous carbon was also detected in large amounts. The amounts are significantly larger than those from proximate and ultimate analyses (as reported in Table

35 | P a g e

2.11). The XRD analysis errors associated with the determination of amorphous material in a sample can be as high as 15 % and, therefore, the lower percentages for the latter.

The major elements present in the coals are in descending order: SiO2, Al2O3, CaO, Fe2O3, and SO3. Hattingh (2012) tested South African coals from the Witbank and Soutpansberg coalfields and made the same observations. The significant amounts of SiO2 and Al2O3 correspond to the large amount of clay and quartz minerals present in the samples.

Table 2.12: XRF analysis results for the ash samples of coal samples and XRD analysis results for the coal samples.

C PS1 C PS2 C PS3 C PS4 C PS5

X-ray fluorescence (wt. %)

SiO2 56.17 52.98 49.86 60.53 45.31

Al2O3 28.98 28.30 31.74 25.03 27.37 CaO 4.24 5.29 3.12 3.02 8.54

Fe2O3 3.35 4.94 7.05 4.37 6.53

SO3 3.29 4.13 3.18 2.88 5.39 MgO 1.37 1.23 0.71 0.98 2.64

TiO2 1.50 1.48 1.89 1.53 1.70

Cr2O3 0.02 0.02 0.03 0.04 0.02

K2O 0.48 0.61 0.69 0.86 0.72 MnO 0.03 0.05 0.03 0.04 0.04

Na2O 0.00 0.02 0.03 0.05 0.31

P2O5 0.43 0.69 0.51 0.31 0.98

V2O5 0.02 0.02 0.03 0.04 0.03

ZrO2 0.06 0.07 0.08 0.06 0.08 BaO 0.12 0.16 0.12 0.12 0.22 SrO 0.12 0.11 0.18 0.09 0.25 ZnO 0.00 0.01 0.00 0.01 0.00

X-ray diffraction (wt. %) Amorphous organic carbon 80.64 77.83 77.69 68.34 80.83 (C)

Quartz (SiO2) 5.38 4.74 4.14 8.52 3.51

Siderite (FeCO3) 0.00 0.36 0.09 0.47 0.04

Kaolinite (Al2Si2O5(OH)4) 12.53 14.84 15.55 17.00 13.12

Pyrite (FeS2) 0.08 0.31 0.90 0.44 0.34 Muscovite 0.00 0.00 2.84 0.00 0.00 (KAl2(AlSi3O10)(FOH)2)

Dolomite (CaMg(CO3)2) 0.32 0.48 0.19 0.67 1.12

Calcite (CaCO3) 0.18 0.53 0.53 0.31 0.56

Gypsum (CaSO4∙2H2O) 0.87 0.91 0.90 1.40 0.48

The petrographic analyses results are presented for the coal samples (Table 2.13). The nomenclature used is that of the International Committee for Coal and Organic Petrology (ICCP) 1994 system (ICCP, 1998; ICCP, 2001; Sýkorová et al., 2005; Pickel et al., 2017).

36 | P a g e

Inertinite percentages are high for all samples with inertodetrinite being dominant. Inertodetrinite is commonly found in South African Witbank and Highveld coals (Everson et al., 2013; Wagner et al., 2018). Inertodetrinite can be described as small (<10 µm) detrital carbon, and is associated with finely dispersed quartz and clays (Figure 2.10) (ICCP, 2001). An inertodetrinite particle finely embedded with quartz and clay is presented in Figure 2.10D. This close association of minerals with coal will lead to unliberated mineral-char particles forming. The unliberated nature will subsequently have an influence on the char beneficiation (Chapter 3).

For southern African coals, semifusinite is divided into reactive and inert semifusinite (Wagner et al., 2018). Reactive semisfusinite has a reflectance value close to that of vitrinite and was formed in semi-anaerobic, waterlogged conditions, where oxidation was limited (Wagner et al., 2018). Inert semifusinite is a paler shade of grey than reactive semifusinite (Wagner et al., 2018).

The total reactive macerals for C PS4 is lower than for the other power stations and implies that the resultant char that will form will be more inertinite rich than the chars from the other samples.

The observable mineral matter contents for C PS4 and C PS5 are relatively high compared to the other samples. Massive quartz particles were seen in C PS4 (Figure 2.11) and are mainly responsible for the higher mineral matter content in this particular sample. All five samples can petrographically be classified as Medium Rank C bituminous coals.

37 | P a g e

Table 2.13: Petrographic analyses results for the coal samples.

C PS1 C PS2 C PS3 C PS4 C PS5

Vitrinite (vol. %) 12.6 15.1 21.4 15.3 13.2 Telinite 0.2 0.0 0.0 0.6 0.2 Collotelinite 8.6 9.8 16.1 8.6 9.0 Vitrodetrinite 0.0 0.0 0.0 0.0 0.0 Collodetrinite 3.2 4.6 5.0 5.7 4.0 Corpogelinite 0.4 0.2 0.0 0.4 0.0 Gelinite 0.2 0.4 0.3 0.0 0.0 Pseudovitrinite 0.0 0.0 0.0 0.0 0.0

Inertinite (vol. %) 67.2 63.5 61.8 48.6 49.8 Fusinite 4.2 2.0 1.2 4.5 3.0 Reactive semifusinite 6.4 6.8 4.7 1.0 4.8 Inert semifusinite 14.2 12.2 9.9 18.4 7.6 Micrinite 0.8 0.6 0.0 0.0 0.0 Macrinite 0.0 0.4 0.0 0.2 0.2 Secretinite 4.4 4.2 3.7 2.9 1.2 Funginite 0.0 0.0 0.0 0.0 0.0 Inertodetrinite R 3.0 4.0 0.9 0.8 1.0 Inertodetrinite I 34.2 33.1 41.3 20.8 32.0

Liptinite (vol. %) 2.8 2.0 0.9 0.8 1.2 Sporinite 2.0 1.6 0.9 0.8 1.2 Cutinite 0.8 0.4 0.0 0.0 0.0 Resinite 0.0 0.0 0.0 0.0 0.0 Alginite 0.0 0.0 0.0 0.0 0.0 Liptodetrinite 0.0 0.0 0.0 0.0 0.0 Suberinite 0.0 0.0 0.0 0.0 0.0 Exsudatinite 0.4 0.0 0.0 0.0 0.0

Mineral Matter (vol. %) 17.4 19.5 15.8 35.3 35.8 Silicate (clay/quartz) 16.0 13.9 10.2 32.0 26.0 Sulfide 0.8 2.8 3.1 1.0 7.8 Carbonate 0.6 2.8 2.5 2.0 2.0 Other 0.0 0.0 0.0 0.4 0.0

Total reactive macerals 24.8 27.9 27.9 17.9 20.2 (vol. %)

Vitrinite reflectance

Rrandom (RoV%) 0.61±0.08 0.68±0.07 0.66±0.09 0.63±0.07 0.66±0.07 Range (RoV%) 0.43-0.83 0.53-0.82 0.45-0.91 0.52-0.97 0.49-0.82 Rank Category Medium Medium Medium Medium Medium Rank C Rank C Rank C Rank C Rank C bituminous bituminous bituminous bituminous bituminous

38 | P a g e

A B

100 µm 100 µm

C D

100 µm 100 µm

Figure 2.10: Inertodetrinite found in the coal samples. A) Vitrinite stringers embedded within an inertodetrinite-rich particle, B and C) Variable grain sizes in an aggregated inertodetrinite-rich particle, D) Silicate minerals (quartz and clay) included in a inertodetrinite-rich particle (Reflected-light, oil immersion, ×500).

A B

100 µm 100 µm

Figure 2.11: Quartz found in the coal samples. A) Massive quartz particle found in C PS4, B) Fine quartz fragments embedded in a massive clay particle (Reflected-light, oil immersion, ×500). 2.5.3 Ash characterisation results The ash samples were visually examined to determine any obvious characteristics (Figure 2.12). The fly ash samples can fittingly be described as “shades of grey”, with the darker tones indicating higher char contents. The fly ash samples from PS1, PS2, PS3, and PS5 were of a light greyish colour with a very fine size distribution (Figure 2.12A). Agglomeration of the fine particles occurred; the samples behaved as a total and not as individual particles

39 | P a g e

(as can be seen from the “clumpiness” evident in the photograph). This agglomeration might be problematic during beneficiation. FA PS4 differed from the other fly ash samples with coarse dark-grey char particles visible (Figure 2.12B). A rapid boiler residence time (Rajoo, 2017) or inefficient milling in this particular power station might be responsible for this coloration, indicating a high LOI.

The bottom ash samples were visibly coarser than the fly ash samples (Figure 2.12C). They consisted of conglomerations of fused, subangular particles. Large, distinct char particles were visible.

Gasification ash is the coarsest of the three ash types (Figure 2.12D). Individual char particles were hand-picked for further investigation. Unreacted coal particles (Figure 2.12E) and char-ash particles that burned according to the “shrinking core model” (Figure 2.12F) were observed. For the latter, it is proposed that coal burns from the outside inwards, leaving a char centre within an ash rim (Everson et al., 2005). The coarse coal feed distribution (<65 mm) (Bunt and Waanders, 2008) to the gasifiers might be responsible for the unreacted coal particles, seeing that large coal particles have a slower burn-out time than small coal particles. Also, a high proportion of carbonaceous shale with <10 %C and high minerals and rock fragments (sandstone, siltstone and mudstone particles) in the feed coals to gasifiers may be responsible for the high proportion of the unburnt carbon in the gasification ash produced during thermo-chemical processes. Wagner et al. (2008) visually distinguished unburned carbon, carbonaceous shales, and “shrinking core” char in gasification ash samples. The unburned carbon particles were further divided into dense carbon, layered carbon, porous carbon, and coal-like carbon (Wagner et al., 2008).

The proximate, XRF, LOI, XRD, and petrography results for the ash samples are presented in Table 2.14.

40 | P a g e

A B

5 mm 5 mm

C D

10 mm 15 mm

E F

5 mm 5 mm

Figure 2.12: Visual examination of ash samples. A) Typical fly ash sample, B) FA PS4 with large, dark char particles clearly visible, C) Typical bottom ash sample, D) Typical gasification ash sample, E) Unreacted coal particles in the gasification ash, F) Char-ash particles in the gasification ash samples burned according to the “shrinking core model”.

41 | P a g e

Table 2.14: Proximate, XRF, LOI, XRD, and petrographic analyses results for the ash samples.

FA FA FA FA FA BA BA BA BA BA GA GA PS1 PS2 PS3 PS4 PS5 PS1 PS2 PS3 PS4 PS5 East West

Proximate Moisture (a.d.b wt. %) 0.2 0.3 0.2 0.5 0.1 0.6 0.5 0.3 0.6 0.8 0.6 0.5 Ash yield (a.d.b wt. %) 96.0 95.6 98.7 92.0 97.7 92.0 96.4 97.5 93.6 93.9 90.7 93.9 Volatile matter (a.d.b wt. %) 1.1 1.2 0.8 0.9 1.3 1.7 2.1 1.2 1.2 3.1 2.4 2.3 Fixed carbon (a.d.b wt. %) 2.7 2.9 0.3 6.6 0.9 5.7 1.0 1.0 4.6 2.2 6.3 3.3

Volatile matter (d.a.f wt. %) 28.9 29.3 72.7 12.0 59.1 23.0 67.7 54.5 20.7 58.5 27.6 41.1 Fixed carbon (d.a.f wt. %) 71.1 70.7 27.3 88.0 40.9 77.0 32.3 45.5 79.3 41.5 72.4 58.9

X-ray fluorescence (wt. %)

SiO2 58.02 56.08 51.24 72.20 45.33 60.28 57.28 52.09 62.90 45.44 53.37 53.47

Al2O3 28.09 27.51 32.76 15.05 29.97 23.04 23.73 29.88 21.30 27.82 24.17 24.97 CaO 4.70 4.88 4.41 2.78 10.57 5.15 5.23 5.30 4.18 11.31 9.22 9.61

Fe2O3 4.46 6.48 6.84 5.64 7.16 7.34 8.66 8.70 6.34 7.91 5.25 4.71 MgO 1.44 1.26 0.74 0.82 2.98 1.52 1.38 0.98 1.12 3.23 2.40 2.25

TiO2 1.53 1.46 1.81 1.15 1.63 1.36 1.36 1.66 1.34 1.54 1.41 1.46

Cr2O3 0.04 0.03 0.01 0.04 0.02 0.04 0.02 0.03 0.07 0.05 0.06 0.03

K2O 0.53 0.73 0.73 0.74 0.92 0.49 0.65 0.64 0.79 0.73 0.69 0.65 MnO 0.04 0.05 0.05 0.04 0.06 0.04 0.07 0.06 0.04 0.07 0.05 0.04

Na2O 0.00 0.00 0.00 0.00 0.08 0.04 0.02 0.02 0.00 0.11 0.54 0.47

P2O5 0.41 0.63 0.41 0.19 0.59 0.30 0.70 0.44 0.34 0.73 0.70 0.72

V2O5 0.03 0.02 0.02 0.03 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.03

ZrO2 0.06 0.07 0.08 0.05 0.07 0.07 0.07 0.08 0.06 0.07 0.07 0.07 BaO 0.12 0.14 0.10 0.08 0.22 0.09 0.16 0.10 0.09 0.20 0.14 0.16 SrO 0.12 0.10 0.19 0.06 0.24 0.11 0.10 0.18 0.10 0.24 0.31 0.30 ZnO 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01

SO3 0.61 0.63 0.33 0.41 0.75 0.54 0.54 0.42 0.45 0.75 1.26 1.09 LOI (wt. %) LOI 3.75 4.01 0.99 7.04 1.93 7.08 3.01 2.26 5.47 5.19 9.04 6.90

42 | P a g e

X-ray diffraction (wt. %) Amorphous 42.26 47.49 38.75 30.16 58.28 48.53 50.33 49.06 44.74 53.34 58.07 55.05

Quartz (SiO2) 17.52 18.66 13.47 52.70 7.18 22.93 25.18 11.17 29.60 7.37 12.51 12.30

Mullite (3Al2O32SiO2 / 2Al2O3SiO2) 30.02 29.70 42.07 15.30 28.65 15.50 20.66 24.55 15.30 17.27 14.70 17.60

Hematite (Fe2O3) 0.64 0.95 1.31 0.68 1.65 0.76 1.28 1.21 0.63 0.59 0.37 0.67

Magnetite (Fe3O4) 1.29 0.68 0.73 0.43 0.35 0.42 0.63 0.35 0.40 0.36 0.41 0.48 Lime (CaO) 0.24 0.30 0.21 0.11 0.96 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Cristoballite (SiO2) 0.20 0.10 0.22 0.03 0.26 0.84 0.15 0.37 0.29 0.12 0.43 0.77

Sillimanite (Al2SiO5) 0.33 0.67 2.50 0.21 0.80 0.90 0.53 1.52 0.05 0.17 0.16 0.59

Calcite (CaCO3) 0.14 0.05 0.00 0.36 0.31 0.19 1.24 0.41 0.77 1.84 0.97 0.90

Anorthite (CaAl2Si2O8) 0.00 0.00 0.00 0.00 0.00 9.91 0.00 11.37 8.21 18.93 12.39 11.69 Periclase (MgO) 3.36 1.41 0.73 0.00 1.56 0.00 0.00 0.00 0.00 0.00 0.00 0.00

Petrography (vol. %) Glass 84.0 70.4 79.7 34.7 92.8 77.8 75.2 86.4 72.3 85.5 79.9 67.0 Quartz 7.6 19.2 9.8 51.4 1.6 10.2 15.8 4.7 16.1 1.8 6.3 5.6 Mullite 0.4 0.0 0.0 0.0 0.0 0.4 0.6 0.8 0.2 1.6 1.0 1.2 Dense iron 1.6 3.3 5.9 3.4 1.6 1.2 3.3 2.5 1.2 2.7 3.3 4.8 Dendritic iron 0.8 2.5 2.6 1.8 0.6 2.0 2.5 1.4 0.0 0.8 2.0 2.6 Baked clay 0.6 0.4 1.0 1.6 0.6 1.2 0.4 1.2 1.6 0.2 0.8 2.4 Anorthite 0.0 0.0 0.0 0.0 0.0 0.2 0.0 0.0 0.2 0.0 0.0 0.0 Other 0.4 0.2 0.0 0.4 0.0 1.0 0.6 0.8 0.0 3.1 3.7 8.2 Total inorganics 95.4 96.0 99.0 93.3 97.2 94.0 98.4 97.8 91.6 95.7 97.0 91.8 Anisotropic char 1.6 2.0 0.0 1.6 1.4 2.8 1.0 1.2 3.1 3.1 2.2 5.0 Isotropic char 0.4 0.2 0.2 0.8 0.2 0.4 0.0 0.0 1.8 0.2 0.0 1.2 Inertinitic char 2.6 1.8 0.8 4.3 1.2 1.8 0.6 1.0 3.1 1.0 0.6 0.8 Unreacted / partially burned coal 0.0 0.0 0.0 0.0 0.0 1.0 0.0 0.0 0.4 0.0 0.2 1.2 Total organics 4.6 4.0 1.0 6.7 2.8 6.0 1.6 2.2 8.4 4.3 3.0 8.2

43 | P a g e

Take note that the standards used for proximate analysis were developed for coal samples; for the “charred” material the analysis temperature might not be sufficient for the complete combustion of carbon. The fixed carbon might be underestimated and the ash yield overestimated by using the standard procedure. There is no proximate standard for coal ash.

The major inorganic elements found in the ash samples are Si, Al, Fe, and Ca, while Mg and Ti can be listed as minor elements. The gasification ash samples also have relatively high

SO3 percentages. Based on the ASTM C618a (2012) classification, the fly ash samples are all categorised as Class F ash, while the bottom ash samples are classified as Class B ash.

The major phases in the ash samples are amorphous material and the minerals quartz, and mullite. Needle mullite forms at temperatures higher than 1400 °C (Bryers, 1986). Due to the high melting point of mullite it may not be completely liquid even at temperatures as high as 1800 °C (Bryers, 1986). Amorphous percentages for the samples were estimated between 30 and 60 wt. %. However, the XRD error on the amorphous count can be as high as 15 wt. % (Verryn, 2017). Although XRD can quantify the amorphous content, it can by no means explain the composition of the amorphous phase itself. More advanced methods are needed such as automated SEM.

Petrographically, the amorphous fraction consisted of glass and baked clay particles (Figure 2.13A and B). The glass particles appear as angular and “cloudy”. This morphology is different than what was seen in some of the other ERA-MIN consortium ash samples. As an example, typical glass particles from Portugal are also illustrated (Figure 2.13C and D). The Portuguese glass particles are mostly solid spheres, cenospheres, plerospheres, iron-glass spheres, and crystal inclusion (mullite) spheres (as classified by Fisher et al., 1978 and Watt and Thorne, 1965). The difference in behaviour might be due to the different clays found in the coal samples. For the Portuguese coal (originating from Colombia) both illite and kaolinite are present. Potash and water in illite clay will flux to form an alumino-silicate melt phase (Hubbard et al., 1984). The formation of carbon dioxide will froth the melt, and upon recrystallization will form hollow, gas-filled cenospheres (Hubbard et al., 1984). For the South African coal samples, only kaolinite is present. Without any potash in the kaolinite to flux, the clay will only dehydrate to form metakaolinite and will subsequently transform into glass and mullite (Hubbard et al., 1984). Hence, the more angular particles found in these ash samples. The baked clay particles consist of a metakaolinite centre and a glassy rim. In some instances, char fragments can also be seen in the core (Hower, 2012). The baked clay volumes in the samples were low, indicative of complete transformation.

Quartz is a coal relic that typically remains unchanged during combustion and gasification (Speight, 2005). Referring back to the coal analyses, it was seen that the quartz contents

44 | P a g e were high and low respectively for C PS4 and C PS5. Therefore, the subsequent quartz contents in FA PS4 and FA / BA PS5 are also respectively high and low (compared to the other samples). The XRF data also shows high and low SiO2 percentages for the mentioned ash samples. The quartz particles in FA PS4 are unusually large; indicative of possible inefficient coal pulverisation in this particular power station. The same large quartz particles were also observed in the corresponding coal samples (Figure 2.11).

A B

100 µm 100 µm

C D

100 µm 100 µm

Figure 2.13: Amorphous particles in the ash samples. A) Typical glass particle found in South African ash samples, angular and “cloudy;, B) Baked clay particle with a metakaolinite centre and a glassy rim; C) Typical glass particle for Portuguese ash sample (ERA-MIN consortium example), plerosphere; D) Typical glass particle for Portuguese ash sample (ERA-MIN consortium example), cenosphere (Reflected-light, oil immersion, ×500).

Mullite forms via the direct crystallisation of clays in coal or during the devitrification of the glass phase in ash (Hubbard et al., 1984). Mullite occurs as needle-like particles, finely interwoven into glass and residue clay particles (Figure 2.14A). The petrographic volume count of the mullite is very low. Hubbard et al. (1984) also reported on the lack of visually discrete mullite grains in their samples. Matjie et al. (2011) commented on mullite’s needle- like embedment within the glass phase. The bottom and gasification ash samples have relatively lower percentages of mullite compared to the fly ash samples. This might be due to

45 | P a g e the presence of anorthite (end member of the plagioclase group) in these samples. The lower residence time, the lower temperature at the bottom of the pulverised boilers (approximately 1200 ºC) and the reaction of metakaolinite and lime produces a melt which crystallises into anorthite upon slow cooling on the boiler / gasifier walls (Matjie et al., 2008). Elongated needles of anorthite can be seen in the samples (Figure 2.14B).

A B

100 µm 100 µm

Figure 2.14: A) Needle-like mullite crystals embedded in a alumino-silicate glass matrix, B) Elongated anorthite crystals observed in the bottom and gasification ash samples (Reflected-light, oil immersion, ×500).

Other minerals present are cristoballite, sillimanite, hematite, and magnetite. Minor amounts of periclase and lime are also found in the fly ash samples. The lime content for FA PS5 is slightly higher than for the other samples. XRF also showed higher CaO percentages for FA PS5 and BA PS5. This might be ascribed to the higher dolomite content in the feed coal to this power station. The gasification ash samples also show higher CaO contents. Although coal data for the gasification ash was not available, this result is typical for gasification ash (Hlatshwayo et al., 2009; Matjie et al., 2005). Hematite and magnetite form the ferrospheric fraction of ash and were morphologically classified into dense and dendritic ferrospheres (Figure 2.15). Due to the presence of the iron on the surface of glass particles, Hubbard et al. (1984) concluded that it must be a neocrystallization phase. Valentim et al. (2016) also observed that iron occurs as finely dispersed crystals trapped within amorphous particles, most likely formed due to the release of iron oxide plumes into the alumino-silicate melt.

Various char morphologies were also identified via petrography. Details on the char morphological composition are presented in Chapter 3. For the majority of the samples, a mix of inertinitic char and anisotropic char is present. The high inertinitic char contents make sense seeing that it is a relic originating from the high inertinite percentages in the corresponding coal samples. The total char in ash percentages are low (<10 vol. %). These low percentages can also be seen from the LOI results (<10 wt. %). This implies that all / a significant amount of the char needs to be recovered if it is to be used as a value-added

46 | P a g e product. The total char volume / LOI for FA PS4 is higher than for the other fly ash samples and might be due to inefficient milling or a rapid residence time experienced in this power station (Rajoo, 2017).

A B

100 µm 100 µm

C D

100 µm 100 µm

E F

100 µm 100 µm

Figure 2.15: A) and B) Dense ferrospheres in ash samples; C) and D) Dendritic ferrospheres in ash samples; E) Dense ferrosphere found in GA East; F) Dense iron fragments mantling a glass particle (Reflected-light, oil immersion, ×500).

The cumulative PSDs are given for the fly ash samples (Figure 2.16A). It can be seen that the PSD for FA PS4 differs from the other four fly ash samples as it is much coarser in size. The other four ash samples are more in line with what is expected. Eskom pulverises their coal feed to 70 % passing 75 μm (Van Alphen, 2017) and as can be seen from the figure,

47 | P a g e the 70 % passing mark for PS1, 2 and 5 fly ash are all 75 μm or lower. The 70 % mark for PS3 is at 80 μm. Normally the fly ash size distribution will follow the coal grain size (Van Alphen, 2005). The slightly over pitched distributions might be due to coalescence or the swelling of vitrinitic and reactive inertinite macerals (Van Alphen, 2005). For PS4, however, the PSD is much coarser and it is suspected that poor coal quality and inefficient milling is the cause of this. The large quartz and char particles observed in this sample strengthen this possibility.

The cumulative PSDs are given for the gasification ash samples (Figure 2.16B). The gasification ash samples are the coarsest of the three types with d70 (70 % passing size) values of 9-10 mm. This is due to the coarser coal feed size distribution of the gasifier. The smaller sizes that are observed in the figure are due to thermal and mechanical fragmentation that is quite pronounced in gasifiers (Bunt, 2006).

The cumulative PSDs are given for BA PS1, BA PS3, and BA PS4 (Figure 2.16C). The PSDs are coarser than for the fly ash samples. Since bottom ash comprises agglomerated ash particles that are too heavy or large to be carried with the flue gasses, this observation makes sense. The cumulative PSDs are given for BA PS2 and BA PS5 (Figure 2.16D). For these two bottom ash samples, the size distributions are finer than observed for the other bottom ash samples. The reason for this is unclear.

48 | P a g e

A 100 B 100

90 90

80 80 d70 d70 70 70 60 FA PS1 60 50 FA PS2 50 GA East 40 FA PS3 40 GA West 30 FA PS4 30 20 20 FA PS5 10 10 0 Cumulative mass passing (%) masspassingCumulative 0

0 50 100 150 200 250 300 350 Cumulative mass passing (%) 0 5 10 15 20 25 30 35 40 45 50 55 60 Size (µm) Size (mm)

C 100 D 100

90 90

80 80 d70 d70 70 70 60 60 50 BA PS1 50 BA PS2 40 BA PS3 40 BA PS5 30 BA PS4 30 20 20 10 10

0 (%) masspassingCumulative 0 Cumulative mass passing (%) masspassingCumulative 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0 50 100 150 200 250 300 350 400 450 500 550 600 Size (mm) Size (µm)

Figure 2.16: Particle size distributions (PSDs) for A) Fly ash samples, B) Gasification ash samples, C) Bottom ash samples 1, 3 and 4, D) Bottom ash samples 2 and 5.

49 | P a g e

2.6 Summary South Africa has 19 coalfields; collectively hosting more than 40 billion tonnes of recoverable ore reserve. The coal is mainly used for electricity generation and synfuels production. Eskom is South Africa’s largest electricity provider with 15 baseload coal power stations currently in operation or the construction phase. Eskom utilises coal from the Witbank Coalfield.

Sasol Synfuels (Secunda) is the only bituminous CTL operation globally and is partially supplying South Africa with synthetic fuels. The company uses the patented FBDB gasifiers for this purpose. Sasol obtains coal from mines in the Highveld Coalfield.

The quality of coal being burned in South Africa is low (average 30 % ash yield). As a result, more than 50 million tonnes of ash waste is produced annually in South Africa. Eskom alone produces more than 34 million tonnes of ash per annum, and only sells 7 %. Sasol produces 7 million tonnes of gasification ash annually. The gasification ash is landfilled.

Power station ash from five different power stations as well as their respective coal feeds was characterised in detail. Two gasification ash samples obtained from Sasol Synfuels were also characterised. This characterisation formed part of the first phase of the current research.

The feed coal samples had high ash yields and low CV values, typical of South African coal, and indicative of a low-grade. The sulphur contents were low, also typically for South African coal. Kaolinite and quartz were the major minerals identified in the coal samples. The coal samples were petrographically classified as Medium Rank C bituminous coal and inertodetrinite was dominant in all coal samples. Large quartz particles were seen in C PS4, and due to its relic status, also in the corresponding ash.

Amorphous material, mullite, and quartz were the major phases identified in the ash samples. Anorthite was also associated with the bottom and gasification ash samples. The LOI percentages of the ash samples were below 10 wt. %, indicating efficient combustion and gasification. The low LOI percentages indicate that, for economic purposes, all or a significant amount of the char will have to be recovered if it is to be used in value-added applications. For FA PS4 the LOI was relatively high compared to the other fly ash samples.

The corresponding PSD for FA PS4 was also relatively coarse with a d70 of 175 µm.

FA PS2, FA PS4, BA PS4, and GA East will be used going forward.

50 | P a g e

Chapter 3: Char-ash separation and char characterisation

3.1 Introduction Char found in coal conversion ash is used as activated carbon, fluid absorber (mercury), or as a precursor for synthetic graphite (Bartoňová, 2015; Hower et al., 2017). However, before utilisation can take place it is necessary to separate the char from the rest of the ash minerals. Due to the low starting char in ash percentages (0.4 to 7.8 wt. % LOI reported for South African sources), the relatively high purities required for value-added applications (~90 wt. % fixed carbon), and the high carbon recoveries that need to be achieved, this is often seen as a challenge. In this chapter, a laboratory char-ash separation process was developed and assessed using fly, bottom, and gasification ash samples obtained from various South African coal conversion utilities (Section 3.3). In Section 3.2 char-ash separation case studies by Hwang et al. (2002), Cabielles et al. (2008), and Maroto-Valer et al. (1999a) are provided. In Section 3.4 characterisastion of the concentrated char from the current research with a discussion on its applicability as a possible precursor for synthetic graphite is provided.

3.2 Case studies 3.2.1 Hwang et al. (2002) Free mercury is known to be a serious health and environmental hazard. Trace amounts of mercury can be found in coal, and this is released in volatile gases upon burning (Garnham and Langerman, 2016). In order to reduce mercury emissions from coal-burning utilities, activated carbons are currently used to capture it (Hwang et al., 2002). Activated carbons are, however, very costly and cheaper alternatives are sought. Due to its porous nature and “waste” status, char found in coal ash is considered as a possible alternative to traditional activated carbon (Hwang and Li 2000; Li et al., 2002; Li and Maroto-Valer 2012; Luo et al., 2004). In the study by Hwang et al. (2002), char was separated from ash and subsequently characterised for mercury adsorption practices. An intertwining, complicated, network of separation steps was followed, with not only char being separated but simultaneously also other valuable components such as cenospheres and ferrospheres. A “clean” ash product was also obtained to be used as an additive in cement and concrete manufacturing.

The separation approach followed by Hwang et al. (2002) is given in Figure 3.1. Three fly ash samples were used:

 Sample 1 - Class F ash;  Sample 2 – Mixed Class F and C ash; and  Sample 3 - Class C ash.

51 | P a g e

Figure 3.1: Char-ash separation schematic followed by Hwang et al. (2002).

52 | P a g e

LOI (950 °C) was used to determine the carbon percentage after each separation stage.

The samples were first subjected to a size separation stage. In this stage, two products were obtained: a +150 µm product and a -150 µm product. Every separation technology has its own size limitations where processing will take place at maximum efficiency (Wills and Napier-Munn, 2006). By dividing the samples into a coarse and fine fraction, the appropriate technologies could thus be exploited with minimum loss of material.

The +150 µm product was introduced to density and subsequent electrostatic separation stages. A Wilfley table (Super Duty RH15S SD) was used for density separation. Due to surface friction and extended settling times, fine particle sizes are problematic to separate with the standard sink-float technique (Wills and Napier-Munn, 2006). Therefore, the use of +150 µm sizes and a Wilfley table was appropriate. Lee et al. (2012) also commented on the usage of a shaking table for char-ash separation, seeing that the additional shaking force can overcome friction problems associated with fines.

An Eriez electrodynamic electrostatic separator (18 kV voltage and 60 rpm drum speed) was used for electrostatic separation trials. For char-ash separation, a triboelectrostatic separator is preferred due to its ability to separate minerals with small differences in their work functions (Bada et al., 2010). Das et al. (2010), however, also utilised and tested char-ash separation via electrodynamic / Corona separators.

The -150 µm fraction (only sample 1) was fed to a froth flotation pilot plant at 91 kg/hr. Froth flotation is known to be able to treat fines (can operate as small as 10 µm) (Wills and Napier- Munn, 2006). The cenospheres were first removed in a water tank (cenospheres are less dense than water) where after the remaining slurry was fed to a magnetic drum separator. Ferrospheres were collected at this stage. Although the overall aim was not to obtain cenospheres and ferrospheres, these two morphologies are known to be value-added products, and removing them simultaneously with the char makes sense.

The residue slurry was conditioned with a fuel oil #2 collector and introduced to a double- stage froth flotation scheme. Froth flotation is a surface phenomenon and rests on the principle that certain minerals are hydrophobic / oily (water repellent) and others are hydrophilic (Fuerstenau and Somasundaran, 2003). In an aerated water tank, the hydrophobic particles will attach themselves to the air bubbles while the hydrophilic particles will reside in the water (Dong, 2010). Due to their low densities, air bubbles (with their additional load) will float to the top of the tank where they can be skimmed off (Dong, 2010). In this manner minerals with different degrees of hydrophobicity are separated.

53 | P a g e

In the study by Hwang et al. (2002), the floating carbon from stage 1 was fed to stage 2, while the rejects from stage 2 were recycled back to stage 1 (continuous mode). In both tanks, the collector dosage was kept at 0.9 kg/tonne. Zhang and Honaker (2015) mentioned that char particles are very porous; implicating that the amounts of collecting / frothing agents added to the flotation process would be high as that the char will absorb most of it.

The major results are provided in Table 3.1. The +150 µm process treatment yielded an 80.90 wt. % LOI product for sample 1, a 73.40 wt. % LOI product for sample 2, and a 70.10 wt. % LOI product for sample 3. Although these percentages are relatively high, the starting LOIs were unfortunately not published and hence an indication of the change could not be obtained. The high carbon recoveries for the gravity separation step are notable - for all samples more than 80 wt. % of the feed carbon was recovered. The -150 µm process treatment yielded a 67.70 wt. % LOI for sample 1. Even though the -150 µm process was run in continuous mode, it still delivered lower LOIs than the batch +150 µm processes. This might be due to the limited abilities of froth flotation with char-ash material and the fine particle sizes.

The discard streams of all separation steps had low LOIs (< 6 wt. % LOI), implying that they can be used for Portland cement manufacturing (ASTM C618a, 2012 regulations). A stream of cenospheres and a stream of ferrospheres were also generated for usage in value-added applications.

54 | P a g e

Table 3.1: Char-ash separation results produced by Hwang et al. (2002).

ID LOI (wt. %) Carbon recovery / yield (%)

Sample 1

Parent sample Not provided

+150 µm 36.30 Not provided

Gravity separation +150 µm 63.44 95.01 recovery

Electrostatic separation (stage 1 and 2) 80.90 73.00 yield

-150 µm 21.71 Not provided

Froth flotation 67.70 Not provided

Sample 2

Parent sample Not provided

+150 µm 41.17 Not provided

Gravity separation +150 µm 65.93 83.96 recovery

Electrostatic separation (stage 1 and 2) 73.40 76.00 yield

Sample 3

Parent sample Not provided

+150 µm 37.42 Not provided

Gravity separation +150 µm 65.03 90.93 recovery

Electrostatic separation (stage 1 and 2) 70.10 79.00 yield

3.2.2 Cabielles et al. (2008) Natural graphite has been classified as a raw critical commodity by the EU, the U.S. Government, and the British Geological Survey (British Geological Survey, 2015; European Commission, 2017; Fortier et al., 2018). It has a high economic value and a high supply risk. For this reason, alternative natural sources or synthetic graphite precursors are highly sought after. A precursor that might be suitable for synthetic graphite manufacturing is the char fraction found in coal conversion ash. In the paper by Cabielles et al. (2008), this suggestion was investigated. As the current research aims to address the same issue, it is appropriate to understand how Cabielles et al. (2008) went about separating the char from the ash.

The experimental approach used by Cabielles et al. (2008) is provided in Figure 3.2. Three fly ash samples, labelled A-C, were obtained from different Spanish power stations burning anthracite. The LOI method was used (815 °C for 2 hours) to determine product purity.

55 | P a g e

As a first iteration, all three samples were subjected to a screening stage where the +80 µm fraction was taken as the product, as the carbon was concentrated in the coarser size fractions. The removal of the -80 µm particles also resulted in the removal of fines, which is known to cause agglomeration and conglomeration problems during separation (Wills and Napier-Munn, 2006).

After size separation, the samples were subjected to an oil agglomeration process. This particular oil agglomeration process was designed by Valdés and Garcia (2006). It has a superior economic benefit over other separation technologies (e.g. froth flotation) seeing that the oil being used is derived from waste household vegetable oils. A Waring blender with a 1L capacity was used (Valdés and Garcia, 2006). Samples (16 g) were first dispersed in water (400 ml) for 5 minutes at 11 000 revolutions/min (Valdés and Garcia, 2006). The oil was then added at different concentrations (1, 3, and 5 wt. %) before mixing for 60 seconds at a blending speed of 11 000 revolutions/min. Gray et al. (2001) also used oil agglomeration in their char-ash separation studies. They were able to obtain 55-70 wt. % LOI products with carbon recoveries of more than 60 wt. %. Although the oil agglomeration separation method can be used for graphitization purposes, using it for other char applications (such as mercury adsorption and usage as a catalyst) will be problematic seeing that char is porous and the oil will penetrate the pores. It, therefore, limits its final usage.

Sample C was also subjected to HCl / HF acid demineralisation (ISO 602-1983) after size separation. In a “real-life” scenario, using acid demineralisation will be impractical seeing that it is environmentally unfriendly and the costs associated with acidification is high. Usually, acidification and leaching are only considered in the beneficiation of extreme valuable commodities, such as gold (Wills and Napier-Munn, 2006).

56 | P a g e

Figure 3.2: Char-ash separation schematic followed by Cabielles et al. (2008).

A summary of the LOIs achieved (850 ºC) is given in Table 3.2. As can be expected, the acid demineralisation process yielded the best separation results. However, the acid-treated sample also delivered the worst graphitization results due to the absence of minerals that can act as catalysts (Cabielles et al., 2008; 2009). Overall, high carbon percentages were reached from the combination of size and oil agglomeration separation. The LOIs of the parent samples were not published; hence an indication of the LOI change could not be obtained. Carbon recovery / sample yield was also not considered.

57 | P a g e

Table 3.2: Char-ash separation results produced by Cabielles et al. (2008).

ID LOI (wt. %)

1 Size separation A 54.64

2 Size separation B 20.93

3 Size separation C 10.41

4 Size and oil agglomeration at 1 wt. % concentration A 87.36

5 Size and oil agglomeration at 1 wt. % concentration B 78.35

6 Size and oil agglomeration at 5 wt. % concentration B 68.02

7 Size and oil agglomeration at 1 wt. % concentration C 76.36

8 Size and oil agglomeration at 3 wt. % concentration C 67.03

9 Size and oil agglomeration at 5 wt. % concentration C 62.06

10 Size and HCl / HF acid demineralisation C 97.43

3.2.3 Maroto-Valer et al. (1999a) Currently, the main market for coal ash utilisation is in Portland cement, where fly ash is used as an additive / binder. However, the presence of char in the ash causes problems as it adsorbs AEAs and surfactants that are added to the cement, consequently destroying its freeze-thaw resistance, increasing its water requirements, increasing the AEA dosage, and altering the colour of the final concrete product (Pedersen et al., 2008). The morphology of char particles differs with anisotropic, isotropic, and inertinitic char being present in ash (Hower, 2012). Maroto-Valer et al. (1999a) believe that one of these forms is responsible for the adsorption of the AEA / surfactants. Therefore, their study was not only aimed at separating the char from the ash, but also into its three different morphological forms. The adsorption capabilities were subsequently tested (Maroto-Valer et al., 1999b). Seeing that anisotropic char also has a higher degree of graphitization, their separation findings might be of importance for the current research.

The experimental setup used by Maroto-Valer et al. (1999a) is provided in Figure 3.3. Two fly ash samples from the U.S., known as Dale and WEPCO, were used.

Petrography was used to obtain the initial char contents, with WEPCO and Dale having total starting volume percentages of 38.5 and 36.4 respectively. The petrographic distribution between anisotropic, isotropic, and inertinitic char for the initial samples are given in Table 3.3. The parent Dale sample was anisotropic-rich, while the WEPCO sample was isotropic- rich.

58 | P a g e

Anisotropic, isotropic, and inertinitic char are known to have different densities. These differences in densities were exploited as the main method of separation by Maroto-Valer et al. (1999a). Density separation in the form of “sink-float” is applied to large particle sizes (Wills and Napier-Munn, 2006). Small particles, such as in fly ash, experience surface friction and therefore settling times are long and efficiency poor (Wills and Napier-Munn, 2006). For this reason, Maroto-Valer et al. (1999a) used Density Gradient Centrifugation (DGC) in which the settling times are sped by the action of centrifugal forces. A Beckman J2- 21 centrifuge with a titanium JCF-Z zonal-core rotor was used. As a dense medium, lithium heteropolytungstate (LST) was chosen. The advantages of using LST over bromoform are that it is non-toxic, recoverable, can be diluted with water, has a low viscosity (10cP), and reaches a high span of densities (Mounteney, 2011). The densities tested ranged between 1.2 to 2.25 SG and the separation was conducted at a rotor speed of 15000 rpm for one hour.

Before the samples were pre-concentrated, Dale was screened at +106 µm and subjected to triboelectrostatic separation. In a separate experiment, Dale was digested with HF / HCl before DGC. For WEPCO the particle size distribution was too fine (< 100 µm) for size separation and therefore it was only subjected to triboelectrostatic separation. The reason for pre-concentration was to increase the char in ash and therefore minimizing the number of DGC runs required.

59 | P a g e

Figure 3.3: Char-ash separation schematic by Maroto-Valer et al. (1999a).

60 | P a g e

A summary of the major results is given (Table 3.3). In terms of pre-concentration, it can be seen that Dale 1 increased to 90.5 vol. %, Dale 2 to 95.0 vol. %, and WEPCO to 87.5 vol. %.

Table 3.3: Char-ash separation results produced by Maroto-Valer et al. (1999a) (using petrography).

