Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Lower mantle diamonds from the Rio Soriso (Juina, Brazil) Hayman, Patrick 2004

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
[if-you-see-this-DO-NOT-CLICK]
ubc_2004-0228.pdf [ 69.27MB ]
[if-you-see-this-DO-NOT-CLICK]
Metadata
JSON: 1.0052467.json
JSON-LD: 1.0052467+ld.json
RDF/XML (Pretty): 1.0052467.xml
RDF/JSON: 1.0052467+rdf.json
Turtle: 1.0052467+rdf-turtle.txt
N-Triples: 1.0052467+rdf-ntriples.txt
Original Record: 1.0052467 +original-record.json
Full Text
1.0052467.txt
Citation
1.0052467.ris

Full Text

LOWER M A N T L E DIAMONDS FROM THE RIO SORISO (JUINA, BRAZIL) by  PATRICK H A Y M A N B.Eng., Queen's University, 1999  A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T O F THE REQUIREMENTS FOR THE D E G R E E OF MASTER OF SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES (Department o f Earth and Ocean Sciences) W e accept this thesis as conforming to the required standard  T H E U N I V E R S I T Y OF BRITISH C O L U M B I A March, 2004 © Patrick Hayman , .•. ^ i  Library Authorization  In presenting this thesis in partial fulfillment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission.  Name of Author (please print)  Louee  Title of Thesis: fi \Q  D e  9  r e e :  So<^l  M  Department of  Date (dd/mm/yyyy)  TMA»MQMtNS  Year: ftP-TH  L  The University of British Columbia Vancouver, B C  T M I ^  <;Q  <c e  F t f t O N A  Canada  r> C E A - M  <>C[£UcE^  2~oct\  Abstract The morphology, colour, fluorescence (FL), cathodoluminescence ( C L ) , nitrogen content and aggregation state, internal morphology and mineral inclusion chemistry have been studied for sixty-nine alluvial diamonds recovered from the R i o Soriso, Juina area, Brazil.  The majority o f the R i o Soriso diamonds are colourless, but'grey, yellow, brown and non-uniform colours are also observed.  Diamonds fluoresce and cathodoluminesce a  variety o f shades and intensities o f blue, turquoise and green.  There is a correlation  between F L and body colour, with most brown diamonds fluorescing turquoise or green. In general, diamonds with brighter C L have higher nitrogen concentrations. Diamond crystals are generally well resorbed, fragmented and plastically deformed, all o f which contributed to the obscuration o f the primary crystal habit. Most diamonds are classified as either tetrahexahedroids or dodecahedroids, but crystals o f undetermined morphology are abundant.  C L examination o f polished diamond surfaces and plates indicate that  some crystals developed through intermittent  episodes o f octahedral growth and  resorption.  Infrared spectroscopic studies show that the diamonds contain trace amounts o f both nitrogen (0-541 ppm, averaging 72 ppm) and hydrogen.  There is strong positive  correlation between nitrogen and hydrogen concentrations. Nitrogen i n most diamonds is fully aggregated as B centres (type IaB, 54%), but there is also a large proportion o f nitrogen free stones (type Ha, 38%). A small population o f diamonds contain nitrogen i n the form o f A centres (type IaA, 1.5% and I a A B , 7%), which is indicative o f residence i n the upper mantle. Nitrogen contents typically decrease from crystal core to rim.  Mineral inclusions recovered and analysed from 30 diamonds include: ferropericlase, MgSi-perovskite,  CaSi-perovskite,  'olivine',  ii  tetragonal-almandine-pyrope-phase,  pyrrhotite, magnetite, pyrope-almandine-grossular garnet and perovskite. Based on the mineralogy o f diamond inclusions and diamond morphology, C L , F L , nitrogen content and aggregation state, the R i o Soriso suite was subdivided into six paragenetic groups which formed i n the lower mantle, transition zone and upper mantle. These paragenetic groups are: 1) ultramafic lower mantle diamonds, 2) ultramafic diamonds sourced from the boundary between the upper and lower mantle, 3) mafic diamonds sourced from 580660 k m , 4) mafic diamonds sourced from any sub-lithospheric depths, 5) eclogitic diamonds sourced from the upper mantle, and 6) peridotitic diamonds sourced from the upper mantle. The preferred explanation for the sampling o f such a large depth interval within the mantle (-200 to >660 km) is that the diamonds were entrained in a plume that originated at the core-mantle boundary.  iii  Table of Contents Abstract List of Figures List of Tables Acknowledgements  ii ix xiv xvi  1.0 Introduction  1  1.1 Motivation for the project 1.2 Location map 1.3 Geology 1.3.1 Continental scale geology 1.3.2 Local geology  2.0 Diamond Morphology  1 3 3 3 4  8  2.1 Introduction 2.1.1 Crystal habit 2.1.1.1 Monocrystalline diamond 2.1.1.2 Polycrystalline diamond 2.1.2 Resorption 2.1.3 Crystal regularity 2.1.4 Crystal intactness 2.1.5 Surface features 2.1.5.1 Surfaces features associated with etching 2.1.5.2 Other surface features 2.1.6 Sequence o f events 2.2 Analytical techniques 2.3 Results 2.3.1 Size 2.3.2 Crystal habit 2.3.3 Resorption 2.3.4 Crystal regularity 2.3.5 Crystal intactness 2.3.6 Etching 2.3.6.1 Trigons 2.3.6.2 Hexagons 2.3.6.3 Tetragons 2.3.6.4 Etch channels 2.3.6.5 Frosting 2.3.7 Deformation Laminations 2.3.8 Hillocks 2.3.9 Fracturing  iv  8 8 10 12 13 16 16 16 17 20 21 22 23 23 24 25 27 27 28 28 28 29 30 30 31 31 33  2.4 Discussion 2.4.1 Summary o f the physical characteristics 2.4.2 Comparison with other diamond studies from the Juina area  3.0 Colour  34 34 34  37  3.1 Introduction 3.1.1 Causes o f colouration i n natural diamond 3.2 Analytical techniques 3.3 Results 3.3.1 Uniform body colours 3.3.2 Non-uniform colours 3.4 Discussion 3.4.1 Comparison with other studies  4.0 Fluorescence of Diamonds  37 37 40 40 41 42 43 43  44  4.1 Introduction 4.2 Analytical methods 4.3 Results  44 44 45  5.0 Cathodoluminescence of Diamonds  48  5.1 Introduction 5.2 Analytical techniques 5.3 Results  48 49 49  6.0 Infrared Spectroscopy of rough diamonds 6.1 Introduction 6.1.1 One-phonon absorption i n diamond related to nitrogen impurities 6.1.2 Process o f nitrogen aggregation 6.1.3 Quantitative calculation o f nitrogen concentration 6.1.4 Time-averaged mantle residence temperatures 6.1.5 IR spectra for some other impurities 6.1.5.1 C H bonds 6.1.5.2 Water ( O H and H O H bonds) 6.1.5.3 Carbon dioxide (CO2 bonds) 6.1.5.4 Carbonate ( C O 3 bonds) 6.2 Analytical techniques 6.2.1 Examination o f error analysis for infrared studies 6.2.1.1 Error from deconvolution software 6.2.1.2 Sensitivity to baseline corrections 6.2.1.3 Reproducibility o f IR spectra 6.2.2 Precision o f IR data 6.2.3 M i n i m u m detection limits 2  v  53 53 54 56 60 61 64 64 65 66 66 66 69 69 70 71 72 73  6.3 Results 6.3.1 Nitrogen concentration and aggregation state measurements 6.3.2 Other impurities detected by PR spectroscopy 6.3.2.1 C - H bonds 6.3.2.2 C H and C H bonds 6.3.3 Unexplained spectra 6.4 Discussion 6.4.1 Relationship between time and temperature 6.4.2 Nitrogen character o f R i o Soriso diamonds 6.4.3 Comparison o f relative hydrogen and total nitrogen concentration 6.4.4 Comparison o f nitrogen characteristics with other studies 2  3  7.0 Growth studies  74 74 76 77 77 78 80 80 81 82 83  85  7.1 Background 7.1.1 Internal structures i n diamond 7.1.2 Infrared spectroscopy o f polished diamond 7.2 Analytical techniques 7.3 Results • 7.3.1 Diamond 1-2 7.3.2 Diamond 1-4 7.3.3 Diamond 2-1 7.3.4 Diamond 2-2 , 7.3.5 Diamond 2-5 7.3.6 Diamond 2-8 7.3.7 Diamond 2-9 7.3.8 Diamond2-11 7.3.9 Diamond 3-1 7.3.10 Diamond 3-5 7.3.11 Diamond 3-8 7.3.12 Diamond 3-10 7.3.13 Diamond 3-11 7.3.14 Diamond 4-17 7.4 Discussion 7.4.1 Summary o f growth studies o f R i o Soriso diamonds 7.4.2 Comparisons with other studies  8.0 Mineral Inclusions  85 85 92 92 94 94 97 102 104 107 108 109 111 114 117 119 120 122 123 125 125 128  130  8.1 Introduction 8.1.1 Composition o f the mantle 8.1.2 Geothermal gradient 8.1.3 Terminology 8.1.4 Relevant mantle minerals and their stability fields  vi  130 131 133 134 135  8.1.4.1 M g S i 0 8.1.4.2 C a S i 0 8.1.4.3 Garnets and highly aluminous silicates 8.1.4.4 M g S i 0 8.1.4.5 S i 0 8.1.4.6 Ferropericlase 8.1.4.7 Stability o f other phases 8.2 Analytical Techniques 8.2.1 Extraction and mounting o f inclusions 8.2.2 Qualitative identification o f inclusions ( E D S ) 8.2.3 Quantitative identification o f inclusions ( E P M A ) 8.3 Results 8.3.1 Inclusions o f prim ary ori gin 8.3.1.1 Ferropericlase 8.3.1.2 M g S i 0 8.3.1.3 C a S i 0 8.3.1.4 M g S i 0 8.3.1.5 Garnet and T A P P 8.3.1.6 Magnetite 8.3.1.7 Sulphides 8.3.2 Inclusions o f uncertain origin 8.3.2.1 S i 0 8.3.2.2 Perovskite 8.3.2.3 Other calcium-bearing minerals 8.3.2.4 Metallic iron 8.3.3 Touching phases 8.3.3.1 M g S i 0 and M g S i 0 composites 8.3.3.2 M g S i 0 and T A P P composites 8.3.3.3 M g S i 0 and ferropericlase composite 8.3.3.4 M g S i 0 , M g S i 0 and T A P P composite 8.3.3.5 C a S i 0 and Ca-rich mineral composites 8.3.4 F e - N i blebs on ferropericlase 8.3.5 Inclusions o f secondary origin 8.3.5.1 Altered ferropericlase grains 8.3.5.2 Local oxidation o f ferropericlase grains 8.3.5.3 Altered Ca-rich grains 8.4 Discussion 8.4.1 Comparison with diamond inclusions from other studies 8.4.1.1 Ferropericlase 8.4.1.2 M g S i 0 8.4.1.3 C a S i 0 8.4.1.4 ' O l i v i n e ' 8.4.1.5 Garnet and T A P P 8.4.1.6 Perovskite 3  3  2  4  2  3  3  2  4  2  3  2  4  3  2  4  3  2  4  3  3  3  vn  8.4.1.7 SiC-2 8.4.1.8 Pyrrhotite, magnetite and native iron 8.4.1.9 F e - N i blebs and magnesioferrite spots on ferropericlase 8.4.2 Inclusion paragenesis 8.4.2.1 Lower mantle 8.4.2.2 Lower mantle/upper mantle 8.4.2.3 Deep transition zone/lower mantle (>~580 km) 8.4.2.4 Peridotitic 8.4.2.5 Eclogitic 8.4.2.6 Paragenesis summary  9.0 Discussion  212 213 214 215 216 217 220 221 222 223  224  9.1 Correlations between diamond body colour, F L , C L and ER. 9.2 Diamond subpopulations 9.2.1 Upper mantle diamonds 9.2.1.1 Upper mantle peridotitic diamonds 9.2.1.2 Upper mantle eclogitic diamonds (type IaA and IaAB) 9.2.2 Eclogitic diamonds (type IaB) 9.2.3 Eclogitic and/or peridotitic diamonds from depths greater than-580 km 9.2.4 Upper mantle/lower mantle (-660 km) diamonds 9.2.5 Lower mantle diamonds 9.2.6 Diamonds o f unknown paragenesis 9.3 Distribution o f paragenetic groups 9.4 Plume origin o f R i o Soriso diamonds 9.5 Origin and distribution o f eclogitic diamonds 9.6 Implications for exploration  224 227 228 229 232 234 236 238 239 241 242 243 245 246  10. Conclusions  249  References  251  Appendix A - Images o f diamond body colour, fluorescence and cathodoluminescence  268  Appendix B - Catalogue o f morphological features and diamond fluorescence Appendix C - Nitrogen concentration, aggregation state and relative hydrogen concentration  286 290  Appendix D - Infrared spectra and deconvoluted curves Appendix E - Frequency o f inclusion phases analysed from each diamond by E P M A method  292 310  Appendix F - Frequency o f inclusion phases analysed from each diamond by E D S method  311  viii  List of Figures  1.1. Location map o f the Juina area 1.2. Location o f cratons, major lineaments and kimberlite provinces in Brazil 1.3. Geochronological provinces and the main lithological associations o f the Amazonian craton 1.4. Topographic map o f Juina mining district and surroundings 1.5. Detailed location map o f Juina diamond mining district 2.1. Pressure-temperature plot o f primary diamond crystal form and diamondgraphite stability 2.2. Octahedral crystal habit 2.3. Cubic crystal habit 2.4. Cubo-octahrdral habit 2.5. Photograph o f diamond aggregate 2.6. Photograph o f a diamond macle twin 2.7. Semi-quantitative resorption classification scheme for crystals that initially grew as octahedrons 2.8. Etch pit orientation on cubic and octahedral faces 2.9. The relative timing o f common features observed on diamond 2.10. Plot o f normalized frequency versus weight i n carats for R i o Soriso diamonds 2.11. Images o f an octahedral crystal (Diamond 4-18) 2.12. S E M image o f aggregate o f octahedral crystals (Diamond 6-8) 2.13. Plot o f frequency versus resorption class for single and polycrystalline diamond 2.14. Plot o f frequency versus external morphology 2.15. S E M images o f different degrees o f diamond resorption 2.16. Plot o f frequency versus degree o f intactness for R i o Soriso diamonds 2.17. Photos and S E M image o f examples from this study for the four classes of'intactness' 2.18. S E M photo o f trigonal pits 2.19. S E M photos o f hexagonal pits 2.20. S E M photos o f tetragons 2.21. S E M photos o f etch channels (ruts) 2.22. S E M image o f fine frosting (Diamond 4-13) 2.23. S E M images o f deformation laminations (Diamond 1-3) 2.24. S E M images o f hillocks 2.25. S E M image o f mechanical wear on edge o f diamond (Diamond 5-9) 2.26. Plot o f morphology distributions for two previous studies on Juina diamonds  ix  3 3 5 6 7 10 10 11 11 12 12 15 18 22 23 24 24 25 25 26 27 27 28 28 29 30 30 31 32 33 35  3.1. 3.2. 3.3. 4.1.  Distribution o f body colours for R i o Soriso diamonds Photographs o f representative diamond colours for R i o Soriso diamonds Photographs o f non-uniform colours Fluorescence colour and fluorescence colour intensity distribution for R i o Soriso diamonds 4.2. Photos o f various F L colours observed for R i o Soriso diamonds 4.3. Compilation photograph o f F L colours for R i o Soriso diamond suite 5.1. Colour distribution for C L colours 5.2. Photographs o f representative diamond C L colours observed 5.3. Close-up greyscale photograph o f C L o f growth features o f resorbed diamond 5.4. Photographs o f C L images o f 47 rough, unpolished diamonds from R i o Soriso 6.1. Difference between IR-active and IR-inactive bonds 6.2. IR spectra for a type n diamond from 500-4000 cm" 6.3. I R spectra for common end-member absorption patterns i n diamond from 900-1500 cm" 6.4. IR absorption spectra o f the development and subsequent degradation oftheB'peak 6.5. Plot o f B/(A+B) centers in diamond versus integrated area under B ' absorption peak 6.6. Transmission electron micrograph showing a cross-section view o f platelets 6.7. Transmission electron micrographs o f dislocation loops and voidites i n diamond 6.8. The progression o f nitrogen aggregation i n diamond 6.9. IR spectra o f C H absorption i n diamond 6.10. I R spectra o f water i n diamond 6.11. IR spectra o f carbon dioxide i n diamond • 6.12. IR spectra o f carbonate i n diamond 6.13. Output results from deconvolution software 6.14. Examination o f base line sensitivity for IR curves 6.15. Distribution o f total nitrogen concentration for R i o Soriso diamonds 6.16. Diamond type distribution for R i o Soriso suite 6.17. Plot o f B centres versus D centres 6.18. Absorption i n the C H stretch region o f R i o Soriso diamonds 6.19. IR spectra o f diamonds 2-5 and 4-17 before and after heating to ~600°C 6.20. I R spectra o f unknown absorption pattern (defect ' X ' ) 6.21. Photograph o f diamond 2-3 and IR spectra o f d e f e c t ' Y ' 6.22. Plot o f minimum time averaged mantle residence temperatures versus minimum residence times for R i o Soriso diamonds 6.23. Plot o f total nitrogen concentration versus % B aggregation for R i o Soriso diamonds 6.24. Plot o f relative hydrogen and total nitrogen concentrations for 1  1  x  40 41 42 45 46 47 50 50 51 52 53 54 55 57 58 58 59 60 65 65 66 66 69 70 74 74 75 77 78 79 79 80 81  R i o Soriso diamonds 6.25. Plot o f total nitrogen concentration versus % B aggregation for selected diamonds and diamond suites worldwide 7.1. C L images o f central plate o f diamond (opposite sides) 7.2. Cartoon o f internal morphology for cuboid and octahedral growth 7.3. X-ray section image o f cubo-octahedral diamond 7.4. C L image illustrating growth-sectorial dependence in synthetic diamond 7.5. S E M - C L image o f diamond core 7.6. Digitally enhanced C L images highlighting some typical internal structures observed i n diamond 7.7. Image o f C L o f diamond 1-2 with IR data for transects 1 to 18 and 10, 22 to 29 7.8. Close-up greyscale C L image o f diamond 1-2 7.9. Photos o f C L , F L and body colour for various sides o f diamond 1-4 7.10. Image o f C L o f diamond 1 -4, side A , with IR data for transect 1-15 7.11. Greyscale photograph o f deformation laminations observed under C L 7.12. Image o f C L o f diamond 1-4, side B (flipped horizontally), with IR data for transects 1-14 and 15-21 7.13. Photograph o f polished surface o f diamond aggregate (Diamond 2-1) 7.14. Image o f C L o f diamond 2-1 with IR data for transect 1 to 11 7.15. Image o f C L o f diamond 2-2 with IR data for transects 7 to 19 and 14, 20 to 30 7.16. Image o f C L o f diamond 2-5 with IR data for transect 1 to 11 7.17. Image o f C L o f diamond 2-8 with IR data for transect 5 to 20 7.18. Image o f C L o f diamond 2-9 with I R data for transect 2 to 24 7.19. Image o f C L o f diamond 2-11 with IR data for transect points indicated 7.20. IR spectrum from point 19 showing large B ' peak (Diamond 2-11) 7.21. Image o f C L o f diamond 3-1, side A , with IR data for transect 3 to 12 7.22. Image o f C L o f diamond 3-1, side B , with IR data for transect 2 to 9, with various photos 7.23. Image o f C L o f diamond 3-5 with IR data for transects 1-12 and 13-21 7.24. Image o f C L o f diamond 3-8 with IR data for transect 2 to 12 7.25. Image o f C L o f diamond 3-10 with IR data for transect 1-14 7.26. Image o f C L o f diamond 3-11 with IR data for transect 2 to 13 7.27. Image o f C L o f diamond 4-17 with I R data for transect A , 1-7 8.1. Seismic velocities for P and S waves through the Earth (0-6370km) 8.2. Seismic velocities for P and S waves through the Earth (200-800km) 8.3. Mineral assemblages and densities for pyrolite and basaltic oceanic crust 8.4. Phase transformations for M g S i 0 3 8.5. Phase transformations for the predicted dominant C a phases i n the mantle 8.6. Phase transformation for (Mgo.89Feo.i i)2Si04 8.7. Phase transformation for Si02 8.8. Photograph o f diamond cracker 8.9. S E M image o f inclusions embedded i n diamond  xi  82 83 86 87 87 88 89 91 95 96 98 99 100 101 102 103 105 107 109 110 112 113 114 116 118 120 121 122 124 131 131 133 137 139 141 143 145 146  8.10. 8.11. 8.12. 8.13. 8.14. 8.15. 8.16. 8.17. 8.18. 8.19. 8.20. 8.21. 8.22. 8.23. 8.24. 8.25. 8.26. 8.27. 8.28. 8.29. 8.30. 8.31. 8.32. 8.33. 8.34.  