ID Inertinitic Isotropic Anisotropic Total (vol. %) (vol. %) (vol. %) (vol. %)

Dale 1

Dale parent 3.8 13.4 19.2 36.4

Dale +106 µm 2.0 13.0 25.5 40.5

Dale triboelectrostatic 5.5 24.0 61.0 90.5

Dale DGC: 1.32-1.65 SG 76.5 7.0 1.0 84.5

Dale DGC: 1.725-1.75 SG 13.1 72.1 8.7 93.9

Dale DGC: 1.92-1.94 SG <0.5 19.5 76.0 96.0

Dale 2

Dale parent 3.8 13.4 19.2 36.4

Dale demineralisation 5.5 37.5 52.0 95.0

Dale DGC 1.50-1.60 SG 85.0 10.0 2.0 97.0

Dale DGC 1.725-1.75 SG 5.0 63.0 30.0 98.0

Dale DGC 1.85-1.975 SG 1.0 28.0 70.0 99.0

WEPCO 3

WEPCO parent 4.0 27.2 7.3 38.5

WEPCO triboelectrostatic 8.5 51.5 27.5 87.5

WEPCO DGC: 1.50-1.60 SG 85.5 8.5 2.5 96.5

WEPCO DGC: 1.725-1.75 SG 9.5 78.5 11.5 99.5

WEPCO DGC: 1.875-1.90 SG 2.0 20.0 75.5 97.5

It seems as though triboelectrostatic separation did not influence char speciation results. For the DGC results, a clear split in char speciation can be seen at different densities. Inertinitic char is concentrated in the 1.3-1.65 SG density range, isotropic char in the 1.725-1.75 SG density range, and anisotropic char in the 1.85-1.975 SG density range. Overall, speciation concentration was the best in WEPCO with all volume percentages above 75. Maroto-Valer et al. (1999a) ascribed this behaviour due to fewer mixed char-ash particles. This is an important result showing that the unliberated nature of minerals can have a substantial effect on the separation efficiency. The maximum yields for Dale 1, Dale 2, and WEPCO were

61 | P a g e found at densities of 1.8-1.925 SG, 1.85 SG, and 1.82 SG respectively. Splitting into carbon species leads to very small volumes of material being recovered. This is undesirable, especially considering the low starting char percentages found in waste ash.

3.2.4 Other

Stanislav Vassilev and co-workers have also done some excellent work regarding the multicomponent utilization of ash. A series of articles were published from their work.

In 2003 (Vassilev et al., 2003) their first article was published in which elaborated characterisation results are provided for five fly ash samples from Spanish pulverised fuel power stations, and their corresponding anthracite coal feeds.

In 2004 (Vassilev et al., 2004a) a second paper was published in which the methods to separate ceramic cenospheres and salt concentrates from the five fly ash samples were provided. The subsequent concentrates were also characterised. The ceramic cenosphere fraction was separated via a sink-float method. Water was used as the dense medium. The salt fraction was evaporated from water-soluble solutions leached from the cenosphere-free fly ash samples.

In the same year (Vassilev et al., 2004b), their paper discussing the methods to separate magnetic and char concentrates was published. They have also characterised the separated fractions in detail. The magnetic extraction step followed the ceramic cenospheres and salt concentrates extraction steps. A hand-held Fe magnet was used. The char concentrates were separated from the non-magnetic fraction by using sieving (>100 μm) and froth flotation methods.

Lastly, Vassilev and Menendez (2005) published a paper concerning the concentration and characterisation of the heavy concentrates from the fly ashes, and the characterisation of the fly ash residues. The extraction of the heavy concentrates followed the isolation of the ceramic cenospheres, the salt concentrates, the magnetic fraction, and the char fraction. The heavy concentrates were extracted by using bromoform as a dense medium in a sink- float process. The fly ash residues were the remaining samples.

Vassilev and Vassileva (2005; 2007) have also published some extensive work on the characterisation of coal ash samples.

62 | P a g e

3.3 Char-ash separation of South African ash samples 3.3.1 Methodology In Chapter 2 a variety of coal conversion ash samples originating from South African combustion as well as gasification utilities were characterised in detail. Based on the results, two fly ash samples, a bottom ash sample, and a gasification ash sample were selected for the char-ash separation trials. A summary of the main properties for each sample is given (Table 3.4).

FA PS4 was chosen due to its high LOI percentage as well as its coarse size distribution. These two properties can make a substantial difference during separation. FA PS2, with a low LOI and fine size distribution, was chosen as a more representative sample for the fly ash samples from South Africa. BA PS4 was collected at the same utility boiler and at the same time as FA PS4. Therefore, this sample was selected to represent the bottom ash samples. For the gasification ash, GA East was selected. This sample will make for interesting separation seeing that it has a very coarse particle size distribution and a high LOI value. In terms of its char characteristics, it is also different from the fly and bottom ash samples. Visual observations showed the presence of unreacted coal particles in this sample.

Table 3.4: Properties of samples selected for the char-ash separation trials.

FA PS2 FA PS4 BA PS4 GA East

Sample type Fly ash Fly ash Bottom ash Gasification ash

Feed coal quality (a.d.b) 23.19 MJ/kg CV, 20.61 MJ/kg CV, 20.61 MJ/kg CV, Not provided, 0.75 wt. % 0.73 wt. % 0.73 wt. % but assumed to sulphur, Med sulphur, Med sulphur, Med be Med Rank C, Rank C, high Rank C, high Rank C, high high ash, and inertinite inertinite inertinite high inertinite

Loss on ignition (wt. %) 4.01 7.04 5.47 9.04

Major phases Glass, mullite, Glass, mullite, Glass, mullite, Glass, mullite, quartz, hematite, quartz (large quartz, hematite, quartz, hematite, magnetite, and frequent), magnetite, magnetite, periclase hematite, calcite, anorthite calcite, anorthite magnetite d70 (μm / mm) 76.80 μm 184.20 μm 0.75 mm 9.80 mm

Visual characterisation Fine, Coarse, large Large with large Large, rock-like, agglomerated, black char char particles unreacted coal, light grey particles can char centre with clearly be seen an ash rim

63 | P a g e

Carbon grade, carbon recovery, and sample yield were used to assess the efficiency of the separation trials. Carbon grade can be defined as the purity of the product stream (Equation 3.1) and, due to cost and time constraints, was measured with the LOI method (combustion at 815 °C - Section 2.5.1). A summary of other methods that can also be used for this purpose is provided (Table 3.5). In the char characterisation section, all these methods are considered. For the ERA-MIN collaboration a 90 wt. % carbon grade was targeted.

퐶푎푟푏표푛 푝푟표푑푢푐푡 푠푡푟푒푎푚 (푔) Equation 3.1 퐶푎푟푏표푛 푔푟푎푑푒 = × 100 푃푟표푑푢푐푡 푠푡푟푒푎푚 (푔)

Carbon recovery can be defined as the percentage of feed carbon distributed to the product stream. Seeing as the initial char in ash percentages are low, it is important to extract all / significant amount of carbon; recovery should, therefore, be as high as possible.

퐶푎푟푏표푛 푝푟표푑푢푐푡 푠푡푟푒푎푚 (푔) Equation 3.2 퐶푎푟푏표푛 푟푒푐표푣푒푟푦 = × 100 퐶푎푟푏표푛 푓푒푒푑 푠푡푟푒푎푚 (푔)

Sample yield can be defined as the percentage of feed distributed to the product stream. The sample yield should initially be kept as low as possible, seeing that the bulk mass consists of non-carbon phases.

푃푟표푑푢푐푡 푠푡푟푒푎푚 (푔) Equation 3.3 푆푎푚푝푙푒 푦푖푒푙푑 = × 100 퐹푒푒푑 푠푡푟푒푎푚 (푔)

The separation process flow diagram is provided in Figure 3.4. The flow diagram was put together based on case studies, equipment availability, and trial and error during optimisation.

64 | P a g e

Table 3.5: Summary of carbon in ash measurement techniques.

Method Description Included components Advantages Disadvantages References

LOI Mass loss when igniting Surface and chemically Quick, easy, affordable Large errors when H2O, Brown and Dykstra

a sample (ASTM D7348, bounded H2O, carbonates, and organic (1995), Fan and Brown 2013) carbonates, organic compounds are present (2001), Mohebbi et al. compounds, and char in major amounts (2015)

Total carbon Used in coal science Carbonates, organic All non-carbon phases Expensive, state-of-the- Maroto-Valer et al. compounds, and char are eliminated, accurate art analysers needed (1999b)

Carbon form Section 3.4.2 - Differentiation between Expensive, time Bjurström et al. (2014), elemental, organic, and consuming, expertise Ferrari et al. (2002), inorganic carbon (carbon needed, acidification, Jezierski (2015), Payá et speciation) large errors for small al. (1998), Schumacher carbon percentages, (2002) temperature sensitive, state-of-the-art equipment

Optical Morphological (volume) - Differentiation between Time consuming, Wagner et al. (2018) techniques count of char anisotropic char, isotropic operator dependent, char, inertinitic char, and state-of-the-art unreacted and partially equipment needed reacted coal

65 | P a g e

Figure 3.4: Separation process flow diagram followed in this thesis (excluding froth flotation as this step was not efficient and therefore discarded in end method).

66 | P a g e

A Zeiss Imager M2M reflected-light petrographic microscope with an oil immersion objective and a combined magnification of ×500 was used for comminution (char-ash association and char liberation possibilities) analyses. The polished blocks described in Section 2.5.1 was used for this purpose.

For the size separation trials, the dry sieving technique was used. The sieving procedure was given in Section 2.5.1. The LOI for each size fraction was obtained to determine the inferred carbon purity of the product. Fractions with high LOIs were selected for further separation of the ash and char.

After the sieving trials, selected size fractions were dried overnight (110 °C) in a drying oven before electrostatic separation commenced, as moisture is known to reverse particle charge and will lower the efficiency of the electrostatic process (Baltrus et al., 2002; Cangialosi et al., 2005; Kelly and Spottiswood, 1982; Wills and Napier-Munn, 2006; Woollacott and Eric, 1994).

An OreKinetics CoronaStat from the University of Pretoria was used for the electrostatic separation trials (Figure 3.5). The CoronaStat used had five collection bins, namely conductive (C), middlings 1 (M1), middlings 2 (M2), middlings 3 (M3), and non-conductive (NC). The rotor speed was initially varied, but it was found that it had no influence on the results, and the speed was fixed at 40 rpm. The voltage control was also initially varied and it was found that for the fly ash samples the best results were obtained at 20 kV, while for the bottom and gasification ash samples a 15 kV voltage was found to be sufficient. The separation was fairly rapid (less than a minute). The bins with the highest LOIs were selected and passed through the electrostatic separator a second time, under the same conditions, to improve the carbon grades. There was a time delay of a week or so between batch one and two due to the need to obtain the LOI results before progressing.

Figure 3.5: OreKinetics CoronaStat used in the electrostatic separation trials.

67 | P a g e

Electrostatic separation exploits the differences in electrical conductivities to separate minerals from one another (Kelly and Spottiswood, 1982). Seeing that carbon is conductive, while typical ash minerals are non-conductive, it is an attractive option for the separation of char from ash. Low investment costs, low operation costs, and a minimal environmental footprint are also associated with electrostatic separation.

For comparison, the electrostatic separation step was replaced with a froth flotation step. Betachem (Pty) Ltd. supplied the hydrocarbon collector (Betacol COM 3 – commercially used in coal and graphite beneficiation) and frother (Betafroth FOM 3 – commercially used in coal beneficiation). The pulp, collector, and frother conditioning times were respectively set at 5, 5, and 1 minute and the trial residence time was set at 5 minutes. The pulp density was set at 10 %. The optimum impeller speed, yielding the best results, was at 1400 rpm and the optimum collector to frother ratio, yielding the best results, was at 50 %.

Investigation of the ash residue after the LOI tests on the electrostatic separation products revealed a substantial amount of darker particles that are still present in the samples (Figure 3.6A). These darker particles are most probably of a magnetic origin. Gray et al. (2002) noted that iron-bearing phases tend to accumulate with the carbon-rich fraction during electrostatic separation. Adding a magnetic separation stage can, therefore, counter this problem. A handheld magnet was used to test this idea. The tinfoil was wrapped around the magnet and passed over a small sample (±10 g). It was found that even by using this relatively weak magnetic intensity, a significant amount of material was collected (Figure 3.6B). A Frantz dry magnetic separator was then used and the magnetic intensity was varied at 0.5 A, 1.0 A, and 1.5 A. Although not a commonly used technique, it was found that magnetic separation can remove large amounts of magnetic material from the char product. The collected magnetic fraction can also subsequently be used in value-added applications.

A B

Figure 3.6: A) Magnetic material in ash residue after LOI tests on electrostatic separated samples and B) Magnetic material collected with a hand held magnet.

68 | P a g e

Separation via density can take place by exploiting the differences in settling velocities of the valuable and gangue minerals (The Southern African Coal Processing Society, 2015). Dense Medium Separation (DMS) is the most common method and involves the “sinking” or “floating” of a mineral in a fluid with a known density (Wills and Napier-Munn, 2006). Minerals heavier than the fluid density will sink, while those lighter than the fluid density will float. In this way, separation of minerals can be achieved. Density separation was conducted by the ERA-MIN Polish partner (Główny Instytut Górnictwa), following the ISO 7936: 1992 standard (Polish equivalent of PN-ISO 7936: 1999). Due to constraints, only FA PS4 was couriered for trials in Poland. A simple sink-float methodology was applied with densities between 1.4 and 2.0 being tested (0.1 intervals). Once the ideal density cut-point was obtained, Bureau Veritas Testing and Inspections South Africa did the separation on the bulk sample. DMS is a wet separation process. Thus far, only dry separation methods were used, and adding a DMS step will require drying and moisture removal steps. Due to the porous nature of char, the dense fluid can also be adsorbed, resulting in usage as a mercury absorber or catalyst not being feasible. Entrainment of the dense liquid in the pores can also lead to a change in apparent density and therefore yielding weakening results (Wills and Napier-Munn, 2006). DMS also requires expensive (and often environmental unfriendly) fluids.

3.3.2 Char-ash separation results Communition analysis A microscope image of a typical char particle found in the samples is given in Figure 3.7. It can be seen that minerals (mainly glass) are finely disseminated and interwoven into the char matrix. This dissemination is most likely related to the high proportion of included minerals and inertodetrinite typically found in the South African feed coals (Section 2.5.2). Inertodetrinite is relatively small in diameter (<10 µm) and is usually associated with finely dispersed quartz and clay (Falcon and Snyman, 1986; ICCP, 2001; Wagner et al., 2018).

The mineral deportment behaviour corresponds to Kelly and Spottiswood (1982) category IV behaviour (Figure 3.8). Lester et al. (2010) found similar inclusions in their char particles. These mixed char-mineral particles will either separate into the product or the discard stream, affecting the recoveries that can be obtained. Milling to liberate the inclusions is therefore important. However, the grid placed over the microscope image (Figure 3.9B) demonstrates that by milling to a size as fine as 50 µm, the smaller particles will remain high in mineral matter, a replicate of the original particle. This is known as the “mini-me” effect (Dorland et al., 2015; McMillan et al., 2015) and renders milling unfeasible. Milling to sizes below 50 μm will also result in the agglomeration of fines, which complicates separation (Wills and Napier-Munn, 2006). The energy costs associated with milling to such fine sizes

69 | P a g e also increase exponentially and should, therefore, be avoided (Wills and Napier-Munn, 2006).

Included glass particle

Char matrix

100 µm

Figure 3.7: Typical char particle found in the ash samples. Included minerals finely disseminated and interwoven into the char matrix can clearly be seen (BA PS4 illustrated) (Reflected-light, oil immersion, ×500).

Figure 3.8: Mineral deportment categories (Kelly and Spottiswood, 1982).

70 | P a g e

100 µm 50 µm

A B

Figure 3.9: Milling of char particles will only lead to the “mini-me” effect pertaining. Even milling to particle sizes as small as A) 100 µm and B) 50 µm will only result in a replicate of the original particle forming (BA PS4 illustrated) (Reflected-light, oil immersion, ×500).

An ultrasonic bath was used to see if the minerals can be “shaken” loose. This idea was suggested by the Portuguese partner, and Zhang et al. (2012) conducted similar trials. In the Portuguese sample, the chars were very porous with infilling minerals (Figure 3.10). In their sample it was thus easy to “shake” the minerals loose; however, they did encounter excessive fragmentation due to this vigorous shaking. For the South African coal ash samples, an ethanol ultrasound bath was used and samples were subjected to treatment for 90 minutes before being microscopically analysed. Although some loose, small glass particles were observed that might have been liberated during ultrasound treatment, the majority of the char particles were still filled with mineral inclusions. The ultrasound technique was thus unsuccessful for the South African samples. It was, therefore, decided to keep the char particles unliberated and to realise that the carbon recoveries will be affected by this.

Figure 3.10: Typical char particle found in the Portuguese ash sample. The chars were porous and infilled with small glassy inclusions (Reflected-light, oil immersion, ×500).

71 | P a g e

Size separation Hurt and Gibbins (1995), Külaots et al. (2004); Lu et al. (2007), and Sharonova et al. (2008) reported on carbon being concentrated in the larger particle sizes (Class F ashes). As volatile matter evolves, the vitrinite and reactive semifusinite in the coal becomes plastic, swell and harden as larger porous char. The macerals in the inertinite dominant particles devolatilise more selectively with limited pore development, forming large mixed porous chars (however, some reactive inertinites behave almost similar to vitrinite). Some of these large chars are not completely consumed during the combustion process and therefore report to the ash as large carbon particles.

Size separation was conducted to establish whether or not this behaviour is also present in the samples from this research thesis. Following sieving, it was noted that the larger fractions of the fly ash samples were darker in colour than the smaller fractions (Figure 3.11); implying that the coarser particles are enriched in char. For the fly ash samples, the LOI percentage for each sieve fraction was determined (Figure 3.12). As expected, the LOI percentage increases with an increase in particle size. Thus, char in fly ash can be pre- concentrated based on size. The same observations for the bottom and gasification ash samples were not seen, and therefore separation based on size was not applicable.

Figure 3.11: Visual confirmation of char enrichment based on particle size (FA PS4 illustrated).

20.00 18.00 16.00 14.00

12.00 10.00 FA PS4 8.00 FA PS2 6.00 LOI (wt.%) 4.00 2.00 0.00 >212 150-212 106-150 75-106 53-75 38-53 <38 Size (µm)

Figure 3.12: Particle size versus LOI (wt. %) for the fly ash samples.

72 | P a g e

Cumulative carbon grade, carbon recovery, and sample yield curves were constructed to determine the size cut-points for the fly ash samples (Figure 3.13). FA PS4 was cut at +150 µm and FA PS2 at +75 µm. At these cut-points, the highest carbon grades were achieved without losing too much carbon (recovery) (Table 3.6). At the same time, the fines were removed. Fines are known to agglomerate, hindering separation.

A

B

Figure 3.13: Cumulative carbon grade, carbon recovery, and sample yield curves for A) FA PS4 and B) FA PS2 at different particle sizes.

73 | P a g e

Table 3.6: Carbon grades, carbon recoveries, and sample yields at selected size cut-points for fly ash samples.

Sample Initial grade Cut-point Final grade Carbon Sample yield (wt. % LOI) (μm) (wt. % LOI) recovery (%) (%)

FA PS4 7.04 150 12.56 77.86 43.67 FA PS2 4.01 75 8.44 64.82 30.82

Electrostatic separation The two size separated fly ash samples (>150 µm and >75 µm) as well as the bottom and gasification ash samples were subjected to an electrostatic separation stage. Particles larger than 600 µm in the bottom ash sample were discarded, losing only 15 % of the initial carbon. For the gasification ash sample, the material loss would have been too great if particles larger than 600 µm were discarded; hence, the gasification ash was manually milled to a top size of 600 µm. The LOI percentages for each bin were determined (Figure 3.14). For the fly ash samples, the carbon was concentrated in the C and M1 bins. For the bottom and gasification ash samples, the carbon was concentrated in the C bin.

50.00

45.00

40.00

35.00

30.00 FA PS4 25.00 FA PS2 LOI (wt.%) 20.00 BA PS4 15.00 GA East 10.00

5.00

0.00 C M1 M2 M3 NC Electrostatic bin

Figure 3.14: Conductivity versus LOI (wt. %) for the ash samples.

The cumulative carbon grade, carbon recovery, and sample yield curves are provided (Figure 3.15). The carbon grades, carbon recoveries, and sample yields at the respective cut-points are given (Table 3.7). At the respective cut-points, the carbon grades tripled, while the majority of the carbon was also recovered. For BA PS4 and GA East, the carbon recoveries were, however, low.

74 | P a g e

100 NC 100 M3

90 M3 90 M2 M2 M3 80 80 70 M1 70 M3 M1 60 60 M2 50 50 M2 C Recovery Recovery 40 40 C Yield Yield 30 30 M1 M1 20 20 C Cumulative Cumulative yield/recovery (%) Cumulative Cumulative yield/recovery (%) 10 10 C 0 0 0 10 20 30 40 50 0 10 20 30 40 Cumulative grade (wt. % C) Cumulative grade (wt. % C) A B 100 NC 100 NC M3 90 M2 90 M3 M2 80 M1 80 M1 70 70 60 60 50 50 Recovery C Recovery 40 40 Yield Yield 30 30 20 C 20

Cumulative Cumulative yield/recovery (%) 10 Cumulative yield/recovery (%) 10 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 Cumulative grade (wt. % C) Cumulative grade (wt. % C) C D

Figure 3.15: Cumulative carbon grade, carbon recovery, and sample yield curves for A) FA PS4; B) FA PS2; C) BA PS4; and D) GA East at different conductivity bins.

75 | P a g e

Table 3.7: Carbon grades, carbon recoveries, and sample yields at selected conductivity cut-points.

Sample Grade after size Cut-point Final grade Carbon Sample yield separation (#bin) recovery (%) (wt. % LOI) (%) (wt. % LOI)

FA PS4 12.56 M1 and C 36.71 72.16 26.10

FA PS2 8.44 M1 and C 25.67 68.77 26.96

BA PS4 8.13 C 26.88 22.79 6.04

GA East 9.04 C 26.80 45.82 13.32

The samples were sent through the electrostatic separator a second time under the same conditions to improve efficiency. For the fly ash samples, the C and M1 fractions were once again separated while for the bottom and gasification ash samples, only the conductive fraction was used (Table 3.8). The carbon grades for all four samples increased upon recycling. The most substantial increase was seen for BA PS4. The carbon recovery for this sample is, however, very low.

Table 3.8: Carbon grades, carbon recoveries, and sample yields at selected conductivity cut-points, 2nd stage electrostatic separation.

Sample Grade after Cut-point Final grade Carbon Sample yield 1st stage (#bin) recovery (%) (wt. % LOI) (%) (wt. % LOI)

FA PS4 36.71 M1 and C 44.30 73.62 55.86

FA PS2 25.67 M1 and C 36.70 78.55 52.43

BA PS4 26.88 C 45.17 36.62 24.76

GA East 26.80 C 37.12 50.21 31.84

The electron voltage of carbon is 4eV and that of Al-Si minerals (glass) 4.7-5eV (Soong et al., 1998). Triboelectrostatic separation is more sensitive than a CoronaStat electrostatic separator in the separation of minerals with small differences in their electron voltages (Bada et al., 2010). This technology might thus be more efficient at separating char from ash. Triboelectrostatic separation is also commercially used for char-ash separation (Baker et al., 2019; Bittner et al., 2012; Bittner et al., 2014; Shilling, 1999). Maroto-Valer et al. (1999a) used size separation in combination with triboelectrostatic separation to obtain carbon grades ranging between 88 and 91 vol. % starting from parent fly ash samples with carbon

76 | P a g e grades ranging between 36 – 39 vol. %. Soong et al. (2002) increased their initial fly ash carbon grades (12 and 14 wt. %) to a maximum of 60 wt. % carbon by using a combination of size and triboelectrostatic separation. Gray et al. (2002) used a double stage triboelectrostatic separation process to obtain a 35 wt. % carbon product from an 18 wt. % carbon fly ash sample. Their carbon recovery was below 50 %. Ban et al. (1997) used a triboelectrostatic separator to obtain carbon products with grades of up to 50 wt. % and recoveries > 50 % from fly ash samples. Unfortunately, a triboelectrostatic separator was not available during the experimentation phase of this study, and therefore its effect on the samples used in this research is unknown but warrants future investigation.

Froth flotation separation

For the froth flotation experiments, the optimum results obtained were similar to that from electrostatic separation. However, the collector dosage amount was very high (3 ml/g). Seeing that hydrocarbon reagents are very costly and capricious, this high dosage is impractical. The microporous structure of char can possibly be blamed for the large amount of collector consumption (Şahbaz et al., 2008, Zhang and Honaker, 2015). During pulp conditioning, water molecules fill the pores and decrease the surface area; making it difficult for the collector to attach and coat the particles (Zhang and Honaker, 2015). More collector is therefore needed to displace the water. Demir et al. (2008) also showed that, due to high temperatures experienced in the boiler, the char particles are oxidised and therefore show a lower degree of hydrophobicity than graphite and coal.

Magnetic separation

For the fly ash samples, increasing the intensity above 0.5 A resulted in a significant drop in carbon recovery without any increase in carbon grade. The reason for this is unclear. For the gasification ash, it was found that at 0.5 A a very low carbon grade but a high carbon recovery is present. The intensity for this sample was therefore increased to 1.5 A. The results are given in Table 3.9. For the fly ash samples, the carbon grades increased substantially. Magnetic separation, however, had a small effect on the bottom and gasification ashes. The carbon recovery for FA PS2 is lower than for FA PS4.

77 | P a g e

Table 3.9: Carbon grades, carbon recoveries, and sample yields for the non-magnetic fractions.

Sample Grade after Magnetic Final grade Carbon Sample yield electrostatic intensity (A) recovery (%) (wt. % LOI) (%) separation

(wt. % LOI)

FA PS4 44.30 0.5 65.74 77.36 60.20

FA PS2 36.70 0.5 56.55 50.83 33.65

BA PS4 45.17 0.5 53.21 86.17 76.00

GA East 37.12 1.5 45.10 73.15 62.60

Density separation In Table 3.10 the DMS results are given. Although the carbon grade increased substantially, the carbon recovery is relatively low. Utilising sink-float on small particles might be a reason for this. Surface friction is problematic for small sizes and settling times are inefficient (Wills and Napier-Munn, 2006). An additional shaking or centrifugal action is therefore required for fines. For this purpose, a reflux classifier situated at North-West University was tested to see if better results can be achieved (see Smith (2015) for equipment details). The efficiency, however, did not improve.

Table 3.10: Carbon grades, carbon recoveries, and sample yields for the density cut-point (FA PS4).

Sample Grade after Density cut Final grade Carbon Sample yield magnetic (SG) recovery (%) (wt. % LOI) (%) separation

(wt. % LOI)

FA PS4 65.74 1.8 82.91 59.23 46.4

Summary A summary of the char-ash separation process results is provided in Table 3.11.

The final carbon grades for the two fly ash samples were the highest. FA PS4 had a higher final grade than FA PS2, but this sample also had a higher starting grade. When comparing the grade increase, it can be seen that FA PS2 increased by a factor of 14.10, while FA PS4 only increased by a factor of 9.34 (excluding density separation) and 11.78 (including density separation). The GA East sample had the lowest final carbon grade.

78 | P a g e

Although the addition of the density separation step showed that the carbon grade can be increased with more than 15 wt. %, density separation is a wet separation process which will require additional drying stages if implemented in an industrial context. The dense liquid used in density separation processes is also known to penetrate the pores of the char product, making it unsuitable for value-added applications such as mercury adsorption, activated carbons, and catalysts. However, for synthetic graphite manufacturing, the additional density separation step should not be a problem. A possible solution to these problems might be to use air fluidization as a density separation method.

The ERA-MIN set a target to obtain a char product with carbon grades above 90 wt. % fixed carbon. This was, however, not achieved in the current research, nor by the collaborating partners at this time. For this study, a CoronaStat electrostatic separator was used, but from previous char-ash separation studies, it was seen that a triboelectrostatic separator is more relevant for this type of material. Utilisation of a triboelectrostatic instead of a CoronaStat electrostatic separator can therefore possibly increase the carbon grade to the sought 90 wt. % fixed carbon. No ERA-MIN partners had access to this technology.

Although the carbon recoveries for the individual separation steps were high (~70 %), the overall recoveries were low, at 7.19 – 32 %. A low carbon recovery is undesirable, seeing that the starting char in ash percentages were low. A low recovery will thus result in low volumes produced, which will be an economic constraint. The reason for the low recoveries might be twofold. Firstly, all trials were conducted in a laboratory / batch mode. In a pilot or process plant, streams would be recycled and higher recoveries would be expected. Secondly, it was seen that small glass particles were embedded in the initial char matrix, giving it an unliberated nature.

Overall, the suggested separation schematic is cost-effective. Electrostatic and magnetic separation are known to be some of the most inexpensive separation techniques available; especially when compared to e.g. froth flotation in which the hydrocarbon flotation reagents (diesel and kerosene) can be very expensive.

Additionally, the inclusion of magnetic separation will yield a ferrosphere product. The ferrosphere fraction can be used in multiple value-added applications, the most promising being the usage as an inexpensive catalyst for processes were traditional catalysts are unstable or deactivate (oxidative dimerization of methane with the formation of ethylene) (Anshits et al., 2011). For the ERA-MIN collaboration, the ferrosphere fraction was analysed to test its applicability.

79 | P a g e

By removing the char, the remaining ash can also be used in cement and concrete applications. According to ASTM standards (ASTM C618a, 2012), the LOI should be <6 wt. % and the particle size distribution <45 μm before usage in cement and concrete can take place. The gangue after size separation fits this description. Once again the ERA-MIN collaboration is testing this initiative, but it will not be further explored in this study.

Table 3.11: Summary on the char-ash separation: Carbon grades, carbon recoveries, and sample yields.

Sample Initial Final grade Grade Carbon Sample grade increase recovery yield (wt. % LOI) factor (%) (wt. % LOI) (%)

FA PS4 7.04 65.74 9.34 32.00 3.83

FA PS4 (density step incl.) 7.04 82.91 11.78 18.95 1.78

FA PS2 4.01 56.55 14.10 17.80 1.47

BA PS1 5.47 53.21 9.73 7.19 1.14

GA East 9.04 45.10 6.54 16.83 2.65

3.4 Characterisation of char samples As the concentrated char is to be used as a potential precursor for synthetic graphite, it is necessary to obtain an understanding of the characteristics of the concentrated samples. The results will aid in deciding whether or not it is worthwhile to pursue graphitization.

3.4.1 The structural progression of carbon with temperature A schematic of the structural progression of carbon with temperature is given in Figure 3.16. It should be mentioned that this temperature intervals and transformations are described for a relatively slow heating rate (few degrees per minute) in a non-reactive atmosphere (typical carbonisation conditions). This needs to be mentioned in particular because the context of the thesis is a combustion context with fast heating rates and an oxidising atmosphere.

80 | P a g e

Figure 3.16: Structural progression of carbon with temperature (adapted from Bourrat, 2000).

At temperatures <500 °C isometric coherent domains are formed. These domains are referred to as basic structural units (BSU) and are relatively small in size (~1 nm) (Bourrat, 2000). Each BSU consists of two to three layers of hexagonal carbon rings with strong bonds existing between the carbon atoms (Bourrat, 2000). BSUs are formed when a carbon precursor undergoes a mesophase. At 350 - 550 °C certain carbon precursors will melt to form small spheres (Oberlin, 1984; Pierson, 1993). The spheres will nucleate, grow, and coalesce to form a bulk mesophase (Oberlin, 1984). With the solidification of the mesophase, a brittle carbon (char / coke) is formed (Oberlin, 1984). The mesophase is important seeing that it gives motion to the carbon atoms, resulting in their realignment to form the BSUs (Tamashausky, 2006). Non-graphitizable carbons have cross-linked or pinned structures that cannot flow / melt easily and therefore the carbon atoms cannot align themselves into this pre-graphite structure (Tamashausky, 2006).

Between 500 and 1500 °C columns of BSUs start to form. Oberlin (1984) compared the BSUs to “bricks” that pile up as they near each other. The pilling occurs as a result of the release of heteroatoms and hydrogen (Oberlin, 1984). Biscoe and Warren (1942) described the piles as having a turbostratic structure; consisting of aggregated crystal groups, each

81 | P a g e having several hexagonal carbon layers roughly orientated parallel and equidistant from each other but still lacking the three-dimensional stacking structure associated with graphite. Franklin (1951) supported the view of Biscoe and Warren (1942) and defined the minimum interlayer spacing (d002) of a turbostratic structure to be 3.44 Å. Single BSUs are often still trapped between the columns and will only disappear at ~1500 °C (Oberlin, 1984).

Up to this point, growth was mainly restricted to the vertical direction. Between 1500 and 2000 °C adjacent turbostratic piles “hook” onto one another to form distorted layers (Oberlin, 1984). However, defects often occur between hooked boundaries, preventing the horizontal growth of the coherent domains (Oberlin, 1984). Maire and Méring (1970) first pointed out the influence of in-plane (horizontal) defects on the graphitization process. Defects include heteroatoms, cross-link and sp3 bonds, Schottky defects (vacancies), and Frenkel defects (interstitial carbon atoms) (Bourrat, 2000; Fischbach, 1970).

At temperatures > 2000 °C, the in-plane defects are annealed to form stiff and straight layers (Oberlin, 1984). The turbostratic structure also disappears to form graphite as shown in Figure 3.17. Individual graphite layers are known as graphene, which essentially is hexagonal carbon rings with strong bonds (524 kJ/mol) existing between the carbon atoms (Bourrat, 2000; Pierson, 1993). Graphene layers are stacked on top of one another through weak bonds (7 kJ/mol) in an ABAB arrangement (with every second layer corresponding)

(Pierson, 1993; Zondlo, 2012). The weak bonds result in graphite having a large d002. In ideal graphite, the d002 is 3.354 Å, while the crystallite sizes (Lc and La) are ∞ (Franklin, 1951; Seehra and Pavlovic, 1993). This last stage is known as the graphitization stage (the former three as the carbonisation stage) and is a kinetically controlled step (Fischbach, 1970). The high activation energy (±250 kcal/mole) is responsible for annealing and defect removal (Fischbach, 1970). Most materials are fully graphitised at 2500 – 3000 °C and will take 2 to 3 hours to complete (Pierson, 1993). If lower temperatures are used the residence time will increase (Pierson, 1993). A catalyst can also be added to speed up the process (Pierson, 1993).

82 | P a g e

Figure 3.17: The crystallography of graphite (Pierson, 1993).

3.4.2 Methodology Due to the costs associated with characterisation and subsequent graphitization, it was decided to only select one char concentrate for these purposes. Based on the efficiency of separation, FA PS4 (density separation included) was chosen as the sample moving forward. However, where costs allowed it, FA PS2 was also characterised. Sample information is given in Table 3.12. The abbreviation CC (char concentrate) was added to sample IDs to distinguish between the parent and concentrated ash samples.

Table 3.12: Sample information of char concentrates (CC) for further characterisation.

Sample ID Purification description Initial grade Product grade Subsequent (wt. % LOI) (wt. % LOI) usage

FA PS2 CC Size, electrostatic, and 4.01 56.55 Selected magnetic separation characterisation FA PS4 CC Size, electrostatic, magnetic, 7.04 82.91 Characterisation and density separation and graphitization

The selected CCs were divided into representative fractions by using a rotary splitter. A 25 g representative fraction of FA PS4 CC was kept for graphitization purposes, while the rest was used for characterisation. The characterisation techniques are given, as well as an indication on which samples they were conducted (Table 3.13).

83 | P a g e

Table 3.13: Char concentrate characterisation techniques.

Analysis FA PS4 CC FA PS2 CC

Proximate ×

Ultimate ×

Carbon form ×

X-ray diffraction (mineralogy) ×

X-ray diffraction (structural) × ×

Raman microspectroscopy × ×

Petrographic × ×

Proximate, ultimate, and carbon form analyses were outsourced to Bureau Veritas Testing and Inspections South Africa. The standards are provided in Table 3.14. Bureau Veritas subcontracted the total carbon, nitrogen, and hydrogen analyses to UIS, due to faulty instrumentation during the period of testing. For carbon form analysis, an in-house designed method was used (Figure 3.18). The total carbon for carbon form analysis was determined via ACT-TPM-027 based on ISO 29541: 2010.

Table 3.14: Proximate, ultimate, and carbon form standards used on the char concentrates.

Analysis Standard

Sample preparation ACT-TPM-001 based on ISO 13909-4: 2001

Moisture ACT-TPM-010 based on ISO 11722: 1999 Volatile matter ACT-TPM-012 based on ISO 562: 2010 Proximate Ash yield ACT-TPM-011 based on ISO 1171: 2010 Fixed carbon By difference

Total sulphur ACT-TPM-013 based on ISO 19579: 2006 Carbon, nitrogen, hydrogen Based on ISO 12902 (2001) – CHN instrumental Ultimate method Oxygen By difference

Carbon form ACT-TPM-028

84 | P a g e

Char concentrate sample

Total carbon Demineralisation Heating in an inert atmosphere ACT-TPM-027 based on ISO 29541: 2010 Removal of inorganic carbon and organic For inorganic carbon removal carbon

Total organic carbon Inorganic carbon

Subtract total organic Combustion in an oxygen carbon from total carbon atmosphere Only contribution is elemental carbon

Experimental

Elemental Determined Organic carbon carbon

Subtract elemental carbon Calculated from total organic carbon

Figure 3.18: Carbon form analysis diagram (ACT-TPM-028).

85 | P a g e

The XRD mineralogy method was given previously (Section 2.5.1).

To determine the XRD structural composition, it was first necessary to demineralise the samples. This ensured that the remaining minerals did not interfere and alter the data. The samples were pulverised to -45 µm before a three-step procedure, conducted by MAK Analytical and following Agrawal and Sharma (2018), was applied: 1) NaOH treatment for 4 hours to rinse sample; 2) HCl treatment for 72 hours to eliminate carbonates; and 3) HF treatment for 24 hours to eliminate silicates. The temperature was kept constant at 90 °C throughout the demineralisation procedure. The final LOI percentages were 98.45 wt. % for FA PS4 CC and 99.31 wt. % for FA PS2 CC.

The acid demineralised samples were prepared for X-ray scanning by adding a silicon standard. The silicon standard is important to adjust for shifting and broadening of diffraction profiles; ensuring that the exact peak positions and full width at half maximum (FWHM) values are obtained (Iwashita et al., 2004). Scanning was done by XRD Analytical & Consulting using a backloading preparation method and a PANalytical AERIS diffractometer with a PIXcel detector and fixed slits with Fe filtered Co-Kα radiation. The results interpretation was done by using built-in X’Pert Highscore plus software functions. The d002 and Lc were respectively obtained by applying Bragg’s law and Scherrer’s equation to the 002 plane (Equations 3.4 and 3.5). In these equations 휆 is the X-ray wavelength (Co = 1.7891 Å), 휃 is Bragg’s angle (peak position), 훽 is the FWHM, and K is a constant chosen as 1 (following Iwashita et al. (2004)).

휆 Equation 3.4 푑 (Å) = 002 2 sin 휃

퐾휆 Equation 3.5 퐿 (Å) = 푐 βcos 휃

Blocks for Raman microspectroscopy were prepared through UJ’s sample preparation facility. Samples were mounted in epoxy resin, and polished (following ISO 7404-2). Although polishing can lead to an increase in observed Raman disorder, the effect on structural disorganised carbonaceous material (as for CCs) is not as pronounced as for organised carbonaceous material (Beyssac et al., 2003; Nasdala et al., 2004; Pasteris, 1989; Wang et al., 1989). Nevertheless, Raman microspectroscopy should be used and interpreted in combination with other structural techniques, such as XRD. For spectral acquisition, a WITec Alpha300 R confocal Raman microspectrometer (UJ Assore Raman Laboratory) was used with a green line of argon (휆 = 532 nm) as the excitation source and a

86 | P a g e laser strength of 3.5 mW. The low laser strength was chosen to avoid sample damage and the shifting of peaks (Beyssac et al., 2003). The microspectrometer was calibrated with a silicon standard before spectra acquisition commenced. A 50× objective lens was used. The integration time was set at 30 seconds with a total of five cycles and with scans extending from 0 to 3600 cm-1. At least 25 spectra were randomly obtained from each sample. These spectra were taken from different particles in the blocks, as well as different spots in individual particles. OriginPro 2019 software was used for curve fitting purposes. The curve fitting method is provided in Appendix B.

For petrographic analysis on the CCs, the Hower (2012) classification, the ICCP char classification, and reflectance analysis were applied to the samples. Block preparation, equipment specifications, and the Hower (2012) classification were discussed in Section 2.5.1. A schematic of the ICCP char classification is provided in Figure 3.19. Reflectance analysis was applied following the procedure described by Li et al. (2018) for amorphous graphite. The mean random reflectance was calculated following ISO 7404-5. To determine maximum reflectance a strontium-titanium standard 5.37 was used for calibration purposes. The mean random reflectance and the maximum reflectance were then used to calculate the minimum reflectance (Equation 3.6), bireflectance (Equation 3.7), and anisotropy (Equation 3.8).

푅푚푖푛 = 2푅푚푒푎푛 − 푅푚푎푥 Equation 3.6

퐵푖푟푒푓푙푒푐푡푎푛푐푒 = 푅푚푎푥 − 푅푚푖푛 Equation 3.7

푅 − 푅 Equation 3.8 퐴푛푖푠표푡푟표푝푦 = 푚푎푥 푚푖푛 푅푚푎푥

87 | P a g e

A – Tenuisphere

B – Crassisphere C – Tenuinetwork

D – Crassinetwork E – Mixed porous

F – Mixed dense G – Inertoid

H – Fusinoid/solid I - Mineroid

Figure 3.19: ICCP char classification (adapted from Wagner et al., 2018).

88 | P a g e

3.4.3 Char characterisation results The proximate, ultimate, carbon form and XRD mineralogy results for FA PS4 CC are presented (Table 3.15).

The moisture percentage is relatively high, especially compared to the volatiles. The volatiles will have to be removed (carbonisation) before graphitization commences.

Ultimate analysis shows that carbon forms the majority of organic elements, followed by oxygen and nitrogen. The sulphur and hydrogen contents are well below one. The atomic H/C and O/C ratios are comparable to the Van Krevelen anthracite / carbon black rank (Van Krevelen, 1993).

Carbon speciation shows that elemental carbon in FA PS4 CC is dominant. This result was also found by Bjurström et al. (2014) and Ferrari et al. (2002) in chars from solid biofuel and solid waste incinerator ash respectively. Elemental carbon consists of a material with strong carbon-carbon bonds, similar to the bonding arrangement typically found in graphite. Organic carbon can be described as “volatile” carbon and consists of material with carbon- organic bonds (e.g. condensed tar, benzene, and toluene) (Bjurström et al., 2014; Brown and Dykstra, 1995). The organic carbon content in the sample is much lower than the elemental carbon content. From the proximate and ultimate results, it was seen that the volatile percentages were also low, therefore justifying this observation. Inorganic carbon describes carbon bonded to inorganics; mostly calcite for coal and char.

From the XRD morphology results, it can be seen that the amorphous phase is dominant, consisting of both amorphous carbon and amorphous glass. From optical observations, it was also seen that the glass phase was the dominant phase present. The other minerals include low counts of quartz, mullite, and gypsum. The mullite percentage is slightly higher than the other two minerals and might be because mullite is closely associated with glass and vitrified clay. Kaolinite transformed to form mullite at elevated temperatures.

89 | P a g e

Table 3.15: Proximate, ultimate (including atomic H/C and O/H ratios), carbon form, and XRD mineralogy analyses results for FA PS4 CC.

FA PS4 CC

Proximate Moisture (a.d.b wt. %) 3.6 Ash yield (a.d.b wt. %) 17.7 Volatile matter (a.d.b wt. %) 4.0 Fixed carbon (a.d.b wt. %) 74.7

Volatile matter (d.a.f wt. %) 5.1 Fixed carbon (d.a.f wt. %) 94.9

Ultimate Total sulphur (a.d.b wt. %) 0.30 Carbon content (a.d.b wt. %) 75.00 Hydrogen content (a.d.b wt. %) 0.18 Nitrogen content (a.d.b wt. %) 1.11 Oxygen content (a.d.b wt. %) 2.11 Moisture and ash (a.d.b wt. %) 21.30

Total sulphur (d.a.f wt. %) 0.38 Carbon content (d.a.f wt. %) 95.30 Hydrogen content (d.a.f wt. %) 0.23 Nitrogen content (d.a.f wt. %) 1.41 Oxygen content (d.a.f wt. %) 2.68

Atomic H/C 0.03 Atomic O/C 0.02

Carbon form Total carbon (wt. %) 75.45 Elemental carbon (wt. %) 66.58 Organic carbon (wt. %) 7.18 Inorganic carbon (wt. %) 1.69

Elemental carbon (% of total carbon) 88.24 Organic carbon (% of total carbon) 9.52 Inorganic carbon (% of total carbon) 2.24

XRD (wt. %)

Quartz (SiO2) 2.0

Mullite (3Al2O32SiO2 / 2Al2O3SiO2) 6.6

Gypsum (CaSO4∙2H2O) 0.1 Amorphous (carbon and glass) 91.4

90 | P a g e

Diffractograms of the he demineralised FA PS4 CC and FA PS2 CC are given (Figure 3.20). The diffractograms are similar to those from chars derived from bituminous coal (Ahn et al., 1999; Hurt et al., 1995, Malumbazo, 2011) and char derived from an oil-fired power plant (Tai et al., 2010). Both FA PS4 CC and FA PS2 CC show the typical graphite 002 peak. This peak is broad and not as “sharp” as seen in graphitic material. The 002 peak is also asymmetrical, indicating that aliphatic compounds are dominant in the samples opposed to aromatic compounds (Feng et al., 2003, Hattingh et al., 2013, Lu et al., 2001, Lu et al., 2002, Okolo, 2010, Van Niekerk et al., 2008, Wu et al., 2008). The other peaks are all diffused. The absence of three-dimensional reflection peaks (101 and 102) indicates a lack of three- dimensional orientation. This absence, as well as the diffusing nature of the peaks, shows a turbostratic structure (Hurt et al., 1995). A turbostratic material can be defined as one consisting of aggregated crystal units, each having several graphene layers (benzene rings) roughly orientated parallel and equidistant from each other, but still lacking the ABAB structure associated with graphite (Biscoe and Warren, 1942; Warren, 1941). During the graphitization process, the ABAB structure will develop and annealing of impurities and imperfections will take place (Pierson, 1993).

Figure 3.20: Diffractograms of FA PS4 CC and FA PS2 CC.

91 | P a g e

The d002, Lc, and La determined with XRD can be used to indicate structural ordering. In ideal graphite, the d002 is 3.354 Å (Franklin, 1951) and the crystallite sizes are ∞ (Seehra and

Pavlovic, 1993). La is difficult to determine for disordered material, seeing that it relies on the

110 peak which is highly diffused in this case (Figure 3.20). The d002 and Lc results for the chars from the current study are provided in Table 3.16. The Lc values are small, especially compared to natural graphite in which Lc values are well above 300 Å. Overall, the samples have a very small crystallite size with a d002 larger than that of a typical turbostratic structure (3.44 Å) (Franklin, 1951). FA PS2 CC is slightly more organised than FA PS4 CC.

Table 3.16: X-ray diffraction structural results for FA PS4 CC and FA PS2 CC.

FA PS4 CC FA PS2 CC

d002 (Å) 3.51 3.48

Lc (Å) 14.25 15.44

A typical Raman spectrum obtained for the samples is presented in Figure 3.21A and consists of 1st order (800-2000 cm-1) and 2nd order (2000-3200 cm-1) spectra. The 2nd order bands are diffused and broad, indicative of a disordered nature. The first order spectra are of importance for this study and are illustrated in more detail in Figure 3.21B.

A B

1st order 2nd order

Figure 3.21: Typical Raman microspectroscopy spectrum obtained in the char concentrates: A) 1st and 2nd order spectra and B) 1st order spectrum close-up.

An example of a curve fitted spectrum is given in Figure 3.22, and a summary of bands fitted to the 1st order spectrum is presented in Table 3.17. Tuinstra and Koenig (1970) first described the presence of the G and D1 bands. The G band occurs at a 1575 cm-1 Raman shift and corresponds to in-plane C-C stretching vibration (Tuinstra and Koenig, 1970). For

92 | P a g e infinite large C-C layers only the G band will be present; thus indicating an ideal graphite structure. The D1 band occurs at a 1355 cm-1 Raman shift and is attributed to a small in- plane crystallite size (Tuinstra and Koenig, 1970). Bény-Bassez and Rouzaud (1985) also ascribed the D1 band to in-plane defects and heteroatoms (O, H, and N). The D2 band occurs at 1620 cm-1 as a shoulder on the G band. It corresponds to in-plane defects and heteroatoms, and in very disordered materials will merge with the G band to form a single wide band at 1600 cm-1 (Beyssac et al., 2002; Lahfid et al., 2010). Kouketsu et al. (2014) mentioned the probability that this “merged” band consists mainly of the D2 band and a neglectable G band. The D3 band is released early in the graphitization process due to out- of-plane (three-dimensional structure) defects such as tetrahedral bonding (Bény-Bassez and Rouzaud, 1985). Nemanich and Solin (1979) also linked a small crystallite size to the formation of this band; it occurs at a Raman shift of 1500 cm-1. Between 1150 and 1265 cm-1 two bands, namely D4 and D5, are found. These two bands are commonly mistaken as one, but researchers such as Ferralis et al. (2016) and Schito et al. (2017) indicated separated D4 and D5-bands at 1150 cm-1 and 1265 cm-1 respectively. Although both bands are poorly understood, they can be narrowed down to out-of-plane defects (three-dimensional structure). Ferralis et al. (2016) relate these two bands to CH species in aliphatic hydrocarbon chains. Sadezky et al. (2005) and Sforna et al. (2014) contributes the D4-band to low crystallinity, sp2-sp3 bonds, ionic impurities, and C-C and C=C stretching vibrations.

Figure 3.22: Raman 1st order spectrum curve fitting example for char concentrates.

93 | P a g e

Table 3.17: Summary on 1st Order Raman microspectroscopy bands fitted in this study for the char samples.

Nomenclature Raman Vibration mode Reference shift (cm-1)

G band 1575 Ideal graphite lattice, E2g Tuinstra and Koenig, 1970 vibration of sp2 bonded hexagonal D1 band 1355 In-plane layer defects (edges, Tuinstra and Koenig, 1970

heteroatoms) and a small La, 2 A1g vibration of sp -bonded hexagonal rings

D2 band 1620 Graphene layer defects, E2g Sadezky et al., 2005; Sforna et al., 2014 vibration of sp2-bonded hexagonal rings D3 band 1500 Out-of-plane defects Bény-Bassez and Rouzaud,1985; (tetrahedral bonds), small Nemanich and Solin, 1979 crystallite size D4 band 1150 Low crystallinity, sp2-sp3 Sadezky et al., 2005; Sforna et al., 2014 bonds, ionic impurities, C-C and C=C stretching vibrations

The curve fitting results are presented in Table 3.18. Both samples had noticeable G (1590- 1593 cm-1) and D1 (1355-1360 cm-1) bands, with the D1 band (0.70-0.71 a.u.) being more intense than the G band (0.54-0.59 a.u.). Due to a higher D1 position for FA PS4 CC, the distance between the G and D1 bands are also smaller than for FA PS2 CC. A lower wavenumber for D1 is attributed to the increase of larger aromatic clusters (Ferrari and Robertson, 2000) and therefore a smaller distance between the D1 and G positions shows a more disordered nature (Schito et al., 2017). The D3 band (1485-1492 cm-1) can be recognised by the elevated “valley” between the G and D1 bands while a left-hand shoulder on the D1 band is indicative of a D4 band (1201-1219 cm-1). Although the D2 band is not visually distinguishable from the G band, Sheng (2007) has remarked that the D2 band is always present when a prominent D1 band is present. From the curve deconvolution, the D2 band position was then determined at a Raman shift of 1628-1630 cm-1 for the char samples. The D1 and G FWHMs for the FA PS2 CC are lower than for FA PS4 CC. This is indicative of a higher structural ordering. The D1 and G band areas for FA PS2 CC were subsequently lower than for FA PS4 CC. As a result, the D1/G FWHM and area ratios, as well as the RA1 and RA2 values, were lower for FA PS2 CC. Following the classification from Kouketsu et al. (2014), the chars were subsequently grouped as “transitional” carbonaceous material, implying they are of an intermediate rank between amorphous and graphitic carbonaceous material and will, therefore, be graphitizable. However, this is a very

94 | P a g e simplistic approach because being amorphous and free of volatiles indicates a very low ability to graphitize.

Table 3.18: Quantitative Raman microspectroscopy curve fitting results for the char concentrates.

Sample FA PS4 CC FA PS2 CC

Band position (cm-1) G 1592.73 ± 2.35 1590.29 ± 3.59 D1 1360.19 ± 3.85 1355.09 ± 3.62 D2 1629.91 ± 0.31 1627.79 ± 2.73 D3 1491.95 ± 2.29 1484.91 ± 3.39 D4 1201.16 ± 8.38 1218.99 ± 8.74

Band FWHM (cm-1) G 116.13 ± 5.90 109.46 ± 3.75 D1 166.39 ± 11.52 140.29 ± 13.57 D2 42.32 ± 3.28 38.47 ± 1.76 D3 101.09 ± 2.87 105.56 ± 3.20 D4 197.17 ± 6.10 200.00 ± 0.00

Band area (cm-1) G 73.55 ± 6.49 63.14 ± 7.68 D1 126.37 ± 12.18 105.01 ± 11.87 D2 2.98 ± 0.22 2.39 ± 0.30 D3 32.16 ± 3.42 33.31 ± 2.61 D4 33.41 ± 5.80 32.77 ± 3.82

Band intensity (normalised, a.u.) G 0.59 ± 0.03 0.54 ± 0.05 D1 0.71 ± 0.05 0.70 ± 0.02 D2 0.05 ± 0.00 0.04 ± 0.00 D3 0.20 ± 0.02 0.20 ± 0.02 D4 0.16 ± 0.03 0.15 ± 0.02

Calculated parameters D-G distance (cm-1) 232.54 ± 3.48 235.21 ± 2.93 D1/G FWHM ratio 1.43 ± 0.06 1.28 ± 0.09 D1/G area ratio 1.72 ± 0.11 1.67 ± 0.07 RA1 0.59 ± 0.02 0.58 ± 0.01 RA2 1.47 ± 0.11 1.39 ± 0.05

The petrographic results are presented in Table 3.19. Quartz occurs in the FA PS2 CC sample, but it is absent in FA PS4 CC. This implies that the addition of the density separation step for FA PS4 CC was able to remove this fraction.

95 | P a g e

Table 3.19: Petrographic analyses results for the char concentrates.

FA PS4 CC FA PS2 CC

Hower (2012) classification Glass (vol. %) 5.8 40.5 Quartz (vol. %) 0.0 4.9 Anisotropic char (vol. %) 43.2 45.4 Isotropic char (vol. %) 15.6 0.4 Inertinitic char (vol. %) 35.0 8.8 Unreacted / partially burned coal (vol. %) 0.4 0.0

Anisotropic char (% of total organics) 45.9 83.2 Isotropic char (% of total organics) 16.6 0.7 Inertinitic char (% of total organics) 37.2 16.1 Unreacted / partially burned coal (% of total organics) 0.4 0.0

ICCP char classification Tenuisphere (vol. %) 0.0 1.8 Crassisphere (vol. %) 0.0 0.2 Tenuinetwork (vol. %) 2.9 2.8 Crassinetwork (vol. %) 23.3 0.4 Mixed porous (vol. %) 40.4 44.0 Mixed dense (vol. %) 22.4 11.4 Inertoid (vol. %) 5.5 10.6 Fusinoid / solid (vol. %) 3.9 6.1 Mineroid (vol. %) 1.6 22.7

Tenuisphere (% of total organics) 0.0 2.3 Crassisphere (% of total organics) 0.0 0.3 Tenuinetwork (% of total organics) 2.9 3.6 Crassinetwork (% of total organics) 23.7 0.5 Mixed porous (% of total organics) 41.1 56.9 Mixed dense (% of total organics) 22.8 14.7 Inertoid (% of total organics) 5.6 13.7 Fusinoid / solid (% of total organics) 4.0 7.9

Reflectance

Rmean (Rr%) 7.23±0.53 8.36±1.80

Rmax (Rr%) 8.13±1.55 11.09±2.47

Rmin (Rr%) 6.32 5.62 Bireflectance (Rr%) 1.82 5.47 Anisotropy 0.22 0.49

Anisotropic char and inertinitic char are dominant in both samples. Inertinite percentages were also high in the corresponding feed coals (Section 2.5.3) and therefore explain the high percentages in the char concentrates. Vitrinite macerals will either form anisotropic or isotropic char upon combustion (Hower, 2012) (however, you also have reactive inertinite macerals that may contribute to it). The predominance of anisotropic over isotropic char depends on whether the vitrinite passed through a softening stage or not during combustion

96 | P a g e

(Hower, 2012). It is, therefore, rank dependent. If vitrinite passes through a softening stage, the carbon atoms will be able to move around and re-align in a graphite-like manner. High anisotropic char contents, therefore, represent an ordered carbon structure. Examples of anisotropic and isotropic char are presented in Figure 3.23 and can be distinguished via a 90º turn of the analyser.

A

100μm

B

100μm

Figure 3.23: A) Anisotropic char and B) Isotropic char. Anisotropic char is distinguished from isotropic char by a spot colour change with a 90° turn of the analyser (reflected-light, ×500, oil objective).

Inertinitic char and isotropic char contents for FA PS2 CC are lower than for FA PS4 CC, while for anisotropic char the reverse occurs. The higher inertinite content in FA PS4 CC might be due to a lower total reactive macerals percentage in the corresponding coal (Table

97 | P a g e

2.13). Isotropic percentages should be higher but was probably burned. The reactive inertinite devolatilises and the carbonaceous residue is transformed into a more inert material.

From the ICCP classification results, it can be seen that both samples contained mostly mixed porous particles, probably related with microlithotypes vitrinite-inertinite-vitrinite (Figure 3.24A and B). FA PS4 CC had a higher crassinetwork percentage which is mostly related with the reactive semifusinite and reactive inertodetrinite. Another observation made regards the rare glassy inclusions (Figure 3.24B). This confirms the suspicion that the majority of combined char-ash particles found in the parent ash were not recovered during separation, explaining the low carbon recoveries.

A B

100 µm 100 µm

Figure 3.24: A) Inertinitic (top) and mixed porous (bottom) char particle and B) Mixed porous char particle with a glassy (rare) inclusion (reflected light, ×500 oil objective).

Bireflectance is a property indicating the change in reflectance (Allaby, 2008; Craig and Vaughan, 1994). The anisotropy is a measurement of graphitization degree. In an amorphous material, the properties along different molecular axes will be the same; indicative of isotropic materials. Materials with different properties along different molecular axes (graphite) can be described as anisotropic. The bireflectance for FA PS2 CC is relatively high, especially considering that for graphite the bireflectance ranges between 6 and 27 (Craig and Vaughan, 1994). As a result, the anisotropy fraction is also high at 0.49. The results for FA PS4 CC are much lower. This supports all the other results in which FA PS4 CC was more disordered than FA PS2 CC. There was also a rank difference between the parental coals.

3.5 Summary A laboratory char-ash separation process was designed and tested on the fly, bottom, and gasification ash samples from South Africa. Product carbon grade and carbon recovery were used to assess the separation efficiency. The process consisted of size, electrostatic, and

98 | P a g e magnetic separation steps. The product grades were 57 to 66 wt. % carbon for the fly ash samples, 53 wt. % carbon for the bottom ash sample, and 45 wt. % carbon for the gasification ash sample. The product grades were substantially increased, especially for the fly ashes, when compared to the initial feed grades (4-7 wt. % carbon in ash). It is believed that the grades can further be increased by using a triboelectrostatic separator instead of a Corona separator (as was used in this study). Triboelectrostatic separators can separate minerals with a small difference in their work functions; thus ideal for carbon (4eV) and ash (4.7-5eV) separation (Bada et al., 2010; Das et al., 2010; Zhang et al., 2012).

The carbon recoveries were low for all samples (6 – 32 %). The reason for the low recoveries might be twofold. All trials were conducted in batch mode; for continuous processes recycling streams will be added which will increase the recovery. Secondly, it was seen from comminution analyses that small ash minerals form part of the char matrix. This “unliberated” nature causes low recoveries. Due to the small milling size needed to liberate the char (< 50 µm), milling is not deemed feasible. An ultrasound bath was implemented to see if the inclusions can be loosened without any fines forming, but this trial was unsuccessful.

To test the influence of density separation, a sink-float step was added to the suggested process and tested on a fly ash sample. The grade increased to 83 wt. % carbon and the overall carbon recovery was 19 %. However, char is very porous and is known to adsorb the dense liquid used in sink-float processes. This limits the applications of char after separation, e.g. as a mercury absorber or in activated carbon applications. Adding a dense liquid step will also necessitate a drying stage, which is energy-intensive.

Characterisation on two concentrated chars (fly ash derived) was conducted. It was found that low volatile percentages were present in the concentrated chars. However, even small percentages of volatile matter will have to be removed in a carbonisation step before graphitization can take place. It was seen that oxygen and nitrogen were the two major volatile species present. Following carbon form analysis, the particles were identified as elemental carbon with the presence of strong carbon-carbon bonds typically found in graphite material.

The concentrated chars contained significant amount of Fe/Mg/Ca-alumino-silicate glass (amorphous materials). These amorphous materials derived from the interactions of the inherent fluxing minerals that associated with included/excluded kaolinite or other aluminium silicate at elevated temperatures to form melt. Anorthite and mullite crystallised out from the melt and amorphous alumino-silicate materials were formed. Other included submicron

99 | P a g e minerals or organic Ca/Fe/Mg could catalyse the graphitization reactions during heat treatment under partial oxidizing or inert conditions.

Structural analyses showed that the concentrated chars consist of very small graphite crystallites with a large interlayer spacing. The material can be described as turbostratic, with a lack of three-dimensional orientation. Raman microspectroscopy classifies the char concentrates as transitional carbonaceous material with the possibility of graphitization.

Textural analyses showed that the concentrated chars consist of inertinitic char and anisotropic char. Anisotropic char has a similar structure to graphite. Reflectance data showed that the char concentrates had lower bireflectance values than graphite.

Overall, it is predicted that the char concentrates will be suitable for graphitization. FA PS2 CC is more ordered than FA PS4 CC.

100 | P a g e

Chapter 4: Char concentrate potential to graphitize

4.1 Introduction As part of the ERA-MIN collaboration objectives, the concentrated char samples from Portugal, Poland, South Africa, and Romania will be graphitized and tested in green energy applications. Seeing as the graphitization steps were still ongoing at the time of the thesis completion, the current chapter only aims to predict whether or not the concentrated chars will be graphitizable or not. This will be done through the characterisation of the char concentrates from the different consortium countries and forms part of Work Package 2 (Section 4.3). A brief summary of the parent coal and ash characterisation results are also provided in Section 4.2. For detailed results, Badenhorst et al. (2019) can be consulted.

4.2 A brief summary on ERA-MIN coal and ash characterisation results 4.2.1 Methodology Ash and corresponding coal samples from Portugal, Poland, South Africa, and Romania were obtained. The sample information is provided in Table 4.1. The South African samples correspond to FA PS4 and C PS4 (Sections 2.5.2 and 2.5.3). The Portuguese coal was imported from Colombia. Romania was not able to supply a coal sample.

Table 4.1: Information of ERA-MIN ash and coal samples.

Sample ID Sample origin

Portuguese coal Pegop power plant Portuguese fly ash

Polish coal Elektrownia Siersza power plant Polish bottom ash

South African coal Eskom power utility South African fly ash

Romanian bottom ash Govora power plant

The samples were divided into representative fractions using a rotary splitter. The characterisation techniques were:

 Proximate and ultimate analyses (Bureau Veritas Testing and Inspections SA);  XRF and XRD analyses (School of Biological, Earth and Environmental Sciences of the University of New South Wales);  Petrographic analysis (University of Johannesburg);  PSD (University of Porto).

101 | P a g e

The standards for the proximate, ultimate, petrographic, and PSD analyses were provided in section 2.5.1. Ultimate analysis was omitted for the ash samples, seeing that their percentage of organic matter is too low for accurate speciation. For the XRF analysis, coal samples were ashed at 815 °C and the ash yield determined in each case. The ashed coals as well as the ash samples were then calcined at 1050 °C, fused with lithium metaborate, and cast into discs (Norrish and Hutton, 1969). Each disc was analysed by XRF spectrometry using a Philips PW 2400 spectrometer and SuperQ software. The results were expressed as percentages of the major elements in each ash sample. For the XRD analysis, the coals were first subjected to low-temperature oxygen-plasma ashing (LFE 4-chamber asher) (according to Australian Standard 1038.22 (2002)), and the weight percentage of low- temperature ash (LTA) determined in each case. The LTA and ash samples were then analysed by a PANalytical Empyrean diffractometer with Co K훼 radiation. Minerals were identified by the ICDD Power Diffraction File, whilst SiroquantTM software was used for quantitative analysis. The ash samples were found to contain significant proportions of amorphous material. A poorly-crystalline alumino-silicate phase (identified as “metakaolin”) was used as a substitute to represent the amorphous component (Ward and French, 2006).

4.2.2 Coal and ash characterisation results for ERA-MIN samples The proximate, ultimate, XRF, XRD, and PSD results are provided in Table 4.2. Fixed carbon and oxygen percentages were determined by difference. The Al2O3 and SiO2 percentages are high for both the coal and ash samples. For the coal samples, this is a direct reflection of the large amount of clay (kaolinite and illite) and quartz minerals present in the samples. For the South African coal sample, the occurrence of illite is not as pronounced as for the other two coal samples. However, the kaolinite content for this sample is markedly higher. The high calcite content for the Polish coal sample is reflected in the high CaO content. For the ash samples, the amorphous and mullite transformation products, as well as the quartz relics, are responsible for the high Al2O3 and SiO2 percentages. The South African fly ash sample has considerably lower amorphous and higher quartz contents compared to the other samples. Other ash minerals include mullite, hematite, and maghemite. Romanian bottom ash also hosted some anorthite crystals.

The Polish bottom ash has a d50 of ~55 μm, which is relatively small compared to typical bottom ash samples as well as the Romanian bottom ash sample. The latter has a d50 of

~120 μm. The South African fly ash has a d50 of ~135 μm, which is relatively coarse compared to typical fly ash samples and the Portuguese fly ash sample (d50 ~45 μm). In fact, this sample has an even larger d50 value than the Romanian bottom ash. The PSDs and carbon distribution results for the ash samples can also be accessed from Cruceru et al. (2017), Santos et al. (2017), and Valentim et al. (2018).

102 | P a g e

Table 4.2: Proximate, XRF, XRD, and PSD analyses results for the ERA-MIN ash samples, proximate and ultimate analyses results for the coal samples, XRF analysis results for the high temperature ashes of the coal samples, and XRD analysis results for the LTA ashes of the coal samples.

Portuguese Polish coal South African Portuguese Polish South African Romanian coal coal fly ash bottom ash fly ash bottom ash Proximate Moisture (a.d.b wt. %) 3.5 3.0 3.8 0.3 0.3 0.5 1.8 Ash yield (a.d.b wt. %) 9.4 22.3 30.2 94.8 88.4 92.0 83.6 Volatile matter (a.d.b wt. %) 36.2 28.0 22.5 1.5 3.0 0.9 5.7 Fixed carbon (a.d.b wt. %) 50.9 46.7 43.5 3.4 8.3 6.6 8.9

Volatile matter (d.a.f wt. %) 41.6 37.5 34.1 30.6 26.5 12.0 39.0 Fixed carbon (d.a.f wt. %) 58.4 62.5 65.9 69.4 73.5 88.0 61.0

Ultimate Total sulphur (a.d.b wt. %) 0.50 0.47 0.73 - - - - Carbon content (a.d.b wt. %) 69.70 60.80 52.50 - - - - Hydrogen content (a.d.b wt. %) 4.76 3.74 3.06 - - - - Nitrogen content (a.d.b wt. %) 1.32 1.09 0.90 - - - - Oxygen content (a.d.b wt. %) 10.92 8.70 8.79 - - - - Moisture and ash (a.d.b wt. %) 12.90 25.30 34.0 - - - -

Total sulphur (d.a.f wt. %) 0.57 0.63 1.11 - - - - Carbon content (d.a.f wt. %) 80.02 81.39 79.55 - - - - Hydrogen content (d.a.f wt. %) 5.46 5.01 4.64 - - - - Nitrogen content (d.a.f wt. %) 1.52 1.46 1.36 - - - - Oxygen content (d.a.f wt. %) 12.54 11.65 13.32 - - - - X-ray fluorescence (wt. %)

SiO2 60.17 41.94 59.38 59.1 46.60 66.95 49.82

Al2O3 22.36 23.21 27.27 19.9 23.44 14.3 18.68

Fe2O3 6.71 6.07 2.38 6.14 5.56 5.36 6.03 CaO 1.26 16.33 3.65 1.38 4.01 2.8 3.47

SO3 1.43 2.73 1.91 0.1 0.1 0.05 0.35

103 | P a g e

K2O 2.52 2.4 0.76 2.25 2.50 0.73 1.8 MgO 1.74 2.43 1.66 1.75 2.67 0.83 1.74

Na2O 1.38 0.98 0.2 1.25 1.53 0.14 0.49

P2O5 0.17 0.29 0.3 0.18 0.22 0.18 0.09

TiO2 1.07 1.02 1.65 0.9 1.06 1.14 0.67

Mn3O4 0.06 0.09 0.05 0.08 0.09 0.05 0.06 LOI - - - 5.85 11.35 6.12 17.21 ASTM C618 class F F F F X-ray diffraction (wt. %) LTA 12.7 28.1 35.2 - - - - Quartz 33.1 14.0 22.6 14.7 9.0 46.2 13.6 Kaolinite 23.2 32.4 59.7 - - - - Illite 32.2 24.4 5.5 - - - Trace Chlorite 5.0 ------Na / Ca feldspar 3.3 ------K feldspar - 4.7 - - - - - Pyrite 1.9 1.2 0.1 - - - - Calcite - 14.6 0.9 - 0.6 - 0.4 Dolomite - 6.2 3.2 - - - - Bassanite 1.2 2.5 1.3 - - - - Gypsum - - 2.4 - - - - Aluminite - - 3.5 - - - - Rutile - - 0.8 - - 0.6 - Mullite - - - 4.2 15.2 18.7 5.2 Cristobalite ------Hematite - - - 0.7 0.9 0.7 0.6 Maghemite - - - 1.5 1.3 2.6 0.7 Anorthite ------4.5 Anhydrite - - - - 0.7 - - Spinel - - - - 0.3 - - Lime - - - - 0.7 - - Amorphous - - - 78.9 71.3 31.2 75.0 d50 (µm) - - - 45 55 135 120

104 | P a g e

The petrographic results for the coal and ash samples are provided in Tables 4.3 and 4.4 respectively. Portuguese fly ash consists mainly of anisotropic char whilst Romanian bottom ash consists mainly of isotropic char. The Portuguese coal sample has a high vitrinite content of 74.6 vol. % and is of bituminous rank origin which softens upon heating, explaining the higher anisotropic char content in the resultant fly ash. Although the coal from Romania was not provided or analysed, it is known that it is the Oltenia lignite (Gorj County, Romania) origin as this is the typical feed to this power plant. Lignite and other low-rank coals do not pass through a softening stage (Hower, 2012), and therefore the dominance of isotropic char in the Romanian bottom ash sample. For the South African fly ash, the organic content consists of mainly inertinitic material. The South African coal has a relatively high inertinite content (48.6 vol. %) that is being reflected directly in the resultant fly ash sample.

The amorphous phase is dominant in the ash samples, in agreement with the XRD results. Solid glassy spheres, cenospheres, pseudoplerospheres, and angular glass particles were all observed in the ash samples. A fraction of the glass particles contained surface elongated and needle crystal inclusions. According to Watt and Thorne (1965), these crystals are mullite (and anorthite in the case of bottom ashes), whilst Fisher et al. (1978) suggested that it might be formed from heterogeneous nucleation on the surface of molten ash. The quartz content for South African fly ash is much higher than for the other samples.

105 | P a g e

Table 4.3: Petrographic results for the ERA-MIN coal samples.

Classification Portuguese coal Polish coal South African coal

Vitrinite (vol. %) 74.6 49.8 15.3 Telinite 0.7 0.4 0.6 Collotelinite 10.5 33.6 8.6 Vitrodetrnitie 0.4 0.0 0.0 Collodetrinite 44.9 14.4 5.7 Corpogelinite 18.1 0.8 0.4 Gelinite 0.0 0.0 0.0 Pseudovitrnite 0.0 0.6 0.0

Inertinite (vol. %) 19.2 30.6 48.6 Fusinite 8.0 8.6 4.5 Reactive semifusinite 2.5 0.4 1.0 Inert semifusinite 5.1 11.0 18.4 Micrinite 0.0 1.2 0.0 Macrinite 0.0 0.0 0.2 Secretinite 0.7 0.4 2.9 Funginite 0.7 0.0 0.0 Inertodetrinite R 0.0 0.0 0.8 Inertodetrinite I 2.2 9.0 20.8

Liptinite (vol. %) 4.3 5.0 0.8 Sporinite 1.4 4.0 0.8 Cutinite 2.2 0.0 0.0 Resinite 0.4 1.0 0.0 Alginite 0.0 0.0 0.0 Liptodetrinite 0.0 0.0 0.0 Suberinite 0.0 0.0 0.0 Exsudatinite 0.4 0.0 0.0

Mineral Matter (vol. %) 1.8 14.6 35.3 Silicate (clay / quartz) 1.8 9.0 32.0 Sulfide 0.0 1.0 1.0 Carbonate 0.0 4.6 2.0 Other 0.0 0.0 0.4

Total reactive macerals (vol. %) 81.4 55.2 17.9

Vitrinite reflectance Rrandom (RoV%) 0.59±0.06 0.73±0.16 0.63±0.07 Rmax (RoV%) 0.75 0.98 0.97 Rmin (RoV%) 0.47 0.33 0.52 Rank Category Medium Rank D, border lining C Medium Rank C Medium Rank C

106 | P a g e

Table 4.4: Petrographic results for the ERA-MIN ash samples.

Technique Portuguese Polish South Romanian fly ash bottom African fly bottom ash ash ash

Organics (vol. %) 7.2 12.2 6.7 8.2 Anisotropic char (vol. %) 5.6 6.1 1.6 0.6 Isotropic char (vol. %) 0.6 2.4 0.8 6.2 Inertinite (vol. %) 1.0 3.7 4.3 0.8 Unreacted / partially burned coal (vol. %) 0.0 0.0 0.0 0.6

Inorganics (vol. %) 92.8 87.8 93.3 91.8 Glass (vol. %) 84.6 78.9 34.7 40.2 Quartz (vol. %) 7.0 4.3 51.4 7.4 Mullite (vol. %) 0.0 0.0 0.0 0.0 Anorthite (vol. %) 0.0 0.0 0.0 0.0 Baked clay (vol. %) 0.2 2.2 1.6 41.6 Dense iron (vol. %) 0.4 1.2 3.4 0.8 Dendritic iron (vol. %) 0.6 1.2 1.8 0.2 Other (vol. %) 0.0 0.0 0.4 1.6 4.3 Char concentrate potential to graphitize 4.3.1 Methodology Ash samples from Portugal, Poland, South Africa, and Romania were collected and the char fractions separated by the respective countries. The sample information is provided in Table 4.5. The South African sample corresponds to FA PS4 CC (Section 3.4.2).

Table 4.5: Sample information of ERA-MIN char concentrates for characterisation.

Sample ID Initial carbon grade Product carbon grade (a.d.b wt. (a.d.b wt. %, fixed carbon) %, fixed carbon)

Portuguese char 3.4 75.1 Polish char 12.7 70.1 South African char 5.7 74.7 (82.91 LOI) Romanian char 8.9 59.2

The samples were divided into representative fractions using a rotary splitter. The char concentrates were submitted to the University of Johannesburg for analysis. The characterisation techniques were:

 Proximate analysis (Bureau Veritas Testing and Inspections South Africa);  Ultimate analysis (Bureau Veritas Testing and Inspections South Africa);  Carbon form analysis (Bureau Veritas Testing and Inspections South Africa);  XRD (mineralogy) analysis (XRD Analytical and Consulting);  Raman microspectroscopy analysis (University of Johannesburg), and  Petrographic analysis (University of Johannesburg)

107 | P a g e

The characterisation methodologies were provided in Section 3.4.2. Unfortunately, the XRD structural and solid-state Nuclear Magnetic Resonance (ss-NMR) results are not yet available. The samples are currently analysed for their XRF composition by the Portuguese partner. The results were not available upon submission of this thesis.

4.3.2 Char characterisation results for ERA-MIN concentrates The proximate, ultimate, carbon form and XRD mineralogy results are presented in Table 4.6. Oxygen and fixed carbon contents are reported by difference.

The volatile matter content for the Romanian char is high compared to the other samples. This result is also reflected in the higher oxygen, hydrogen, and sulphur contents, and the higher organic carbon percentage of this sample. The inclusion of heteroatoms in the Romanian char matrix indicates elemental impurities and will have to be removed in a carbonisation step before graphitization commences. The volatile matter content for the Portuguese char is relatively low compared to the other samples. This sample also has a low oxygen content and a high elemental carbon percentage. The latter is of importance as it shows the inclusion of strong carbon-carbon bonds (similar to the bonding arrangement of graphite) in the char matrix. The South African and Polish chars also have high elemental carbon percentages but not as dominant as in the Portuguese char.

The amorphous phase is dominant in all samples and includes both amorphous glass and amorphous carbon. Mullite and quartz are present in lesser amounts in all the samples. The mullite percentages are slightly higher, possibly due to its close association (interlocked nature) with the glass phase. Silicon and iron are known to act as catalysts during graphitization (Cabielles et al., 2008; 2009). Both these elements are present in the glass phase, while silicon is also present in the mullite and quartz minerals. Other low count minerals include gypsum (South African and Romanian chars), calcite (Polish and Romanian chars), and bassanite (Romanian char). The presence of calcite in the Polish and Romanian chars was also seen in the higher inorganic carbon percentages of these two samples. Calcite was also detected as a major mineral in the Polish coal sample.

108 | P a g e

Table 4.6: Proximate, ultimate, carbon form, and XRD mineralogy results for the ERA-MIN char concentrates.

Portuguese Polish South African Romanian char char char char

Proximate Moisture (a.d.b wt. %) 0.5 3.3 3.6 7.6 Ash yield (a.d.b wt. %) 22.4 22.2 17.7 13.6 Volatile matter (a.d.b wt. %) 2.0 4.4 4.0 19.6 Fixed carbon (a.d.b wt. %) 75.1 70.1 74.7 59.2

Volatile matter (d.a.f wt. %) 2.6 5.9 5.1 24.9 Fixed carbon (d.a.f wt. %) 97.4 94.1 94.9 75.1

Ultimate Total sulphur (a.d.b wt. %) 0.32 0.36 0.30 1.29 Carbon content (a.d.b wt. %) 74.50 70.70 75.00 67.10 Hydrogen content (a.d.b wt. %) 0.37 0.28 0.18 1.70 Nitrogen content (a.d.b wt. %) 0.98 0.90 1.11 0.56 Oxygen content (a.d.b wt. %) 0.93 2.26 2.11 8.15 Moisture and ash (a.d.b wt. %) 22.90 25.50 21.30 21.20

Total sulphur (d.a.f wt. %) 0.42 0.48 0.38 1.64 Carbon content (d.a.f wt. %) 96.63 94.90 95.30 85.15 Hydrogen content (d.a.f wt. %) 0.48 0.38 0.23 2.16 Nitrogen content (d.a.f wt. %) 1.27 1.21 1.41 0.71 Oxygen content (d.a.f wt. %) 1.21 3.03 2.68 10.34

Atomic H/C 0.06 0.05 0.03 0.30 Atomic O/C 0.01 0.02 0.02 0.09

Carbon form Total carbon (wt. %) 75.37 74.54 75.45 69.50 Elemental carbon (wt. %) 69.83 56.30 66.58 3.30 Organic carbon (wt. %) 4.12 14.01 7.18 63.10 Inorganic carbon (wt. %) 1.42 4.23 1.69 3.10

Elemental carbon (% of total carbon) 92.65 75.53 88.24 4.75 Organic carbon (% of total carbon) 5.47 18.80 9.52 90.79 Inorganic carbon (% of total carbon) 1.88 5.67 2.24 4.46

XRD (wt. %)

Quartz (SiO2) 2.4 1.1 2.0 0.4

Mullite (3Al2O32SiO2 / 2Al2O3SiO2) 5.6 5.1 6.6 0.9

Gypsum (CaSO4∙2H2O) 0.0 0.0 0.1 0.1

Calcite (CaCO3) 0.0 0.6 0.0 0.5 Bassanite 0.0 0.0 0.0 0.4

(CaSO4∙0.5H2O/2CaSO4∙H2O) Amorphous 92.0 93.2 91.4 97.7

The petrographic results are presented in Table 4.7. The ICCP char classification results are illustrated in Figure 4.1. The Portuguese sample consists predominantly of anisotropic char

109 | P a g e

(Hower (2012) classification). Anisotropic char forms from vitrinite macerals in coal which undergo a mesophase during combustion (Hower, 2012). The mesophase allows the carbon atoms to realign themselves in an ordered graphite-like manner. The high anisotropic percentage in the Portuguese sample is therefore indicative of a higher degree of graphitization in this sample. The Romanian sample consists mainly of isotropic char and unreacted / partially burned coal. Isotropic char is associated with low-rank coal precursors (lignite) in which the vitrinite (technically huminite) macerals do not go through a mesophase during combustion, and therefore the carbon atoms do not realign in a graphite-like manner. The Romanian sample, therefore, has a lower degree of pre-graphitization. The inclusion of unreacted / partially burned coal particles is also indicative of a disordered carbonaceous structure. This also explains why the volatile matter and organic carbon contents are higher for this particular sample, as it consists of a high percentage of coal particles.

The Polish sample consists of a mix of anisotropic and isotropic char particles, and the South African sample consists of a mix of anisotropic and inertinitic char (inertoids and fusinoids) particles. The inclusion of high percentages of anisotropic char shows a higher degree of order than seen in the Romanian sample, but not as pronounced as in the Portuguese sample.

The bireflectance determined for the samples ranges between 0.21 (Romania) and 2.15 (Poland). The values are very low, seeing that for graphite the bireflectance ranges between 6 and 27 (Craig and Vaughan, 1994). The anisotropy percentages range between 0.05 (Romania) and 0.25 (Poland). The anisotropic char visually shows very incipient anisotropy, not the "mosaic" of coke. The low anisotropy for Romanian char is due to this sample’s high isotropic content.

110 | P a g e

Table 4.7: Petrographic analysis results for the ERA-MIN char concentrates.

Portuguese Polish South Romanian char char African char char

Hower (2012) classification Glass (vol. %) 5.0 8.2 5.8 4.2 Anisotropic char (vol. %) 83.3 57.8 43.2 0.6 Isotropic char (vol. %) 9.5 24.0 15.6 46.4 Inertinitic char (vol. %) 2.2 9.8 35.0 3.6 Unreacted / partially burned coal (vol. %) 0.0 0.2 0.4 45.2

Anisotropic char (% of total organics) 87.7 63.0 45.9 0.6 Isotropic char (% of total organics) 10.0 26.1 16.6 48.4 Inertinitic char (% of total organics) 2.3 10.7 37.2 3.8 Unreacted / partially burned coal (% of total 0.0 0.2 0.4 47.2 organics)

ICCP char classification Tenuisphere (vol. %) 17.9 0.0 0.0 0.2 Crassisphere (vol. %) 34.6 7.1 0.0 0.0 Tenuinetwork (vol. %) 14.3 26.0 2.9 8.5 Crassinetwork (vol. %) 18.4 31.3 23.3 24.2 Mixed porous (vol. %) 7.8 23.2 40.4 32.6 Mixed dense (vol. %) 0.0 2.8 22.4 11.7 Inertoid (vol. %) 0.5 0.6 5.5 2.7 Fusinoid / solid (vol. %) 1.7 1.5 3.9 14.9 Mineroid (vol. %) 4.8 7.5 1.6 5.2

Tenuisphere (% of total organics) 18.8 0.0 0.0 0.2 Crassisphere (% of total organics) 36.3 7.7 0.0 0.0 Tenuinetwork (% of total organics) 15.0 28.1 2.9 9.0 Crassinetwork (% of total organics) 19.3 33.8 23.7 25.5 Mixed porous (% of total organics) 8.2 25.1 41.1 34.4 Mixed dense (% of total organics) 0.0 3.0 22.8 12.3 Inertoid (% of total organics) 0.5 0.6 5.6 2.8 Fusinoid / solid (% of total organics) 1.8 1.6 4.0 15.7

Reflectance

Rmean (Rr%) 8.05±0.82 7.40±0.74 7.23±0.53 3.96±1.59

Rmax (Rr%) 8.75±1.11 8.48±1.32 8.13±1.55 4.06±1.54

Rmin (Rr%) 7.34 6.33 6.32 3.86 Bireflectance (Rr%) 1.40 2.15 1.82 0.21 Anisotropy 0.16 0.25 0.22 0.05

111 | P a g e

A B

C D

Figure 4.1: A) A high percentage of crassispheres was observed in the Portuguese char, indicating possible application in fluid adsorption, B) Large network and mixed porous particles were seen in the Polish char, C) Mixed porous particles with rare glassy inclusions were seen in the South African char, D) The Romanian char sample had a higher percentage of fusinoid / solid and was also fragmented due to pulverisation during beneficiation (Reflected-light, oil immersion, ×500).