S E M images o f ferropericlase grains Plot o f FeO versus M g O for fPer grains by diamond S E M images o f M g S i 0 grains S E M images o f C a S i 0 grains S E M images o f 01 grains S E M images and accompanying E D S spectra o f aluminous silicate grains Plot o f CaO versus C r 2 0 (wt%) for aluminous silicates i n this study Plot o f A l + C r versus S i for aluminous silicates i n this study S E M images o f magnetite grains " S E M images o f sulphide grains S E M images o f S i 0 Photographs o f Si02 grains under U V light and i n the absence o f U V light S E M images and E D S spectra o f perovskite grains S E M images and E D S spectra o f exotic Ca-Si grains S E M image and E D S spectrum o f native iron grain S E M images o f touching olivine and M g S i 0 inclusions S E M images o f touching T A P P and M g S i 0 inclusions S E M image o f touching olivine and ferropericlase inclusion S E M image o f three-phase composite (Inclusion 1.5 - J ) S E M images o f composite grains o f C a S i 0 and 'exotic' Ca-rich phases S E M images and E D S spectra for F e - N i blebs on ferropericlase gains S E M images o f F e - N i blebs and linear features on ferropericlase grains Photograph o f altered ferropericlase grain (Inclusion 3.9 - A ) S E M images o f weathered ferropericlase grains S E M images and E D S spectrum o f secondary magnesioferrite spots on ferropericlase (from Diamond 3-6) 8.35. S E M images o f weathered Ca-rich grains 8.36. Plot o f Fe versus M g for ferropericlase grains from Juina, B r a z i l and Guinea, West Africa 8.37. Plot o f mg vs. frequency for M g S i 0 inclusions worldwide 8.38. Plot o f mg vs. frequency for 'olivine' inclusions worldwide 8.39. Plot o f A l + C r versus S i for aluminous silicates worldwide 8.40. Plot o f C a O versus C r 0 (wt%) for aluminous silicates worldwide 8.41. Plot o f M g versus F e versus S i for M g S i 0 , O l and fPer grains in association i n this study 8.42. Plot o f M g versus F e versus S i for M g S i 0 and fPer grains from experimental studies and associations i n this study 8.43. Distribution o f diamond paragenetic groupings based on mineral inclusion data 9.1. Comparison o f F L colour distribution by diamond body colour 9.2. Photos comparing diamond C L and diamond F L 9.3. Plot o f average nitrogen concentration versus fluorescence intensity 9.4. Plot o f mantle residence time versus mantle residence temperature for xenoliths and various diamond nitrogen characters 3  3  3  3 +  3 +  4 +  2  3  3  3  +  158 159 164 166 169 172 173 174 175 177 179 180 182 184 185 187 189 191 192 194 196 197 198 198 199 200  2 +  3  3 +  3 +  4 +  2  2 +  3  2 +  203 205 207 209 210  4 +  3  2 +  2 +  219  4 +  3  xii  220 223 224 225 226 229  9.5. 9.6. 9.7. 9.8. 9.9. 9.10. 9.11. 9.12. 9.13.  Photos o f diamond body colour, F L and C L for upper mantle diamonds o f peridotitic paragenesis Plot o f temperature versus pressure for peridotitic xenoliths and diamonds Photos o f diamond body colour, F L and C L for upper mantle eclogitic diamonds Plot o f temperature versus pressure for eclogitic xenoliths and diamonds Photos o f diamond body colour, F L and C L for eclogitic type IaB diamonds Photos o f diamond body colour, F L and C L for diamonds from >580 k m depth Photos o f diamond body colour, F L and C L for U M / L M diamonds Photos o f diamond body colour, F L and C L for L M diamonds Distribution o f paragenetic groups for R i o Soriso diamonds based on all studies  xiii  230 231 232 233 235 237 238 240 242  List of Tables  2.1. Conversion between various diamond morphology terminology classification schemes 6.1. Base line sensitivity study 6.2. Reproducibility study 6.3. M D L ' s for I R data 6.4. B ' absorption and D centre concentrations for type IaA and I a A B diamonds 7.1. Summary o f results for growth studies o f R i o Soriso diamonds 8.1. M g S i 0 3 polymorphs 8.2. C a S i 0 polymorphs 8.3. Aluminous silicates 8.4. M g S i 0 polymorphs 8.5. SiC-2 polymorphs 8.6. Statistics on oxide analyses 8.7. Statistics on sulphide analyses 8.8. List o f standards used i n electron microprobe analyses 8.9. M i n i m u m detection limits for weight percent values 8.10. M i n i m u m detection limits for cation values 8.11. Precision for weight percent values at 95% confidence level 8.12. Precision for cation values at 95% confidence level 8.13. Major oxide chemistry for ferropericlase grains 8.14. Cation calculations for ferropericlase 8.15. Major oxide data for M g S i 0 3 grains (wt%) 8.16. Cation calculations for M g S i 0 8.17. Major oxide chemistry for C a S i 0 grains (wt%) 8.18. Cation calculations for C a S i 0 grains 8.19. Major oxide data for olivine grains (wt%) 8.20. Cation calculations for olivine grains 8.21. Major oxide data for T A P P and eGrt grains (wt%) 8.22. Cation calculations for T A P P and eGrt grains 8.23. Major oxide data for magnetite grains (wt%) 8.24. Cation calculations for magnetite grains 8.25. Chemical data for sulphide grains listed as weight percent for each element (wt%) 8.26. Cation calculations for sulphide grains 8.27. Major oxide data for perovskite grains (wt%) 8.28. Cation calculations for perovskite grains 8.29. Major oxide data for 'exotic' C a - S i - O grains (wt%) 8.30. Cation calculations f o r ' e x o t i c ' C a - S i - 0 grains 8.31. Touching phases and their associations for R i o Soriso diamonds 8.32. Major oxide data for touching inclusions o f M g S i 0 and olivine (wt%) 8.33. Cation calculations for touching inclusions o f M g S i 0 and olivine 3  2  4  3  3  3  3  3  xiv  15 70 71 73 76 126 136 138 140 141 142 147 148 148 149 150 150 151 160 161 164 164 167 167 170 170 172 173 176 176 177 178 181 181 183 183 186 188 188  8.34. 8.35. 8.36. 8.37. 8.38. 8.39. 8.40. 8.41. 8.42. 8.43. 8.44. 8.45. 8.46. 8.47.  Major oxide data for touching inclusions o f M g S i C h and T A P P (wt%) Cation calculations for touching inclusions o f M g S i 0 3 and T A P P Major oxide data for touching inclusions o f ferropericlase and olivine (wt%) Cation calculations for touching inclusions o f ferropericlase and olivine Major oxide data for inclusions i n diamond 1-5 (wt%) Cation calculations for inclusions in diamond 1-5 E M P A and E D S data for composite grains o f CaSiC»3 and other C a phases Major oxide data for secondary magnesioferrite spots on fPer (wt%) Cation calculations for secondary magnesioferrite spots on ferropericlase Published data on ferropericlase diamond inclusions Published data on MgSiC»3 grains with a probable deep origin (>660 km) Published data on C a S i 0 diamond inclusions Published data on olivine grains with a probable deep origin (>400 km) Aluminous silicates and T A P P data from select localities with a deep origin 8.48. Si02 data from select localities worldwide 8.49. M g ' s for olivine-ferropericlase-MgSiCh associations i n this study 3  xv  190 190 191 191 192 193 195 198 198 202 203 206 207 . 208 212 219  Acknowledgements  First and foremost, I want to thank my mother, father and brother who have always supported me during my time at U B C . Although contact with them has become increasingly sporadic and they know little about my thesis, they certainly deserve most o f my thanks.  There are many people who I am indebted to for help in all aspects o f completing my thesis while at U B C .  The guidance, patience and generosity o f my supervisor, M a y a  Kopylova, have helped me immensely towards completing my thesis.  M a t i Rudsuepp  deserves thanks for teaching me to use much o f the analytical equipment in the department - and although I didn't initially appreciate his constant heckling, I soon came to accept it, and, strangely, now appreciate his humour! I would also like to thank people in the E O S building that help with the general efficiency o f the running o f the department.  In particular, I would like to thank Bryon Cranston, for his ability to find  any equipment with little notice, and Ray and Doug for their expert construction o f gadgets that became vital for the collection o f data for my thesis. I would also like to thank Mark Hutchison for his eagerness and willingness to help and to answer questions regarding Juina diamonds.  There are many people that have kept me sane and helped me see and enjoy many o f the wonderful things that Vancouver has to offer. A s my project came closer to completion, I worked more days and longer hours, and I am grateful to Heidi for putting up with my ridiculous schedule and being such a great girlfriend.  Chad and Alastair have been my  housemates for over two years and have become great friends over that time. People that deserve thanks are too numerous to list individually, so I offer a general thank you to all the graduate students that I have met while studying here, most o f which I now consider as friends.  xvi  1.0 Introduction  1.1 Motivation for the project  There are two main motivations for this thesis.  The first is to study the rare mineral  inclusions i n the diamonds and the second is to characterise the suite for exploration purposes.  Diamonds recovered from the Juina area o f Brazil contain a rare set o f mineral inclusions which suggests that the diamonds crystallised i n the lower mantle (e.g. Harte and Harris (1994); W i l d i n g (1990); Hutchison (1997); Kaminsky et al,  (2001a)).  The mineral  inclusions found are extremely rare and are recovered from less than 1% o f diamonds worldwide (Stachel, 2001).  These rare inclusions have also been recovered  from  diamonds from other continents, e.g. Australia (Scott-Smith et al., 1984), South Africa (Scott-Smith et al, 1984), Western Africa (Stachel et al, 2000b), central U S A (Otter and Gurney, 1989) and the Northwest Territories, Canada (Davies et al, generally rare.  1999a), but are  There are three localities where these rare inclusions make up a  significant proportion o f the diamond population: D O - 2 7 , N W T , where they make up - 2 5 % o f the total population (Davies et al,  1999a); Kankan, Guniea, where the  proportion is unreported but significant (Stachel et al, 2000b); and Juina, where most diamonds are considered to be from the lower mantle (Hutchison, 1997). This study w i l l examine 69 diamonds from the region that has historically produced the highest proportion o f diamonds containing these rare inclusions.  M a n y rocks found on the Earth's surface have origins i n the deep Earth (e.g. xenoliths, orogenic massifs, and diamonds, among others). O f these materials available for study, diamonds are unique i n that they are the only medium that can preserve a pristine sample o f deep mantle at the Earth's surface. Diamond is a relatively inert mineral and may act as' an impermeable seal around mineral inclusions that were accidentally trapped during  1  diamond crystallisation. Thus, diamonds have the ability to preserve an uncontaminated part o f the mantle, albeit very small.  Material from all other sources do not have the  benefit o f having been isolated for their existence outside o f the mantle and may have their chemistry changed through mixing or alteration. Pristine diamond inclusions are truly our only samples from the deep Earth that geologists can collect for study, and for this reason, they are extremely valuable. M a n y o f the inclusions found in diamonds from Juina are interpreted as having origins at depths >660 k m , which is considerably deeper than where most diamonds are sourced (-200-250 km). These diamonds are indeed rare and are o f extreme importance i n terms o f scientific study. Because o f the rarity o f these inclusions, the published database is very limited. This study w i l l increase the size o f the current database for further interpretation.  The diamonds for this study were recovered from the R i o Soriso, located i n the diamond mining district o f Juina. Juina is the largest producer o f diamonds i n Brazil, at roughly 10 m i l l i o n carats per year (Teixeira, N . ( R T Z Mineracao), 1997 personal communication to M . T . Hutchison). A l l economic operations recover diamonds from channels, paleochannels, flats and terraces i n the region. Large amount o f diamonds are being mined out of this drainage system and yet geologists cannot pinpoint the source for the abundant diamond deposits. There are a number o f kimberlites i n the region, which are an obvious source, but they are generally poorly studied. Another candidate for the diamond source is that o f local secondary collectors.  Characterisation o f diamond suites is a necessary step towards determining i f the alluvial diamonds are indeed sourced from the local kimberlites, the secondary collectors or some other, yet undiscovered, source. Characterisation o f diamond suites (also referred to as a diamond fingerprinting)  can involve any o f a number o f techniques.  This study w i l l  characterise the R i o Soriso diamonds i n terms o f morphology, colour, fluorescence, cathodoluminescence, internal morphology, impurities and inclusions. Through these studies, it was possible to subdivide the suite into subpopulations.  2  1200 km  1.2 Location map  R i o Soriso is located i n the Juina Province, Mato Grosso State, Brazil, which is approximately i n the centre o f South America at the west-central margin o f Brazil (Fig. 1.1). The river is located approximately 550 kilometres northwest o f Cuiaba (by air) or 724 k m by ground transit. Fig. 1.1. Location map of the Juina area.  1.3 Geology  1.3.1 Continental scale geology  In 2001, Brazil was the 1 1  th  largest  diamond producer i n the world i n terms o f volume ( 1 4 (Mining 2003).  Journal, London,  A u g 23,  A l l economic quantities are  recovered other  i n terms o f value)  th  from  placer deposits and  secondary  collectors.  Four  cratons are recognised i n Brazil (Fig. 1.2),  with  most  diamond  deposits  occurring on or slightly o f f either the Amazonica  (Amazonian)  or  Sao  Francisco cratons. There are two large continental scale lineaments defined by aeormagnetics  surveys  and  aerial  photography: lineament 125°AZ, which  Fig. 1.2. Location of cratons, major lineaments and kimberlite provinces (numbered) in Brazil. 1 Ariquemes, 2 - Pimenta Bueno, 3 - Vilheno, 4 - Juina (Aripuana), 5 - Paranatinga (Batovi), 6 - Poxoreu, 7 Amorinopolis, 8 - Alto Paranaiba, 9 - Presidente Olegario, 10 - Bambui, 11 - Lajes, 12 - Redondao, 13 - Santa Filomena-Bom Jesus (Gilbues), 14 - Picos, and 15 - Jaguari-Rosario do Sul. Modified from Tompkins (1992) and Hutchison (1997).  3  trends N W - S E (Bardet, 1977), and the Transbrasiliano lineament, which trends S W - N E and continues into Africa (Schobbenhaus and Campos, 1984). The Blumenau lineament, although much smaller i n scale, is also recognised (Hartman et al, 1980). A l l lineaments were reactivated during the opening o f the South Atlantic, which resulted i n the emplacement o f numerous alkaline intrusions along these lineaments (Tompkins, 1992). Lineament 125°AZ has been interpreted as a continental extension o f oceanic fractures i n the South Atlantic (Bardet, 1977).  The Amazonian craton is surrounded by Neoproterozoic orogenic belts and is divided into six geochronological provinces (Fig. 1.3). Juina is situated i n the R i o Negro-Juruena Province (1.8-1.55 Ga) and is bounded by the older Ventuari-Tapajos Province (1.951.80 Ga) to the north, northeast, and the younger Sunsas Province (1.25-1.0 Ga) to the southwest (Tassinari et al., 1999).  The basement rocks o f the R i o Negro-Juruena  Province are mostly composed o f granitic gneisses and granitoids o f tonalitic and granodioritic composition (Tassinari et al., 1999).  There are fifteen recognised kimberlite or alkaline rock provinces i n B r a z i l and each falls along one o f the three lineaments previously introduced (Fig. 1.2) (Tompkins, 1992; Svisero, 1995). Diamonds are recovered from two main districts, Mato Grosso (includes numbers 1-4, although most diamonds are recovered from Juina, F i g . 1.2) and Minas Gerais (numbers 8-10, F i g . 1.2).  1.3.2 L o c a l geology  The Junia mining district lies between 59° and 60° West and 11° and 12° South (Fig. 1.4). Lineament AZ°125 does not have a topographic expression i n the Juina area but forms a basement feature that passes just to the south o f the mining district.  Diamonds are  recovered from streams that lie in the more rugged terrain to the north, as well as the less rugged areas to the south that are mostly obscured by Phanerozoic cover.  4  I 50°W Atlantic Ocean  Geochronological Provinces  Geological Units  Central Amazonian >2.3Ga Maroni-ltacaiunas • • 2 . 2 - 1 . 9 5 Ga • Ventuari-Tapajos ••l.95-1.8Ga Rio Negro-Juruena 1.8-1.55 Ga Rondonian-San Ignacio 1.5-1.3 Ga • Sunsas • 1.25-1.0 Ga  Phanerazoic covers  '"  § Granitoids I Precambrian Sedimentary covers Acid-Int. Volcanic covers I Basic Volcanism | Greenstone Belts I Granulitic complex m  Neoproterozoic mobile belt Basement Structural high  Fig. 1.3. Geochronological provinces and the main lithological associations of the Amazonian craton (reproduced from Tassinari et al., 1999). The star indicates the location of the Juina mining district.  5  R i o Soriso is the second northernmost river in the Juina mining district. It is fed by the Chicoria Creek and i n turn feeds into the Aripuana River, which flows into the Madera River and eventually drains out through the A m a z o n  river.  The coordinates o f the  junction between R i o Soriso and Chicoria Creek are approximately 59° 10' West and 11°20' South. Diamonds have been recovered from most streams in the area, particularly from many o f the small streams that feed into the R i o Cintra Larga. Diamonds from many o f these streams have been the focus o f several studies. The largest collection o f data is for diamonds recovered from R i o Sao L u i z (Wilding et al., 1991; Hutchison, 1997; Harte and Harris, 1994). A more recent publication by K a m i n s k y et al. (2001a) reports data from R i o Sao Luiz, R i o Mutum, Corrigo Chicoria and R i o Vermelho. Diamonds from several rivers (Sao L u i z , Porcao, Duas Barms, Cinta Larga) as well as diamonds from three local kimberlites, are the focus o f an ongoing study by Araujo et al. (2003).  Rio Negro Juruena Mobile Belt /  Ventuari-Tapajos Province  Neo-proterozoic Mobile Belt  Lineament  Fig. 1.4. Topographic map of Juina mining district and surroundings. Three geochrolological provinces are visible: the Rio Negro Juruena mobile belt, in which Juina sits; the Ventuari-Tapajos Province to the north; and the Sunsas Province to the south.  