The Raman microspectroscopy first order spectra are of importance for this study, and are illustrated in more detail (Figure 4.2). The spectra for the chars from Portugal, Poland, and South Africa (mainly anisotropic char) were similar (because the coal rank is similar) and an example is illustrated in Figure 4.2A. For the sample from Romania, char (mainly isotropic char), as well as more disordered carbon types (possibly corresponding to the partially and unreacted coal observed via petrography), were seen. The disordered carbon types were divided into Carbon I and Carbon II. These all had different spectra as illustrated in Figures 4.2B, C, and D respectively.

112 | P a g e

Figure 4.2: Typical 1st order Raman microspectra obtained from the anisotropic char particles from Portugal, Poland, and South Africa (A), the isotropic char particles from Romania (B), the Carbon I particles from Romania (C), and the Carbon II particles from Romania (D).

Curve fitting results are presented in Table 4.8. There is not much difference between the anisotropic char from Portugal, Poland, and South Africa and the isotropic char from Romania. The D1 FWHM for the char from Portugal is slightly lower than for the other samples, while the char from Romania has a slightly higher D1 FWHM. This is indicative of respectively a higher and lower structural ordering. The D1 band area for the Portuguese char was subsequently also lower than the other samples. As a result, the D1/G FWHM and area ratios, as well as the RA2 value, were lower for the Portuguese char and higher for the Romania char. Following the classification from Kouketsu et al. (2014), the chars from Portugal, Poland, South Africa, and Romania were subsequently grouped as “transitional” carbonaceous material; implying they are of an intermediate rank between amorphous and graphitic carbonaceous material.

There are however some major differences between the chars (Poland, Portugal, South Africa, and Romania) and the Carbon I and II particles from Romania. For the latter, the G band intensities were higher than the D1 band intensities. Due to these low D1 band

113 | P a g e intensities, the area and subsequent area ratios and RA2 were lower than for the char samples. Due to a higher D1 position (1363-1374 cm-1), the distance between the G and D1 bands are also smaller than for the char samples. A lower wavenumber for D1 is attributed to the increase of larger aromatic clusters (Ferrari and Robertson, 2000) and therefore a smaller distance between the D1 and G positions shows a more disordered nature (Schito et al., 2017). The D2 (1603-1617 cm-1) and D3 (1469-1477 cm-1) positions are lower than for the chars. The G FWHM and D1 FWHM for both Carbons I and II are much larger than for the char samples. Subsequently, their FWHM ratios are also higher than for the char samples. This is once again indicative of a more disordered nature. Following the classification by Kouketsu et al. (2014), the Carbons I and II char from Romania were subsequently grouped as “amorphous” carbonaceous material.

4.4 Summary A prediction is made that the char concentrates from Portugal, Poland, and South Africa will be graphitizable. These samples all have high elemental carbon percentages, low volatile matter percentages, high anisotropic char contents, and can be classified as “transitional carbonaceous material” (varying between amorphous carbon and graphite). Silicon and iron species, which are known to act as catalysts during graphitization, are also present in these samples (Cabielles et al., 2008; 2009). The Portuguese char seems the most graphitizable.

A prediction is made that the char concentrate from Romania will be non-graphitizable. This sample has a high organic carbon percentage, a high volatile matter content, high isotropic char, and unreacted / partially burned coal percentages, and can be classified as “amorphous carbonaceous material” (non-graphitizable).

114 | P a g e

Table 4.8: Raman microspectroscopy curve fitting results for the ERA-MIN char samples.

Sample Portugal char Poland char South Africa Romania char Romania Carbon I Romania Carbon II char

Band position (cm-1) G 1593.72 ± 2.92 1585.85 ± 4.42 1592.73 ± 2.35 1585.86 ± 2.96 1586.03 ± 1.49 1587.94 ± 0.84 D1 1358.75 ± 3.51 1355.93 ± 4.58 1360.19 ± 3.85 1350.49 ± 0.97 1363.05 ± 4.05 1373.87 ± 3.63 D2 1629.58 ± 1.41 1623.76 ± 4.99 1629.91 ± 0.31 1625.88 ± 4.36 1617.03 ± 8.09 1603.46 ± 3.63 D3 1487.42 ± 3.44 1484.21 ± 3.19 1491.95 ± 2.29 1485.31 ± 3.59 1476.56 ± 4.40 1468.65 ± 5.20 D4 1209.50 ± 13.92 1196.90 ± 12.69 1201.16 ± 8.38 1186.68 ± 14.42 1212.56 ± 9.85 1202.48 ± 21.99 Band FWHM (cm-1) G 113.31 ± 4.42 117.85 ± 5.49 116.13 ± 5.90 114.73 ± 3.54 121.36 ± 4.01 126.63 ± 2.44 D1 148.43 ± 11.09 166.06 ± 13.52 166.39 ± 11.52 181.37 ± 11.84 186.51 ± 9.08 205.06 ± 15.66 D2 41.34 ± 2.33 42.22 ± 3.22 42.32 ± 3.28 40.58 ± 1.58 52.54 ± 13.00 65.43 ± 8.22 D3 104.88 ± 5.62 100.48 ± 4.28 101.09 ± 2.87 101.19 ± 6.23 109.41 ± 10.51 103.14 ± 9.15 D4 197.51 ± 5.85 197.20 ± 7.50 197.17 ± 6.10 183.58 ± 12.00 199.47 ± 10.74 189.64 ± 22.48 Band area (cm-1) G 67.40 ± 7.66 73.67 ± 6.79 73.55 ± 6.49 71.98 ± 4.07 79.64 ± 4.12 71.06 ± 5.13 D1 103.12 ± 11.29 119.02 ± 12.48 126.37 ± 12.18 128.53 ± 6.10 104.08 ± 10.05 90.50 ± 13.81 D2 2.63 ± 0.21 2.24 ± 0.27 2.98 ± 0.22 2.39 ± 0.43 2.34 ± 1.53 4.04 ± 2.54 D3 32.93 ± 4.53 30.89 ± 5.41 32.16 ± 3.42 29.99 ± 6.28 22.55 ± 7.72 9.43 ± 2.59 D4 28.48 ± 4.13 29.18 ± 4.27 33.41 ± 5.80 28.13 ± 7.54 30.79 ± 7.88 20.21 ± 6.06 Band intensity (normalised, a.u.) G 0.56 ± 0.05 0.59 ± 0.04 0.59 ± 0.03 0.59 ± 0.04 0.62 ± 0.02 0.53 ± 0.03 D1 0.65 ± 0.06 0.67 ± 0.04 0.71 ± 0.05 0.67 ± 0.05 0.52 ± 0.04 0.41 ± 0.04 D2 0.04 ± 0.00 0.03 ± 0.00 0.05 ± 0.00 0.04 ± 0.01 0.03 ± 0.01 0.04 ± 0.02 D3 0.20 ± 0.02 0.18 ± 0.02 0.20 ± 0.02 0.19 ± 0.03 0.14 ± 0.03 0.06 ± 0.01 D4 0.14 ± 0.02 0.15 ± 0.02 0.16 ± 0.03 0.14 ± 0.01 0.13 ± 0.04 0.10 ± 0.02 Calculated parameters D-G distance (cm-1) 234.97 ± 4.17 229.92 ± 4.77 232.54 ± 3.48 235.37 ± 2.30 222.98 ± 3.47 214.07 ± 3.94 D1/G FWHM ratio 1.31 ± 0.07 1.41 ± 0.09 1.43 ± 0.06 1.58 ± 0.06 1.54 ± 0.08 1.62 ± 0.12 D1/G area ratio 1.53 ± 0.12 1.62 ± 0.11 1.72 ± 0.11 1.79 ± 0.17 1.31 ± 0.08 1.27 ± 0.15 RA1 0.56 ± 0.02 0.59 ± 0.02 0.59 ± 0.02 0.60 ± 0.02 0.56 ± 0.02 0.57 ± 0.03 RA2 1.28 ± 0.09 1.43 ± 0.11 1.47 ± 0.11 1.51 ± 0.12 1.29 ± 0.08 1.31 ± 0.13

115 | P a g e

Chapter 5: Natural graphite in southern Africa

5.1 Introduction Graphite is at the forefront of technological development, with Li-ion batteries, fuel cells, and pebble-bed nuclear reactors all relying on graphite to be able to function. Concurrently, China (the world’s largest natural graphite producing country) is drastically decreasing its natural graphite output due to environmental concerns. As a result, the EU, U.S., and British Geological Survey have all listed natural graphite as one of their top critical materials; exhibiting it as both a high supply risk and of great economic importance (British Geological Survey, 2015; European Commission, 2017; Fortier et al., 2018). Although synthetic graphite can be used, the high costs associated with manufacturing (~U.S. $10 000 – 20 000 per tonne selling price) can lead to its natural counterpart (~U.S. $800 – 900 per tonne selling price) being preferred (Spencer and Hill, 2016). Africa is known to host the largest untapped deposits of natural graphite globally and might present an attractive solution to the current graphite crisis. In this chapter, southern Africa will be considered as a possible source of natural graphite via a report on natural graphite in this region (Sections 5.2 to 5.11) and the characterisation of selected occurrences for quality determination (Section 5.12).

5.2 An introduction to natural graphite Natural graphite is formed through regional or contact metamorphism of organic carbonaceous sediments and occurs as flake, amorphous, or vein-type graphite. The characteristics of each differ substantially (Table 5.1).

Flake graphite is formed from the regional metamorphism of methane and fine crude oil droplets in sedimentary rocks and is syngenetic to its host rock. As the name suggests, this type of graphite occurs as “flakes” (average 2.5 mm), disseminated in layers or as lenses / pockets in marble, schist, quartzite, and gneiss. Flake graphite contributes ~40 % to the natural graphite market (Otto, 2011) and is used in refractory, brake lining, lubricant, and Li- ion battery applications (Fogg and Boyle, 1987; Mitchell, 1993). Expensive processing techniques, such as froth flotation and chemical purification, are often used to purify the ore (ore grade 5-30 % graphite) (Krauss et al., 1988). Froth flotation yields a 75-97 % graphite grade while chemical treatment yields a 98-99.9 % graphite grade (Krauss et al., 1988). Although purity is important for graphite consumers, the flake size can be seen as just as relevant. Flake graphite is divided into jumbo (> 300 µm), large (180-300 µm), medium (150- 180 µm), small (75-150 µm), and fine / amorphous (< 75 µm) sizes, with the market price increasing with size (Table 5.2). Austria, Brazil, India, Canada, China, Germany, Zimbabwe, and Madagascar are the major producing countries (Feytis, 2010; Mitchell, 1993).

116 | P a g e

Amorphous graphite, contrary to the nomenclature, is a crystal, but with a very low degree of order. It is formed from the syngenetic contact metamorphism of coal seams, or highly carbonaceous shales. Amorphous graphite consists of aggregates of fine graphite crystals and is finely interlaced with impurities. The aggregates are usually very small (40-70 μm) and the retail price is much lower than for flake or vein graphite. Amorphous graphite is priced at U.S. $400 to 600 per tonne (Hayes, 2016; Spencer and Hill, 2016). It contributes ~60 % to the natural graphite market and is used in refractory, steel, paint, and Li-ion battery applications (Fogg and Boyle, 1987; Mitchell, 1993; Otto, 2011). Productive amorphous graphite mines are located in China, Europe, Mexico and the USA (Feytis, 2010). Ore grades are high, varying between 25 and 85 % graphite, and therefore purification consists of simple techniques such as handpicking (Fogg and Boyle, 1987; Krauss et al., 1988; Mitchell, 1993; Otto, 2011). Purified grades range between 60 and 90 % graphite (Fogg and Boyle, 1987).

Vein graphite is the scarcest (~1 %) of the three natural types, but can also be seen as the most valuable (Otto, 2011). Only Sri-Lanka is a competitive producer at present, but Moss (2016) reports on economically vein graphite mining commencing in Namibia. The geological formation of the vein graphite is unique. There is definite evidence that the graphite has migrated to its host rock and is epigenetic. Graphite precipitated from CO2 and CH4 fluids / melts into fissures and cavities of metamorphic (granulite) and igneous rocks. This type of graphite occurs as large bodies and can have carbon percentages of up to 95 %. Therefore, beneficiation and processing techniques are simple and inexpensive. Hand sorting, washing, winnowing, and screening are used to produce product grades ranging between 98 and 99.9 % graphite (Fogg and Boyle, 1987; Krauss et al., 1988). Vein graphite is used in carbon brushes, Li-ion batteries, and seals and gaskets (Fogg and Boyle, 1987; Mitchell, 1993).

117 | P a g e

Table 5.1: Properties of flake, amorphous, and vein natural graphite (compiled from Feytis, 2010; Fogg and Boyle, 1987; Krauss et al., 1988; Luque et al., 2014; Mitchell, 1993; Otto, 2011; Simandl et al., 2015).

Flake Amorphous Vein

Description Flaky texture, disseminated in layers 40-70 μm aggregates which can Occurs as interlocked aggregated or as lenses / pockets in contain anthracite thin films / massive bodies (3 m) in metamorphic rocks, average 2.5 mm fissures / cavities of metamorphic / particle size igneous rocks, ~4 cm particle size

Origin Syngenetic, regional metamorphism Syngenetic, contact metamorphism Epigenetic, precipitation of CO2 and

of methane / crude oil in sediments of coal seams CH4 fluids / melts Host rocks Gneisses, quartzites, schists, Coal Granulites, igneous rocks e.g. marbles, granulites plutonic and volcanic Ore grade (wt. % graphite) 5-30 25-85 >95 Mining methods Surface and underground Surface Underground Beneficiation Crushing, screening, wet grinding, Handpicking, crushing, screening, air Hand sorting, washing, screening, flotation, chemical purification classification winnowing Product grade (wt. % graphite) 75-97 (chemical treatment can yield 60-90 98-99.9 98-99.9) with coarse >150 µm and fine <150 µm flake sizes being sold Main uses Refractories, brake linings, Refractories, steel industry, paint, Carbon brushes, batteries, seals, lubricants, and batteries and batteries and gaskets Major producers Austria, Brazil, India, Canada, China, China, Europe, Mexico, and United Sri-Lanka Germany, Zimbabwe, and States of America (USA) Madagascar % of world production ~40 ~60 lowest priced ~1 most valuable

118 | P a g e

Table 5.2: Natural graphite price (U.S. $ per tonne) for different flake sizes with a 95 % purity (2016-2025) (Spencer and Hill, 2016).

Size (µm) 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025

>300 2 000 2 069 1 866 1 813 1 860 1 936 1 978 1 998 2 000 2 005

150-300 1 000 1 016 943 949 956 981 995 1 002 1 001 1 003

75-150 750 764 720 730 729 742 750 753 752 752

Fines 550 562 525 540 540 549 554 555 552 551

“Basket price” 802 824 770 838 882 908 924 926 913 905

The major natural graphite producing countries and production volumes are provided (Table 5.3). China is the largest producer of natural graphite, with over 65 % of the total global production (U.S. Geological Survey, 2019). However, due to environmental concerns (both mining and processing), China is drastically decreasing its output. It is estimated that by 2025, China’s production will decrease to 300 000 tonnes / annum (Spencer and Hill, 2016).

Table 5.3: Natural graphite producing countries and production volumes in 2018 (U.S. Geological Survey, 2019).

Country Natural graphite production Percentage of total (%) (2018; tonnes)

China 630 000 67.73 Brazil 95 000 10.21 Canada 40 000 4.30 India 35 000 3.76 Mozambique 20 000 2.15 Ukraine 20 000 2.15 Russia 17 000 1.83 Norway 16 000 1.72 Pakistan 14 000 1.51 Madagascar 9000 0.97 Mexico 9000 0.97 North Korea 6000 0.65 Vietnam 5000 0.54 Sri Lanka 4000 0.43 Namibia 2200 0.24 Turkey 2000 0.22 Zimbabwe 2000 0.22 Other 4000 0.43 Total 930 200 100.00

119 | P a g e

5.3 Natural graphite occurrences in southern Africa Africa is known for its untapped deposits of natural graphite. As such, it could be considered as a possible substitute supplier, making up for the decrease in supply from China while the demand for graphite continues to increase. It can be seen (Table 5.3) that Madagascar, Zimbabwe, Namibia, and Mozambique are currently the only African countries reported to be mining natural graphite (and where production only commenced in 2017 for the latter two) (U.S. Geological Survey, 2019). Other economic deposits have been identified in Angola, Botswana, Ethiopia, Kenya, Malawi, South Africa, Swaziland, Tanzania, Uganda, and Zambia (Figure 5.1). Political and economic instabilities are, however, problematic in Africa, affecting investment in graphite mining, as well as other commodities.

Research on natural graphite occurrences in Africa is limited, with only Reimer (1984) having published anything substantial. However, Reimer (1984) only considered South Africa, Namibia, Botswana, and Zimbabwe, and the publication can be considered as “outdated”. An updated literature review on natural graphite occurrences in Africa is provided in the current research. The southern African countries of South Africa, Swaziland, Lesotho, Namibia, Botswana, Zimbabwe, Mozambique, and Madagascar are considered (Figure 5.1). The idea is to create a consolidated database, which will aid and inform governmental officials, mining houses, scientists, academics, environmentalists, and other role-players as to the graphite potential in southern Africa. Information on occurrences, as well as their mining potential, is given.

Figure 5.1: African countries with known natural graphite occurrences as well as African countries with known natural graphite occurrences to be discussed in text.

120 | P a g e

5.4 South Africa In South Africa, natural graphite occurs in rocks located in the Provinces of Limpopo, KwaZulu-Natal, Mpumalanga, Western Cape, Northern Cape, North West, and the Free State. A geological map of South Africa is given in Figure 5.2, indicating the major graphite occurrences and associated lithologies to be discussed.

Figure 5.2: Geological map of South Africa with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008).

5.4.1 Northern Cape Province Graphite deposits in the Northern Cape Province occur as part of the Bushmanland Group of the Namaqua-Natal Belt, which host Middle-to-Late Proterozoic metamorphic rocks (Reimer, 1984). Graphite occurrences have been reported on the farm Oup 80 (Figure 5.2, location 2), at the Aggeneys-Gamsberg zinc deposits (Figure 5.2, location 3), and in the Garies area (Figure 5.2, location 4) (Cairncross and Dixon, 1995; Cairncross, 2004; Department of Mines, 1940; Praekelt and Schoch, 1997; Praekelt et al., 1997; Reimer, 1984). At Oup 80 and Gams, the graphite occurs as stringers in schists, or as flakes in sphalerite ore (Praekelt and Schoch, 1997; Praekelt et al., 1997; Reimer, 1984). At Garies, the graphite is found as disseminated flakes in feldspathic quartzite, but the concentration of graphite is considered to be too low for exploration (Department of Mines, 1940).

Graphite is known to occur as a trace component in mantle-derived xenoliths occurring in various kimberlites in southern Africa (Figure 5.2, location 1) (Gellatly, 1966; Robinson, 1977; Schulze et al., 1997; Viljoen, 1995).

121 | P a g e

There is no past, nor proposed, graphite mining activities in the Northern Cape Province as of the date of this paper.

5.4.2 Western Cape Province Cole et al. (2014) reported that at least six occurrences of graphite occur in the Western Cape Province. These occurrences are located in the Vanrhynsdorp and Malmesbury Groups, and in the Cape Supergroup (lower part of Bokkeveld Series) (Belcher, 2003; Department of Mines, 1940; Reimer, 1978; Reimer, 1984).

In the Vanrhynsdorp area (Figure 5.2, location 5), graphite occurs as <2 mm thick stringers, laminae, and elliptical specks in bluish-grey phyllite rocks. In the richer areas (~46 % graphite), the graphite stringers can reach a thickness of 2 cm. Microscopic and mineralogy studies revealed that the carbon consists of small granules of graphite and reconstituted kerogen that strongly resembles boghead coal (Reimer, 1978; Reimer, 1984).

In the Malmesbury Group (Figure 5.2, location 6), flake graphite is found in schists and the metamorphic temperature of the graphite formation was estimated at 300 °C (Belcher, 2003).

In the Cape Supergroup (Figure 5.2, location 7), near the towns of Ceres, Laingsburg, Prince Albert, Robertson, and Worcester, beds of black shales occur, which have only been reported as being carbonaceous and not graphitic, apart from the black shales at location 8 (Figure 5.2), near the towns of Caledon, Bredasdorp, and Swellendam. The graphitic shales are fine clay / silicates, giving it a soft, black, and shiny state, and are commonly mistaken for being enriched in graphite. The actual graphitic percentages are low and are not suitable for mining (Department of Mines, 1940).

Overall, the graphite in the Western Cape is not seen as being economically viable (Cole et al., 2014).

5.4.3 KwaZulu-Natal Province Graphite is found on the eastern side of the Namaqua-Natal Belt (KwaZulu-Natal Province) in Middle-to-Late Proterozoic metamorphic rocks (Reimer, 1984).

In the Margate Terrane of this Namaqua-Natal Belt (Figure 5.2, location 9), the graphite occurs as quartz-graphite schists in the calc-silicate rich dolomite marbles of the Marble Delta Formation (Cairncross and Dixon, 1995; Cairncross, 2004; McCourt et al., 2006; Otto 1973; Reimer, 1984, Simpson and Tregidga, 1956). The graphite was described occurring as irregular pockets and veins, and disseminated flakes in a coarsely crystalline marble. It has excellent quality, but its low quantity hinders exploitation. The associated limestone,

122 | P a g e however, can be extracted and graphite could be produced as a by-product (Department of Mines, 1940). Cairncross and Dixon (1995) and Thomas et al. (1994) also reported the rare association of graphite with spodumene in the Highbury pegmatites of this area. The Department of Mines (1940) also mentions the occurrence of graphite pellets, set in a matrix of feldspar and micropegmatite enclosing crystals of cordierite and enstatite, in this Terrane. The graphite represents fragments of carbonaceous shale that underwent partial fusion. It is considered as being uneconomical due to the intergrowth of associated minerals.

In the Mzumbe Terrane of the Namaqua-Natal Belt (Figure 5.2, location 10), the graphite occurs in granitic gneisses and graphite rich marbles (~40 % carbon) (Kerr et al., 1987; Reimer, 1984).

In the Tugela Terrane of the Namaqua-Natal Metamorphic Belt (Figure 5.2, location 11), graphite schists occur (Bisnath et al., 2008; McCourt et al., 2006; Scogings, 1991).

The Department of Mines (1940) mentioned the occurrence of amorphous graphite near Ladysmith (Doornkloof farm), while Krauss et al. (1988) mentioned the occurrence of amorphous graphite near Mtubatuba (Figure 5.2, location 12). Graphite is a result of a dolerite intrusion and is associated with coal (Department of Mines, 1940; Snyman and Barclay, 1989). However, the graphitic conversion is not complete, and therefore the deposits have low economic value (Department of Mines, 1940).

The graphite occurrences of KwaZulu-Natal have been described as uneconomical (Steenkamp, 2017).

5.4.4 Mpumalanga Province Within the Archaean schist belts of the Barberton Supergroup (Fig Tree Group), graphite has been found in dormant gold mines (Figaro and French’s Bob) and current gold mines (Sheba, Fairview, and New Consort) (Figure 5.2, location 13) (Altigani et al., 2016; Cairncross and Dixon, 1995; Reimer, 1984). The graphite occurs in stringers of graphitic schists embedded in ~6 m wide zones of carbonaceous cherts and shales (Hofmann and Bolhar, 2007; Reimer, 1984). Graphitization most likely took place through shearing and associated hydrothermal processes / alteration (Reimer, 1984).

Karkhanis (1975) also mentioned the occurrence of modified rhombohedral (ABC stacking arrangement) graphite in the Barberton Supergroup (Onverwacht Group) (Figure 5.2, location 14). The graphite occurs as flakes with an abundance of 1.5 – 4 % in chert and carbonate samples and formed due to a shearing mechanism.

Graphite is also found in the sedimentary formations of the Transvaal Supergroup, lower Pretoria Group (Machadorp area) (Figure 5.2 location 15). The graphite occurs in

123 | P a g e carbonaceous shales and metapelites and formed due to contact metamorphism along the edge of the southern Bushveld Igneous Complex, where intrusive diabase sheets graphitized the country rock (Reimer, 1984; Uken, 1998). According to Reimer (1984), small- scale mining of graphite has taken place in this area, with mining activity on the Twyfelaar 11 IT farm being well-known. On this farm, 3-4 m thick graphitic shales with 24-28 % carbon content were mined.

No future graphite mining activities in the Mpumalanga Province have been reported.

5.4.5 Free State Province In the sedimentary formations of the Witwatersrand Supergroup (Orange Free State Goldfield), graphite occurs as sheaves in auriferous conglomerates (Figure 5.2, location 16) (Cairncross and Dixon, 1995; Reimer, 1984; Schidlowski, 1967). Regional metamorphism of the carbonaceous matter is the main mechanism of formation (Schidlowski, 1967).

Graphite has also been associated with eclogite xenoliths (and, more rarely, peridotite xenoliths) from the Jagersfontein (Figure 5.2, location 17), Blaauwbosch, and Roberts Victor kimberlites (Figure 5.2, location 18) (Jacob and Jagoutz, 1994; Schulze et al., 1997; Viljoen et al., 1991; Wagner, 1916).

No past or future graphite mining activities have been reported in the Free State.

5.4.6 North West Province In the Merensky Reef of the Bushveld Complex, graphite occurs in pyroxenitic pegmatites in peculiar pothole structures (Figure 5.2, location 19). The graphitic content can reach 80 % in some areas and is associated with minerals such as amphibole, biotite, low-K phyllosilicates, chlorite, sulphides, as well as platinum-group minerals (Ballhaus, 1988; Ballhaus and Stumpfl, 1985). Under a microscope, the graphite textures range from lamellae, concentric, and globules, to a clustered “ball” of graphite needles (Ballhaus and Stumpfl, 1985). The graphite is described as being vein-type (Buntin et al., 1985; Nicholson and Mathez, 1991; Mathez et al., 1989).

Graphite occurrences were also described just above the Daspoort Horizon in the Magaliesberg shales. The graphitic content is ~10 %. Local mining took place for the production of paints, stove polish, and the colouring of slabs and tiles. Mining was, however, small-scale and due to the nature of the occurrences, economic exploration would lead to high transportation costs (Department of Mines, 1940).

124 | P a g e

5.4.7 Limpopo Province Limpopo hosts South Africa’s most important and economically viable graphite deposits, with previous mining activities reported from the 1940’s to 1980’s (Wilke, 1969).

Graphite occurs in the Proterozoic sedimentary formations of the Bushveld Complex (Figure 5.2, location 20). The graphite is found in dunite pipes (olivine-rich) and occurs as globules, films between olivine grains, and along fractures transecting Cr-Fe-Ti spinel grains. It is believed that the graphite was deposited in the fissures by magmatic fluids after the crystallisation of the dunite (Stumpfl and Rucklidge, 1982). Reimer (1984) also mentions the occurrence of graphite on the farm Maandagshoek 254 KT in this area, with graphite occurring in carbonaceous shale xenoliths in gabbro. No past or future graphite mining activities have been reported in this specific area of the Limpopo Province.

Graphite is also found in the Archaean schist belts south of the Soutpansberg Mountain range (Figure 5.2, location 21). Occurrences on the farms Stranger’s Rest 431 LT, Goedehoop 489 LS, Goedehoop 120 LT, and Nooitgedacht 489 LS have been reported (Reimer, 1984). At Goedehoop 489 LS, the graphite occurs in the chlorite schists of schist belt remnants in gneisses (Reimer, 1984).

Jonkel Carbon and Grafites Pty Ltd. are considering Goedehoop 120 LT as a prospective mining area (Khoza, 2016 (personal communication)). The occurrence is located 32 km east of Groot Spelonken and the graphite was described as lenses (3 meters thick) of amorphous / massive graphite interspersed with patches and traversed by veinlets of flake and columnar graphite (Department of Mines, 1940). This site was worked between 1910 and 1940 with a 4 to 6 tonnes / month output. The graphite was used locally in furnace linings, paint manufacturing, the colouring of slabs and tiles, foundry facing, boiler composition, greases and oils, and pipe-joint compounds (Department of Mines, 1940).

Amorphous graphite is reported in the coals of the Ecca Group (Karoo Supergroup) (Figure 5.2, location 22). This graphite was formed as a result of the intrusion of olivine dolerite through the coal seams. Between 1943-1984, more than 2500 tonnes of graphite were produced from the Mutale Mine, situated close to the town of Musina, with graphite occurring in 3-6 m thick seams, with alternating graphite-shale layers. The graphitic content ranges between 36 - 55 %, and a 40 - 50 % graphitic content product was produced via hand sorting (Wilke, 1969). It is estimated that more than 35 000 tonnes of reserve remains at this mine, and the likelihood of similar occurrences in the area (notably Pafuri) is high (Ashton et al., 2011; Wilke, 1969; Wilson, 1989).

125 | P a g e

Numerous graphite deposits can be found in the Archaean gneissic terrane of the Beit Bridge Complex, north of the Soutpansberg Mountain range (Figure 5.2, location 23). A summary of these occurrences is given in Table 5.4. The farms Dawn 71 MT, Gumbu Mine, and the neighbouring farms Steamboat 306 MS, Inkom 305 MS, and Arrie, host economically viable deposits.

Table 5.4: Graphite occurrences north of the Soutpansberg Mountain Range, Limpopo Province, South Africa (Wilke, 1969).

Description Geology Potential

Steamboat 306 MS, Inkom See text 305 MS, Arrie Albasini 524 MS Close to Mopane <2 mm flakes in Not economic viable marbles (1 %) and granite gneisses (2 %) Krige 495 MS Close to Mopane <4 mm flakes in Academic interest only Prachtig 538 MS garnet-biotite- S’Gravenhage 496 MS sillimanite gneisses, <1 Kempshall 497 MS % graphite content Dawn 71 MT See text Gumbu Mine See text Khononga Kop 5 miles east from 1) 3 mm flakes in 1) 900 tonne reserves, Gumbu Mine graphite-bearing hard ore will result in garnet-hypersthene fines forming during granulite, 10 % processing. 2) Possible graphite content. 2) 2 mining but prospecting mm flakes and powder needed in graphite gneisses and iron-bearing graphite gneisses, 5-15 % graphite content Wendy 86 MT West of Gumbu Flakes and powder in Not economic viable Bali 84 MT Mine garnet-biotite gneisses (2 %) and granitic gneisses (4 %), iron oxides present in latter Madimbo 10 miles east from Flakes in graphite Not economic viable Gumbu Mine gneisses, 5-10 % graphite content

At Dawn 71 MT, 3 mm graphite flakes occur in graphitic gneisses. The graphite content ranges between 20 and 25 %, and 29 tonnes of graphite were mined in 1942. Reserves are estimated at 13 000 tonnes, with the potential for future mining (Wilke, 1969).

Gumbu Mine (just east of Musina) has been the biggest graphite producer in South Africa, with over 9000 tonnes mined from 1942-1978 (Wilke, 1969). The remaining reserves amount to 100 000 tonnes, and there exists the potential for the re-establishment of a small-scale

126 | P a g e mine. Flakes (<4 mm) occur in the graphite gneisses, with an average graphite content of 30 % (Wilke, 1969).

Graphite occurs as large flakes (2 mm) and powder in biotite-graphite schists at the Steamboat deposit (located just south-west from Alldays) (Wilke, 1969). Jonkel Carbons and Grafites Pty Ltd. aim to commence with a small-scale mining operation here. Exploration in the late 1980’s by Mintek and the South African Development Trust at Steamboat found that it hosts 2.5 million tonnes of ore, with a total graphitic carbon content of 8.8 % (Taylor, 1991; Wilke, 1969). Tests show that a 90 % graphite grade, with 90 % graphite recovery, can be beneficiated here (Taylor, 1991). According to Feytis (2010), Steamboat will only be able to handle small volumes of production (± 100 tonnes per annum). If successful, the Steamboat graphite project can create jobs for 50 people and push an investment capital of R30 million into Limpopo Province (Trade and Investment Limpopo, 2012). There is also the potential for the neighbouring farms of Inkom and Arrie to open mines.

5.4.8 Summary of mining potential in South Africa In South Africa, natural graphite occurs in seven provinces: Limpopo, KwaZulu-Natal, Mpumalanga, Western Cape, Northern Cape, North West, and the Free State. Limpopo hosts the most economical deposits, with previous graphite mining activities reported at Dawn 71 MT (flake), Gumbu Mine (flake graphite), and Mutale Mine (amorphous graphite). There is potential for small-scale re-establishment of mining at all three deposits. Jonkel Carbons and Grafites Pty Ltd. are planning to start with small-scale flake graphite mining at the Goedehoop 120 LT and Steamboat 306 MS deposits in Limpopo. The first production was set for 2010, but logistic issues have prevented this (Feytis, 2010). Although large volumes of high-quality graphite can be found here, both deposits are situated on community land (Khoza, 2016 (personal communication)). This creates problems of an ethical and legislation nature if it is to be mined. Electricity and water are available at both deposits, but roads are not accessible at Goedehoop and should be opened and tarred before production can commence (Khoza, 2016 (personal communication)). Graphite production volumes in South Africa are not enough for exporting, and, as such, local utilisation is targeted (Feytis, 2010).

5.5 Swaziland A geological map of Swaziland is given (Figure 5.3). In Swaziland, graphite schists have been reported on the Sterkstroom farm in the Hlatikulu District (Figure 5.3, location 1) (Cairncross, 2004; Hunter, 1962). Hunter (1962) commented on the potential for mining due to the high graphitic content in this area.

127 | P a g e

Figure 5.3: Geological map of Swaziland with major graphite occurrence and associated lithologies indicated (map modified from Schlüter, 2008).

5.6 Lesotho A geological map of Lesotho is given (Figure 5.4).

Figure 5.4: Geological map of Lesotho with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008).

Boyd and Nixon (1975) and Pearson et al. (1991) reported on the occurrence of coarse graphite flakes (2–4 mm) in xenoliths occurring in kimberlites found in northern Lesotho

128 | P a g e

(Figure 5.4, location 1). These are similar to those located in the adjacent Free State Province. Fine-grained graphite intergrown with pyrrhotite and serpentine occurs at location 2 (Figure 5.4) in ilmenite-rich nodules of xenolith kimberlites (Boyd and Nixon, 1975).

5.7 Namibia In Namibia, natural graphite occurs in rocks of the !Karas, Erongo, Khomas, Hardap, and Otjozondjupa regions. A geological map of Namibia is given in Figure 5.5 and the major graphite occurrences and associated lithologies are indicated.

Figure 5.5: Geological map of Namibia with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008).

5.7.1 !Karas region In the !Karas Region, graphite occurs on the farms Aukam 104 and Harichab 121 (Figure 5.5, location 1) (Moss, 2016). These farms are located in an erosional window of the Nama Group, exposing the Middle-to-Late Proterozoic rocks of the Namaqua Metamorphic Complex (Moss 2016, Reimer, 1984). The graphite occurs as extremely large seams (up to 30 cm in width) in a body of kaolinised granite-gneiss and as specks in the kaolinised rock (Martin, 1965; Ministry of Mines and Energy, 1992). It was precipitated into fissures and fractures by a passing CO2–rich hydrothermal fluid, forming the rare vein-type graphite (Moss, 2016).

Due to this vein-type graphite, the deposit was and still is extensively exploited. Between 1940-1956 and 1964-1974 more than 22 000 tonnes of graphite were produced from the Aukam site (Ministry of Mines and Energy, 1992; Moss, 2016). The graphite ore was railed

129 | P a g e to South Africa for processing (Ministry of Mines and Energy, 1992). Currently, Gratomic Inc. (formerly CKR Carbon Corporation), Next Graphite Inc., and Gazania 242 Pty Ltd. are in a joint venture to produce graphite from this site. In April 2018, the first concentrate since 1974 was produced (5 tonnes of 88-95 % graphite) (GlobeNewswire, 2018; Moss, 2016), and currently the processing plant is operating at a capacity of 650 tonnes per annum (GlobeNewswire, 2018). Upgrades on the processing plant are in place, which will increase the capacity to an estimated 11 000 tonnes per annum. This production is threefold that of Sri-Lanka, the only other notable vein graphite producer globally (GlobeNewswire, 2018; U.S. Geological Survey, 2019).

5.7.2 Erongo region In the Erongo region, graphite occurs predominantly in marbles of the Swakop Group (Figure 5.5, location 2).

Lehtonen et al. (1996) described the occurrence of 1 - 2 mm graphite flakes in the Karibib Formation (Arises River Member) of the Swakop Group, where the calcite marble contains 5% graphite, giving it a speckled appearance. Gross et al. (2000) and Gross (2006) describe the graphite as being 3 - 4 mm thin flakes occurring along calcite grain contacts, “nests” of small clusters, or as occurring sporadically in sparry calcite. Optical observations of the graphite show “growth spirals” and “hillocks”, which indicates that it formed from a fluid phase at low graphite supersaturation (unrestricted by surrounding calcite) (Rakovan and Jaszczak, 2002a, b; Walter, 2004). Badenhorst (1987), Liu et al. (2002), and Swart (1986) also confirmed the presence of graphite in the Karibib Formation, and Nörtemann et al. (2000) mentioned similar graphite crystals and formation mechanisms in skarns found here.

In the Rössing Formation of the Swakop Group, Lehtonen et al. (1996) and Berning et al. (1976) reported on disseminated graphite in quartzites and marbles. Steven (1987) reported on disseminated graphite flakes in calcitic marble near the Otjua tungsten deposit, <0.5 mm in size and makeup between 0 - 2 % of the host marble.

Finely disseminated graphite has also been found in the dolomitic marbles of the Abbabis Inlier of the Swakop Group (Brandt, 1987; Reimer, 1984; Smith, 1965). In the coarser- grained marbles of this area, the graphite occurs as flakes (Reimer, 1984) and its content can be as high as 10 %, however, grades are generally not uniform (Smith, 1965).

Although no past mining has been reported in the Erongo region, Argosy Minerals Ltd. is currently acquiring mining rights for their Erongo Graphite Project (Area 51) (275 km north- west of Windhoek) (Figure 5.5, location 3) (Argosy Minerals Ltd., 2019). Gecko Namibia Ltd is also currently acquiring rights to mine at Black Range 72 (Figure 5.5, location 4) (40 km

130 | P a g e west of Usakos), where 13.75 million tonnes of 4.52 % flake graphite are hosted in mica- bearing rocks (Gecko, 2018). The Ministry of Mines and Energy (1992) report that due to the association with mica, a mere 60 % rate of processing recovery can be achieved at Black Range and the product will mostly be in powder form due to excessive milling needed for liberation.

5.7.3 Khomas region Kukla (1988; 1992) reviewed graphite schist occurrences on the farm Kaan 309 (Figure 5.5, location 5), and reported a high graphitic content schist unit with a 60 m thickness.

Graphite schists also overlay the marble rocks surrounding the Rietfontein inlier of the Swakop Group, south-east of Windhoek (Figure 5.5, location 6) (Gevers, 1934; Reimer, 1984).

The Ministry of Mines and Energy (1992) reported on high-quality graphite schists with several lenses and bands alternating with sericitic and amphibolitic schists on the farms Lichtenstein 366 and Melrose 368, and high-quality graphite schists with 46 – 37 % graphite content, on Portsmut 664 farm. Unfortunately, in the latter case, the deposit includes mica, quartz, and clay, which make beneficiation difficult.

No past or future graphite mining activities have been reported in the Khomas region.

5.7.4 Hardap region In the Hardap region (Figure 5.5, location 7), graphitic schist deposits occur close to the farm Aroams 315, located near the Hakos Mountains and the ore grades here range between 37 and 46 % graphite. The graphite can be described as a good quality flake-type. The graphite is associated with mica, quartz, and clay, all of which make the mining and processing challenging (Ministry of Mines and Energy, 1992).

5.7.5 Otjozondjupa region Graphite occurs on the farms Okanjande 145, Good Hope 298, and Highlands 311, near the town of Otjiwarongo in the Otjozondjupa region (Figure 5.5, location 8) (Ministry of Mines and Energy, 1992). The graphite occurs as well crystallised small flakes and lenticular bodies within gneissic and schistose rocks (Cairncross, 2004; Gevers, 1934; Ministry of Mines and Energy, 1992). According to Frimmel and Miller (2009), the graphite metamorphosed from crude oil in sapropelic shales. The graphite content ranges from 5 - 6 %, has an average flake size of 6 mm, and the deposit has a measured mineral resource of ~12.83 million tonnes (Gecko, 2018; Ministry of Mines and Energy, 1992). Gecko Namibia Ltd. started production at Okanjande in early 2017, with a production rate of 20 000 tonnes per annum, to be expanded to produce 30 000 tonnes per annum (Gecko, 2018).

131 | P a g e

5.7.6 Summary of mining potential in Namibia Natural graphite occurs in the !Karas, Erongo, Khomas, Otjozondjupa, and Hardap regions of Namibia. Various mines have recently commenced in these regions, and the potential for expansion exists. In the !Karas Region, the scarce vein-type graphite is mined at the historical Aukam site (2018 start-up). Moss (2016) reported that there is good infrastructure in this area for mining, with sufficient electricity and water available, as well as a nearby rail link. In the Erongo region, flake graphite can be found within calcite marbles. Argosy Minerals Limited is in the process of acquiring mining rights for a project in this region. Although no mining has been reported in the Khomas region thus far, a 60 m thick graphite schist occurs on the farm Kaan 309 that might hold some potential for future projects. In 2017, Gecko Namibia Ltd commenced with flake production on farm Okanjande which is located in the Otjozondjupa region. The company is also considering mining on farm Black Range, located within the Erongo district. The major problem associated with mining in the Otjozondjupa Region, and in Namibia in general, is the lack of sufficient groundwater, while at the Okanjande mine, groundwater is currently obtained from a site outside of the project area (Sarma and Hattle, 2014).

5.8 Botswana Graphite occurs in the Central and Southern Districts of Botswana. A detailed geological map of Botswana is given in Figure 5.6 and the major graphite occurrences and associated lithologies are indicated.

5.8.1 Central District In the Central District graphite occurs in the Archaean schist belt of the Bushman Shear Zone (Figure 5.6, location 1) in mantle-derived xenoliths occurring in the kimberlites of the Orapa diamond mine (Figure 5.6, location 2); and in gneisses close to Pencil Hill (Figure 5.6, location 3).

132 | P a g e

Figure 5.6: Geological map of Botswana with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008).

The Bushman Shear Zone (Figure 5.6, location 1) occurs in eastern Botswana, between Mosetse and Dukwe, and forms a part of the Basement Complex (Massey, 1973). The Bushman Shear Zone formed part of prehistoric copper mining in Botswana and graphite was mined as a by-product (Van Waarden, 2014). Graphite occurs as schists and phyllites here and is closely associated with sheared dolomitic limestone (Cairncross, 2004; Massey, 1973). The graphite occurs where the limestone makes contact with the surrounding undifferentiated granite gneiss (Massey, 1973). Barton et al. (1994) observed that the graphite-bearing rocks vary from quartz-chlorite phyllites with limited graphite to a gray graphitic phyllite, and finally to black graphitic schist. The graphite schist zones can be as thick as 2.5 m and the graphite quality is comparable to that of commercial sources. Graphite has also been reported in spoil from a well, where the schist belt crosses the Mosetse River (Massey, 1973).

At location 2 (the Orapa kimberlite) (Figure 5.6), graphite occurs as part of an eclogite- dominated xenolith suite and these rocks are similar to those found in the South African provinces of the Northern Cape and Free State (Field et al., 2008; Robinson, 1977; Robinson et al., 1984; Schulze et al., 1997).

At location 3 (Figure 5.6), Bisan Ltd is currently investigating the mining potential of graphite near to Pencil Hill. Initial tests have indicated a flake-type graphite and a graphitic content of up to 22 % in ore (Pencil Hill, 2015).

133 | P a g e

5.8.2 Southern District Pyrite-rich graphite zones can be found in sheared dolomites located near the town of Moshaneng (Figure 5.6, location 4) (Cairncross, 2004; Massey, 1973; Reimer, 1984).

5.8.3 Summary of mining potential in Botswana Natural graphite deposits occur in the Central and Southern Districts of Botswana. The economic potential of graphite found in the Southern District is low. The Central District, however, hosts two economically-favourable deposits. These are located at Pencil Hill and in the Bushman Shear Zone. Bisan Ltd. is currently investigating the mining potential of graphite at Pencil Hill, while Massey (1973) reported on finding good quality graphite in the Bushman Shear Zone, with potential for mining.

5.9 Zimbabwe In Zimbabwe, graphite occurs in the Provinces of Matabeleland North, Mashonaland West, Midlands, Harare, Manicaland, Masvingo, and Bulawayo. In the Mashonaland West Province, the Lynx Graphite Mine is currently producing and exporting flake graphite (Ministry of Mines and Mining Development, 2018). A detailed geological map of Zimbabwe, with the graphite locations and associated lithologies indicated, is provided in Figure 5.7.

Figure 5.7: Geological map of Zimbabwe with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008).

134 | P a g e

5.9.1 Matabeleland North Province In the Matabeleland North Province, graphite is located in the Dete-Kamativi Inlier, specifically in the Inyantue (Figure 5.7, location 1) and Malaputese (Figure 5.7, location 2) Formations.

In the Inyantue Formation, graphite occurs intercalated in garnetiferous gneisses and schists (Master et al., 2010; Master, 2013). The graphite here ranges from single bands to three or more, ranging from 1 to 20 m in thickness, associated with coarse-grained diopsidic rocks and biotite-hypersthene granulites. The common understanding is that the graphite-diopside bands formed from carbonaceous shales containing lenticular beds of limestone (Master, 2013).

Muchemwa (1987) listed Black Diamond, Downright, Upright, Forthright, Graf, Lloyd, Ncho, Ward, and Globe as graphite claims in the Inyantue Formation. According to this author only the Black Diamond and Globe deposits contain high-grade flake graphite that has the potential to be mined: at the Black Diamond Claims, the graphite occurs as schists with pegmatitic veins, and in small knots or lumps in a siliceous magnesian limestone. The ore here has an average graphite content of 33 %; while at the Globe Claims the graphite content in the ore is 50 %, and flakes of up to 1 mm can be found.

In the Malaputese Formation, graphite occurs as schists intercalated in metapelites, regarded as a metamorphosed black shale (Master et al., 2010; Master, 2013). Here, the graphite is classified as low-grade, and extraction is uneconomical due to the large quantities of associated quartz, mica, and iron oxides (Muchemwa, 1987).

5.9.2 Mashonaland West Province In the Mashonaland West Province, graphite occurs as part of the Early Proterozoic sedimentary formations of the Lomagundi (Figure 5.7, location 3) and Piriwiri Groups (Figure 5.7, location 4).

Master (1996), Master et al. (2010), and Tennick and Phaup (1976) discussed the occurrence of graphite shales in the argillaceous sediments of the Nyagari Formation of the Lomagundi Group. Freiberg (1907) also briefly mentions the association of graphite with gold in the Ayrshire Mine located in the Lomagundi Group.

Master et al. (2010) reported graphite occurrences in the Umfuli and Chenjiri Formations of the Piriwiri Group. The graphitic slates also contain thin layers of black cherty maganiferous quartzite (Master, 1996; Master et al., 2010; Reimer, 1984). The Piriwiri Group hosts the Lynx Mine, currently Zimbabwe’s only producing and exporting graphite mine (Figure 5.7, location 5) (Ministry of Mines and Mining Development, 2018).

135 | P a g e

The Lynx Mine is situated 35 km north-west of Karoi and has been operational since 1965. As from late 2017, the mine is solely owned by the Zimbabwe Mining Development Corporation (Mzamo, 2018). Four graphite ore bodies, namely: Amalie, Barbara, Maria, and Glück Aüf, occur at the Lynx mine and are composed of flake graphite (25 %), quartz (50 %), sillimanite (15 %), plagioclase (minor), garnet (minor), and biotite (minor) (Muchemwa, 1987). The graphite at the Lynx Mine consists of 40 % large flakes and 60 % small flakes, both needing to be upgraded to 80 - 90 % before exporting. Processing takes place through crushing, milling, classification, flotation, centrifuging, drying, screening, and bagging operations, and the final product consists of flakes between 1 and 500 µm (Muchemwa, 1987).

Other graphite mines and claims in the Mashonaland West Province have also been reported (Table 5.5). Previously mining occurred at Graphite King, Juma Claims, Silaka Kaswaya Mine, Zororo Claims, and Lucky Day Mine (Muchemwa, 1987). There is the potential to re-open mining activities at Graphite King, as well as to start production at the Madadza Claims.

Graphite King is situated just above the Lynx Mine and produced small volumes of graphite between 1944 and 1945. Reserves here are thought to be fairly large, but no assessment of readily available tonnages was made. The graphite is finely associated with mica, making milling and liberation practices difficult. However, beneficiation tests have shown that a 74 % graphite concentrate can be reached if the correct techniques are used (Muchemwa, 1987).

The Madadza Graphite Claims are located 6 km south-west of the Lynx Mine and are divided into eastern and western sections. Graphite in the western section occurs as flake- type in sillimanite schist, underlain by biotite gneiss country rock. On the eastern side, graphite is intimately associated with pegmatite in biotite gneisses. Previous exploration work on these claims concluded that the small flake size, fine silica associations, low recovery, and hardness of the seams would prevent mining. The graphite deposit on the eastern side is known to be very shallow and only extends 200 m in length (Muchemwa, 1987).

5.9.3 Midlands Province In the Midlands Province of Zimbabwe, graphite occurs in the Chirumanzu, Lalapanzi, and Lower Gweru regions (Figure 5.7, location 6). At Chirumanzu, low-grade flake graphite of little commercial value occurs. In the Lalapanzi region, graphite schists at the Rydal Claims are also considered uneconomical due to association with quartz, feldspar, mica, and iron oxides. In the Lower Gweru region, graphite seams are thin (<1 mm), and graphite mining is not plausible (Muchemwa, 1987).

136 | P a g e

Table 5.5: Summary of graphite occurrences in the Mashonaland West Province, Zimbabwe (Muchemwa, 1987).

Mine/claim Description Geology Potential

Catkin Graphite Claims Four small deposits, situated 15 km Metamorphosed graphitic shale Average carbon content 30 %, but deposits are sout-west of the Lynx mine horizons too small and lenticular for mining Chidziro Graphite Claims Close to Karoi, situated on a privately Graphite associated with Lenticular nature, high ferruginous percentages, owned farm phyllites and garnetiferous and mica presence make mining impractical gneiss Glen Isla Graphite Claims Close to the Gwira Hill, eastern margin Lenses of metamorphosed Carbon content of 35 %, but deposits are too of the Urungwe Klippe Piriwiri graphitic shales in small with low flake quality granitic gneisses Good Willie Mine - - - Graphite King Mine See text Kadziru Deposit Close to Karoi, next to the Kadziru Graphite occurs in contorted Average carbon content 33%, development was River quartz-mica schists considered but later abandoned Lucky Day Mine Two Tree Hill Extension No. 1 farm Graphitic schists in the Piriwiri 234 tonnes of graphite during the period 1972 to Group 1973 Lynx Graphite Mine See text Madadza Graphite Claims See text Manyangau Group Six claims registered in 1959 and an - In 1982, 5 tonnes of high-grade flake graphite additional five later registered. The were mined at Zororo. Due to thin graphite Zororo claims are the most promising horizons and a remote locality, mining was and are situated on the Umpara River abandoned Mwami Group The Juma Claims are the most Graphite schists of the In 1982, 5 tonnes graphite was mined and 5 promising in this group, located close to Lomagundi Group tonnes stockpiled at Juma Mwami Silaka Kaswaya Mine Close to the confluence of the Rukuwe - 37 tonnes mined during 1981 - 1982 and Nyaodza rivers Swempi Graphite Claims Close to the Mwami River Amorphous graphite in pelitic Exploitation unlikely due to amorphous nature rocks Trokiaza Claims Close to Lynx Mine, one block of 25 Graphitic schists Presence of mica and hardness of graphite claims schists make mining impractical

137 | P a g e

5.9.4 Harare Province In the Harare Province, a graphite deposit within kaolinised granite is located at a site 71 km east-north-east of Harare city (Figure 5.7, location 7). The graphite is the amorphous type with a 30 % graphite content. However, the clay content in the ore is very high, which will cause problems during mining (Muchemwa, 1987).

5.9.5 Manicaland Province In the Odzi region of the Manicaland Province (Figure 5.7, location 8), a high-grade graphite deposit occurs (Muchemwa, 1987).

5.9.6 Masvingo Province In the Masvingo Province (Figure 5.7, location 9), flake graphite occurs in graphitic quartzite, but the grade is low (55.66 % after treatment) and therefore it is deemed to be uneconomical (Muchemwa, 1987).

5.9.7 Bulawayo Province Small quantities of graphite were noted near Bulawayo (Figure 5.7, location 10), southern Zimbabwe, which is found in limestones of the Bembesi gold belt (Macgregor, 1940).

5.9.8 Summary of mining potential in Zimbabwe Various natural graphite deposits occur in Zimbabwe. The majority of economically-viable deposits are situated in the Provinces of Matabeleland and Mashonaland West. However, due to the ongoing political instability and current economic crises in Zimbabwe, the Lynx Graphite Mine (in the Mashonaland West Province) is currently the only operating mine in the country. In 2009, the Lynx mine had to reduce its production drastically due to the combined effect of the world-wide recession, as well as political instability within the country (Feytis, 2010). This mine has been producing graphite since 1965 but is currently only operating at 75-80 % capacity (Nkala, 2016). According to the U.S. Geological Survey (2019), Zimbabwe’s graphite production in 2018 was 2000 tonnes, which is 0.22 % of the total world production. This is significantly less than what was mined in the years from 1992- 1998. During these years, an average of 10 000 tonnes per annum was mined (Kalyoncu, 1996; 1998). In 1983, a record 20 000 tonnes of graphite were produced (Krauss et al., 1988). The Lynx Mine recently made a plea to possible investors for equipment upgrades (Nkala, 2016, Steenkamp, 2017).

5.10 Mozambique Graphite deposits can be found in seven of Mozambique’s ten provinces: Tete, Manica, Sofala, Zambezia, Nampula, Niassa, and Cabo Delgado (Schlüter, 2008). Of these, Cabo Delgado is the most promising, with a variety of mining projects in the pipeline (U.S.

138 | P a g e

Geological Survey, 2019). A detailed geological map of Mozambique, with the graphite locations and associated lithologies indicated, is provided in Figure 5.8.

Figure 5.8: Geological map of Mozambique with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008).

5.10.1 Tete Province The Tete Province is divided into a northern and southern region by the Sanangoe Thrust Zone. On the northern side, graphite-bearing marbles are associated with the Angonia (Figure 5.8, location 1), Luia (Figure 5.8, location 2), and Zambue (Figure 5.8, location 3) Groups (Evans et al., 1999). Graphite has previously been mined in the Angonia Group (Jourdan, 1990). In the Angonia Group, anorthite is extensively kaolinised and contains veins and stringers of graphite (Barr and Brown, 1987). Voortman and Spiers (1986) mention the association of graphite with anorthite, as well as with biotite, in this Group.

On the southern side of the Sanangoe Thrust Zone, graphitic calc-silicate gneisses occur in the Barue (Figure 5.8, location 4) and Chidue Groups (Figure 5.8, location 5) (Evans et al., 1999).

5.10.2 Manica Province Manhica (1998) and Pekkala et al. (2008) mention the occurrence of graphite-bearing phyllites and schists in the Vengo sequence of the M’Beza / Vengo Formation of the Manhica greenstone belt (Figure 5.8, location 6). Pekkala et al. (2008) also state that graphite is to be found at location 7 (Figure 5.8).

139 | P a g e

5.10.3 Sofala Province Schlüter (2008) and Pekkala et al. (2008) mention the occurrence of graphite in the Sofala Province (Figure 5.8, location 8).

5.10.4 Zambezia Province Schlüter (2008) and Pekkala et al. (2008) mention the occurrence of graphite in the Sofala Province (Figure 5.8, location 9).

5.10.5 Nampula Province In Nampula Province, graphitic gneisses occur in the marbles (carbonates) of the Evate rocks within the Monapo Klippe (Figure 5.8, location 10) (Gajdošová et al., 2017; Hurai and Huraiová, 2015; Manhica, 1991). De Klerk (2015) describes the graphite as being flaky, with ore grades ranging between 11 % and 55 %. It is estimated that there are approximately 4 million tonnes of ore here (De Klerk, 2015). The deposit was discovered by a Russian team, and has previously been mined (on-and-off from 1911 to 1950) (Jourdan, 1990; Manhica, 1991; Rocha et al., 2017).

On a microscopic level, Hurai et al. (2017) observed the inclusion of this graphite in massive sulphidic samples as either: i) randomly-oriented crystals and ribbons in pyrite; ii) isolated spherules in carbonates, or along carbonate-pyrite grain boundaries; iii) aggregates of deformed graphite crystals; iv) molybdenite-graphite-zeolite along carbonate-pyrite boundaries; or iv) graphite-molybdenite-chalcopyrite-carbonate-zeolite veinlets and stringers in pyrite.

5.10.6 Niassa Province An estimated 5 million tonnes of 50 % grade graphite ore occur at Nipepe, in the Niassa Province (Figure 5.8, location 11) and the graphite deposit could be exploited for over 27 years at a rate of 80 000 tonnes mined per annum (Mandlate, 2018). Boyd et al. (2010) also report on finding graphite in the paragneisses of the Unango Complex.

5.10.7 Cabo Delgado Province The largest untapped deposits of graphite in the world are located in the Cabo Delgado Province, with an estimated reserve of 17 million tonnes of graphite (Figure 5.8, locations 12 and 13) (U.S. Geological Survey, 2019). Although this deposit has only attracted attention recently, a variety of exploration projects are underway, potentially enabling Mozambique to overtake China as the biggest graphite producer in the world (Du Venage, 2014). In general, the graphite occurs in graphite-bearing mica schists and gneisses and is described as coarsely grained with graphite abundances ranging between 5 % and 20 %. The graphite- bearing horizons extend up to several kilometres, and their thickness ranges from 10 m to

140 | P a g e

100 m (Bingen et al., 2007). According to Boyd et al. (2010) and De Klerk (2015), the graphite deposits occur in the Xixano, Lalamo, and Montepuez Complexes, as well as in the Lurio Belt of the Cabo Delgado Province.

In the Xixano Complex (Figure 5.8, location 12), graphite schists and gneisses occur, with carbon contents ranging between 9 % and 21 % (Bingen et al., 2006; Boyd et al., 2010). Syrah Resources Balama West and East prospects, and Triton Minerals Balama North and South prospects are located here (Syrah Resources, 2018; Triton Minerals, 2018).

In the Balama West region, graphitic schist surrounds a core of graphitic psammite (Syrah Resources, 2018). The estimated 170 million tonnes of ore reserve, with an average flake- graphite grade of 19 % graphite and 0.43 % vanadium pentoxide, place the Balama West deposit as one of the largest high-grade graphite deposits globally (Chafy et al., 2018). Syrah Resources commenced with production in 2017 (U.S. Geological Survey, 2018). In Balama East, coarse, high-grade flake graphite occurs. This is an excellent source of jumbo graphite. Syrah Resources’ combined prospects in Balama East and West are set to last 42 years, with a plant feed rate of 2 million tonnes per annum, an average grade of 16 % graphite, an average recovery of 93 %, and an average graphite production rate of 313 000 tonnes per annum (Syrah Resources, 2018).

Limited exploration activities by Triton Minerals have commenced on Balama South, but activities on Balama North are well documented (Triton Minerals, 2018). In the Balama North prospect, two projects are currently operating, namely Cobra Plains and Nicanda Hill. Both sites are said to contain large, high-grade flake graphite, and are associated with vanadium. At Cobra Plains, there are 103 million tonnes of inferred ore with an average graphite content of 5.2 %. At Nicanda Hills (including Black Hills and Charmers), there is 730 – 1200 million tonnes of inferred ore with an average graphite content of 5 – 6 % (Triton Minerals, 2018).

In the Montepuez Complex (Figure 5.8, location 13), gneiss with 11 - 13 % graphite content occurs (Boyd et al., 2010). In 2018, Battery Minerals commenced with feasibility studies for the application of possible graphite mines in the Balama Central and Montepeuz areas (Battery Minerals, 2018).

In the Lurio Belt (Figure 5.8, location 13), over 30 graphite occurrences are recorded in gneisses interbedded with quartzite, leptite, and marble (De Klerk, 2015).

In the Lalamo Complex, Ancuabe region (Figure 5.8, location 13), graphite flakes and fine grains are concentrated within biotitic migmatitic gneisses (De Klerk, 2015). Previous mining at Ancuabe took place from 1994 to 1999. Triton Minerals are currently considering

141 | P a g e producing from deposits surrounding this historic mine (Triton Minerals, 2018). Reserves at Ancuabe are estimated to be at 35.5 million tonnes, with 1 million tonnes remaining at the historic mine (De Klerk, 2015; Jourdan, 1990; Triton Minerals, 2018). The graphitic grade at Ancuabe ranges between 15 % and 20 % (Jourdan, 1990), although Boyd et al. (2010) also detail finding samples with a 26 % graphite content in this area. The graphite flakes at Ancuabe are described as “jumbo” and “super-jumbo”, with >92 % of flakes >150 µm and >84 % of flakes > 212 µm (Triton Minerals, 2018).

5.10.8 Summary of mining potential in Mozambique Mozambique can be considered as the graphite hotspot of southern Africa and even globally. The Cabo Delgado Province hosts an estimated 17 million tonnes of graphite, with several projects currently being launched in the province (U.S. Geological Survey, 2019). The project at Balama West commenced its mining operations in the summer of 2017 and already contributed ~2 % to world production for 2018. This is more than the combined production achieved at Africa’s other producing countries (Madagascar, Namibia, and Zimbabwe) (U.S. Geological Survey, 2019). Resources at Balama West are estimated to be more than the rest of the world’s annual combined graphite production, and it is considered as one of the most promising graphite projects to-date (Du Venage, 2014). The project at the historical Ancuabe Mine (future project by Triton Minerals) is located ±60 km from the port of Pemba, making it easily accessible for exporting purposes. The Ancuabe deposit is valuable due to the large, jumbo, and super-jumbo flakes that can be found near the outer surface of the deposit (Triton Minerals, 2018). These large particle sizes are attracting various investors, and Triton Minerals claim this to be one of the best sources of graphite globally (Triton Minerals, 2018).

5.11 Madagascar Madagascar, together with Namibia, Mozambique, and Zimbabwe is one of four countries in southern Africa that are at present actively producing and exporting graphite; and Madagascar is especially known for its high-quality flake graphite. The graphite occurs in the Graphite Sequence, also known as the Système du Graphite (Figure 5.9) (Berglund and Touret, 1976; Cameron and Weis, 1960).

142 | P a g e

Figure 5.9: Geological map of Madagascar with major graphite occurrences and associated lithologies indicated (map modified from Schlüter, 2008).

The Graphite Sequence consists of four groups, namely: The Ampanihy (south); Ambatolampy (centre); Andriba (north-west); and Manampotsy (east) Groups (Randrianasolo, 1996).

In the Ampanihy Group, graphite occurs as graphite-bearing hornblende-biotite gneisses in the 20 km wide area known as the Ampanihy shear belt (Figure 5.9, location 1) (Dissanayake and Chandrajith, 1999; Windley et al., 1994). The Molo graphite project, run by NextSource Materials Inc. and scheduled to begin operations in the last quarter of 2019, is situated in this belt (NextSources Materials Inc., 2017). The area is believed to contain 141.28 million tonnes of ore, with a 6.13 % flake graphite content (NextSources Materials Inc., 2017). Janardhan (1999) discussed graphite gneisses occurring in the anorthosites of the Ampanihy shear belt (Figure 5.9, location 2). De Wit et al. (2001) and Parthasarathy et al. (2006) also mention graphite occurrences in psammitic and pelitic paragneisses, located toward the eastern end of the Amphanihy Group (Figure 5.9, location 3). The graphite occurs as highly crystalline flakes, believed to have formed at metamorphic temperatures ranging between 750 °C and 850 °C (Parthasarathy et al., 2006). Graphite occurrences were also noted in the Androyan (Tucker et al., 2011) and Vohibory (Bingen et al., 2006) Sequences. These are, respectively, located just north and west of the Ampanihy Group.

In the Ambatolampy Group (Figure 5.9, location 4), graphite occurs as discontinuous layers associated with mica schists, and as graphite-sillimanite gneisses (Brenon, 1972; Chantraine and Radelli, 1970; Randrianasolo, 1996). Mining in this group occurred between 1910 and

143 | P a g e

1952 in the towns of Antsiriribe, Andavabato, and Andravoravo (Chantraine and Radelli, 1970).

The Manampotsy Group is considered another hotspot for graphite production. According to Chantraine and Radelli (1970), mining previously occurred in the upper 30-to-50 m of the Group, hosting 5 – 10 % graphite. The ore in Manampotsy consists of biotite and green hornblende gneisses, as well as graphite gneisses (Brenon, 1972; Müller, 2000). Bass Metals has several graphite mines in the Manampotsy Group, located in the Toamasina area (Figure 5.9, location 5). The most notable of these is the Graphmada Mine, which Bass Metals acquired from Stratmin Global Resources in 2016 (Padhy, 2017). Revamping of the mine is currently underway, and it is hoped that by 2019 it will be producing 20 000 tonnes of high-quality flake graphite per annum (Padhy, 2017). Currently, 500 tonnes are produced per month from a 4.5 % graphite flake ore. Other Bass Metals mines in this area include Andapa, Mahela, Mahefedok, Ambatofafana, and Loharano (Bass Metals, 2018).

In the Andriba Group (Figure 5.9, location 6), graphite is scarce, especially in the uppermost part (Brenon, 1972; Nédélec et al., 2000), but graphite gneisses have been reported (Müller, 2000).

5.11.1 Summary of mining potential in Madagascar Graphite is one of Madagascar’s main export commodities. According to the U.S. Geological Survey (2019), graphite production in 2018 was 9000 tonnes, amounting to 0.97 % of global production. In comparison, the graphite production of Madagascar was 5000 tonnes in 2014 and 2015 and increased to 8000 tonnes in 2016 (U.S. Geological Survey, 2015; U.S. Geological Survey, 2016; U.S. Geological Survey, 2018). At present, the Ampanihy and Manampotsy Groups are being mined / considered for mining. In the former, the Molo graphite project from NextSource Materials Inc. is notable. This deposit contains 141.28 million tonnes of ore, with a 6.13 % flake graphite content and production will start by the end of 2019 (NextSources Materials Inc., 2017). In the Manampotsy Group, Bass Metals is mining near the port town of Toamasina. Their most notable mine is Graphmada, which is currently under revamping and expansion. This mine will be able to produce 20 000 tonnes per annum of high-quality graphite by the end of 2019. Some advantages of mining in Madagascar include the high-grade of graphite flakes found here, the low cost exploration, and the low-cost of mining and processing. Some difficulties include political and investment risks, limited infrastructure, remote locations, and a shortage of skilled labour.

144 | P a g e

5.12 Characterisation of natural graphite from two localities in southern Africa One of the objectives of the ERA-MIN project was to produce synthetic graphite from char in ash. To understand if the graphitization trials successfully produced synthetic graphite – Charphite – an appreciation of natural graphite is required.

5.12.1 Methodology Two samples were obtained from current exploration deposits. The samples were supplied by Jonkel Carbons and Grafites (Pty) Ltd. The first sample originates from the Goedehoop 120 LT deposit situated in the Soutpansberg mountain range in the Limpopo Province of South Africa. The sample was ground and pre-concentrated (~30 wt. % carbon) via flotation by Jonkel Carbons and Grafites (Pty) Ltd before purchase. The second sample was obtained from a Rantwood Enterprise deposit located in Karoi Zimbabwe close to the current operating Lynx Graphite Mine. The sample was ground, but not concentrated / purified before purchase. Nomenclature and other sample information are provided (Table 5.6).

Table 5.6: Natural graphite sample nomenclature, descriptions, and other information.

Sample ID Sample description Mass obtained (kg)

SA Graphite Goedehoop Deposit, Limpopo South Africa 10.0

Zim Graphite Rantwood Enterprise Deposit, Karoi Zimbabwe 10.0

The samples were divided into smaller, representative fractions using the cone and quartering technique as a first means, followed by rotary splitting. A representative sample from each parent was characterised in detail.

The characterisation techniques are given in Table 5.7. Proximate, XRF and carbon form analyses were outsourced to Bureau Veritas Testing and Inspections South Africa. The standards and methods used were provided in Sections 2.5.1 (proximate and XRF) and 3.4.2 (carbon form).

A PANalytical X’Pert Pro diffractometer with Cu radiation (housed at UJ Spectrau) was used for the XRD mineralogy analysis. Samples were prepared with the backloading procedure. X’Pert Highscore plus software was used for mineral identification. Only the minerals were identified, and not quantified.

For XRD structural analysis it was first necessary to demineralise the samples, to ensure that minerals did not interfere and alter the carbon diffractogram. Due to financial and sample quality (low graphite content in Zim graphite) constraints, only the SA Graphite

145 | P a g e sample was demineralised and analysed. The final carbon percentage was 90.38 wt. % (LOI basis). The demineralisation and XRD analysis methods were given in Section 3.4.2.

Raman microspectroscopy sample preparation and image acquisition were provided in Section 3.4.2. Due to the absence of D3 and D4 bands, it was decided to fit only the G, D1, and D2 bands. An example of peak fitting done on the graphite is given in Appendix B.

Petrography was used for image acquisition (morphology and flake size) and reflectance measurements. Block preparation and instrument specifications were discussed in Section 2.5.1. Reflectance measurement techniques and calculations were given in Section 3.4.2.

Table 5.7: Natural graphite characterisation techniques used in the current study.

Analysis SA Graphite Zim Graphite

Proximate × ×

X-ray fluorescence × ×

Carbon form × ×

X-ray diffraction (mineralogy) × ×

X-ray diffraction (structural) ×

Raman microspectroscopy × ×

Petrographic × ×

5.12.2 Natural graphite characterisation results The proximate, XRF, and carbon form results are provided (Table 5.8). The standards used for proximate analysis were developed for coal samples; it may be that for the graphite samples, the combustion temperature is not sufficient due to the presence of strong carbon- carbon bonds. The fixed carbon may then be an underestimation of the true carbon content, and therefore explain the differences in total and fixed carbon percentages. Also, take note that the graphite samples are geological samples without any processing, and therefore, the low fixed carbon and high ash yield percentages.

The major elements are Al2O3, Fe2O3, SiO2, and K2O. The inclusion of potassium might be due to the mica and feldspar that are commonly associated with graphite (Mitchell, 1993). SA Graphite has a higher CaO percentage than Zim Graphite, while Zim Graphite has a higher Fe2O3 percentage.

Elemental carbon is dominant in both samples. Seeing that elemental carbon represents strong carbon-carbon bonds that are typically associated with graphite, this result makes

146 | P a g e sense. Organic carbon can be described as “volatile” carbon, and in graphite, it is mainly ascribed to contaminants from the environment (Schumacher, 2002). From the proximate results, it can be seen that the volatile matter contents for the two samples are low, resulting in the respective organic carbon contents also being low. The inorganic carbon contents are relatively high for the two samples and are attributed to the presence of calcite.

Table 5.8: Proximate, XRF, and carbon form analyses results for the natural graphite samples.

SA Graphite Zim Graphite

Proximate Moisture (a.d.b wt. %) 0.7 0.5 Ash yield (a.d.b wt. %) 72.6 93.6 Volatile matter (a.d.b wt. %) 4.0 4.4 Fixed carbon (a.d.b wt. %) 22.7 1.5

Volatile matter (d.a.f wt. %) 15.0 74.6 Fixed carbon (d.a.f wt. %) 85.0 25.4

X-ray fluorescence (wt. %)

SiO2 72.96 67.54

Al2O3 14.21 20.59 CaO 2.97 0.08

Fe2O3 2.70 5.37

K2O 3.67 4.06

TiO2 1.19 0.71

Cr2O3 0.09 0.06 MgO 0.81 1.04 MnO 0.02 0.04

Na2O 0.00 0.34

P2O5 0.07 0.05

V2O5 0.05 0.02

ZrO2 0.04 0.03 BaO 0.07 0.10 SrO 0.03 0.01 ZnO 0.00 0.01

SO3 0.51 0.01

Carbon form Total carbon (wt. %) 38.94 3.38 Elemental carbon (wt. %) 33.27 2.67 Organic carbon (wt. %) 0.88 0.10 Inorganic carbon (wt. %) 4.79 0.61

Elemental carbon (% of total carbon) 85.44 79.00 Organic carbon (% of total carbon) 2.26 2.96 Inorganic carbon (% of total carbon) 12.30 18.05

The XRD mineralogy results are provided (Table 5.9). Results show that graphite, quartz, and calcite were the major minerals identified in the SA Graphite sample. For the Zim

147 | P a g e

Graphite sample, the major minerals were identified as graphite, quartz, an amorphous iron phase (Figure 5.10), and feldspar. The presence of calcite in the former, and absence in the latter, is also evident in the XRF results. The higher Fe2O3 percentage for the Zim Graphite explains the inclusion of the amorphous iron phase.

A B

100 µm 100 µm

C D

100 µm 100 µm

Figure 5.10: A collection of minerals found in the Zim Graphite sample: A) and B) Intense reddening can be ascribed to amorphous Fe-oxide (Geuna et al., 2015); C) Graphite embedded in an amorphous Fe-oxide phase; D) Honeycomb mineral (left), graphite (middle), and quartz (right)(Reflected-light, oil immersion, ×500).

Included in Tale 5.9 for comparison is data for a Mozambican ore, a Zimbabwean (Lynx mine) concentrate (Sandmann et al., 2014), as well as a flake sample from Madagascar (Sun et al., 2017). Quartz, feldspar, and muscovite are the main minerals in these samples. Thao et al. (2017) and Gautneb and Tveten (2000) also noted the occurrence of micas in their graphitic ores.

148 | P a g e

Table 5.9: XRD mineral identification for natural graphite samples.

Analysis SA Graphite Zim Graphite Mozambique Zimbabwe Madagascar (Sandmann et (Sandmann et (Sun et al., al., 2014) al., 2014) 2017)

Mineral 1 Graphite Quartz Quartz Graphite Quartz Mineral 2 Quartz Graphite Feldspar Quartz Kaolinite Mineral 3 Calcite Iron-rich phase Muscovite Muscovite Muscovite Mineral 4 Other Feldspar Graphite Other Graphite

The diffractogram for the demineralised SA Graphite is given, indicating the major peaks (Figure 5.11). These peaks are typical for natural graphite as reported by Inagaki (2013) and Pierson (1993). Note the sharpness of the 002, 004, and 101 peaks, indicating a well- crystallised structure (Pierson, 1993). Due to a ~10 % inorganic residue following demineralisation, additional low count peaks were also observed.

Figure 5.11: Diffractogram for demineralised SA Graphite.

If graphene layers are stacked in an ABAB arrangement (every second layer corresponds) it is referred to as hexagonal (2H) graphite, a thermodynamic stable form of graphite (Krauss et al., 1988). If graphene layers are stacked in an ABCABC arrangement (every third layer corresponds), it is referred to as rhombohedral (3R) graphite; an extended stacking fault of

149 | P a g e hexagonal graphite (Kwiecińska and Petersen, 2004; Pierson, 1993). Surrounding the 101 peak, additional peaks corresponding to rhombohedral graphite were observed (Figure 5.12). According to Parthasarathy et al. (2006), rhombohedral graphite can be ascribed to the presence of epigenetic / vein-type graphite. Marsh and Rodríguez-Reinso (2006) however, reported on the occurrence of rhombohedral graphite due to shear deformation caused by intensive milling and grinding. Seeing that the current sample was pulverised to - 75 µm prior to testing, as well as the indisputable evidence of flake graphite from petrographic analyses, the latter explanation is favoured. Tectonic shear stress might also have had an influence (Crespo et al., 2006).

Figure 5.12: Diffractogram for demineralised SA Graphite highlighting the occurrence of rhombohedral graphite.

For quantification, the d002, Lc, graphitisation degree, and metamorphic temperature were determined. Equations for calculation of d002 and Lc were given in Section 3.4.2. Tagiri (1981) proposed a numerical expression (Equation 5.1) for the graphitization degree (GD) of natural graphite. Both d002 and Lc are taken into consideration.

푑 − 3.70 Equation 5.1 퐺퐷 = 002 ∗ 100 log(퐿푐/1000)

150 | P a g e

Wada et al. (1994) established a linear relationship between GD and metamorphic temperature for graphitic pelites and limestones from Japan. The relationship for pelites from the Ryoke metamorphic terrain will be used to obtain the metamorphic temperature for this study (Equation 5.2). The relationship was not tested on samples outside Japan and for rocks other than pelites, and therefore can only be seen as a crude estimate. Baiju et al. (2005) and Parthasarathy et al. (2006) also applied it to natural graphite from India and Madagascar respectively.

푇푒푚푝푒푟푎푡푢푟푒 (°퐶) = (3.2 × 퐺퐷) + 280 Equation 5.2

The quantification results are provided in Table 5.10 and compared to literature values for natural graphite from Mozambique (Marques et al., 2009), Madagascar (Fukuda et al., 1997), and the Ampanihy Group in Madagascar (Parthasarathy et al., 2006).

For ideal graphite, the d002 is believed to be 3.35 Å; however, researchers such as Howe et al. (2003) used modernised techniques to amend this value to 3.38 Å. For a complete random structure / no graphitic order, the d002 is 3.44 Å. The d002 results for all four samples in Table 5.10 are clustered around the two ideal values. The Lc will be ∞ for ideal graphite

(Seehra and Pavlovic, 1993). The Lc values for SA Graphite, Mozambique, and Madagascar (Fukuda et al., 1997) are relatively low compared to flakes from Madagascar (Parthasarathy et al., 2006) and India (Baiju et al., 2005) in which Lc is in excess of 500 Å. The results are, however, well above amorphous graphite tested from China, in which Lc values were all below 138 Å (Li et al., 2018).

Tashiro et al. (2017) investigated 3.95 Ga rocks from Canada and found that the graphite metamorphic temperatures ranged between 585 and 800 °C. Similarly, Baiju et al. (2005) reported temperatures between 650 and 800 °C for Indian graphite samples, and Wada and Oana (1975) between 600 and 650 °C for Japanese graphite samples. The results presented in Table 5.10 suggest metamorphic temperatures between 450 and 550 °C for southern African graphite samples (excluding those from Parthasarathy et al., 2006). Crespo et al. (2006) suggested that for small Lc values (SA Graphite), tectonic shear stress might have had a significant impact on the crystallite size. Temperatures can be underestimated with up to 100 °C when using small Lc values in metamorphic temperature calculations.

151 | P a g e

Table 5.10: X-ray diffraction structural results for SA Graphite.

SA Graphite Mozambique Madagascar Madagascar (Marques et al., (Fukuda et al., (Parthasarathy 2009) 1997) et al., 2006)

d002 (Å) 3.35 3.37 3.35 3.33-3.34

Lc (Å) 381 243 292 568-585

GD 84 54 65 148-180

Metamorphic 547 452 489 754-857 temperature (°C)

Tagiri and Oba (1986) classified carbonaceous matter as coaly, disordered graphite, graphite, or fully-ordered graphite based on their d002 and Lc values (based on work from Honda et al. (1968) and Tagiri (1981)). This classification scheme is illustrated in Figure 5.13 with SA Graphite and previously mentioned reference samples indicated. It can be seen that SA Graphite is classified as “graphite”. The same classification is obtained from work by Landis (1971) (not discussed here).

Figure 5.13: Carbonaceous material classification based on d002 and Lc (adapted from Tagiri and Oba, 1986).

Raman microspectroscopy spectra obtained in the graphite samples are given in Figure 5.14. The spectra varied with very ordered graphite (Figure 5.14A), moderately ordered graphite (Figure 5.14B), and disordered graphite (Figure 5.14C) being present. The differences in the spectra might be due to different laser penetrating planes (edge and basal planes) (Compagnini et al., 1997; Katagiri et al., 1988; Kawashima and Katagiri, 1999; Tan

152 | P a g e et al., 2004; Wang et al., 1989), due to imperfections caused by mechanical polishing (Beyssac et al., 2003; Nasdala et al., 2004; Pasteris, 1989; Wang et al., 1989), or due to different sampling locations.

A

B

C

Figure 5.14: Raman microspectroscopy spectra for the natural graphite samples. A) Very ordered graphite; B) Moderately ordered graphite, and C) disordered graphite.

The Raman microspectroscopy curve fitting results (average) for SA Graphite and Zim Graphite are presented in Table 5.11. The D3, D4, and D5 bands had low counts and were therefore not included in the fitting. The G, D1, and D2 bands for both samples occur at 1588, 1356, and 1627 cm-1 respectively. The band shifts are slightly higher than those reported by Tuinstra and Koenig (1970) for typical graphite material. The G band FWHM, areas, and intensities are comparable for both samples. The G bands are pronounced and “sharp” with intensities ranging between 0.92 and 0.98 and FWHMs ranging between 21 to 24 cm-1. The D1 band area and intensity are much larger / higher for Zim Graphite - this indicates a more disordered nature for this sample. The D1/G band intensity ratio is therefore also higher for Zim Graphite.

153 | P a g e

Table 5.11: Quantitative Raman microspectroscopy curve fitting results for the natural graphite samples.

Sample SA Graphite Zim Graphite

Band position (cm-1) G 1588.88 ± 0.67 1588.46 ± 0.60 D1 1356.04 ± 0.59 1355.94 ± 0.72 D2 1626.37 ± 6.77 1627.10 ± 1.44

Band FWHM (cm-1) G 21.16 ± 3.58 23.76 ± 1.18 D1 40.63 ± 2.06 42.12 ± 1.27 D2 16.24 ± 4.56 18.46 ± 1.05

Band area (cm-1) G 31.98 ± 3.29 34.15 ± 1.82 D1 41.05 ± 24.78 64.95 ± 6.47 D2 3.92 ± 2.48 6.02 ± 0.93

Band intensity (normalised, a.u.) G 0.98 ± 0.11 0.92 ± 0.08 D1 0.63 ± 0.36 0.98 ± 0.08 D2 0.14 ± 0.07 0.21 ± 0.03

Calculated parameters D1/G intensity ratio 0.65 1.07

R2 0.53 0.62 Metamorphic temperature (°C) 398 360

Raman microspectroscopy is an excellent tool to determine the degree of metamorphism for carbonaceous materials. For low-grade metamorphic terrains (200-330 °C) (clear presence of D3, D4, and D5 bands), Lahfid et al. (2010) used the areas of 1st order bands to determine the degree of metamorphism as well as the metamorphic temperature. For high-grade metamorphic terrains (330-650 °C) (neglectable small D3, D4, and D5 bands), Beyssac et al.

(2002) proposed a relationship (R2) between the areas of the G, D1, and D2 bands (Equation 5.3) to determine the peak metamorphic temperature (Equation 5.4).

퐴퐷1 Equation 5.3 푅2 = 퐴퐷1 + 퐴퐷2 + 퐴퐺

푇(°퐶) = (−455 × 푅2) + 641 (±50 °퐶) Equation 5.4

The metamorphic temperatures for Zim Graphite and SA Graphite were calculated as 360 and 398 °C respectively. This is much lower than the 547 °C calculated via XRD for SA Graphite. The reason for this is unclear. Both samples metamorphosed at relatively low temperatures. Kouketsu et al. (2014) classified carbonaceous materials metamorphosing at 280-400 °C as “medium-grade carbonaceous material”.