6  There are a number o f kimberlites in the area (Fig. 1.5) two o f which have been dated at 92-95 M a (U/Pb dating o f zircons  from  kimberlitic  breccia)  (Heaman et al., 1998). Kimberlites are located near the southwestern margin o f the Amazonian craton and are mostly emplaced in the Permo-Carboniferous sedimentary rocks o f the Fazenda da Casa  Branca  Formation.  Other  kimberlites have intruded the older R i o Negro-Juruena Province (Tassinari et al,  1999).  It is speculated that the  Chicoria Creek diamonds are sourced from at least four local kimberlite pipes  F i g . 1.5. Detailed location map o f Juina diamond mining district. Included are the local drainage, kimberlite locations, and several rivers that have been the focus o f previous studies. Reproduced from Juina M i n i n g website. R i o Soriso is located just o f f the map and drains into R i o A r i p u a n a .  ( M . Tremblay, personal communication,'.  N o kimberlites are currently being mined and it is unclear whether or not they contain economic quantities o f diamonds. It is also unclear i f they are indeed the source for all the diamonds recovered from local rivers. The Chapadao sediments (Cretaceous-Tertiary sandstones o f the Parecis Formation, Heaman et al., 1998) located at the headwaters o f the R i o Sao L u i z , are thought to be a possible source for alluvial diamonds.  7  2.0 Diamond Morphology 2.1 Introduction  Morphological studies are considered essential for fingerprinting diamond populations (e.g. Harris et al., 1975; Robinson, 1979; Orlov, 1977; Gurney et al., in print), and is o f particular importance for the suite represented here as the primary source is unknown. Through coupling morphological studies with previous experimental work, much can be determined about a diamond's history (e.g. Robinson et al., 1989). This study documents features described by others together with experimental findings to develop a history for the population, and to determine i f there is any basis for considering i f this suite comprises two or more population sub-sets, each with unique histories.  2.1.1 Crystal habit  The variety o f primary crystal habits i n diamond has been the focus o f numerous studies (e.g. Orlov, 1977; Sunagawa, 1984b). Orlov (1977) divided single crystal forms into five types: 1) octahedral, 2) cubic, 3) certain cubes and combinations o f cubes, octahedra, and dodecahedra, 4) coated stones and 5) black or dark stones (due to the presence o f inclusions o f graphite). Orlov (1977) also divided the polycrystalline habits o f diamond into five categories. They are: 1) ballas, 2) aggregates, 3) bort comprised o f small euhedral crystals, 4) bort characterized by irregular granular crystals and 5) carbonado. It is unclear why Orlov created some classes based on properties seemingly irrelevant to crystal habit, such as the presence o f graphite inclusions.  Crystal habit is a product o f numerous factors, such as growth rate o f crystal faces, temperature and pressure conditions, and the chemistry o f parental fluid (Klein and Hurlbut, 1985). Sunagawa (1984b) states that primary diamond morphology is strongly controlled by the level o f carbon supersaturation (a) between the liquid and solid phase.  8  Although Sunagawa's (1984b) classification scheme for diamond is similar to Orlov (1977), he bases divisions  on the  level  o f supersaturation.  Above  a  critical  supersaturation value (a**), unstable and abnormal growth conditions prevail which are favourable to the formation o f radiating, granular and concentric habits. conditions, polycrystalline aggregates  Under these  o f cryptocrystalline diamond such as bort,  carbonado, framesite and ballas, develop. B e l o w a critical supersaturation level (a*), stable conditions prevail i n which single crystals develop through layer-by-layer growth (discussed i n greater detail i n chapter 7.0 on growth studies).  Aggregates o f euhedral  crystals also form under these stable conditions. Between these critical supersaturation levels, hopper crystals (hollow crystals that have a skeletal texture by failing to grow faces) are expected. Most natural cubes exhibit a radiating structure (Lang and Moore, 1972) and are thus believed to form under conditions near but slightly below a** (Sunagawa, 1984b).  Growth experiments on synthetic diamond i n controlled environments have demonstrated that crystal habit is i n part controlled by pressure and temperature (Clausing, 1997) (Fig. 2.1 A ) .  Cubes crystallize at lower temperatures while octahedra crystallize at greater  temperatures.  There is a transition zone between these forms where cubo-octahedra are  stable. These experimental observations are likely the basis for Haggerty's (1986) model o f crystal form stratification within the lithospheric root, where octahedral forms crystallise i n the lowest reaches o f the root, cubes form at the most shallow depths possible for diamond stability and transitional forms are found between these extremes. However, Sunagawa (1984a) cautions application o f experimental results from synthetic diamonds to natural systems as there are still properties o f the mantle that remain poorly understood. The stability fields for diamond and graphite are shown i n F i g . 2 . I B .  9  B)  A)  Temperature (°C) 800 1  Temperature (°C) 1400 1600  1  1  1200 1  1i  1600 1~  1  c ">o  11  is  -  nd  P-T range of expcriemtnal work by Clausing (1997)  1  1  -  -  transition zone  -  -lower mantle -  Fig. 2.1. Pressure-temperature plot of primary diamond crystal form and diamond-graphite stability. A) After Clausing (1997). B) Diamond-graphite field after Kennedy et al. (1976). Horizontal lines at -14.5 GPa and 23.5 GPa mark the approximate upper limits of the transition zone and lower mantle respectively and the thick bent curve indicates an approximate geothermal gradient (from Joswig et al., 1999)  2.1.1.1 Monocrystalline diamond The three primary single crystal forms described here (octahedron, cube, and cubooctahedron) are the result o f growth.  The external morphology o f diamond, however,  does not reflect just growth; it also reflects the post crystallization history, including resorption, brittle fracturing and deformation. A s well, there are cases involving multiple stages o f growth, fracturing and dissolution.  Nevertheless, some background on the  crystal structure o f diamond is warranted.  The octahedron is the most common primary form o f diamond observed i n nature (Orlov, 1977). triangular  faces  or  three-point  Octahedra comprise eight surfaces  with  three-point  symmetry (Fig. 2.2). In diamond literature, these faces are often described using M i l l e r indices. The surface labelled (111) (Fig. 2.2) is the growth face that intersects the imaginary axes at  10  Fig. 2.2. Octahedral crystal form.  coordinates (1,0,0), (0,1,0) and (0,0,1). Because every face intersects the ai-a2-a3-axes at either positive or negative I, all faces belong to the {111} form. Throughout this report when discussing octahedral growth surfaces or features that form on these surfaces, the term '(111) surface' w i l l be used for brevity to describe all eight surfaces.  A similar  convention w i l l be adopted to describe cubic faces (belonging to the {100} form), which w i l l be referred to as (100) surfaces.  In terms o f growth mechanics, a diamond is bound by the crystal faces that take the longest time to grow; the face that nucleates and grows quickly, grows to extinction (Clausing, 1997).  Under most conditions where diamonds form, the cubic face (100)  grows to extinction, thus letting the octahedral faces develop and control the morphology of the stone.  Octahedral faces may be flat and smooth (Fig. 2.2) or they may have  stepped development.  Although not as common as octahedral crystals, cubic crystals often make up a significant proportion o f total diamond populations.  ^*(010) (001)  They are characterized by six square faces or  four-point surfaces with four-fold symmetry (Fig. 2.3). Cubic  ""aT  crystals rarely exhibit smooth faces, rather, their surfaces are usually undulating and rough.  Phaal (1965) attributes this  roughness  for  to  the  tendency  Fig. 2.3. form.  (100) aT  T  Cubic crystal  dissolution to concentrate on (100) faces.  The  cubo-octahedron  combination o f forms.  is  a  It exhibits  eight triangular faces and six square faces, for a total o f 14 (Fig. 2.4). The  ratio o f size o f octahedral to  Fig. 2.4. Cubo-octahrdron, a combination form of cube and octahedron. Crystal morphology varies by relative size between (111) and (100) faces.  11  cubic faces can vary so that the crystal appears from nearly cubic to nearly octahedral. Laboratory experiments have shown that there is a full transition from octahedron to cubo-octahedron to cube and that a combination form with a large (111):(100) ratio likely formed at higher temperatures than a crystal with a lower (111):(100) ratio.  2.1.1.2 Polycrystalline d i a m o n d  Polycrystalline diamonds can be broadly divided into two classes, twins and aggregates.  Aggregates, or  what Orlov (1977) refers to as variety V I I diamond, are the coalescence o f multiple octahedral  crystals  (Fig. 2.5). Harris et al. (1975) describe aggregates as being composed o f two or more diamonds in some form o f conjunction. This includes a diamond entirely enclosed within another, a diamond embedded within the  surface  of  another,  or  multiple  stones  Fig. 2.5. Photograph of diamond aggregate (this study).  unconformably aggregated together. A twin, however, is the symmetrical intergrowth o f two or more crystals o f the same substance (Klein and Hurlbut, 1985). Unlike aggregations, twinning is crystallographically controlled. The most common diamond twin observed has a triangular morphology and is called a made (Fig. 2.6).  The classification o f an aggregate in this study  essentially follows that o f Harris et al. (1975); any stone containing two or more crystals that do not share the  same  crystallographic  axis  is  considered  an  aggregate. When possible, the individual crystals that make up an aggregate are described i n terms o f single crystals.  12  Fig. 2.6. Photograph of a diamond macle twin. The flattened triangular shape is characteristic of macles. The shared crystallographic axis is parallel to the plane of the page, and hence the reason why the second crystal is not visible.  Other forms o f polycrystalline diamond, such as ballas, bort, framesite, stewartite and carbonado (Orlov, 1977; Sunagawa, 1984b) and yakutite (Kaminsky, 1992) were not observed i n this study.  2.1.2  Resorption  Rounded crystals are common to most diamond suites and, as such, led many to believe that rounded morphology was a primary crystallographic form o f diamond. The failure o f modern experiments to crystallize rounded diamond (e.g. Bovenkerk, 1961; Clausing, 1997) has weakened the growth argument.  The ability o f experiments to reproduce  rounded crystals through diamond dissolution (e.g. Kanda et al, 1977) strongly suggests that  rounded  diamonds  are  a  product  of  resorption.  The  development  of  cathodoluminescence as a tool to examine the internal growth habit o f diamond has shown conclusively that all rounded diamonds are a result o f dissolution (see Moore and Lang (1974) for a more comprehensive history o f this debate).  It is generally accepted that rounded diamonds are the result o f dissolution. However, there is still no consensus on the corrosive agent responsible for the resorption o f diamond. Examples o f diamonds partially exposed i n mantle xenoliths often exhibit nonuniform resorption, with the exposed part o f the diamond more resorbed than the portion enclosed by the xenolith. This observation is cited as evidence that kimberlite magma dissolves diamond. Robinson et al (1989) proposed that the wide variation o f resorption observed i n one kimberlite reflects the time at which each diamond was liberated from its hosting xenolith. Diamonds released at great depths during magmatic ascent are more resorbed than diamonds that are liberated near the Earth's surface.  However, Haggerty  (1986) and Pattison and Levinson (1995), among others, have observed euhedral microdiamonds i n some diamond suites. younger  than  macrodiamonds  from  They suggest that these diamonds are much  the  same  kimberlite.  It  is  unclear  why  microdiamonds, which one would expect to be most resorbed due to their small surface to  13  volume ratio, lack signs o f dissolution.  Pattison and Levinson (1995) propose that  microdiamonds crystallise from kimberlite magma.  In any case, it is likely that  kimberlite magma is corrosive to diamond when specific conditions are met (perhaps when oxygen fugacity levels are conducive for carbon dissolution), however, other corrosive agents (i.e. carbon dioxide, steam, and oxygen, among others) may play a role in diamond dissolution. Cathodoluminescence ( C L ) studies o f polished diamond surfaces reveal that many crystals have experienced numerous periods o f growth and dissolution, indicating that kimberlitic magma is not the only possible corrosive agent, and that dissolution can occur in the mantle where diamonds reside.  Because the rounded shape o f diamond was once believed by many to be a product o f growth, we are left with confusing terminology to describe these rounded grains. It was recognized that unlike true crystal faces, the surfaces on rounded diamonds are curved. A s such, it was agreed that any crystallographic term used should end i n the suffix ' o i d ' , to signify that the form lacks flat faces. Robinson (1979) and many others use the term tetrahexahedroid ( T H H ) , which has 24 trigonal faces, to describe rounded crystals o f initially octahedral habit. Others, such as Kaminsky et al. (2000) and Moore and Lang (1974), use the term dodecahedroid (12 rhombic faces) to describe these rounded crystals. The term 'combination O - D ' is used by Kaminsky et al., (2000) to describe transitional crystals between the dodecahedroid and the octahedron.  There is no consensus as to  which terms should be used to describe rounded and partially rounded diamonds, and there likely never w i l l be i f resorbed shapes are to be described using crystallographic terminology. However, it is recognized that most published data includes these terms and for the purposes o f comparison, they are included i n this study.  Otter and Gurney (1989) developed a classification scheme, first proposed by D . Robinson, which avoids the problem o f describing shapes unrelated to crystal growth with crystallographic terms.  They divided the degree o f resorption based on percent  preservation into 5 classes, with class 1 describing diamonds that have between 1-55% o f  14  their  initial  preserved  and  greater  than  crystal class  5  95%  preserved. M c C a l l u m et  95%  >99%  85%  75%  62.5%  1-55%  al.  Fig. 2.7. Semi-quantitative resorption classification scheme for crystals that initially grew as octahedrons. Numbers on top refer to the resorption class while the percent values on bottom represent the amount of crystal preservation (after McCallum et al, 1994).  (1994)  developed  this  classification to include a sixth includes  greater than 99% preservation.  further  class,  which  forms  with  F i g . 2.7, from M c C a l l u m et al. (1994) illustrates the  percent preservation for each class. Although no such classification scheme exists for cubic or cubo-octahedral forms, certainly a similar scheme could be developed.  It is suggested here that the terms dodecahedroid and tetrahexahedroid be dropped from scientific papers as descriptive terms for diamond morphology. The lack o f consistency in terminology, coupled with what is deemed an incorrect usage o f a crystallographic term, makes these terms confusing and unscientific.  Crystallographic terms  say  something intrinsic about a crystal and should not be used to describe a secondary, superficial modification o f a crystal. The resorption class scale suggested by M c C a l l u m et al. (1994) avoids confusion and is semi-quantitative. Similar resorption classification  Table 2.1. Conversion between various terminology classification schemes Resorption Class (McCallum et al, 1994)  Percent Preservation  McCallum et a/.'s, (1994) suggested terminology  Terms used in this study  1 2 3 4 5 6  1-55 55-70' 70-80 80-90 90-99 >99  Tetrahexahedroid Octahedral tetrahexahedroid Transitional octa-THH Transitional THH-octa Tetrahexahedroidal octahedron Octahedron  Tetrahexahedroid Dodecahedroid Transitional O-D Transitional O-D Transitional O-D Octahedron  Semi-quantitative resorption classification and their equivalent qualitative terms, from McCallum et al, (1994) and terminology used in this study.  15  schemes should be developed for the other primary, single crystal forms o f diamond; namely, for cubic and cubo-octahedral forms.  2.1.3 Crystal regularity  Crystal regularity is a measurement based on comparison o f the lengths o f three orthogonal axes.  Following terminology used by Robinson (1979), equidimensional  grains have three axes o f equal lengths, slightly distorted grains have axes o f similar length, flat grains have two axes o f equal length and one that is less than 1/3 o f the others, elongate grains have two axes that are less than 1/3 the length o f the third axis, and irregular grains are those that fall outside o f any o f the above classes.  2.1.4 Crystal intactness  Intactness is a term used to describe the amount o f the original crystal lost to brittle fracturing.  It does not consider the loss o f diamond to etching or resorption. The four  divisions used in this classification are: intact, broken, fragment, and fraction (after F . V . Kaminsky, personal communication).  A n intact crystal retains a l l growth and/or  resorption surfaces, a broken grain retains greater than 2/3 's o f the original crystal, a fragment comprises between 1/3 and 2/3's o f the initial crystal, and a fraction represents less than a third o f the initial crystal. Although this classification scheme is subjective, the divisions convey a general idea o f crystal intactness.  2.1.5 Surface features  Numerous surface features o f diamond have been documented most thoroughly b y Robinson (1979). M a n y features reveal significant events i n a diamonds history while others appear to be manifestations o f one event affecting another. F o r example, shieldshaped laminae are the result o f partial resorption o f stepped growth on octahedral faces.  16  A s such, the presence o f shield-shaped laminae are not recorded, instead, the growth and resorption events are documented.  