154 | P a g e

Flake graphite particles found in the SA Graphite sample are shown (Figure 5.15). Flake sizes in this sample were optically measured between 120 and 230 µm (length). Typical graphite found in the Zim Graphite sample, consisting of very small, flake graphite particles are illustrated (Figure 5.16). Flake sizes in this sample were optically measured between 15 and 74 µm (length). Take note however that both SA Graphite and Zim Graphite were milled before purchase. The sizes are therefore not representative of the size in the ore.

A B

100 µm 100 µm

C D

100 µm 100 µm

E F

100 µm 100 µm

Figure 5.15: A collection of flake graphite particles found in the SA Graphite sample: A) Coarse flake graphite (left) and quartz (right) particles; B) Coarse flake graphite particle; C) Basal plane view of a flake graphite particle; D) Agglomerated graphite flakes with a quartz inclusion (left); E) Agglomerated graphite flakes (basal plane); F) Variety of coarse flake graphite particles (Reflected-light, oil immersion, ×500).

155 | P a g e

A B

100 µm 100 µm

Figure 5.16: A collection of flake graphite particles found in the Zim Graphite sample: A) Small graphite flakes; B) Small graphite flakes embedded within an amorphous Fe-oxide (Reflected-light, oil immersion, ×500).

Reflectance analysis was only undertaken on the SA Graphite sample (due to the limited amount of graphite particles in Zim Graphite). The results are given in Table 5.12. The bireflectance is given as 9.28 and the anisotropy as 0.86. Anisotropy is an indication of the ability of a mineral to reflect plane-polarised light when the specimen is rotated while bireflectance is the difference between maximum and minimum reflectance (Allaby, 2008). Graphite is known to have a high bireflectance (6-27); the higher this value the better the graphitization (Craig and Vaughan, 1994). The same is applicable for anisotropy. A value of 1 indicates complete anisotropy.

Sweeney and Burnham (1990) and Barker (1988) developed vitrinite maturation models in which the metamorphic temperature can be determined from mean vitrinite reflectance (Mukoyoshi et al., 2006). In Equations 5.5 and 5.6, Sweeney and Burnham’s model at 1 million years and 10 million years respectively are given. In Equation 5.7 Barker’s time- independent model is given.

푇(°퐶) = 174 + (93(ln 푝푒푟푐푒푛푡푎푔푒푅푚푒푎푛)) Equation 5.5

푇(°퐶) = 158 + (90(ln 푝푒푟푐푒푛푡푎푔푒푅푚푒푎푛)) Equation 5.6

푇(°퐶) = 148 + (104(ln 푝푒푟푐푒푛푡푎푔푒푅푚푒푎푛)) Equation 5.7

All three equations were used to determine the metamorphic temperature for the SA Graphite sample and the results are given in Table 5.12. The metamorphic temperatures are in-line with that determined from Raman microspectroscopy, but not with XRD.

156 | P a g e

Table 5.12: Petrography reflectance analysis results for SA Graphite sample.

Analysis SA Graphite

Rmean (Rr%) 6.13±0.78

Rmax (Rr%) 10.77±1.41

Rmin (Rr%) 1.49

Bireflectance (Rr%) 9.28

Anisotropy 0.86

Metamorphic temperature (°C) – Equation 5.5 343

Metamorphic temperature (°C) – Equation 5.6 321

Metamorphic temperature (°C) – Equation 5.7 337

5.13 Summary Natural graphite is currently identified as a high risk, critical global commodity, and alternative sources are being sought. In the current research, natural graphite occurrences and mining potential in southern Africa have been extensively discussed. The countries of South Africa, Swaziland, Lesotho, Namibia, Botswana, Zimbabwe, Mozambique, and Madagascar were considered. The review should aid and inform governmental officials, mining houses, scientists, academics, environmentalists, and any other role-players, as to the natural graphite potential found in southern Africa.

At the time of writing, deposits in Namibia, Madagascar, Mozambique, and Zimbabwe are being exploited for their graphite reserves. Mozambique is seen as an emerging hotspot for graphite mining, due to its high-quality and large volumes of flake graphite reserves. More than 17 million tonnes of minable graphite are reported for Mozambique.

In Namibia, the scarce vein-type graphite occurs at the historical Aukam Mine in the !Karas region. Sri-Lanka is currently the only notable vein graphite producer globally.

In South Africa, the Limpopo Province contains over 67 natural graphite deposits. Jonkel Carbons and Grafites Pty Ltd. are planning a small-scale flake graphite mine at the Goedehoop and Steamboat deposits in Limpopo Province. However, graphite volumes are not large enough for export, therefore only local utilisation is targeted.

Political and economic instability in many of the southern African countries results in the reluctance of investors to commit to exploration and exploitation in the region. For example, the Lynx Mine in Zimbabwe had to reduce its production drastically in 2009 due to the combined effect of the global recession and ongoing political instability.

157 | P a g e

Although every effort was made to create a comprehensive and complete list of natural graphite deposits in southern Africa based on available public domain literature, it is still possible that some undocumented graphite deposits exist. It would be beneficial to undertake a regional field-based project to determine the extent of the reserves, the qualities, and current status of the reviewed occurrences.

To comment on the quality of graphite in southern Africa, samples from potential South African and Zimbabwean deposits were obtained and characterised.

The volatile percentages were low, indicating limited impurities. The elemental carbon percentages, strong carbon-carbon bonds, are dominant in both samples. This makes sense seeing that graphene layers in graphite are essentially hexagonal carbon rings with strong sigma bonds (524 kJ/mol) between the carbon atoms.

Quartz and calcite were the major minerals identified in the SA Graphite sample, while for the Zim Graphite sample the major minerals were quartz, feldspar, and an iron-rich phase.

Structural analyses showed that the crystallite size of SA Graphite was relatively small at 381 Å, especially compared to other sources in which the crystallite size exceeds 500 Å.

Textural characterisation showed that both samples can be described as flake-type graphite. The flake size of graphite particles found in the SA Graphite sample was optically measured between 120 and 230 µm (length) and in Zim Graphite between 15 and 74 µm (length).

The metamorphic temperatures were determined with XRD, Raman microspectroscopy, and vitrinite reflectance respectively. It was found that XRD overestimated the metamorphic temperature. The other two analyses resulted in temperatures ranging between 321 and 398 °C. This is relatively low graphite metamorphism, and can explain the small crystallite size. Based on the metamorphic temperature the graphite can be described as “medium grade carbonaceous material”.

The results obtained from this chapter are important seeing as it identifies possible sources of natural graphite in southern Africa to help bridge the gap between supply and demand. To understand if the graphitization trials for the ERA-MIN Charphite project were successful, the characterisation of natural graphite was required for comparison purposes.

158 | P a g e

Chapter 6: Summary, conclusions and recommendations

6.1 Summary This thesis forms part of the third ERA-MIN collaboration between Portugal, Poland, Romania, Argentina, and South Africa under the project Charphite. The aim of the Charphite project was to determine if char found in coal ash can be used as a substitute for natural graphite in green energy applications. This thesis covered only part of the overall Charphite project and focussed specifically on the South African output. The aim of this thesis was to test the applicability of char in South African ash to act as a possible precursor for synthetic graphite.

The characterisation of twelve South African coal ash sources, the separation of char from these sources, and the characterisation / evaluation of the extracted char as a possible precursor for synthetic graphite were presented. A desktop study on natural graphite occurrences in southern Africa, and characterisation of selected samples were also provided.

6.2 Conclusions The aim and objectives were met:

1. The desktop study on coal conversion ash in South Africa provided information on the conversion utilities, the ash volumes produced by these utilities, the current usage of coal ash in South Africa, and the accessibility of coal ash in South Africa. More than 50 million tonnes of ash are produced annually in South Africa. The majority of the ash is produced from Eskom’s 15 coal-fired power station (~36 million tonnes) and Sasol’s CTL process (~7 million tonnes). Eskom has the capacity to sell 26 % of their produced ash (ash not being used as salt effluent sink), but are currently only selling 7 %. Sasol is currently utilising a very small percentage of their ash. The reason for the low utilisation percentages is twofold: i) ash in South Africa is classified as a waste and subsequent regulations associated with this classification hinders application; and ii) high transportation costs limit offtake to users in Mpumalanga only. 2. Ash, and the corresponding feed coal, from power plants and a gasification unit in South Africa were obtained and characterised in detail. Characterisation included proximate, ultimate, CV, XRF, LOI, XRD, petrography, and PSD analyses. Based on the results, suitability of the ash for char concentration was considered, and four ash samples were selected for char beneficiation.

159 | P a g e

3. A desktop study on the separation of char from coal ash was conducted via examining case studies from Hwang et al. (2002); Cabielles et al. (2008), and Maroto-Valer et al. (1999a). The desktop study provided information on how to produce and recover high-grade char concentrates using size, density, electrostatic, magnetic, froth flotation, oil agglomeration, and acid demineralisation purification techniques. 4. A process was then developed to separate char from the previously obtained ash samples using size, electrostatic, magnetic, and density separation techniques. Carbon grades of <85 wt. % and carbon recoveries of <35 % were obtained. The project objective (to achieve grades >90 wt. % with high recoveries) was thus not met with simple techniques. Acid demineralisation, however, yielded carbon grades in excess of 90 wt. %. The char concentrates from both simple and acid demineralisation techniques were submitted for graphitization trials. 5. The extracted chars were characterised in detail (proximate, ultimate, carbon form, XRD mineralogy, XRD structural, Raman microspectroscopy, and petrographic analyses). The results supplied information on the possible usage of char in graphitic applications. The char from South Africa were classified as “transitional carbonaceous material” with the potential for graphitization.

The secondary objectives were met:

6. The South African char concentrate characteristics were compared to those of the other ERA-MIN consortiums. The comparison commented on the graphitization ability of the concentrated chars. The char concentrates from South Africa, Portugal, and Poland were classified as “transitional carbonaceous material” with the possibility for graphitization. The char concentrate from Romania was classified as amorphous or non-graphitizable. 7. The desktop study on natural graphite occurrences in southern Africa informed on the possible usage / sustainability of graphite for future generations, and a review paper was produced. 8. Natural graphite samples from South Africa and Zimbabwe were obtained and characterised. Characterisation included proximate, XRF, carbon form, XRD mineralogy, XRD structural, Raman microspectroscopy, and petrographic analyses. Based on the results, the graphite can be described as “medium grade carbonaceous material”.

The major results came from Objectives 4 and 5. Separation of the char from ash yielded product grades between 45 and 66 wt. % LOI. A density separation step was added to one

160 | P a g e of the fly ash samples and a final grade of 83 wt. % LOI was achieved. Although a significant increase from the initial char in ash percentages was obtained, the carbon recoveries were very low (6 to 32 %) due to the high amount of inherent mineral matter. The evaluation of the extracted chars showed the presence of strong carbon-carbon bonds, similar to those found in graphite, and limited impurities (oxygen, nitrogen, sulphur, and hydrogen). The anisotropy of the samples ranged between 22 and 49 %; the reference natural graphite sample had an anisotropy of 86 %. The three-dimensional structure of the chars can be described as turbostratic, with randomly orientated carbon layers, small graphite crystallite sizes, and large interlayer spacings. Raman microspectroscopy classified the chars as being “transitional” (being somewhere between amorphous and graphitic carbon) with the possibility acting as a precursor for synthetic graphite. It is proposed that the South African char is a suitable candidate for graphitization based on its carbon purity and structural order.

6.3 Recommendations The following recommendations pertaining to future research are suggested:

1. Except for newly built Medupi and Kusile, all Eskom power stations are scheduled to be mothballed by 2050 (Department of Energy, 2018). After the mothballing of the power stations, fresh ash would not be available anymore. It is therefore recommended that the study be repeated on weathered or landfilled ash. 2. Also it is recommended that further test work should be conducted on waste coals (>60 million tones waste coals are produced per year) from coal mines and thermo- chemical processes to prepare synthetic graphite. 3. Due to the low starting char in ash percentages, a significant amount of carbon has to be recovered to make economic sense. However, the carbon recoveries obtained were low. The reasons for the low recoveries might be due to all trials being conducted in batch mode, or due to small, unliberated ash minerals that formed part of the char matrix. It is recommended that the separation process be upscale to continuous-mode to assess if the recoveries can be increased. Additional tests should also be conducted regarding the unliberated behaviour of the char-ash particles. 4. A final char grade of 90 wt. % carbon was sought from the separation trials. However, the final product grades achieved ranged between 45 and 83 wt. % carbon. It is recommended that a triboelectrostatic separator (instead of a CoronaStat electrostatic separator) should be used, seeing as it is more effective on minerals with small differences in their work functions (such as char and ash) (Bada et al., 2010; Das et al., 2010; Zhang et al., 2012). Air fluidization as a separation technology can also be tested.

161 | P a g e

5. The economic evaluation on the proposed process of recovering chars or synthetic graphite from South African coal ash samples should be conducted. 6. Although Charphite (char to graphite) can be used as a synthetic solution for the current supply and demand issues in the graphite market, southern Africa is known to host vast amounts of untapped natural graphite (a cheaper alternative to synthetic graphite). Although every effort was made to create a comprehensive and complete list of natural graphite deposits in southern Africa based on available public domain literature, it is still possible that some undocumented graphite deposits exist. It would be beneficial to undertake a regional field-based project to determine the extent of the reserves, the qualities, and current status of the reviewed occurrences. It is also recommended to conduct a detailed desktop study on other African countries (especially East and Central Africa). 7. The graphite from southern African characterised was pre-concentrated and it is recommended that samples should be obtained directly from the field to comment more accurately on flake size, graphitic content, beneficiation possibilities, and possible applications.

162 | P a g e

References Agrawal, V., Sharma, S., 2018. Molecular characterization of kerogen and its implications for determining hydrocarbon potential, organic matter sources and thermal maturity in Marcellus Shale. Fuel 228, 429–437. https://doi.org/10.1016/j.fuel.2018.04.053

Ahmaruzzaman, M., 2010. A review on the utilization of fly ash. Progress in Energy and Combustion Science 36, 327–363. https://doi.org/10.1016/j.pecs.2009.11.003

Ahn, Y., Kil, D., Yang, J.I., Chung, H.S., 1999. Characteristics of unburned carbon particles recovered from fly ash [WWW Document]. URL https://pdfs.semanticscholar.org/0471/f9543d73afdd32b8e07a64b020a69fc94c2a.pdf (accessed 7.10.19).

Allaby, M., 2008. Bireflectance. A Dictionary of Earth Sciences.

Altigani, M.A.H., Merkle, R.K.W., Dixon, R.D., 2016. Geochemical identification of episodes of gold mineralisation in the Barberton Greenstone Belt, South Africa. Ore Geology Reviews 75, 186–205. https://doi.org/10.1016/j.oregeorev.2015.12.016

Andrews, M., 2012. Combustion optimisation [WWW Documents]. URL http://www.eandcspoton.co.za/resources/docs/Enviro- handbook/Combustion_optimisation.pdf (accessed 7.4.19).

Anshits, A.G., Sharonova, O.M., Anshits, N.N., Vereshchagin, S.N., Rabchevskii, E.V., Solovjev, L.A., 2011. Ferrospheres from fly ashes: Composition and catalytic properties in high-temperature oxidation of methane, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, Denver, CO. http://www.flyash.info/

Argosy Minerals Ltd, 2019. Erongo Graphite Project - Namibia [WWW Document]. URL http://www.argosyminerals.com.au/erongo-graphite-project-namibia (accessed 7.4.19).

Arnoldi, M., 2017. UCG power plant developer eagerly awaits request for proposal [WWW Document]. URL https://www.miningweekly.com/article/ucg-power-plant-developer- eagerly-awaits-request-for-proposal-2017-03-17/rep_id:3861 (accessed 7.3.19).

Ashton, P., Love, D., Mahachi, H., Dirks, P., 2011. An overview of the impact of mining and mineral processing operations on water resources and water quality in the Zambezi,

163 | P a g e

Limpopo and Olifants catchments in southern Africa (Report). Pretoria.

ASTM (American Society for Testing and Materials). 2012. Standard specification for coal fly ash and raw or calcined natural pozzolan for use in concrete. West Conshohocken: ASTM International. (ASTM C618a).

ASTM (American Society for Testing and Materials). 2013. Standard test method for major and minor elements in coal and coke ash by X-Ray fluorescence. West Conshohocken: ASTM International. (ASTM D4326).

ASTM (American Society for Testing and Materials). 2013. Standard test methods for Loss on Ignition (LOI) of solid combustion residues. West Conshohocken: ASTM International. (ASTM D7348).

ASTM (American Society for Testing and Materials). 2018. Standard classification of coals by rank. West Conshohocken: ASTM International. (ASTM D388).

Australian Standard. 2002. Coal and coke – Analysis and testing. Part 22: Higher rank coal – Mineral matter and water of constitution. NSW: Standards Australia International. (AS 1038.22-2000).

Bada, S.O., Falcon, R.M.S., Falcon, L.M., 2010. The potential of electrostatic separation in the upgrading of South African fine coal prior to utilization - a review. Journal of the Southern African Institute of Mining and Metallurgy 110, 691–702.

Badenhorst, F.P., 1987. Lithostratigraphy of the Damara Sequence in the Omaruru area of the Northern Central Zone of the Damara Orogen and a proposed correlation across the Omaruru Lineament. Communications of the Geological Survey of South West Africa/Namibia 3, 3-9.

Badenhorst, C.J., Wagner, N.J., Valentim, B.R.V., Santos, A.C., Guedes, A., Białecka, B., Całus, J., Popescu, L.G., Cruceru, M., Predeanu, G., Viljoen, K.S., Lázaro-Martínez, J.M., Abagiu, T.A., 2019. Char from coal ash as a possible precursor for synthetic graphite – Recent developments of the Charphite project, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, St. Louis, MO. http://www.flyash.info/

Baiju, K.R., Satish-Kumar, M., Kagi, H., Nambiar, C.G., Ravisankar, M., 2005. Mineralogical characterization of graphite deposits from Thodupuzha-Kanjirappally Belt, Madurai Granulite Block, Southern India. Gondwana Research 8, 223–230.

164 | P a g e

https://doi.org/10.1016/S1342-937X(05)71120-5

Baker, L., Gupta, A., Hrach, F., 2019. Fly ash reclamation and beneficiation using a triboelectric belt separator, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, St. Louis, MO. http://www.flyash.info/

Ballhaus, C.G., Stumpfl, E.F., 1985. Occurrence and petrological significance of graphite in the Upper Critical Zone, western Bushveld Complex, South Africa. Earth and Planetary Science Letters 74, 58–68. https://doi.org/10.1016/0012-821X(85)90166-9

Ballhaus, C.G., 1988. Potholes of the Merensky Reef at Brakspruit Shaft, Rustenburg platinum mines; primary disturbances in the magmatic stratigraphy. Economic Geology 83, 1140–1158. https://doi.org/10.2113/gsecongeo.83.6.1140

Baltrus, J.P., Wells, A.W., Fauth, D.J., Diehl, J.R., White, C.M., 2001. Characterization of carbon concentrates from coal-combustion fly ash. Energy Fuels 15, 455–462. https://doi.org/10.1021/ef000201o

Baltrus, J.P., Diehl, J.R., Soong, Y., Sands, W., 2002. Triboelectrostatic separation of fly ash and charge reversal. Fuel 81, 757–762. https://doi.org/10.1016/S0016- 2361(01)00196-X

Baludikay, B.K., François, C., Sforna, M.C., Beghin, J., Cornet, Y., Storme, J.-Y., Fagel, N., Fontaine, F., Littke, R., Baudet, D., Delvaux, D., Javaux, E.J., 2018. Raman microspectroscopy, bitumen reflectance and illite crystallinity scale: comparison of different geothermometry methods on fossiliferous Proterozoic sedimentary basins (DR Congo, Mauritania and Australia). International Journal of Coal Geology 191, 80–94. https://doi.org/10.1016/j.coal.2018.03.007

Ban, H., Li, T.X., Hower, J.C., Schaefer, J.L., Stencel, J.M., 1997. Dry triboelectrostatic beneficiation of fly ash. Fuel 76, 801–805. https://doi.org/10.1016/S0016- 2361(97)00045-8

Barker, O.B., 1986. An overview of the coal reserve position and certain technical and geological constraints to coalfield development in South Africa. S.A. Mining World October, 94-107.

Barker, C.E., 1988. Geothermics of petroleum systems: implication of the stabilization of kerogen thermal maturation after a geologically brief heating duration at peak temperature (Report), U.S. Geological Survey Bulletin, 1870: 26-29.

165 | P a g e

Barnes, D.I., Sear, L.K.A., 2006. Ash utilisation from coal-based power plants, in: AshTech 2006. United Kingdom Quality Ash Association, Birmingham.

Barr, M.W.C., Brown, M.A., 1987. Precambrian gabbro–anorthosite complexes, Tete Province, Mozambique. Geological Journal 22, 139–159. https://doi.org/10.1002/gj.3350220612

Barton, J.M., Blaine, J.L., Doig, R., Byron, C.L., 1994. The geological setting and style of copper mineralization at the Bushman group of deposits, northeastern Botswana. Journal of African Earth Sciences 18, 87–97. https://doi.org/10.1016/0899- 5362(94)90022-1

Bartoňová, L., 2015. Unburned carbon from coal combustion ash: An overview. Fuel Processing Technology 134, 136–158. https://doi.org/10.1016/j.fuproc.2015.01.028

Bass Metals, 2018. Bass Metals [WWW Document]. URL https://www.bassmetals.com.au/ (accessed 7.4.19).

Battery Minerals, 2018.Battery Minerals [WWW Document]. URL https://www.batteryminerals.com/ (accessed 7.4.19).

Belaid, M., Falcon, R., Vainikka, P., 2014. Pulverized coal versus circulating fluidized-bed boilers - perspectives and challenges for South Africa. South African Journal of Chemical Engineering 19, 72–81.

Belcher, R.W., 2003. Tectonostratigraphic evolution of the Swartland region and aspects of orogenic lode-gold mineralisation in the Pan-African Saldania Belt, Western Cape, South Africa (Thesis). Stellenbosch: Stellenbosch University.

Benson, S.A., Hurley, J.P., Zygarlicke, C.J., Steadman, E.N., Erickson, T.A., 1993. Predicting ash behavior in utility boilers. Energy Fuels 7, 746–754. https://doi.org/10.1021/ef00042a008

Bény-Bassez, C., Rouzaud, J.N., 1985. Characterization of Carbonaceous Materials by Correlated Electron and Optical Microscopy and Raman Microspectroscopy. Scanning Electron Microscopy 1985, 119-132.

Berglund, L., Touret, J., 1976. Garnet—biotite gneiss in ‘Système du Graphite’ (Madagascar): Petrology and fluid inclusions. Lithos 9, 139–148. https://doi.org/10.1016/0024-4937(76)90031-1

166 | P a g e

Berning, J., Cooke, R., Hiemstra, S.A., Hoffman, U., 1976. The Rössing uranium deposit, South West Africa. Economic Geology 71, 351–368. https://doi.org/10.2113/gsecongeo.71.1.351

Beyssac, O., Goffé, B., Chopin, C., Rouzaud, J.N., 2002. Raman spectra of carbonaceous material in metasediments: a new geothermometer. Journal of Metamorphic Geology 20, 859–871. https://doi.org/10.1046/j.1525-1314.2002.00408.x

Beyssac, O., Goffé, B., Petitet, J.-P., Froigneux, E., Moreau, M., Rouzaud, J.-N., 2003. On the characterization of disordered and heterogeneous carbonaceous materials by Raman spectroscopy. Spectrochimica Acta Part A 59, 2267–2276. https://doi.org/10.1016/S1386-1425(03)00070-2

Bingen, B., Viola, G., Henderson, I.H.C., Smethurst, M., Boyd, R., Thomas, R.J., Bjerkgård, T., Feito, P., Hollick, L.M., Jacobs, J., Key, R.M., Rossi, D., Sandstad, J.S., Skår, Ø., Smith, R., Solli, A., Tveten, E., 2006. Geochronology of Pan-African terrain assembly in NE Mozambique, in: 21st Colloquium on African Geology. Maputo.

Bingen, B., Bjekkgård, T., Boyd, R., Grenne, T., Henderson, I., Lutro,O., Melezhik, V., Motuza, G., Nordgulen, Ø., Often, M., Sandstad, J.S., Smelror, M., Solli, A., Stein, H., Saether, O.M., T. Thorsnes, T., Tveten, E., Bauer, W., Dunkley, P., Gonzalez, E., Hollick, L., Jacobs, J., Key, R.M., Smith, R., Thomas, R.J., Jamal, D., Catuane, F., De Azavedo, S., Feitio, P., Manhica, V., Manuel, S., Moniz, A., Njange, F., Rossi, D., De S. Soares, H., Tembe, D., Uachave, B., Viola, G., Zandamela, E., 2007. The geology of Niassa and Cabo Delgado Provinces with parts of Zambesia and Nampula Provinces, Mozambique. Maputo: Ministry of Mineral Resources and Energy, National Directorate of Geology (Report).

Biscoe, J., Warren, B.E., 1942. An X‐Ray study of Carbon Black. Journal of Applied Physics 13, 364–371. https://doi.org/10.1063/1.1714879

Bisnath, A., McCourt, S., Frimmel, H.E., Buthelezi, S.B.N., 2008. The metamorphic evolution of mafic rocks in the Tugela Terrane, Natal Belt, South Africa. South African Journal of Geology 111, 369–386. https://doi.org/10.2113/gssajg.111.4.369

Bittner, J.D., Gasiorowski, S.A., Hrach, F.J., Lewandowski, W., 2012. Separation technologies’ 17 years of commercial experience in fly ash processing, in: EUROCOALASH 2012 Conference. Thessaloniki. http://www.ecoba.com/eurocoalash.html

167 | P a g e

Bittner, J.D., Hrach, F.J., Gasiorowski, S.A., Canellopoulus, L.A., Guicherd, H., 2014. Triboelectric belt separator for beneficiation of fine minerals. Procedia Engineering 83, 122–129. https://doi.org/10.1016/j.proeng.2014.09.021

Bjurström, H., Lind, B.B., Lagerkvist, A., 2014. Unburned carbon in combustion residues from solid biofuels. Fuel 117, 890–899. https://doi.org/10.1016/j.fuel.2013.10.020

Blissett, R.S., Rowson, N.A., 2012. A review of the multi-component utilisation of coal fly ash. Fuel 97, 1–23. https://doi.org/10.1016/j.fuel.2012.03.024

Boral, 2018. Fly ash slides for investors – As at 29 May 2018 [WWW Document]. URL https://www.boral.com/sites/corporate/files/media/field_document/180528-Flyash- slides-for-investors-as-at-29May2018.pdf (accessed 7.4.19).

Bourrat, X., 2000. Structure in Carbons and Carbon Artifacts, in: Marsh, H., Rodríguez- Reinoso, F. (Eds.), Sciences of Carbon Materials. Universidad de Alicante, Spain, pp. 1-97.

Boyd, F.R., Nixon, P.H., 1975. Origins of the ultramafic nodules from some kimberlites of northern Lesotho and the Monastery Mine, South Africa, in: Ahrens, L.H., Dawson, J.B., Duncan, A.R., Erlank, A.J. (Eds.), Physics and chemistry of the earth: Volume 9. Pergamon Press Ltd., Great Britian, pp. 431-454.

Boyd, R., Nordgulen, Ø., Thomas, R.J., Bingen, B., Bjerkgård, T., Grenne, T., Henderson, I., Melezhik, V.A., Often, M., Sandstad, J.S., Solli, A., Tveten, E., Viola, G., Key, R.M., Smith, R.A., Gonzalez, E., Hollick, L.J., Jacobs, J., Jamal, D., Motuza, G., Bauer, W., Daudi, E., Feitio, P., Manhica, V., Moniz, A., Rosse, D., 2010. The geology and geochemistry of the east African Orogen in northeastern Mozambique. South African Journal of Geology 113, 87–129. https://doi.org/10.2113/gssajg.113.1.87

Brandt, R., 1987. A revised stratigraphy for the Abbabis Complex in the Abbabis Inlier, Namibia. South African Journal of Geology 90, 314-323.

Brenon, P., 1972. The Geology of Madagascar, in: Battistini, R., Richard-Vindard, G. (Eds.), Biogeography and Ecology of Madagascar. Springer Science+Business Media, Dordrecht, pp. 27-86. https://doi.org/10.1007/978-94-015-7159-3

British Geological Survey, 2015. Risk List 2015 (Report). National Environmental Research Council, England.

168 | P a g e

Brown, R.C., Dykstra, J., 1995. Systematic errors in the use of loss-on-ignition to measure unburned carbon in fly ash. Fuel 74, 570–574. https://doi.org/10.1016/0016- 2361(95)98360-Q

Bryers, R.W., 1986. Influence of Segregated Mineral Matter in Coal on Slagging, in: Vorres, K.S. (Ed.), Mineral Matter and Ash in Coal. American Chemical Society, Washington, pp. 236-248. https://doi.org/10.1021/bk-1986-0301.fw001

Bunt, J.R., 2006. A new dissection methodology and investigation into coal property transformational behaviour impacting on a commercial-scale Sasol-Lurgi MK IV fixed- bed gasifier (Thesis). Potchefstroom: North-West University.

Bunt, J.R., Waanders, F.B., 2008. An understanding of lump coal physical property behaviour (density and particle size effects) impacting on a commercial-scale Sasol- Lurgi FBDB gasifier. Fuel 87, 2856–2865. https://doi.org/10.1016/j.fuel.2008.03.022

Buntin, T.J., Grandstaff, D.E., Ulmer, G.C., Gold, D.P., 1985. A pilot study of geochemical and redox relationships between potholes and adjacent normal Merensky Reef of the Bushveld Complex. Economic Geology 80, 975–987. https://doi.org/10.2113/gsecongeo.80.4.975

Cabielles, M., Montes-Morán, M.A., Garcia, A.B., 2008. Structural study of graphite materials prepared by HTT of unburned carbon concentrates from coal combustion fly ashes. Energy Fuels 22, 1239–1243. https://doi.org/10.1021/ef700603t

Cabielles, M., Rouzaud, J.-N., Garcia, A.B., 2009. High-Resolution Transmission Electron Microscopy studies of graphite materials prepared by High-Temperature Treatment of unburned carbon concentrates from combustion fly ashes. Energy Fuels 23, 942– 950. https://doi.org/10.1021/ef800763s

Cairncross, B., Dixon, R., 1995. Minerals of South Africa (Report). The Geological Society of South Africa, Johannesburg.

Cairncross, B., 2001. An overview of the Permian (Karoo) coal deposits of southern Africa. Journal of African Earth Sciences 33, 529–562. https://doi.org/10.1016/S0899- 5362(01)00088-4

Cairncross, B., 2004. Field Guide to Rocks and Minerals of southern Africa. Penguin Random House South Africa, Johannesburg.

169 | P a g e

Cameán, I., Garcia, A.B., 2011. Graphite materials prepared by HTT of unburned carbon from coal combustion fly ashes: Performance as anodes in lithium-ion batteries. Journal of Power Sources 196, 4816–4820. https://doi.org/10.1016/j.jpowsour.2011.01.041

Cameron, E.N., Weis, P.L., 1960. Strategic graphite: A survey (Report), Geological Survey Bulletin 1082-E. United States Government Printing Office, Washington.

Cangialosi, F., Notarnicola, M., Liberti, L., Caramuscio, P., Belz, G., Gurupira, T.Z., Stencel, J.M., 2005. Characteristics of Fly Ashes and Processing Conditions Affecting Carbon-Ash Separation under Pneumatic Transport, Triboelectric Processing, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, Lexington, KY. http://www.flyash.info/

Carlson, C.L., Adriano, D.C., 1993. Environmental impacts of coal combustion residues. Journal of Environmental Quality 22, 227–247. https://doi.org/10.2134/jeq1993.00472425002200020002x

CBS News, 2008a. TVA Tennessee coal ash disaster near Harriman [WWW Document]. YouTube. URL https://www.youtube.com/watch?v=c_-EL5Jrrmk (accessed 7.4.19).

CBS News, 2008b. Tennessee coal sludge disaster [WWW Document]. YouTube. URL https://www.youtube.com/watch?v=BZhT4jx2XZ8 (accessed 7.4.19).

Chabalala, V.P., Wagner, N., Potgieter-Vermaak, S., 2011. Investigation into the evolution of char structure using Raman spectroscopy in conjunction with coal petrography; Part 1. Fuel Processing Technology 92, 750–756. https://doi.org/10.1016/j.fuproc.2010.09.006

Chafy, A., Tidswell, M., Lane, C., 2018. Case study: project start-up for thickened graphite tailings, storage and water recovery strategies, in: 21st International Seminar on Paste and Thickened Tailings. Perth.

Chamber of Mines of South Africa, 2014. National coal strategy for South Africa (Report). Chamber of Mines of South Africa, Johannesburg.

Chantraine, J., Radelli, L., 1970. Tectono-minerogenetic units of the basement of Madagascar. Economic Geology 65, 690–699. https://doi.org/10.2113/gsecongeo.65.6.690

170 | P a g e

Cole, D.I., Ngcofe, L., Halenyane, K., 2014. Mineral commodities in the Western Cape province, South Africa (Report). Council for Geoscience, Bellville.

Compagnini, G., Puglisi, O., Foti, G., 1997. Raman spectra of virgin and damaged graphite edge planes. Carbon 35, 1793–1797. https://doi.org/10.1016/S0008-6223(97)00141- 3

Craig, J.R., Vaughan, D.J., 1994. Ore Microscopy and Ore Petrography, second ed. John Wiley and Sons, New York.

Crespo, E., Luque, F.J., Barrenechea, J.F., Rodas, M., 2006. Influence of grinding on graphite crystallinity from experimental and natural data: implications for graphite thermometry and sample preparation. Mineralogical Magazine 70, 697–707. https://doi.org/10.1180/0026461067060358

Cruceru, M., Valentim, B., Predeanu, G., Abagiu, T.A., Slavescu, V., Anghelescu, L., 2017. Concentration of the residual carbon from bottom ash by particle size separation. International Journal of Geology 11, 1-7.

Cui, Y., Stubington, J.F., 2001. In-bed char combustion of Australian coals in PFBC. 3. Secondary fragmentation. Fuel 80, 2245–2251. https://doi.org/10.1016/S0016- 2361(01)00109-0

Dacombe, P., Pourkashanian, M., Williams, A., Yap, L., 1999. Combustion-induced fragmentation behavior of isolated coal particles. Fuel 78, 1847–1857. https://doi.org/10.1016/S0016-2361(99)00076-9

Dakič, D., van der Honing, G., Valk, M., 1989. Fragmentation and swelling of various coals during devolatilization in a fluidized bed. Fuel 68, 911–916. https://doi.org/10.1016/0016-2361(89)90129-4

Das, S., Köhnlechner, R., Aman, F., Dascalescu, L., 2010. Corona separation of fly ash. IEEE Transactions on Industry Applications 46, 2157–2164. https://doi.org/10.1109/TIA.2010.2071170

De Klerk, A., 2015. Mozambique: A highlighted summary [WWW Document]. URL https://www2.deloitte.com/content/dam/Deloitte/za/Documents/energy- resources/ZA_Mozambican_Cue_Card_221015.pdf (accessed 7.10.19).

De Wit, M.J., Bowring, S.A., Ashwal, L.D., Randrianasolo, L.G., Morel, V.P.I., Rambeloson,

171 | P a g e

R.A., 2001. Age and tectonic evolution of Neoproterozoic ductile shear zones in southwestern Madagascar, with implications for Gondwana studies. Tectonics 20, 1- 45. https://doi.org/10.1029/2000TC900026

Demir, U., Yamik, A., Kelebek, S., Oteyaka, B., Ucar, A., Sahbaz, O., 2008. Characterization and column flotation of bottom ashes from Tuncbilek power plant. Fuel 87, 666–672. https://doi.org/10.1016/j.fuel.2007.05.040

Department of Energy, 2018. Integrated Resource Plan 2018 (Report). Government printer, Pretoria.

Department of Environmental Affairs, 2018a. South Africa – State of waste report (Report), 2nd draft report. Government printer, Pretoria.

Department of Environmental Affairs, 2018b. National Environmental Management: Waste Act, 2008 (ACT NO. 59 of 2008) – Regulations regarding the exclusion of a waste stream or a portion of a waste stream from the definition of waste. (Notice 715). Government gazette, 41777:4, 18 Jul.

Department of Mines, 1940. The mineral resources of the union of South Africa. Government Printer, Pretoria.

Desjardins, J., 2012. Graphite: The driving force behind green technology [WWW Document]. URL https://www.visualcapitalist.com/portfoliographite-driving-force- behind-green-technology-graphene/ (accessed 7.3.19).

Dissanayake, C.B.C., Chandrajith, R., 1999. Sri Lanka–Madagascar Gondwana Linkage: Evidence for a Pan‐African Mineral Belt. The Journal of Geology 107, 223-235. https://doi.org/10.1086/314342

Dong, N.S., 2010. Reducing carbon-in-ash (Report). SIEA Clean Coal Centre, London.

Dorland, M.I., Campbell, Q.P., Le Roux, M., Erasmus, P., McMillan, K., Dorland, H.C., 2015. Predicting the washability behaviour of coal by numerical simulation of washabilities and product liberation from photographs, in: Proceedings of the South African Coal Processing Society Conference. South African Coal Processing Society, Secunda.

Du Venage, G., 2014. The race for graphite is on in Africa [WWW Document]. URL https://www.e-mj.com/features/the-race-for-graphite-is-on-in-africa/ (accessed 7.4.19).

172 | P a g e

Eberhard, A., 2011. The future of South African coal: market, investment, and policy challenges (Report). Stanford University: Program on Energy and Sustainable Development, Stanford. eNCA, 2014. South African air quality at risk if Eskom gets its way [WWW Document]. URL https://www.enca.com/south-africa/south-african-air-quality-risk-if-eskom-gets-its-way (accessed 7.10.19).

Eskom, 2017a. Fact sheet – Coal in South Africa [WWW Document]. URL http://www.eskom.co.za/AboutElectricity/FactsFigures/Documents/CO0007CoalSAR ev14.pdf (accessed 7.4.19).

Eskom, 2017b. COP17 fact sheet – Kusile and Medupi coal-fired power stations under construction [WWW Document]. URL http://www.eskom.co.za/OurCompany/SustainableDevelopment/ClimateChangeCOP 17/Documents/Kusile_and_Medupi_coal- fired_power_stations_under_construction.pdf (accessed 7.4.19).

Eskom, 2018. Eskom [WWW Document]. URL http://www.eskom.co.za/Pages/Landing.aspx (accessed 7.4.19).

Eskom Rotek Industries, 2019. Eskom Rotek Industries [WWW Document]. URL http://www.rotekindustries.co.za/Pages/Landing.aspx (accessed 7.4.19).

European Commission, 2017. Communication from the commission to the European parliament, the council, the European economic and social committee and the committee of the regions on the 2017 list of critical raw materials for the EU (Report), COM (2017) 490 final. Brussels.

Evans, R.J., Ashwal, L.D., Hamilton, M.A., 1999. Mafic, ultramafic, and anorthositic rocks of the Tete Complex, Mozambique: petrology, age, and significance. South African Journal of Geology 102, 153–166.

Everson, R., Neomagus, H., Kaitano, R., 2005. The modeling of the combustion of high-ash coal–char particles suitable for pressurised fluidized bed combustion: shrinking reacted core model. Fuel 84, 1136–1143. https://doi.org/10.1016/j.fuel.2005.01.011

Everson, R.C., Koekemoer, A., Bunt, J.R., Neomagus, H.W.J.P., Schwarz, C., 2013. Detailed characterization of South African high mineral matter inertinite–rich coals and density fractions and effect on reaction rates with carbon dioxide: macerals,

173 | P a g e

microlithotypes, carbominerites and minerals. South African Journal of Chemical Engineering 18, 1-16.

Falcon, M.S., Snyman, C.P., 1986. An introduction to coal petrography: Atlas of petrographic constituents in the bituminous coals of southern Africa (Report), Review Paper Numbr 2. The Geological Society of South Africa, Johannesburg.

Fan, M., Brown, R.C., 2001. Comparison of the loss-on-ignition and thermogravimetric analysis techniques in measuring unburned carbon in coal fly ash. Energy Fuels 15, 1414–1417. https://doi.org/10.1021/ef0100496

Feng, B., Bhatia, S.K., Barry, J.C., 2003. Variation of the crystalline structure of coal char during gasification. Energy Fuels 17, 744–754. https://doi.org/10.1021/ef0202541

Ferralis, N., Matys, E.D., Knoll, A.H., Hallmann, C., Summons, R.E., 2016. Rapid, direct and non-destructive assessment of fossil organic matter via microRaman spectroscopy. Carbon 108, 440–449. https://doi.org/10.1016/j.carbon.2016.07.039

Ferrari, A.C., Robertson, J., 2000. Interpretation of Raman spectra of disordered and amorphous carbon. Physical Review B 61, 14 095-14 107. https://doi.org/10.1103/PhysRevB.61.14095

Ferrari, S., Belevi, H., Baccini, P., 2002. Chemical speciation of carbon in municipal solid waste incinerator residues. Waste Management 22, 303–314. https://doi.org/10.1016/S0956-053X(01)00049-6

Feytis, A. 2010. The bright side of graphite. Industrial Minerals 7, 31-39.

Field, M., Stiefenhofer, J., Robey, J., Kurszlaukis, S., 2008. Kimberlite-hosted diamond deposits of southern Africa: A review. Ore Geology Reviews 34, 33–75. https://doi.org/10.1016/j.oregeorev.2007.11.002

Fischbach, D.B., 1971. The graphitization process. TANSO 1970, 115–120. https://doi.org/10.7209/tanso.1970.115

Fisher, G.L., Prentice, B.A., Silberman, D., Ondov, J.M., Biermann, A.H., Ragaini, R.C., McFarland, A.R., 1978. Physical and morphological studies of size-classified coal fly ash. Environmental Science and Technology 12, 447–451. https://doi.org/10.1021/es60140a008

Flagan, R.C., Seinfeld, J.H., 1988. Fundamentals of Air Pollution Engineering. Prentice-Hall,

174 | P a g e

Inc., Englewood Cliffs, NJ.

Fogg, C.T., Boyle, E.H., 1987. Flake and high-crystalline graphite availability – market economy countries – A minerals availability appraisal (Report), Bureau of Mines information circular 9122/1987. United States Bureau of Mines, Washington.

Fortier, S.M., Nassar, N.T., Lederer, G.W., Brainard, J., Gambogi, J., McCullough, E.A., 2018. Draft critical mineral list – Summary of methodology and background information – U.S. Geological Survey technical input document in response to Secretarial Order No. 3359 (Report). Reston, VA.