Features which were deemed important i n terms o f  adding more to the story were documented, while those that were deemed superfluous were not recorded. For this reason many o f the 41 pristine surface features described i n Robinson (1979) are not mentioned here.  In this study, surface features examined for  include: etch pits (trigons, hexagons, and quadrons); etch channels; corrosion sculptures; frosting; deformation laminations; hillocks; green spots; fracturing; and mechanical wear.  There is no obvious way to divide surface features.  Some clearly form earlier than  others, some are produced by chemical dissolution while others form as a result o f mechanical wear, and some are restricted to growth faces while others form only on resorbed surfaces.  The approach adopted here was to divide surface features into two  categories, those that formed as a result o f etching, and all other features.  2.1.5.1 Surfaces features associated with etching  Etching is a common event that affects the vast majority o f diamonds. C o m m o n features observed are etch pits, etch channels, and frosting. Some features occur during residence in the mantle, others occur immediately after magma emplacement, some features are restricted to particular faces while others show no preferential development on particular faces. The coupling o f detailed examination o f diamonds and laboratory experiments has led to the establishment o f constraints on some environmental conditions.  For this  reason, etching observations may be considered important i n terms o f piecing together the history o f the diamonds.  Trigons and hexagons are found on (111) faces while tetragons are restricted to (100) faces.  Most other features associated with etching are found on resorbed surfaces.  Trigons and tetragons may be flat-bottomed or point-bottomed (pyramidal) while only flat-bottomed hexagons  exist (Robinson, 1979).  17  Trigons exhibit either positive  orientation, whereby the apices  Positive orientation  o f the trigon points i n the same direction as the apices o f the  Cubic {100} faces  (111) face, or more commonly, they  exhibit  •  Negative orientation  o  negative  orientation, whereby the apices o f the trigon point to the long edges o f the octahedral  face  (Fig. 2.8). In a similar manner,  Octahedral {111} faces  / v\  Fig. 2.8. Etch pit orientation on cubic and octahedral faces.  tetragons display both positive and negative orientations.  Tolansky (1955) proposed that trigons, hexagons and tetragons are growth features, or more precisely, a result o f growth failure. However, experiments by Sunagawa et al. (1984) have demonstrated quite conclusively that these features are products o f etching and that they are focused on the outcrop o f screw dislocations i n the crystal structure. The direct cause o f this attack on the structure o f diamond is unclear, but there have been numerous proposals as to the corrosive agent(s) responsible. kimberlite magma, steam,  Some hypotheses are  carbon dioxide, oxygen, chlorine, bromine, hydrogen,  hydrogen fluoride and hydrogen bromide (Robinson, 1979).  Various experiments have been done to induce etch features o n diamond (Harris and Vance, 1974; Robinson, 1979). Results vary considerably depending on etchant used and temperature and pressure conditions. Robinson (1979) summarises the results o f many o f these experiments.  T w o o f these points are considered pertinent to this study and are  reproduced here: 1) negatively oriented features require temperature i n excess o f 950°C and etchants are likely carbon dioxide and steam, and 2) at low pressures, oxygen gas and strong oxidizing agents are the only etchants capable o f creating positively oriented etch features at temperatures between 450 and 1000°C. From these observations, Robinson et  18  al. (1989) concluded that negatively oriented etch features form earlier than positively oriented ones and that different orientations.  etchants and temperatures may explain different  Experiments by Phaal (1965) demonstrate that the orientation o f etch  features is controlled by the same conditions for diamonds o f both cubic and octahedral habit. A s well, it was noted that cubic faces are more easily etched and that when both tetragons and trigons occurred on the same diamond, tetragons were more pronounced and better developed (this point was mentioned i n section 2.1.1.1 on the roughness o f natural cubic forms).  The formation o f hexagons is interpreted by Phaal (1965) to be the product o f two etchants acting simultaneously.  Oxygen gas is responsible for the positively oriented  component while steam or wet carbon dioxide is responsible for the negative component. Hexagons have been produced i n etching experiments under low pressure at temperatures between 950 and 1000°C (Evans and Sauter, 1961), although Robinson (1979) states that they can be formed at temperatures above 1000°C.  Etch channels, also referred to as ruts, are straight or, more commonly, sinuous grooves that penetrate the diamond as a result o f etching (Orlov, 1977; Robinson, 1979). Ruts postdate resorption and typically: 1) trace octahedral to subconchoidal planes on resorbed surfaces,  2)  radiate  from  interpenetrantly-twinned  inclusion  crystals  pits,  or  3)  (Robinson, 1979).  develop Orlov  along (1977)  seams  between  attributes  the  formation o f ruts to oxidizing fluids coming i n contact with diamond. H e proposes that these fluids penetrate cracks within a xenolith hosting diamond, thus locally etching the diamond.  Frosting is another common surface feature, and is described as a clouding or frosting o f the diamond's appearance.  It typically forms on rounded (resorbed) diamonds and  roughens smooth surfaces; i n the process, an otherwise transparent crystal becomes semitransparent. Etching experiments by Robinson (1979) demonstrate that frosting can be  19  reproduced by chemical dissolution o f diamond. Although frosting preferentially forms on rounded diamond, it is not restricted to resorbed faces and is sometimes observed on flat faces and fracture surfaces.  Robinson's (1979) experiments produced two types o f  frosting, 1) coarse frosting, which results from rapid etching by either wet carbon dioxide or steam with subordinate free oxygen at temperatures between 950 and 1000°C, and 2) fine frosting, which results from rapid etching b y oxygen gas at temperatures at or below ~950°C.  Based on cross-cutting relationships, Robinson (1979) and Robinson et al.  (1989) interpret frosting as a late-stage etch feature that occurs at a similar time to positively oriented etch pits.  2.1.5.2 Other surface features  Surface features not associated with etching are described here. These are: deformation laminations, hillocks, fracture surfaces and mechanical wear.  Deformation laminations (also called deformation lamellae) are lines or striations that form as a result o f plastic deformation o f diamond (Urusovskaya and Orlov, 1964) and are usually visible only after resorption (Robinson et al., 1989). Laminations form as the crystal structure glides along the (111) plane, and thus are crystallographically controlled. Striations are more resistant to dissolution than the rest o f the diamond (due to work hardening), and become positive features after resorption (De Vries, 1975). They may form as one, two or even three sets o f parallel lines on a resorption surface. Experiments by D e Vries (1975) have shown that ductile deformation o f diamond begins  at  temperatures between 900 and 1100°C and pressures between 10 and 60 kbars.  Hillocks are a loosely defined surface feature.  Orlov (1977) describes hillocks as  pyramidal and drop-shaped features o f positive-relief which are controlled by crystal habit but are visible only on rounded faces. Robinson (1979) divided hillocks into five subcategories: 1) elongate hillocks, 2) ellipsoidal hillocks, 3) semi-cylindrical hillocks  20  with hexagonal pits, 4) transverse hillocks, and 5) pyramidal hillocks. Robinson (1977) states that the first four types are controlled by the same factors as suggested by Orlov (1977), but asserts that some pyramidal hillocks are associated with deformation laminations and may form at the intersection o f two sets o f lines.  Other workers  associate elongate hillocks with deformation laminations (e.g. M c C a l l u m et al., 1994; M c K e n n a , 2001).  A t one time diamond was believed to be so strong that it could not be broken (Krajick, 2001). W e now know this not to be true; most diamonds have experienced at least partial brittle fracturing i n the mantle, during kimberlite ascent or during residence i n the surficial environment. In this study two types o f brittle fractures were observed: 1) 'preemption' fracture surfaces, those formed during residence in the mantle that appear to pre-date at least some resorption; 2) and late-stage fractures, which are formed either during magma ascent or in the surficial environment and are characterised by conchoidial surfaces.  The latter category is somewhat related to the degree o f intactness and this  study aims to be consistent between the two classifications.  