Franklin, R.E., 1951. The structure of graphitic carbons. Acta Crystallographica 4, 253–261. https://doi.org/10.1107/S0365110X51000842

Franus, W., Wiatros-Motyka, M.M., Wdowin, M., 2015. Coal fly ash as a resource for rare earth elements. Environ Science and Pollution Research 22, 9464–9474. https://doi.org/10.1007/s11356-015-4111-9

Freiberg, R.B., 1907. Notes on the gold-bearing rock of the Ayrshire mine, Lomagundi, Rhodesia. South African Journal of Geology 10, 13–16.

Frimmel, H.E., Miller, R., 2009. Chapter 5.7 Mineral Deposits, in: Gaucher, C., Sial, A.N., Frimmel, H.E., Halverson, G.P. (Eds.), Developments in Precambrian Geology, Neoproterozoic-Cambrian Tectonics, Global Change and Evolution: A Focus On South Western Gondwana. Elsevier, pp. 227–229. https://doi.org/10.1016/S0166- 2635(09)01616-8

Fuerstenau, M.C., Somasundaran, P., 2003. Flotation, in: Fuerstenau, M.C., Han, K.N. (Eds.), Principles of Mineral Processing. Society for Mining, Metallurgy, and Exploration, Littleton, CO, pp. 245-306.

Fukuda, K., Kikuya, K., Isono, K., Yoshio, M., 1997. Foliated natural graphite as the anode material for rechargeable lithium-ion cells. Journal of Power Sources 69, 165–168. https://doi.org/10.1016/S0378-7753(97)02568-8

Fulekar, M.H., Dave, J.M., 1986. Disposal of fly ash—an environmental problem. International Journal of Environmental Studies 26, 191–215. https://doi.org/10.1080/00207238608710257

Gajdošová, M., Huraiová, M., Hurai, V., Konečný, P., 2017. In situ chemical dating of new

175 | P a g e

accessory minerals from the Evate deposit in Mozambique, in: Conference on Accessory Minerals. Vienna and Bratislava.

Garnham, B.L., Langerman, K.E., 2016. Mercury emissions from South Africa’s coal-fired power stations. Clean Air Journal 26, 14–20. https://doi.org/10.17159/2410- 972X/2016/v26n2a8

Gautneb, H., Tveten, E., 2000. The geology, exploration and characterisation of graphite deposits in the Jennestad area, Vesterålen, northern Norway. NGU Bulletin 436, 67- 74.

Gecko, 2018. Project – Okanjande Graphite [WWW Document]. URL https://www.gecko.na/?page_id=27 (accessed 7.4.19).

Gellatly, D.C., 1966. Graphite in natural and experimental carbonate systems. Mineralogical Magazine 35, 963-970. https://doi.org/10.1180/minmag.1966.035.275.08

Gevers, T.W., 1934. The geology of the Windhoek district in South-West Africa. Transactions of the Geological Society of South Africa 37, 221-251.

Ginster, M., Matjie, R.H., 2005. Beneficial utilisation of Sasol coal gasification ash, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, Lexington, KY. http://www.flyash.info/

Glasser, F.P., 2004. Coal Combustion Wastes: Characterization, Reuse and Disposal, in: Gieré, R., Stille, P. (Eds.), Energy, Waste and the Environment: A Geochemical Perspective. The Geological Society, London, pp. 211-222. https://doi.org/10.1144/GSL.SP.2004.236.01.01

GlobeNewswire, 2018. Next Graphite announces successful startup of Aukam processing plant and initial shipment [WWW Document]. URL https://www.globenewswire.com/news-release/2018/05/15/1502494/0/en/Next- Graphite-Announces-Successful-Startup-of-Aukam-Processing-Plant-and-Initial- Shipment.html (accessed 7.4.19).

Gray, M.L., Champagne, K.J., Finseth, D.H., 2001. Continuous air agglomeration method for high carbon fly ash beneficiation. Patent US A9161673.

Gray, M.L., Champagne, K.J., Soong, Y., Killmeyer, R.P., Maroto-Valer, M.M., Andrésen, J.M., Ciocco, M.V., Zandhuis, P.H., 2002. Physical cleaning of high carbon fly ash.

176 | P a g e

Fuel Processing Technology 76, 11–21. https://doi.org/10.1016/S0378- 3820(02)00006-1

Gross, C.J., Weber, K., Siegesmund, S., Vollbrecht, A., 2000. Abnormal grain growth and Fe/Mn stained fluid paths in metacarbonates of the Damara Orogen (Namibia). Communications of the Geological Survey of Namibia 12, 117-129.

Gross, C.J., 2006. Grain coarsening and hydrothermal alteration in metacarbonates of the Damara orogeny, Namibia, in: 11th Symposium Tektonik, Struktur- und Kristallingeologie. Universitätsdrucke Göttingen, Göttingen.

Guedes, A., Valentim, B., Prieto, A.C., Rodrigues, S., Noronha, F., 2010. Micro-Raman spectroscopy of collotelinite, fusinite and macrinite. International Journal of Coal Geology 83, 415–422. https://doi.org/10.1016/j.coal.2010.06.002

Guedes, A., Valentim, B., Prieto, A.C., Noronha, F., 2012. Raman spectroscopy of coal macerals and fluidized bed char morphotypes. Fuel 97, 443–449. https://doi.org/10.1016/j.fuel.2012.02.054

Hancox, P.J., Götz, A.E., 2014. South Africa’s coalfields — A 2014 perspective. International Journal of Coal Geology 132, 170–254. https://doi.org/10.1016/j.coal.2014.06.019

Hancox, P.J., 2016. The Coalfields of South-Central Africa: A current perspective. Episodes 39, 407-428. https://doi.org/10.18814/epiiugs/2016/v39i2/95785

Harris, D., Heidrich, C., Feuerborn, J., 2019. Global aspects on coal combustion products, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, St. Louis, MO. http://www.flyash.info/

Harvey, R.D., Ruch, R.R., 1984. Overview of mineral matter in U.S. coals. ACS Division of Fuel Chemistry Preparation 29, 2-8.

Hattingh, B., 2012. Product evaluation and reaction modelling for the devolatilization of large coal particles (Thesis). Potchefstroom: North-West University.

Hattingh, B.B., Everson, R.C., Neomagus, H.W.J.P., Bunt, J.R., van Niekerk, D., Jordaan, J.H.L., Mathews, J.P., 2013. Elucidation of the structural and molecular properties of typical South African coals. Energy Fuels 27, 3161–3172. https://doi.org/10.1021/ef400633d

Hayes, T., 2016. Graphite for batteries [WWW Document].

177 | P a g e

https://www.edisoninvestmentresearch.com/?ACT=19&ID=17404&dir=sectorreports& field=19. (accessed 7.3.19).

Heidrich, C., Weir, A., Feuerborn, H.-J., 2013. Coal combustion products: a global perspective, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, Lexington, KY. http://www.flyash.info/

Helmuth, R., 1987. Fly ash in cement and concrete. Portland Cement Association, Skokie, IL.

Hlatshwayo, T.B., Matjie, R.H., Li, Z., Ward, C.R., 2009. Mineralogical characterization of Sasol feed coals and corresponding gasification ash constituents. Energy Fuels 23, 2867–2873. https://doi.org/10.1021/ef8010806

Hofmann, A., Bolhar, R., 2007. Carbonaceous cherts in the Barberton Greenstone Belt and their significance for the study of early life in the Archean record. Astrobiology 7, 355–388. https://doi.org/10.1089/ast.2005.0288

Honda, H., Kobayashi, K., Sugawara, S., 1968. X-ray characteristics of non-graphitizing-type carbon. Carbon 6, 517–523. https://doi.org/10.1016/0008-6223(68)90091-2

Howe, J.Y., Rawn, C.J., Jones, L.E., Ow, H., 2003. Improved crystallographic data for graphite. Powder Diffraction 18, 150–154. https://doi.org/10.1154/1.1536926

Hower, J.C., 2012. Petrographic examination of coal-combustion fly ash. International Journal of Coal Geology 92, 90–97. https://doi.org/10.1016/j.coal.2011.12.012

Hower, J.C., Groppo, J.G., Graham, U.M., Ward, C.R., Kostova, I.J., Maroto-Valer, M.M., Dai, S., 2017. Coal-derived unburned carbons in fly ash: A review. International Journal of Coal Geology 179, 11–27. https://doi.org/10.1016/j.coal.2017.05.007

Hubbard, F.H., McGill, R.J., Dhir, R.K., Ellis, M.S., 1984. Clay and pyrite transformations during ignition of pulverised coal. Mineralogical Magazine 48, 251–256. https://doi.org/10.1180/minmag.1984.048.347.09

Hunter, D.R., 1962. The mineral resources of Swaziland (Report), Bulletin No. 2. Geological Survey and Mines Department, Mbabane.

Hunter, M., 2018. South African fly ash used to help build the continent’s longest suspension bridge. Ash at work 2018, 18-19.

178 | P a g e

Hurai, V., Huraiová, M., 2015. Fluid inclusions in apatite indicate low-temperature, metasomatic-infiltration origin of the Evate carbonatite deposit (Mozambique), in: European Current Research On Fluid Inclusions (ECROFI-XXIII). Leeds.

Hurai, V., Paquette, J.-L., Huraiová, M., Slobodník, M., Hvožďara, P., Siegfried, P., Gajdošová, M., Milovská, S., 2017. New insights into the origin of the Evate apatite- iron oxide-carbonate deposit, Northeastern Mozambique, constrained by mineralogy, textures, thermochronometry, and fluid inclusions. Ore Geology Reviews 80, 1072– 1091. https://doi.org/10.1016/j.oregeorev.2016.09.017

Hurt, R.H., Gibbins, J.R., 1995. Residual carbon from pulverized coal fired boilers: 1. Size distribution and combustion reactivity. Fuel 74, 471–480. https://doi.org/10.1016/0016-2361(95)98348-I

Hurt, R.H., Davis, K.A., Yang, N.Y.C., Headley, T.J., Mitchell, G.D., 1995. Residual carbon from pulverized-coal-fired boilers. 2. Morphology and physicochemical properties. Fuel 74, 1297–1306. https://doi.org/10.1016/0016-2361(95)00100-J

Hwang, J., Li, Z., 2000. Control of mercury emissions using unburned carbon from combustion by-products. Patent US 6027551.

Hwang, J.Y., Sun, X., Li, Z., 2002. Unburned carbon from fly ash for mercury adsorption: I. Separation and characterization of unburned carbon. Journal of Minerals and Materials Characterization and Engineering 1, 39–60. https://doi.org/10.4236/jmmce.2002.11004

ICCP (International Committee for Coal and Organic Petrology)., 1998. The new vitrinite classification (ICCP system 1994). Fuel 77, 349-358. https://doi.org/10.1016/S0016- 2361(98)80024-0

ICCP (International Committee for Coal and Organic Petrology)., 2001. The new inertinite classification (ICCP system 1994). Fuel 80, 459-471. https://doi.org/10.1016/S0016- 2361(00)00102-2

Inagaki, M., 2013. Advanced Carbon Materials, In: Somiya, S. (Ed.), Handbook of Advance Ceramics: Materials, Applications, Processing, and Properties. Academic Press, Waltham, MA, pp. 25-222. https://doi.org/10.1016/C2010-0-66261-4

Insley, H., Ewell, R.H., 1935. Thermal behaviour of the kaolin minerals. Journal of Research of the National Bureau of Standards 14, 615-627.

179 | P a g e

ISO (International Organisation for Standardisation). 1983. Coal – Determination of mineral matter. Geneva: ISO. (ISO 602).

ISO (International Organisation for Standardisation). 1992. Hard coal - Determination and presentation of float sink characteristics - General directions for apparatus and procedures. Geneva: ISO. (ISO 7936).

ISO (International Organisation for Standardisation). 1999. Solid mineral fuels – Hard coal – Determination of moisture in the general analysis test sample by drying in nitrogen. Geneva: ISO. (ISO 11722).

ISO (International Organisation for Standardisation). 2001. Solid mineral fuels – Determination of total carbon, hydrogen and nitrogen – Instrumental methods. Geneva: ISO. (ISO 12902).

ISO (International Organisation for Standardisation). 2001. Hard coal and coke – Mechanical sampling – Part 4: Coal – Preparation of test samples. Geneva: ISO. (ISO 13909-4).

ISO (International Organisation for Standardisation). 2006. Solid mineral fuels – Determination of sulfur by IR spectrometry. Geneva: ISO. (ISO 19579).

ISO (International Organisation for Standardisation). 2009. Solid mineral fuels – Determination of gross calorific value by the bomb calorimetric method and calculation of net calorific value. Geneva: ISO. (ISO 1928).

ISO (International Organisation for Standardisation). 2009. Methods for the petrographic analysis of coals. Part 2: Methods of preparing coal samples. Geneva: ISO. (ISO 7404-2).

ISO (International Organisation for Standardisation). 2009. Methods for the petrographic analysis of coals. Part 3: Method of determining maceral group composition. Geneva: ISO. (ISO 7404-3).

ISO (International Organisation for Standardisation). 2009. Methods for the petrographic analysis of coals. Part 5: Method of determining microscopically the reflectance of vitrinite. Geneva: ISO. (ISO 7404-5).

ISO (International Organisation for Standardisation). 2010. Hard coal and coke – Determination of volatile matter. Geneva: ISO. (ISO 562).

ISO (International Organisation for Standardisation). 2010. Solid mineral fuels –

180 | P a g e

Determination of ash. Geneva: ISO. (ISO 1171).

ISO (International Organisation for Standardisation). 2010. Solid mineral fuels – Determination of total carbon, hydrogen and nitrogen content – Instrumental method. Geneva: ISO. (ISO 29541).

ISO (International Organisation for Standardisation). 2016. Hard coal and coke – Mechanical sampling – Part 2: Coal – Sampling from moving streams. Geneva: ISO. (ISO 13909- 2).

Iwashita, N., Park, C.R., Fujimoto, H., Shiraishi, M., Inagaki, M., 2004. Specification for a standard procedure of X-ray diffraction measurements on carbon materials. Carbon 42, 701–714. https://doi.org/10.1016/j.carbon.2004.02.008

Jacob, D., Jagoutz, E., 1994. A diamond-graphite bearing eclogitic xenolith from Roberts Victor (South Africa): Indications for petrogenesis from Pb-, Nd-, and Sr-isotopes, in: Fifth International Kimberlite Conference: Extended Abstracts. Companhia de Pesquisa de Recursos Minerais, Araxá, pp. 190-192. https://doi.org/10.29173/ikc2508

Janardhan, A.S., 1999. Southern Granulite Terrain, South of the Palghat-Cauvery Shear Zone: Implications for India-Madagascar Connection. Gondwana Research 2, 463– 469. https://doi.org/10.1016/S1342-937X(05)70284-7

Jeffrey, L.S., 2005. Characterization of the coal resources of South Africa. Journal of the Southern African Institute of Mining and Metallurgy 105, 95–102.

Jeffrey, L., Henry, G., McGill, J.M., 2014. Introduction to South African Coal Mining and Exploration, in: Van Schoor, M., Fourie, C.J.S. (Eds.), A Guide for Applying Geophysics to Coal Mining Problems in South Africa. Struik Publishers, Johannesburg, pp. 4-21.

Jezierski, S., 2015. Determination of different carbon species in solids – Using the real chemistry to obtain the right results [WWW Document]. https://www.envirotech- online.com/article/environmental-laboratory/7/analytik-jena-ag/determination-of- different-carbon-species-in-solids-ndash-using-the-real-chemistry-to-obtain-the-right- results/1938 (accessed 7.3.19).

Jourdan, P.P., 1990. Strategies for the regional planning of the minerals industry in southern Africa: The case of the SADCC (Thesis). Leeds: University of Leeds.

181 | P a g e

Kalenga, P. M., Cukrowska, E., Tutu, H., Chimuka, L., 2011. Characterization of South African coal for metals, inorganic and organic sulfur compounds. South African Journal of Chemistry 64, 254–262.

Kalyoncu, R.S., 1996. Graphite (Report), Minerals Information. U.S. Geological Survey, Reston, VA.

Kalyoncu, R.S., 1998. Graphite (Report), Minerals Information. U.S. Geological Survey, Reston, VA.

Karkhanis, S.N., 1975. Evidence of graphite from the Onverwacht series Precambrian cherts, Swaziland system of South Africa. Chemical Geology 16, 233–238. https://doi.org/10.1016/0009-2541(75)90031-5

Katagiri, G., Ishida, H., Ishitani, A., 1988. Raman spectra of graphite edge planes. Carbon 26, 565–571. https://doi.org/10.1016/0008-6223(88)90157-1

Kataka, M.O., Matiane, A.R., Odhiambo, B.D.O., 2018. Chemical and mineralogical characterization of highly and less reactive coal from Northern Natal and Venda- Pafuri coalfields in South Africa. Journal of African Earth Sciences 137, 278–285. https://doi.org/10.1016/j.jafrearsci.2017.10.019

Kawashima, Y., Katagiri, G., 1999. Observation of the out-of-plane mode in the Raman scattering from the graphite edge plane. Physical Review B 59, 62–64. https://doi.org/10.1103/PhysRevB.59.62

Kelly, E.G., Spottiswood, D.J., 1982. Introduction to Mineral Processing. Wiley, New York.

Kerr, A., Eglington, B.M., Milne, G.C., 1987. The geochemistry and geochronology of some Proterozoic Granitoid rocks from the Natal Structural and Metamorphic Province, southeastern Africa. Revista Brasileira de Geociências 17, 543-549.

Khoza, M. 2016. Steamboat and Grootegeluk graphite deposits in South Africa [Personal communications]. 13 Oct., Johannesburg.

Kouketsu, Y., Mizukami, T., Mori, H., Endo, S., Aoya, M., Hara, H., Nakamura, D., Wallis, S., 2014. A new approach to develop the Raman carbonaceous material geothermometer for low-grade metamorphism using peak width. Island Arc 23, 33– 50. https://doi.org/10.1111/iar.12057

Krauss, U.H., Schmidt, H.W., Taylor, H.A., Sutphin, D.M., 1988. International strategic

182 | P a g e

minerals inventory summary report – natural graphite (Report), U.S. Geological Survey circular 930-H. U.S. Geological Survey, Denver, CO.

Krishnamoorthy, V., Pisupati, S.V., 2015. A critical review of mineral matter related issues during gasification of coal in fixed, fluidized, and entrained flow gasifiers. Energies 8, 10430–10463. https://doi.org/10.3390/en80910430

Kruger, R.A., Krueger, J.E., 2005. Historical development of coal ash utilisation in South Africa, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, Lexington, KY. http://www.flyash.info/

Kruger, R., 2017. International developments in ash utilisation, in: One-day Workshop on Coal Ash – Production and Utilisation with a Multitude of Opportunities. Johannesburg.

Kukla, P.A., 1988. New aspects of the sedimentology and structure of the Kuiseb formation in the western Khomas Trough, Damara Orogen, Swa/Namibia. Communications of the Geological Survey of South West Africa / Namibia 4, 37-48.

Kukla, P.A., 1992. Tectonics and sedimentation of a late Proterozoic Damaran convergent continental margin, Khomas Hochland, Central Namibia (Report), Memoir 12 of the Geological Survey of Namibia. Geological Survey of Namibia, Windhoek.

Külaots, I., Hurt, R.H., Suuberg, E.M., 2004. Size distribution of unburned carbon in coal fly ash and its implications. Fuel 83, 223–230. https://doi.org/10.1016/S0016- 2361(03)00255-2

Kwiecińska, B., Petersen, H.I., 2004. Graphite, semi-graphite, natural coke, and natural char classification—ICCP system. International Journal of Coal Geology 57, 99–116. https://doi.org/10.1016/j.coal.2003.09.003

Lahfid, A., Beyssac, O., Deville, E.P., Negro, F., Chopin, C., Goffé, B., 2010. Evolution of the Raman spectrum of carbonaceous material in low-grade metasediments of the Glarus Alps (Switzerland). Terra Nova 22, 354-360. https://doi.org/10.1111/j.1365- 3121.2010.00956.x

Landis, C.A., 1971. Graphitization of dispersed carbonaceous material in metamorphic rocks. Contributions to Mineralogy and Petrology 30, 34–45. https://doi.org/10.1007/BF00373366

183 | P a g e

Lee, S.-J., Cho, H.-C., Kwon, J.-H., 2012. Beneficiation of coal pond ash by physical separation techniques. Journal of Environmental Management 104, 77–84. https://doi.org/10.1016/j.jenvman.2012.03.034

Lehtonen, M.I., Manninen, T.E.T., Schreiber, U.M., 1996. Report: Lithostratigraphy of the area between the Swakop, Khan and lower Omaruru Rivers, Namib Desert. Communications of the Geological Survey of Namibia 11, 71-82.

Lester, E., Alvarez, D., Borrego, A.G., Valentim, B., Flores, D., Clift, D.A., Rosenberg, P., Kwiecinska, B., Barranco, R., Petersen, H.I., Mastalerz, M., Milenkova, K.S., Panaitescu, C., Marques, M.M., Thompson, A., Watts, D., Hanson, S., Predeanu, G., Misz, M., Wu, T., 2010. The procedure used to develop a coal char classification— Commission III Combustion Working Group of the International Committee for Coal and Organic Petrology. International Journal of Coal Geology 81, 333–342. https://doi.org/10.1016/j.coal.2009.10.015

Li, Z., Sun, X., Luo, J., Hwang, J.Y., 2002. Unburned carbon from fly ash for mercury adsorption: II. Adsorption isotherms and mechanisms. Journal of Minerals and Materials Characterization and Engineering 1, 79–96. https://doi.org/10.4236/jmmce.2002.12006

Li, X., Hayashi, J., Li, C.-Z., 2006a. Volatilisation and catalytic effects of alkali and alkaline earth metallic species during the pyrolysis and gasification of Victorian brown coal. Part VII. Raman spectroscopic study on the changes in char structure during the catalytic gasification in air. Fuel 85, 1509–1517. https://doi.org/10.1016/j.fuel.2006.01.011

Li, X., Hayashi, J., Li, C.-Z., 2006b. FT-Raman spectroscopic study of the evolution of char structure during the pyrolysis of a Victorian brown coal. Fuel 85, 1700–1707. https://doi.org/10.1016/j.fuel.2006.03.008

Li, J., Maroto-Valer, M.M., 2012. Computational and experimental studies of mercury adsorption on unburned carbon present in fly ash. Carbon 50, 1913–1924. https://doi.org/10.1016/j.carbon.2011.12.042

Li, K., Rimmer, S.M., Liu, Q., 2018. Geochemical and petrographic analysis of graphitized coals from Central Hunan, China. International Journal of Coal Geology 195, 267– 279. https://doi.org/10.1016/j.coal.2018.06.009

Liu, J., Walter, J.M., Weber, K., 2002. Fluid-enhanced low-temperature plasticity of calcite

184 | P a g e

marble: Microstructures and mechanisms. Geology 30, 787–790. https://doi.org/10.1130/0091-7613(2002)030<0787:FELTPO>2.0.CO;2

Lu, L., Sahajwalla, V., Kong, C., Harris, D., 2001. Quantitative X-ray diffraction analysis and its application to various coals. Carbon 39, 1821–1833. https://doi.org/10.1016/S0008-6223(00)00318-3

Lu, L., Kong, C., Sahajwalla, V., Harris, D., 2002. Char structural ordering during pyrolysis and combustion and its influence on char reactivity. Fuel 81, 1215–1225. https://doi.org/10.1016/S0016-2361(02)00035-2

Lu, Y., Rostam-Abadi, M., Chang, R., Richardson, C., Paradis, J., 2007. Characteristics of fly ashes from full-scale coal-fired power plants and their relationship to mercury adsorption. Energy Fuels 21, 2112–2120. https://doi.org/10.1021/ef070145s

Luo, J., Hein, A.M., Hwang, J.Y., 2004. Adsorption of vapor phase mercury on various carbon. Journal of Minerals and Materials Characterisation and Engineering 3, 13-22. http://dx.doi.org/10.4236/jmmce.2004.31002

Luque, F.J.F., Huizenga, J.-M., Crespo-Feo, E., Wada, H., Ortega, L.C., Barrenechea, J.M.F., 2014. Vein graphite deposits: geological settings, origin, and economic significance. Mineralium Deposita 49, 261–277. https://doi.org/10.1007/s00126-013- 0489-9

Macgregor, A.M., 1940. A Pre-Cambrian Algal limestone in Southern Rhodesia. South African Journal of Geology 43, 9–15.

Mahlaba, J.S., 2011. The influence of brine chemistry on the workability, strength development and mineralogy of fly ash pastes (Thesis). Pretoria: University of Pretoria.

Mahlaba, J.S., Kearsley, E.P., Kruger, R.A., 2011. Physical, chemical and mineralogical characterisation of hydraulically disposed fine coal ash from SASOL Synfuels. Fuel 90, 2491–2500. https://doi.org/10.1016/j.fuel.2011.03.022

Maire, J., Méring, J. 1970. Graphitization of soft carbons, in: Walker, P.L. Jr., (Ed.), Chemistry and physics of carbon – Volume 6. Marcel Dekker Inc., New York, pp. 125-189.

Malaza, N., 2014. Basin analysis of the Soutpansberg and Tuli coalfields, Limpopo province

185 | P a g e

of South Africa (Thesis). Fort Hare: University of Fort Hare.

Malhotra, V.M., 2008. Mineral Admixtures, in: Nawy, E.G. (Ed.), Concrete Construction Engineering Handbook. CRC Press, Florida, pp. 2.1-3.1.

Malumbazo, N., 2011. Chemical and physical structural studies on two inertinite-rich lump coals (Thesis). Johannesburg: University of the Witwatersrand.

Mandlate, F., 2018. Graphite exploration might start soon in Niassa, Mozambique [WWW Document]. https://clubofmozambique.com/news/graphite-exploration-may-start- soon-in-niassa-mozambique/ (accessed 7.3.19).

Manhica, A., 1991. Phosphates and apatite of Mozambique. Fertilizer research 30, 167–175. https://doi.org/10.1007/BF01048649

Manhica, A.d.S.T., 1998. The geology of the Mozambique belt and the Zimbabwe craton around Manica, Western Mozambique (Thesis). Pretoria: University of Pretoria.

Maroto-Valer, M.M., Taulbee, D.N., Hower, J.C., 1999a. Novel separation of the differing forms of unburned carbon present in fly ash using Density Gradient Centrifugation. Energy Fuels 13, 947–953. https://doi.org/10.1021/ef990029s

Maroto-Valer, M.M., Taulbee, D.N., Schobert, H.H., Hower, J.C., 1999b. Characterization of differing forms of unburned carbon present in fly ash, in: 1999 International Ash Utilization Symposium. University of Kentucky Center for Applied Energy Research, Lexington, KY. http://www.flyash.info/

Maroto-Valer, M.M., Zhang, Y., Lu, Z., Andrésen, J.M., Schobert, H.H., 2002. Development of Value-Added Products from Fly Ash Carbons, in: Maroto-Valer, M.M., Song, C., Soong, Y. (Eds.), Environmental Challenges and Greenhouse Gas Control for Fossil Fuel Utilization in the 21st Century. Springer Science+Business Media, New York, pp. 431-444. https://doi.org/10.1007/978-1-4615-0773-4

Marsh, H., Rodríguez-Reinoso, F., 2006. Activated Carbon. Elsevier Science, Amsterdam. https://doi.org/10.1016/B978-0-08-044463-5.X5013-4

Martin, H., 1965. The Precambrian Geology of South West Africa and Namaqualand. Precambrian Research Unit, University of Cape Town, Cape Town.

Marques, M., Suárez-Ruiz, I., Flores, D., Guedes, A., Rodrigues, S., 2009. Correlation between optical, chemical and micro-structural parameters of high-rank coals and

186 | P a g e

graphite. International Journal of Coal Geology 77, 377–382. https://doi.org/10.1016/j.coal.2008.06.002

Massey, N.W.D., 1973. Resources inventory of Botswana: Industrial rocks and minerals (Report), Mineral Resources Report No. 3. Government Printer, Gabarone.

Master, S., 1996. Post-conference field excursion No 2: The Palaeoproterozoic Magondi Mobile Belt, NW Zimbabwe (Report), IGCP Project 363: Lower Proterozoic of sub- equatorial Africa.

Master, S., Bekker, A., Hofmann, A., 2010. A review of the stratigraphy and geological setting of the Palaeoproterozoic Magondi Supergroup, Zimbabwe – Type locality for the Lomagundi carbon isotope excursion. Precambrian Research 182, 254–273. https://doi.org/10.1016/j.precamres.2010.08.013

Master, S., 2013. An introduction to the geology and geochronology of the Dete-Kamativi Inlier, in: Geological Society of Zimbabwe Annual Summer Symposium. Victoria Falls.

Mathews, C., 2016. Fly ash: Hazardous heaps mount up [WWW Document]. https://www.businesslive.co.za/fm/fm-fox/2016-04-14-fly-ash-hazardous-heaps- mount-up/ (accessed 7.3.19).

Mathez, E.A., Dietrich, V.J., Holloway, J.R., Boudreau, A.E., 1989. Carbon distribution in the Stiliwater Complex and evolution of vapor during crystallization of Stillwater and Bushveld magmas. Journal of Petrology 30, 153–173. https://doi.org/10.1093/petrology/30.1.153

Matjie, R.H., Ginster, M., Van Alphen, C., Sobiecki, A., 2005. Detailed characterisation of Sasol ashes, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, Lexington, KY. http://www.flyash.info/

Matjie, R.H., van Alphen, C., Pistorius, P.C., 2006. Mineralogical characterisation of Secunda gasifier feedstock and coarse ash. Minerals Engineering 19, 256–261. https://doi.org/10.1016/j.mineng.2005.06.010

Matjie, R.H., Li, Z., Ward, C.R., French, D., 2008. Chemical composition of glass and crystalline phases in coarse coal gasification ash. Fuel 87, 857–869. https://doi.org/10.1016/j.fuel.2007.05.050

187 | P a g e

Matjie, R.H., French, D., Ward, C.R., Pistorius, P.C., Li, Z., 2011. Behaviour of coal mineral matter in sintering and slagging of ash during the gasification process. Fuel Processing Technology 92, 1426–1433. https://doi.org/10.1016/j.fuproc.2011.03.002

McConnell, J.D.C., Fleet, S.G., 1970. Electron optical study of the thermal decomposition of kaolinite. Clay Minerals 8, 279-290.

McCourt, S., Armstrong, R.A., Grantham, G.H., Thomas, R.J., 2006. Geology and evolution of the Natal belt, South Africa. Journal of African Earth Sciences 46, 71–92. https://doi.org/10.1016/j.jafrearsci.2006.01.013

McLennan, A.R., Bryant, G.W., Bailey, C.W., Stanmore, B.R., Wall, T.F., 2000. Index for iron-based slagging for pulverized coal firing in oxidizing and reducing conditions. Fuel 14, 349–354. https://doi.org/ 10.1021/ef990127d

McMillan, K., Campbell, Q.P., Le Roux, M., Wagner, N., Dorland, H.C., 2015. Composition, component size, fracture patterns and product liberation – Washability behaviour of coals from different coalfields of the world, in: Proceedings of the South African Coal Processing Society Conference. South African Coal Processing Society, Secunda.

Ministry of Mines and Energy., 1992. The mineral resources of Namibia (Report).

Ministry of Mines and Mining Development., 2018. Mineral potential: Procedures and requirements of acquiring licences and permits in terms of the mines and minerals act (Report), Chapter 21:05. Government of Zimbabwe, Zimbabwe.

Mitchell, C.J., 1993. Industrial minerals laboratory manual: Flake graphite (Report), British Geological Survey Technical Report WG/92/30. NERC, Nottingham.

Mitchell, R.E., Akanetuk, A.E.J., 2002. Mechanisms of pyrite oxidation to non-slagging species (ReportNational Energy Technology Laboratory, California.

Mohebbi, M., Rajabipour, F., Scheetz, B.E., 2015. Reliability of loss on ignition (LOI) test for determining the unburned carbon content in fly ash, in: World of Coal Ash (WOCA) Conference. University of Kentucky Center for Applied Energy Research, Nashville, TN. http://www.flyash.info/

Moss, R., 2016. Technical report on the Aukam graphite deposit, Bethanie district, Karas region, Namibia (Report). CKR Carbon Corporation, Toronto.

Mounteney, I., 2011. The use of lithium heteropolytungstate as an alternative to bromoform

188 | P a g e

for heavy media separations (Report), British Geological Survey International Report IR/11/049. British Geological Survey, Nottigham.

Muchemwa, E., 1987. Graphite in Zimbabwe (Report), Mineral Resources Series No. 20. Zimbabwe Geological Survey, Harare.

Mukoyoshi, H., Sakaguchi, A., Otsuki, K., Hirono, T., Soh, W., 2006. Co-seismic frictional melting along an out-of-sequence thrust in the Shimanto accretionary complex. Implications on the tsunamigenic potential of splay faults in modern subduction zones. Earth and Planetary Science Letters 245, 330–343. https://doi.org/10.1016/j.epsl.2006.02.039

Müller, B.G.J., 2000. The evolution and significance of the Bongolava-Ranotsara shear zone, Madagascar (Thesis). Johannesburg: University of Johannesburg.

Mzamo, P., 2018. Zimbabwe graphite miners strike at former German-owned mine [WWW Document]. URL https://miningnews.co.za/2018/01/09/zimbabwe-graphite-miners- strike-at-former-german-owned-mine/#.%20D (accessed 7.3.19).

Nasdala, L., Smith, D.B., Kaindl, R., Ziemann, M., 2004. Raman spectroscopy: Analytical perspectives in mineralogical research. EMU Notes in Mineralogy 6, 1-63. https://doi.org/10.1180/EMU-notes.6.7

Nédélec, A., Ralison, B., Bouchez, J.-L., Grégoire, V., 2000. Structure and metamorphism of the granitic basement around Antananarivo: A key to the Pan-African history of central Madagascar and its Gondwana connections. Tectonics 19, 997–1020. https://doi.org/10.1029/2000TC900001

NextSources Materials Inc., 2017. Graphite [WWW Document]. URL http://www.nextsourcematerials.com/graphite/ (accessed 7.3.19).

Ndhlalose, M., Malumbazo, N., Wagner, N., 2015. Coal quality and uranium distribution in Springbok Flats Coalfield samples. Journal of the Southern African Institute of Mining and Metallurgy 115, 1167–1174. https://doi.org/10.17159/2411- 9717/2015/v115n12a4

Nemanich, R.J., Solin, S.A., 1979. First- and second-order Raman scattering from finite-size crystals of graphite. Physical Review B 20, 392–401. https://doi.org/10.1103/PhysRevB.20.392

189 | P a g e

Nicholson, D.M., Mathez, E.A., 1991. Petrogenesis of the Merensky Reef in the Rustenburg section of the Bushveld Complex. Contributions to Mineralogy and Petrology 107, 293–309. https://doi.org/10.1007/BF00325100

Nkala, O., 2016. Graphite mine seeking $5.9m for plant upgrades, mechanisation and exploration [WWW Document]. URL https://www.miningweekly.com/article/graphite- mine-seeking-59m-for-plant-upgrades-mechanisation-and-exploration-2016-06-24 (accessed 7.3.19).

Norrish, K., Hutton, J.T., 1969. An accurate X-ray spectrographic method for the analysis of a wide range of geological samples. Geochimica et Cosmochimica Acta 33, 431-453. https://doi.org/10.1016/0016-7037(69)90126-4

Nörtemann, M.F-J., Mücke, A., Weber, K., Meinert, L.D. 2000. Mineralogy of the Navachab skarn deposit, Namibia: and unusual Au-bearing skarn in high-grade metamorphic rocks. Communications of the Geological Survey of Namibia 12, 169-177.

Oberlin, A., 1984. Carbonization and graphitization. Carbon 22, 521–541. https://doi.org/10.1016/0008-6223(84)90086-1

Okolo, G.N., 2010. The effects of chemical and physical properties of chars derived from inertinite–rich, high ash coals on gasification reaction kinetics (Thesis). Potchefstroom: North-West University.

Openshaw, S.C., 1992. Utilization of coal fly ash (Thesis). University of Florida, Gainesville.

Otto, J.D.T., 1973. The geology and petrology of the Marble Delta (Thesis). Stellenbosch: Stellenbosch University.

Otto, E., 2011. Growing demand from electric vehicles & mobile electronics (Report). Cormark Securities, Toronto.

Padhy, S., 2017. Graphite in Africa: Madagascar, Mozambique, Namibia and Tanzania [WWW Document]. URL https://investingnews.com/daily/resource-investing/battery- metals-investing/graphite-investing/graphite-africa-madagascar-mozambique- namibia-tanzania-energizer/ (accessed 7.3.19).

Parthasarathy, G., Sreedhar, B.H., Chetty, T., 2006. Spectroscopic and X-ray diffraction studies on fluid deposited rhombohedral graphite from the Eastern Ghats Mobile Belt, India. Current Science 90, 995-1000.

190 | P a g e

Pasteris, J.D., 1989. In situ analysis in geological thin-sections by laser Raman microprobe spectroscopy: A cautionary note. Applied Spectroscopy 43, 567–570. https://doi.org/10.1366/0003702894202878

Payá, J., Monzó, J., Borrachero, M.V., Perris, E., Amahjour, F., 1998. Thermogravimetric methods for determining carbon content in fly ashes. Cement and Concrete Research 28, 675–686. https://doi.org/10.1016/S0008-8846(98)00030-1

Pearson, D.G., Boyd, F.R., Field, S.W., Pasteris, J.D., Haggerty, S.E., Nixon, P.H., 1991. Graphite-Bearing Peridotites from the Kaapvaal Craton their Carbon Isotopic Compositions and Implications for Peridotite Thermobarometry, in: Fifth International Kimberlite Conference: Extended Abstracts. Companhia de Pesquisa de Recursos Minerais, Araxá, pp. 323–325. https://doi.org/10.29173/ikc2556

Pedersen, K.H., Jensen, A.D., Skjøth-Rasmussen, M.S., Dam-Johansen, K., 2008. A review of the interference of carbon containing fly ash with air entrainment in concrete. Progress in Energy and Combustion Science 34, 135–154. https://doi.org/10.1016/j.pecs.2007.03.002

Pekkala, Y., Lehto, T., Mäkitie, H., 2008. GTK Consortium geological surveys in Mozambique 2002-2007 (Report), Special Paper 48. Geological Survey of Finland, Espoo.

Pencil Hill, 2015. Pencil Hill [WWW Document]. YouTube. URL https://www.youtube.com/watch?v=TeyPVTr9JIc (accessed 7.4.19).

Pickel, W., Kus, J., Flores, D., Kalaitzidis, S., Christanis, K., Cardott, B.J., Misz-Kennan, M., Rodrigues, S., Hentschel, A., Hamor-Vido, M., Crosdale, P., Wagner, N., ICCP., 2017. Classification of liptinite – ICCP System 1994. International Journal of Coal Geology 169, 40-61. https://doi.org/ 10.1016/j.coal.2016.11.004

Pierson, H.O., 1993. Handbook of Carbon, Graphite, Diamond and Fullerenes – Properties, Processing and Applications. Noyes Publications, Park Ridge, NJ.

PN-ISO (Polska Norma). 1999. Węgiel kamienny -- Oznaczanie i przedstawianie charakterystyk wzbogacania grawitacyjnego -- Ogólne wytyczne dotyczące aparatury i procedur. Poland: PN. (PN-ISO 7936).

Praekelt, H.E., Schoch, A.E., 1997. Stratigraphical synthesis of the metasediments of the Aggeneys terrane in the Namaqua Mobile Belt. South African Journal of Geology

191 | P a g e

100, 85–100.

Praekelt, H.E., Schoch, A.E., Visser, J.N.J., 1997. The metasediments of the Aggeneys terrane in the Namaqua Mobile Belt: sedimentary response to extensional- compressional variations in a continental environment. South African Journal of Geology 100, 101–110.

Pretorius, C.C., Boshoff, H.P., Pinheiro, H.J., 2002. Analyses of coal product samples of South African collieries 2001-2002 (Report), Bulletin 114. Department of Minerals and Energy, Pretoria.

Pretorius, I., 2015. Impacts and control of coal-fired power station emissions in South Africa (Thesis). Potchefstroom: North-West University.

Prévost, X.M., 2003. SA coal resources and reserves, a present-day outlook, in: 31st Internatinal Symposium on Applications of Computers and Operations Research in the Minerals Industries. The South African Institute for Mining and Metallurgy, Cape Town. http://www.saimm.co.za/Conferences/Apcom2003/

Rajoo, P. 2017. Boiler operations at Eskom. [Personal communications]. 13 Apr., Johannesburg.

Rakovan, J., Jaszczak, J.A. 2002a. Graphite with growth spirals from Arises River Marbles, Wlotzkas Baken, western Namibia. The Mineralogical Record 33, 79.

Rakovan, J., Jaszczak, J.A., 2002b. Multiple length scale growth spirals on metamorphic graphite {001} surfaces studied by atomic force microscopy. American Mineralogist 87, 17–24. https://doi.org/10.2138/am-2002-0103

Randrianasolo, L.G., 1996. Structure and tectonics of the Ampanihy Group in the vicinity of the Ankafotia and Saririaky anorthosite bodies, Southwestern Madagascar (Thesis). Johannesburg: University of Johannesburg.

Reimer, T.O., 1984. Graphite in Precambrian rocks of Southern Africa: implications on the carbon content of metamorphic rocks. Precambrian Research 26, 223–234. https://doi.org/10.1016/0301-9268(84)90002-0

Reimer, T.O., 1978. Stromatolitic dolomites in the late Precambrian Malmesbury Group, Cape Province, South Africa. Sedimentary Geology 20, 29–40. https://doi.org/10.1016/0037-0738(78)90047-7

192 | P a g e

Reynolds-Clausen, K. 2016a. Eskom coal supply, ash management, and ash quality. [Personal communications]. 14 Jul., Johannesburg.

Reynolds-Clausen, K. 2016b. Eskom coal quality. [Personal communications]. 28 Jul., Johannesburg.

Reynolds-Clausen, K., Singh, N., 2019. South Africa’s power producer’s revised coal ash strategy and implementation progress. Coal Combustion and Gasification Products 11, 10-17. http://www.flyash.info/

Robinson, D.N., 1977. Diamond and graphite in eclogite xenoliths from kimberlite, in: Second International Kimberlite Conference: Extended Abstracts. American Geophysical Union, Santa Fe, pp. 279–281. https://doi.org/10.29173/ikc1021

Robinson, D.N., Gurney, J.J., Shee, S.R., 1984. Diamond Eclogite and Graphite Eclogite Xenoliths from Orapa, Botswana. Developments in Petrology 11, 11–24. https://doi.org/10.1016/B978-0-444-42274-3.50008-1

Rocha, A., Maria, R., Waite, U.S., Cassimo, U.A., Falinski, K., Yost, R., 2017. Improving grain legume yields using local Evate rock phosphate in Gùrué District, Mozambique. African Journal of Agricultural Research 12, 1889–1896. https://doi.org/10.5897/AJAR2017.12331

Russell, N.V., Méndez, L.B., Wigley, F., Williamson, J., 2002. Ash deposition of a Spanish anthracite: effects of included and excluded mineral matter. Fuel 81, 657–663. https://doi.org/ 10.1016/S0016-2361(01)00155-7

Sadezky, A., Muckenhuber, H., Grothe, H., Niessner, R., Pöschl, U., 2005. Raman microspectroscopy of soot and related carbonaceous materials: Spectral analysis and structural information. Carbon 43, 1731–1742. https://doi.org/10.1016/j.carbon.2005.02.018

Şahbaz, O., Öteyaka, B., Kelebek, Ş., Uçar, A., Demir, U., 2008. Separation of unburned carbonaceous matter in bottom ash using Jameson cell. Separation and Purification Technology 62, 103–109. https://doi.org/10.1016/j.seppur.2008.01.005

Sandmann, D., Haser, S., Gutzmer, J., 2014. Characterisation of graphite by automated mineral liberation analysis. Mineral Processing and Extractive Metallurgy 123, 184– 189. https://doi.org/10.1179/1743285514Y.0000000063

193 | P a g e

Santos, A., Abagiu, T., Anghelescu, L., Badenhorst, C., Bialecka, B., Calus-Moszk, J., Cempa, M., Cruceru, M., Flores, D., Freire, C., Guedes, A., Klupa, A., Martinez, J.L., Popescu, L., Predeanu, G., Ribeiro, J., Slavescu, V., Wagner, N., Wrana, A., Valentim, B., 2017. Project CHARPHITE: presentation and up to date of Porto University results, in: Jornadas Do ICT.

Sarma, D., Hattle, A., 2014. Environmental impact assessment for the proposed development of the Okanjande graphite mine and exploration activities (Report), Specialist Study: Hydrological and Hydrogeological Assessments Package 2. Gecko Graphite, Windhoek.

Sasol., 2018. Sasol [WWW Document]. URL https://www.sasol.com/. (accessed 7.3.19).

Schidlowski, M. 1967. Note on graphite in the Witwatersrand conglomerates. South African Journal of Geology 70, 65-66.

Schito, A., Romano, C., Corrado, S., Grigo, D., Poe, B., 2017. Diagenetic thermal evolution of organic matter by Raman spectroscopy. Organic Geochemistry 106, 57–67. https://doi.org/10.1016/j.orggeochem.2016.12.006

Schlüter, T., 2008. Geological Atlas of Africa: With Notes on Stratigraphy, Tectonics, Economic Geology, Geohazards, Geosites and Geoscientific Education of Each Country, second ed. Springer-Verlag, Berlin Heidelberg. https://doi.org/10.1007/978- 3-540-76373-4

Schulze, D.J., Valley, J.W., Viljoen, K.S., Stiefenhofer, J., Spicuzza, M., 1997. Carbon isotope composition of graphite in mantle Eclogites. The Journal of Geology 105, 379–386. https://doi.org/10.1086/515933

Schumacher, B.A., 2002. Methods for the determination of total organic carbon (TOC) in soils and sediments (Report). United States Environmental Protection Agnecy, Las Vegas.

Scogings, A.J., 1991. Alkaline intrusives from the Tugela terrane, Natal Metamorphic Province (Thesis). Durban: University of KwaZulu-Natal.

Seehra, M.S., Pavlovic, A.S., 1993. X-Ray diffraction, thermal expansion, electrical conductivity, and optical microscopy studies of coal-based graphites. Carbon 31, 557–564. https://doi.org/10.1016/0008-6223(93)90109-N

194 | P a g e

Seriti., 2017. Fact Sheet 2017 [WWW Document]. URL https://seritiza.com/wp- content/uploads/2017/10/Seriti-Fact-sheet-updated-with-91.pdf (accessed 7.3.19).

Sforna, M.C., van Zuilen, M.A., Philippot, P., 2014. Structural characterization by Raman hyperspectral mapping of organic carbon in the 3.46 billion-year-old Apex chert, Western Australia. Geochimica et Cosmochimica Acta 124, 18–33. https://doi.org/10.1016/j.gca.2013.09.031

Sha, X., 2009. Coal gasification. Encyclopaedia of Life Support Systems.

Shah, K.V., Cieplik, M.K., Vuthaluru, H.B., 2015. A Review on Ash Formation During Pulverized Fuel Combustion: State of Art and Future Research Needs, in: Ravindra, P. (Ed), Advances in Bioprocess Technology. Springer, Cham, Switzerland, pp. 27- 56. https://doi.org/10.1007/978-3-319-17915-5_3

Sharonova, O.M., Anshits, N.N., Yumashev, V.V., Anshits, A.G., 2008. Composition and morphology of char particles of fly ashes from industrial burning of high-ash coals with different reactivity. Fuel 87, 1989–1997. https://doi.org/10.1016/j.fuel.2007.10.007

Sheng, C., 2007. Char structure characterised by Raman spectroscopy and its correlations with combustion reactivity. Fuel 86, 2316–2324. https://doi.org/10.1016/j.fuel.2007.01.029

Shilling, M., 1999. Carolina Power & Light Co. Carbon / ash separation, in: 1999 International Ash Utilization Symposium. University of Kentucky Center for Applied Energy Research, Lexington, KY. http://www.flyash.info/

Simandl, G.J., Paradis, S., Akam, C., 2015. Graphite deposit types, their origin, and economic significance, in. Symposium on critical and strategic materials. British Columbia Ministry of Energy and Mines, Victoria, pp. 163-172.

Simpson, E.S.W., Tregidga, J.A., 1956. The Archaen rocks of the Marble Delta District, Natal. South African Journal of Geology 59, 237–258.

Smith, D.A.M., 1965. The geology of the area around the Khan and Swakop Rivers in South West Africa (Report), Memoir 3 South West Africa Series. Staatsdrukker, Pretoria.

Smith, I.G.T., 2015. Correlating laboratory and pilot scale reflux classification of fine coal (Thesis). Potchefstroom: North-West University.

195 | P a g e

Smith, M.W., Dallmeyer, I., Johnson, T.J., Brauer, C.S., McEwen, J.-S., Espinal, J.F., Garcia-Perez, M., 2016. Structural analysis of char by Raman spectroscopy: Improving band assignments through computational calculations from first principles. Carbon 100, 678–692. https://doi.org/10.1016/j.carbon.2016.01.031

Snyman, C.P., Barclay, J., 1989. The coalification of South African coal. International Journal of Coal Geology 13, 375–390. https://doi.org/10.1016/0166-5162(89)90100-6

Snyman, C.P., Botha, W.J., 1993. Coal in South Africa. Journal of African Earth Sciences (and the Middle East) 16, 171–180. https://doi.org/10.1016/0899-5362(93)90165-M

Soong, Y., Link, T.A., Schoffstall, M.R., Turčániová, L., Baláž, P., Marchant, S., Schehl, R.R., 1998. Triboelectrostatic separation of mineral matter from Slovakian coals. Acta Montanistica Slovaca 3, 393-400.

Soong, Y., Schoffstall, M.R., Gray, M.L., Knoer, J.P., Champagne, K.J., Jones, R.J., Fauth, D.J., 2002. Dry beneficiation of high loss-on-ignition fly ash. Separation and Purification Technology 26, 177–184. https://doi.org/10.1016/S1383-5866(01)00162- 9

South African Coal Roadmap, 2011. Overview of the South African coal value chain (Report).

Speight, J.G., 2005. Handbook of Coal Analysis. John Wiley & Sons Inc., Hoboke, NJ. https://doi.org/10.1002/9781119037699

Speight, J.G., 2013. The Chemistry and Technology of Coal, third edition. CRC Press, Boca Raton.

Spencer, R., Hill, L., 2016. A flake’s chance in cell: quantifying graphite demand (Report). Canaccord Genuity, Australia.

Steenkamp, N., 2017. Graphite glut stirs SADC slumbers [WWW Document]. URL https://www.miningafricaonline.co.za/index.php/mining-features/mining-in- africa/3306-graphite-glut-stirs-sadc-slumber (accessed 7.3.19).

Steven, N.M., 1987. Tungsten mineralisation at the Otjua prospect: a geological and geochemical investigation (Thesis). Cape Town: University of Cape Town.

Stumpfl, E.F., Rucklidge, J.C., 1982. The platiniferous dunite pipes of the eastern Bushveld. Economic Geology 77, 1419–1431. https://doi.org/10.2113/gsecongeo.77.6.1419

196 | P a g e

Sun, K., Qiu, Y., Zhang, L., 2017. Preserving flake size in an African flake graphite ore beneficiation using a modified grinding and pre-screening process. Minerals 7, 2-13. https://doi.org/10.3390/min7070115

Swart, R., 1986. Report on reconnaissance geological mapping of a portion of the Damara Orogen (Area 2015D). Communications of the Geological Survey of South West Africa / Namibia 2, 27-37.

Sweeney, J.J., Burnham, A.K., 1990. Evaluation of a simple model of vitrinite reflectance based on chemical kinetics. The American Association of Petroleum Geologists Bulletin 74, 1559-1570.

Sýkorová, I., Pickel, W., Christanis, K., Wolf, M., Taylor, G.H., Flores, D., 2005. Classification of huminite – ICCP System 1994. International Journal of Coal Geology 62, 85-106. https://doi.org/ 10.1016/j.coal.2004.06.006

Syrah Resources., 2018. Balama graphite operation, Mozambique [WWW Document]. URL http://www.syrahresources.com.au/balama-project (accessed 7.3.19).

Tai, F.C., Wei, C., Chang, S.H., Chen, W.S., 2010. Raman and X‐ray diffraction analysis on unburned carbon powder refined from fly ash. Journal of Raman Spectroscopy 41, 933-937. https://doi.org/10.1002/jrs.2532

Tagiri, M., 1981. A measurement of the graphitizing-degree by the X-Ray Powder Diffractometer. The Journal of the Japanese Association of Mineralogists, Petrologists and Economic Geologists 76, 345–352. https://doi.org/10.2465/ganko1941.76.345

Tagiri, M., Oba, T., 1986. Hydrothermal syntheses of graphite from bituminous coal at 0.5-5 kbar water vapor pressure and 300-600ºC. Journal of the Japanese Association of Mineralogists, Petrologists and Economic Geologists 81, 260-271.

Tamashausky, A.V., 2006. An introduction to synthetic graphite [WWW Document]. URL https://asbury.com/resources/education/science-of-graphite/synthetic-graphite/ (accessed 7.3.19).

Tan, P., Dimovski, S., Gogotsi, Y., 2004. Raman scattering of non-planar graphite: arched edges, polyhedral crystals, whiskers and cones. Philosophical transactions. Series A, Mathematical, physical, and engineering sciences 362, 2289–2310. https://doi.org/10.1098/rsta.2004.1442

197 | P a g e

Tashiro, T., Ishida, A., Hori, M., Igisu, M., Koike, M., Méjean, P., Takahata, N., Sano, Y., Komiya, T., 2017. Early trace of life from 3.95 Ga sedimentary rocks in Labrador, Canada. Nature 549, 516–518. https://doi.org/10.1038/nature24019

Taylor, H.A., 1991. Minerals Yearbook - Graphite (Report). Reston, VA.

Tennick, F.P., Phaup, A.E., 1976. The geology of the country around Magondi, Lomagundi, Hartley and Gatooma Districts (Report), Bulletin 65. Rhodesia Geological Survey, Salisbury.

Thao, H.M., Hien, T.T., Anh, D.D., Nga, P.T., 2017. Mineralogical characteristics of graphite ore from Bao Ha deposit, Lao Cai Province and proposing a wise use. Vietnam Journal of Earth Sciences 39, 324–336. https://doi.org/10.15625/0866- 7187/39/4/10728

The Aberdeen Group., 1985. Fly ash. How much to use? What are the cost savings? [WWW Document]. URL https://www.concreteconstruction.net/_view-object?id=00000153- 8b9c-dbf3-a177-9fbd58630000 (accessed 7.3.19).

The Southern African Coal Processing Society, 2015. Coal Preparation in Southern Africa, fifth ed. The Southern African Coal Processing Society, Glenvista.

The Star, 2017. Eskom plant 50 year life and air quality retrofit schedule. The Star.

Thomas, R.J., Bühmann, D., Bullen, W.D., Scogings, A.J., De Bruin, D., 1994. Unusual spodumene pegmatites from the Late Kibaran of southern Natal, South Africa. Ore Geology Reviews 9, 161–182. https://doi.org/10.1016/0169-1368(94)90026-4

TimesLive., 2018. Eskom: the power giant at the core of SA’s state rot [WWW Document]. URL https://www.timeslive.co.za/news/south-africa/2018-02-04-eskom-the-power- giant-at-the-core-of-sas-state-rot/ (accessed 7.10.19).

Tishmack, J.K., Burns, P.E., 2004. The Chemistry and Mineralogy of Coal and Coal Combustion Products, in: Gieré, R., Stille, P. (Eds.), Energy, Waste and the Environment: A Geochemical Perspective. Geological Society, London, pp. 223-246. https://doi.org/10.1144/GSL.SP.2004.236.01.14

Tomeczek, J., Palugniok, H., 2002. Kinetics of mineral matter transformation during coal combustion. Fuel 81, 1251–1258. https://doi.org/10.1016/S0016-2361(02)00027-3

Trade and Investment Limpopo, 2012. Annual Report Trade and Investment Limpopo

198 | P a g e

(Report). Trade and Investment Limpopo, Polokwane.

Triton Minerals., 2018. Projects [WWW Document]. URL https://www.tritonminerals.com/projects/ (accessed 7.3.19).

Tucker, R.D., Roig, J.Y., Macey, P.H., Delor, C., Amelin, Y., Armstrong, R.A., Rabarimanana, M.H., Ralison, A.V., 2011. A new geological framework for south- central Madagascar, and its relevance to the “out-of-Africa” hypothesis. Precambrian Research 185, 109–130. https://doi.org/10.1016/j.precamres.2010.12.008

Tuinstra, F., Koenig, J.L., 1970. Raman spectrum of graphite. Journal of Chemical Physics 53, 1126–1130. https://doi.org/10.1063/1.1674108

Uken, R., 1998. The geology and structure of the Bushveld Complex metamorphic aureole in the Olifants River area (Thesis). Durban: University of KwaZulu-Natal.

U.S. Geological Survey, 2015. Mineral commodity summaries 2015 (Report). Reston, VA. https://doi.org/10.3133/70140094

U.S. Geological Survey, 2016. Mineral commodity summaries 2016 (Report). Reston, VA. https://doi.org/10.3133/70170140

U.S. Geological Survey, 2018. Mineral commodity summaries 2018 (Report). Reston, VA. https://doi.org/10.3133/70194932

U.S. Geological Survey, 2019. Mineral commodity summaries 2019 (Report). Reston, VA. https://doi.org/10.3133/70202434

Valdés, A.F., Garcia, A.B., 2006. On the utilization of waste vegetable oils (WVO) as agglomerants to recover coal from coal fines cleaning wastes (CFCW). Fuel 85, 607– 614. https://doi.org/10.1016/j.fuel.2005.08.011

Valentim, B.R.V., 2015. Coal char as a substituting material of natural graphite in green energy technologies – Charphite (Third ERA-MIN Joint Call (2015) on Sustainable Supply of Raw Materials in Europe Project Proposal) (Report). Porto, Portugal.

Valentim, B., Shreya, N., Paul, B., Gomes, C.S., Sant’Ovaia, H., Guedes, A., Ribeiro, J., Flores, D., Pinho, S., Suárez-Ruiz, I., Ward, C.R., 2016. Characteristics of ferrospheres in fly ashes derived from Bokaro and Jharia (Jharkand, India) coals. International Journal of Coal Geology 153, 52–74. https://doi.org/10.1016/j.coal.2015.11.013

199 | P a g e

Valentim, B., Bialecka, B., Calus-Moszko, J., Cruceru, M., Freire, C., Guedes, A., Lazaro- Martinez, J., Popescu, L., Predeanu, G., Santos, A.C., Wagner, N., 2018. Charphite project: Coal char as a substituting material of natural graphite in green energy technologies, in: Carbon Conference 2018.

Van Alphen, C., 2005. Factors influencing fly ash formation and slag deposit formation (slagging) on combusting a south african pulverised fuel in a 200 MWe boiler (Thesis). Johannesburg: University of the Witwatersrand.

Van Alphen, C., 2017. Ash formation and mineralogy, in: One-day Workshop on Coal Ash – Production and Utilisation with a Multitude of Opportunities. Johannesburg.

Van Dyk, J.C., Keyser, M.J., Coertzen, M., 2006. Syngas production from South African coal sources using Sasol–Lurgi gasifiers. International Journal of Coal Geology 65, 243– 253. https://doi.org/10.1016/j.coal.2005.05.007

Van Dyk, J., Brand, J.F., Strydom, C., Waanders, F., 2015. Update on the 50-MW e Theunissen Underground Coal Gasification project. Cornerstone 3, 63-67.

Van Krevelen, D.W., 1993. Coal, third ed. Elsevier Science, Amsterdam.

Van Niekerk, D., Pugmire, R.J., Solum, M.S., Painter, P.C., Mathews, J.P., 2008. Structural characterization of vitrinite-rich and inertinite-rich Permian-aged South African bituminous coals. International Journal of Coal Geology 76, 290–300. https://doi.org/10.1016/j.coal.2008.08.014

Van Waarden, C., 2014. Prehistoric Copper Mining in Botswana. Encyclopaedia of the History of Science, Technology, and Medicine in Non-Western Cultures. https://doi.org/10.1007/978-94-007-3934-5_9871-1

Vanderbruggen, A., Gilbricht, S., Möckel, R., Rudolph, M., 2018. Characterization and beneficiation of pyrolyzed black mass: Increasing the recycling rate of spent lithium ion battery, in: Minerals Engineering International: Hi-Tech Metals ’18. Cape Town.

Vassilev, S.V., Menendez, R., Alvarez, D., Diaz-Somoano, M., Martinez-Tarazona, M.R., 2003. Phase-mineral and chemical composition of coal fly ashes as a basis for their multicomponent utilization. 1. Characterization of feed coals and fly ashes. Fuel 82, 1793-1811. https://doi.org/10.1016/S0016-2361(03)00123-6

Vassilev, S.V., Menendez, R., Diaz-Somoano, M., Martinez-Tarazona, M.R., 2004a. Phase-

200 | P a g e

mineral and chemical composition of coal fly ashes as a basis for their multicomponent utilization. 2. Characterization of ceramic cenosphere and salt concentrates. Fuel 83, 585-603. https://doi.org/10.1016/j.fuel.2003.10.003

Vassilev, S.V., Menendez, R., Borrego, A.G., Diaz-Somoano, M., Martinez-Tarazona, M.R., 2004b. Phase-mineral and chemical composition of coal fly ashes as a basis for their multicomponent utilization. 3. Characterization of magnetic and char concentrates. Fuel 83, 1563-1583. https://doi.org/10.1016/j.fuel.2004.01.010

Vassilev, S.V., Menendez, R., 2005. Phase-mineral and chemical composition of coal fly ashes as a basis for their multicomponent utilization. 4. Characterization of heavy concentrates and improved fly ash residues. Fuel 84, 973-991. https://doi.org/10.1016/j.fuel.2004.11.021

Vassilev, S.V., Vassileva, C.G., 2005. Methods for characterization of composition of fly ashes from coal-fired power stations: A critical overview. Energy & Fuels 19, 1084- 1098. https://doi.org/ 10.1021/ef049694d

Vassilev, S.V., Vassileva, C.G., 2007. A new approach for the classification of coal fly ashes based on their origin, properties, and behaviour. Fuel 86, 1490-1512. https://doi.org/ 10.1016/j.fuel.2006.11.020

Verryn, S. 2017. Amorphous phase percentages measured via XRD [Personal communications]. 27 March., Johannesburg.

Viljoen, K.S., Robinson, D.N., Swash, P.M., 1991. Diamond- and Graphite-Peridotite Xenoliths from the Roberts Victor Mine, in: Fifth International Kimberlite Conference: Extended Abstracts. Companhia de Pesquisa de Recursos Minerais, Araxá, pp. 440– 442. https://doi.org/10.29173/ikc2597

Viljoen, K.S., 1995. Graphite- and diamond-bearing eclogite xenoliths from the Bellsbank kimberlites, Northern Cape, South Africa. Contributions to Mineralogy and Petrology 121, 414–423. https://doi.org/10.1007/s004100050106

Voortman, R.L., Spiers, B., 1986. Angonia: Landscape ecological survey and land evaluation for rural development planning Angonia District, Tete Province. Vol. 4 Landscapes and land units (Report), Comunicação No. 33. Do Instituto Nacional De Investigação Agronómica, Maputo.

Wada, H., Oana, S., 1975. Carbon and oxygen isotope studies of graphite bearing

201 | P a g e

carbonates in the Kasuga area, Gifu prefecture, central Japan. Geochemical Journal 9, 149–160. https://doi.org/10.2343/geochemj.9.149

Wada, H., Tomita, T., Matsuura, K., Tuchi, K., Ito, M., Morikiyo, T., 1994. Graphitization of carbonaceous matter during metamorphism with references to carbonate and pelitic rocks of contact and regional metamorphisms, Japananese Contributions to Mineralogy and Petrology 118, 217–228. https://doi.org/10.1007/BF00306643

Wagner, P.A., 1916. Graphite-bearing xenoliths from the Jagersfontein diamond mine. South African Journal of Geology 19, 54–56.

Wagner, N.J., Matjie, R.H., Slaghuis, J.H., van Heerden, J.H.P., 2008. Characterization of unburned carbon present in coarse gasification ash. Fuel 87, 683–691. https://doi.org/10.1016/j.fuel.2007.05.022

Wagner, N., Malumbazo, N., Falcon, R., 2018. Southern African Coals and Carbons. Struik Nature, Cape Town.

Walter, J.M., 2004. Fabric development, electrical conductivity and graphite formation in graphite-bearing marbles from the Central Damara Belt, Namibia (Thesis). Göttingen: University of Göttingen.

Wang, A., Dhamenincourt, P., Dubessy, J., Guerard, D., Landais, P., Lelaurain, M., 1989. Characterization of graphite alteration in an uranium deposit by micro-Raman spectroscopy, X-ray diffraction, transmission electron microscopy and scanning electron microscopy. Carbon 27, 209–218. https://doi.org/10.1016/0008- 6223(89)90125-5

Ward, C.R., French, D., 2006. Determination of glass content and estimation of glass composition in fly ash using quantitative X-ray diffractometry. Fuel 85, 2268-2277. https://doi.org/10.1016/j.fuel.2005.12.02

Warren, B.E., 1941. X-Ray diffraction in random layer lattices. Physical Review 59, 693–698. https://doi.org/10.1103/PhysRev.59.693

Watt, J.D., Thorne, D.J., 1965. Composition and pozzolanic properties of pulverised fuel ashes. I. Composition of fly ashes from some British power stations and properties of their component particles. Journal of Applied Chemistry 15, 585–594. https://doi.org/10.1002/jctb.5010151207

202 | P a g e

Wells, J.J., Wigley, F., Foster, D.J., Gibb, W.H., Willaimson, J., 2004. The relationship between excluded mineral matter and the abrasion index of a coal. Fuel 83, 359– 364. https://doi.org/ 10.1016/S0016-2361(03)00262-X

Wigley, F., Williamson, J., Gibb, W.H., 1997. The distribution of mineral matter in pulverised coal particles in relation to burnout behaviour. Fuel 76, 1283–1288. https://doi.org/10.1016/S0016-2361(97)00139-7

Wilke, D.P., 1969. Grafietafsettings noord van die Soutpansberg, Transvaal - ‘n Kerngrondstofondersoek hoofsaaklik ten behoewe van die raad op atoomkrag (Report), Bulletin 51. Staatsdrukker, Pretoria.

Wills, B.A., Napier-Munn, T.J., 2006. Wills’ Mineral Processing Technology, seventh ed. Butterworth-Heinemann, Oxford. https://doi.org/10.1016/B978-0-7506-4450-1.X5000- 0

Wilson, M.G.C., 1989. A preliminary appraisal of the mineral potential of Venda based on a reconnaissance geochemical soil sampling survey and literature review (Thesis). Grahamstown: Rhodes University.

Windley, B.F., Razafiniparany, A., Razakamanana, T., Ackermand, D., 1994. Tectonic framework of the Precambrian of Madagascar and its Gondwana connections: a review and reappraisal. Geologische Rundschau 83, 642–659. https://doi.org/10.1007/BF00194168

Woollacott, L.C., Eric, R.H., 1994. Mineral and Metal Extraction – An Overview. South African Institute of Mining and Metallurgy, Johannesburg.

Wu, S., Gu, J., Zhang, X., Wu, Y., Gao, J., 2008. Variation of Carbon crystalline structures

and CO2 gasification reactivity of Shenfu coal chars at elevated temperatures. Energy Fuels 22, 199–206. https://doi.org/10.1021/ef700371r

Yao, Z.T., Ji, X.S., Sarker, P.K., Tang, J.H., Ge, L.Q., Xia, M.S., Xi, Y.Q., 2015. A comprehensive review on the applications of coal fly ash. Earth-Science Reviews 141, 105–121. https://doi.org/10.1016/j.earscirev.2014.11.016

Yeh, T.-S., Wu, Y.-S., Lee, Y.-H., 2011. Graphitization of unburned carbon from oil-fired fly ash applied for anode materials of high power lithium ion batteries. Materials Chemistry and Physics 130, 309–315. https://doi.org/10.1016/j.matchemphys.2011.06.045

203 | P a g e

Zhang, H., Cen, K., Yan, J., Ni, M., 2002. The fragmentation of coal particles during the coal combustion in a fluidized bed. Fuel 81, 1835–1840. https://doi.org/10.1016/S0016- 2361(02)00111-4

Zhang, Y., Granite, E.J., Maroto-Valer, M.M., Tang, Z., 2003. Activated carbons produced from unburned carbon in fly ash and their application for mercury capture. Fuel Chemistry Division Preprints 48, 32-33.

Zhang, L., Hou, J., Bi, X.T., Grace, J.R., Janke, T., Arato, C., 2012. Electrostatic beneficiation of fly ash in a free-falling system. Particuology 10, 154–160. https://doi.org/10.1016/j.partic.2011.07.006

Zhang, W., Honaker, R., 2015. Studies on carbon flotation from fly ash. Fuel Processing Technology 139, 236–241. https://doi.org/10.1016/j.fuproc.2015.06.045

Zondlo, J.W., 2012. Graphite: Structure, Properties, and Applications, in: Mukhopadhyay, P., Gupta, R.K. (Eds.), Graphite, Graphene, and Their Polymer Nanocomposites. CRC Press, Boca Raton, pp. 1-58. https://doi.org/10.1201/b13051-5

204 | P a g e

Appendix A – Automated SEM methodology and results

Automated SEM was applied to the ash samples, to gain an understanding of the amorphous phase composition as well as the char-ash associations. Due to financial and time constraints, it was decided to only analyse GA East. The sample was crushed to -3 mm, mounted in polished carnauba wax blocks and coated with carbon (SJT MetMin). A total of five carnauba wax blocks were tested to get statistical representative data. The equipment specifications are provided (Table A.1). To aid with the phase identification process, research by Matjie et al. (2006) as well as Webmineral’s (www.webmineral.com) comprehensive database was used. Analyses were assisted by Dr. Bradley Guy from the University of Johannesburg.

Table A.1: Automated SEM instrument specifications.

Automated SEM type Quanta 600F Detectors Bruker EDS program Esprit 1.8.2

The phases identified, as well as their respective weight percentages, are provided (Table A.2). The minerals correspond to those obtained with XRD namely quartz, anorthite and mullite being the major crystals. For the amorphous phases, Si-Al dominant glasses with varying proportions of Ca, Fe, K, Ti, Mg, and Na were found. These glassy particles form when rapid cooling or quenching prevents the crystallisation of the melt. Material trapped halfway between mineral and ash crystals are thus formed. The glasses were often filled with anorthite, mullite and quartz crystals. A mineral / glass similar structure to microcline / orthoclase and rich in potassium was also found and was most probably derived from muscovite or illite (Matjie et al., 2006). Upon heating kaolinite in coal will go through a series of irreversible dehydration and decomposition reactions. The evaporation and dehydration of moisture take place between 100 °C and 600 °C, ultimately forming metakaolinite. From the automated SEM data, a spongy material was observed with ratios of aluminium and silicon similar to that of kaolinite. It is believed that this material might be the metakaolinite product formed from the dehydration of kaolinite (not depicted via XRD as a crystal). Matjie et al. (2006) confirmed this by observing that the cavities, giving this material its spongy texture, resulted from the loss of moisture. The elemental assay is given as well as the distribution of each phase (Table A.3). From this table, it can be seen that Al and Si are present in the majority of the phases making it, as seen with XRF, the main elements in ash. C, Ca, Fe, and Ti are also common, all contributing more than 1 wt. % to the total assay. More than 85 % of the carbon is distributed in the char while the rest is present as calcite.

205 | P a g e

Table A.2: Ash phase identification via automated SEM.

Phase Crystalline/Amorphous Description Wt. %

Char Crystalline Unburned carbon 3.07 Quartz Crystalline Unaltered quartz grains 14.58 Kaolinite Amorphous/Crystalline Spongy Si-Al material most probably formed from 14.62 kaolinite, also known as metakaolinite Mullite/Sillimanite Crystalline Transformation product from clays and the 11.31 devitrification of glass Si-Al glass Amorphous Amorphous Si-Al oxides with Ca, Fe, K and Ti traces 22.8 Orthoclase Amorphous/Crystalline K-bearing glass with similar structure to 1.29 microcline/orthoclase Si-Al-Ca glass Amorphous Amorphous Si-Al-Ca oxides with Fe, Mg, Na traces 11.09 Si-Al-Ca-Fe glass Amorphous Amorphous Si-Al-Ca-Fe oxides with Mg traces 4.78 Anorthite Crystalline Present only in gasification and bottom ashes 7.16 Lime/Calcite Crystalline Inorganic carbon 3.53 Fe-oxides Crystalline Hematite and magnetite 1.55 Other Crystalline Phases e.g. apatite, halite, mayenite, pyrite, Fe-Ti-O, 1.83 dolomite, and tilleyite. Present in quantities <1 wt. % Unknown - - 2.38

206 | P a g e

Table A.3: Elemental assay and elemental distribution, automated SEM.

Elemental distribution (wt. %)

Fe

-

Ca - Ca - Al oxides - Char - Total glass glass Al glass Lime/ Other Al Assay Quartz (wt. %) - Calcite Mullite/ - Si Element Kaolinite Anorthite Unknown Fe Sillimanite Si Orthoclase Si

Al 13.59 22.48 31.62 21.95 0.92 9.02 4.45 9.38 0.18 100 C 3.56 86.25 11.9 1.85 100 Ca 7.37 12.02 33.69 12.17 17.12 19.15 5.85 100 Cl 0.02 100 100 F 0.01 100 100 Fe 2.98 20.71 7.74 25.17 36.38 10.01 100 H 0.24 93.41 6.45 0.14 100 K 0.61 70.4 29.6 100 Mg 0.43 70.49 25.54 3.97 100 Mn 0.01 100 100 Na 0.48 98.9 1.1 100 O 45.09 17.23 18.08 12.25 23.54 1.32 10.27 3.97 7.27 3.75 1.03 1.29 100 P 0.1 100 100 S 0.23 100 100 Si 21.81 31.25 14.58 6.84 28.47 1.8 7.98 2.88 6.09 0.11 100 Ti 1.08 98.47 1.53 100 Unknown 2.38 100 100

207 | P a g e

The particles seldom consisted of a single phase. Particles with binary and ternary phases are illustrated (Figures A.1-A.5). The phase legend is provided in Figure A.1.

Figure A.1: Phase legend for automated SEM images.

A massive quartz particle with kaolinite, calcite, mullite, glass, and carbon inclusions are shown (Figure A.2).

Figure A.2: Massive quartz particle with mullite, kaolinite, glass, calcite, and carbon inclusions.

Glass particles (SiAlCa and CaSiAlFe) with large, elongated anorthite needles are shown in Figures A.3 and A.4. Traces of Fe-oxides are also present.

208 | P a g e

Figure A.3: Glass particle with large, elongated anorthite needles.

Figure A.4: Glass particle with anorthite needle inclusions.

A quartz / mullite / kaolinite / glass particle is shown in Figure A.5.

209 | P a g e

Figure A.5: Quartz/kaolinite/glass/mullite particle.

Mineral association data is given (Table A.4). More than 50 % of the char is associated with a free surface. The rest is mainly associated with kaolinite (metakaolinite), which makes sense seeing that clays are usually closely associated with the parent coal. Another interesting observation is the association of kaolinite with mullite. Mullite is a transformation product from kaolinite and metakaolinite and therefore this association. Anorthite is disseminated as needles in the calcium-rich glasses (this mineral is therefore usually found in calcium-rich ash).

A carbon particle with inclusions is given in Figure A.6. These carbon-mineral particles are of importance, seeing that they will influence the separation of char from ash. In Figure A.7 the char fractional distribution versus liberation class is given. It can be seen that 25.55 % of the particles are liberated (assuming >80 wt. % char per particle can be described as liberated) while 36.35 % is locked (<30 wt. % char per particle). The middlings fraction is 38.11 %. In Figure A.8 the fractional char distribution versus free surface class is given. Free surface in this sense meaning the exposure of the phase to the surrounding area. Three classes were defined; not exposed (particles with 0-30 % of char being exposed), middlings (particles with 30-80 % of char being exposed), and exposed (particles with 80-100 % of char being exposed). It can be seen that 23.27 wt. % of all char is distributed in the latter class. These particles will be easy to treat with surface separation techniques such as froth flotation. The remainder is not exposed.

210 | P a g e

Table A.4: Phase association / locking data.

Phase association (wt. % of total phase)

Fe

-

Ca -

Ca - Al oxides - Free Free Char - Total Al glass glass glass Lime/ Other Al Phase Quartz - Calcite Mullite/ - surface Si Kaolinite Anorthite Unknown Fe Sillimanite Si Orthoclase Si

Char 0 7.56 17.74 1.48 5.34 0.28 4.63 1.19 1.29 1.14 0.23 0.91 5.96 52.23 100 Quartz 6.92 0 10.48 2.51 14.60 2.21 5.75 1.77 1.72 0.71 0.24 0.96 6.72 45.39 100 Kaolinite 13.31 8.58 0 5.46 16.17 0.51 4.77 1.63 3.16 0.48 0.16 0.37 4.56 40.85 100 Mullite/ Sillimanite 3.58 6.75 17.47 0 11.38 0.94 3.48 1.61 1.53 0.59 0.20 0.85 3.74 47.87 100 Si-Al glass 4.30 12.83 17.38 3.82 0 0.60 5.82 2.93 2.76 0.55 0.20 0.43 4.05 44.32 100 Orthoclase 2.75 23.98 6.71 3.88 7.32 0 2.98 1.70 0.67 0.26 1.01 1.85 10.92 35.98 100 Si-Al-Ca glass 5.53 7.70 7.95 1.76 9 0.37 0 6.06 15.20 2.27 0.47 1.42 8.19 34.10 100 Si-Al-Ca-Fe glass 3.01 4.92 5.80 1.67 9.34 0.45 12.23 0 15.91 3.15 3.15 1.46 10.56 28.35 100 Anorthite 2.87 4.25 9.21 1.40 7.25 0.15 22.53 12.59 0 1.65 0.62 0.85 6.93 29.69 100 Lime/Calcite 4.10 3.12 2.43 0.96 2.68 0.10 6.81 4.65 3.00 0 0.17 12.12 17.64 42.23 100 Fe-oxides 3.02 3.25 2.87 0.96 2.93 1.20 4.07 12.12 2.83 0.53 0 2.95 34.33 28.94 100 Other 5.18 9.88 4.39 3.40 4.47 1.27 4.92 2.81 2.16 14.38 1.26 0 20.18 25.69 100 Unknown 2.59 3.44 2.66 0.69 2.30 0.49 3.02 1.86 1.47 2.05 1.26 1.22 0 76.94 100

211 | P a g e

Figure A.6: Carbon particle with mineral inclusions.

18.0

16.0 14.0 12.0 10.0 8.0 6.0 4.0 2.0

Fractional distribution (wt.%) 0.0

Char composition of particle (wt. %)

Locked = 36.35% Middlings = 38.11% Liberated = 25.55%

Figure A.7: Fractional distribution for different liberation classes.

212 | P a g e

18

16 14 12 10 8 6 4

Fractional distribution (wt. %) 2 0

Char free surface of particle (wt. %)

Not exposed = Middlings = 43.57% Fully exposed = 33.16% 23.27%

Figure A.8: Fractional distribution for different free surface classes.

213 | P a g e

Appendix B – Raman microspectroscopy curve fitting procedure

B.1 Char samples Only the 1st order spectra were taken into account (800-2000 cm-1), seeing that the 2nd order spectra (2000-3200 cm-1) were not resolved well enough for fitting purposes. The 1st order spectra were normalized to a 0-1 range and then smoothed using the Adjacent Averaging function (20 points per window). The background due to fluorescence was subtracted by using a 4th order polynomial function (Figure B.1A). According to Schito et al. (2017), a polynomial baseline is preferred over a linear baseline, due to overestimation of the band intensities from the latter. The baseline anchor points were chosen at positions below 1000 cm-1 (3 points) and above 1800 cm-1 (3 points).

Numerous 1st order band assignments exist for coal, char, and related carbonaceous material. Li et al. (2006a; b) and Smith et al. (2016) have assigned ten bands to their respective brown coal and biomass char samples. These samples were however much more disordered and amorphous than the samples studied in this paper. Guedes et al. (2012) also used ten bands to understand the structure of different coal macerals and char morphotypes. Ferralis et al. (2016) and Schito et al. (2017) considered six bands for their kerogen samples, while Guedes et al. (2010) fitted the same bands to their coal samples (sub- bituminous to anthracite rank). Chabalala et al. (2011) only considered four bands for coal and char samples from the South African Witbank coalfield. Sadezky et al. (2005), Sforna et al. (2014), and Sheng (2007) assigned five bands to their soot (carbon black), chert, and coal char samples respectively. Seeing that the spectra for this study showed similarities to those from the last-mentioned three researchers, it was decided to follow their approach and also fit five bands.

As a first approach, the fitting method suggested by Sadezky et al. (2005) was followed in which the G, D1, D2, and D4 bands were fitted with a Lorentzian function, while the D3 band was fitted with a Gaussian function. It was however found that a large deviation between the fitted curve and actual curve existed, leading to a low R2 value of ~0.9980 (Figure B.1B). Although the D1 band was deconvoluted appropriately, it was found that the G band was lacking in intensity as well as width. Therefore, alternative fitting functions and combinations were tried and tested. It was found that fitting the D1, G, and D4 bands with a Gaussian function, and the D2 and D3 bands with a Lorentzian function, yielded better results. Restrictive bounds were added (as suggested by Sforna et al., 2014) to converge appropriately (Table B.1). An example of a curve fitted with this method is given in Figure

214 | P a g e

B.1C. It can be seen that by following this approach, limited fitting deviation was observed with an R2 value of ~0.9997. Both the D1 and G bands were deconvoluted appropriately, with intensities and widths following the actual curve shape. In Figure B.1D a fitting example of a very disordered particle is given. These types of particles were however seldom seen, and therefore neglected in the analyses.

Table B.1: Restrictive bounds added to the char Raman microspectopy curve fitting procedure.

Band Fitting function Band position (cm-1) FWHM (cm-1)

G Gaussian 1580-1610 None D1 Gaussian 1340-1370 None D2 Lorentzian 1610-1630 None D3 Lorentzian 1470-1520 None D4 Gaussian 1150-1250 <200

Once all spectra were fitted, the average band position, average band full width at half maximum (FWHM), average band area, and average band intensity were obtained for each sample. To determine the error, the standard deviations were taken. Furthermore, the distance between the G and D1 bands (Baludikay et al., 2018; Schito et al., 2017), the FWHM ratio between the D1 and G bands (Baludikay et al., 2018; Schito et al., 2017; Sforna et al., 2014), the area ratio between the D1 and G bands (Schito et al., 2017; Sforna et al., 2014), and the RA1 and RA2 values as established by Lahfid et al. (2010) and subsequently used by Schito et al. (2017) and Sforna et al. (2014), were calculated. These calculated parameters are frequently used to indicate maturity in disorganised carbonaceous material, as opposed to the G and D1 band intensity ratios used in organised carbonaceous material. The calculation of RA1 and RA2 are provided in Equations B.1 and B.2.

퐴 + 퐴 Equation B.1 푅퐴1 = 퐷1 퐷4 퐴퐷1 + 퐴퐷2 + 퐴퐷3 + 퐴퐷4 + 퐴퐺

퐴 + 퐴 Equation B.2 푅퐴2 = 퐷1 퐷4 퐴퐷2 + 퐴퐷3 + 퐴퐺

215 | P a g e

A B

C D

Figure B.1: A) Baseline subtraction; B) Curve fitting using the method from Sadezky et al. (2005); C) Curve fitting method followed in this study; D) Curve fitting method followed in this study for disordered carbon. (Actual curve=black, fitted curve=red, G band=light blue, D1 band=green, D2 band=orange, D3 band=dark blue, D4 band=purple).

216 | P a g e

B.2 Graphite samples Only the 1st order spectra were taken into account (800-2000 cm-1). The 1st order spectra were normalized to a 0-1 range. The background due to fluorescence was subtracted by using a linear polynomial function. The G, D1, and D2 bands were fitted with a Lorentzian function. All spectra were fitted with an R2 value > 0.99. In Figure B.2 an example of curve fitting done on the natural graphite samples is given.

Normalised intensity (a.u.) intensity Normalised

Raman shift (cm-1)

Figure B.2: Raman peak fitting example (SA Graphite illustrated).

217 | P a g e