Mechanical wear, like dissolution, preferentially attacks the corners and edges o f diamond.  However, unlike dissolution, mechanical wear produces a rough texture on  these surfaces and is clearly distinguishable from the chemical wear .of diamond. Abrasion occurs during residence i n the surficial environment and is a common feature on alluvial and paleo-alluvial diamonds that have been transported over long distances (i.e., diamonds found off the west coast o f Namibia).  2.1.6 Sequence of events  A summary o f a diamond's history can be determined by placing all the  features  documented in this chapter in order o f occurrence, mainly based on cross-cutting relationships. The sequence o f events (Fig. 2.9) is reported by Robinson et al. (1989): 1)  21  diamond  growth,  2)  plastic  1. G r o w t h face 4. Resorption  deformation, 3) etching i n the form  of  negatively  features, 4) resorption and 5) etching i n the form o f positively oriented  features  development  of  ""^3. N e g a t i v e l y oriented e t c h i n g  oriented  and  the  frosting.  However,  this  'sequence  of  events'  is  clearly  an  oversimplification  as  3?v- 2. L a m i n a t i o n lines  /LJ W  jlfer^jSPffl  2~>"is£X  5. P o s i t i v e l y oriented etching and frosting  Fig. 2.9. The relative timing of common features observed on diamond. The number refers to the relative timing of each feature. Features are discussed in detail in this chapter. Growth faces (section 2.3.3), lamination lines (section 2.3.7.2a), negatively oriented etching (section 2.3.7.1a), resorption (section 2.3.4) and positively oriented etching (section 2.3.7.1a). (after Robinson et al., 1989).  cathodoluminescence studies on polished diamond surfaces  often reveal that the diamond experienced  numerous  alternating episodes o f growth and dissolution.  2.2  Analytical techniques  Diamonds were analysed using a transmitted light optical microscope and a scanning electron microscope ( S E M ) .  A Leica M Z FLILT optical microscope with a lOx zoom lens and l x objective lens was used for macroscopic examination o f diamonds.  Observations were made i n both  transmitted and reflected light mode. Digital images were collected using a Spot Insight Colour 3.2.0 digital camera and when required, were enhanced using Adobe Photoshop 6.0.  Detailed examination o f small surface features on diamond was done using a S E M . Diamonds were placed on standard aluminium stubs without being carbon coated (for fear o f creating future complications during infrared analysis). Without a carbon coat, it was uncertain whether or not the stones would charge up. The fraction o f small stones  22  (those less than 0.5 carats or ~4 m m i n size) generally did not charge up on the instrument, thus permitting detailed examination. The fraction o f larger stones, however, frequently charged up and are thus poorly represented i n S E M images. The S E M was operated in back-scattered electron ( B S E ) mode using an accelerating voltage o f 15 k V and an estimated beam current o f 1 n A .  2.3 Results  The results for the morphological characterization o f the R i o Soriso diamonds are summarised i n Appendix B . A more detailed description o f the features catalogued is presented below.  2.3.1  Size  The  diamonds  range  studied  considerably  weight,  from  0.003  in  0.7  0.6  to  m  wis  0.404 grams (0.015-2.02 carats).  A  plot  Small  • Large  of  diamond weight against normalized  frequency  (Fig. 2.10) demonstrates  0.2  Log normal |\^<" distribution  the bimodal character o f the diamond suite that it does not lognormal  and fit a  distribution,  as would be expected for a randomly selected suite of  one  ML  J 0.4  0.8  1 1.2 WoigW (carats)  1.6  Fig. 2.10. Weight in carats versus normalized frequency. A log normal distribution is superimposed on the graph to illustrate that the data does not fit. For some correlations the population is divided into two halves, large (>0.5 carats) and small (<0.5 carats).  population  23  (Boggs, S., 1987). The size distribution of the samples does not reflect the diamond population as a whole, but is an artefact of sampling bias.  Large diamonds were  preferentially selected for study as there is a greater probability that they contain more and larger inclusions than small diamonds.  2.3.2 Crystal habit  Of the 69  diamonds  available for study, 64 were  monocrystalline  and  5  were  polycrystalline.  None  of the monocrystalline diamonds  exhibits  well-formed  primary  crystal morphology, as described 2.1.1.  in  section  Fig. 2.11. Images of an octahedral crystal (Diamond 4-18). A ) S E M image of flat-faced crystal growth. Step development is minor and resoprtion is absent. B) Photograph of same diamond showing the complex nature of the crystal.  Octahedral  (111) faces are rare or obscured and cubic (100) faces are absent. Diamond 4-18 is the closest crystal to an octahedron (Fig. 2.11 A and B), but has a somewhat more  complex  morphology  than  for  a  perfect  octahedron (e.g. Fig. 2.2). It should be stressed here that this does not imply that octahedral growth did not occur. The results of this work will show that most crystals likely grew as octahedrons (or at least grew octahedral faces) but that these faces have since been rounded by dissolution, broken by brittle fracturing, and/or deformed by plastic deformation.  The five  24  Fig. 2.12. S E M image of aggregate of octahedral crystals (Diamond 68).  polycrystalline stones are classified (e.g.  as  16  aggregates  F i g . 2.12).  Single Crystals  14  12  Two  10  aggregates are comprised of  only  two  crystals  2J 6  a.  4  (diamonds 2-1 and 4-11)  2  while the other three are  0  class 1  comprised o f more five  individual  class 2  class 3  than  class 5  class 6  nonuniform  unknown  class 6  nonuniform  unknown  crystals Aggregates  (diamonds 4-15, 5-8 and 68).  class 4 Resorption  N o crystals appear to  share  a  common  crystallographic  axis  and  u.  l  are thus aggregates and not twins.  class 1  class 2  class 3  class 4 Resorption  2.3.3  class 5  Fig. 2.13. Frequency versus resorption class. Top graph is for single crystals (n=64) and bottom graph is for aggregates (n=5). Nonuniform stones possess two resorption classes. Resorption classes are from McCallum et al. (1994).  Resorption  The degree o f resorption o f monocrystalline  and  polycrystalline stones was estimated using F i g 2.7 as a guide. Data is presented in  Table 2.2  heading category'  under  the  'resorption and  the  population distribution is graphically Fig. 2.13.  presented  in  Some crystals,  Fig. 2.14. Frequency versus external morphology. The abbreviations used are: THH - tetrahexahedroid; D - dodecahedroid; O-D transitional octahedron-dodecahedroid; O - octahedron; U unknown; and A - aggregate. Table 2.1 explains the conversion between these terms and the resorption scale used in Fig. 2.9.  25  typically those which are considered fractions (see section 2.3.5), could not be classified and are labelled 'unknown'.  Some crystals possess non-uniform resorption and are  classified as such. Stones were also described using morphological terms (described i n section 2.1.2) so that the data can be compared with other data sets (Fig. 2.14). Consistency between the two classification schemes was retained whenever possible, although many more crystals were classified as having an unknown morphology using the descriptive classification. (See Table 2.1 for conversion between the M c C a l l u m et al.,  1994  semi-quantitative  classification scheme).  classification scheme  and  the  qualitative descriptive  Diamonds i n this study are strongly resorbed, with the average  degree o f dissolution falling between classes 2 and 3 for single crystals while aggregates tend to be less resorbed. S E M images o f typical examples o f rounded forms can be seen in Fig. 2.15.  Fig. 2.15. SEM images of diamond resorption. A) resorption class 3, etch features on flat crystal face, resorbed corners on right (diamond 6-4); B) resorption class 3, black surface in middle of photograph is crystal face (diamond 2-1); C) resorption class 1, no flat crystal faces (diamond 4-12); D) resorption class 1 (diamond 4-16); E) resorption class 1 (diamond 5-2) and; F) resorption class 1 (diamond 4-13). Resorption scale is from McCallum et al., (1994) and is reproduced in Fig. 2.9.  26  2.3.4 Crystal regularity  Twenty-nine percent of grains studied are either equidimensional or slightly distorted, 67% are considered irregular and only 4% are considered flat.  There are no elongate  grains. However, note that the irregular forms are not necessarily products of growth but may be a manifestation of other processes such as non-uniform dissolution, plastic deformation and brittle fracturing.  2.3.5 Crystal intactness  Most crystals are considered to be either 35  broken (45%) or fragments (29%), with fraction and fully intact crystals making  >2/3  ~1  30  >l/3  25  up the remainder of the population  20 8 g.15 &  (11% and 14% respectively) (Fig. 2.16). Examples of each degree of intactness can be seen in Fig. 2.17.  Brittle Intact  fracturing, either during residence in the mantle,  residence  in  the  surficial  environment or during mining are the reason(s) for the small percentage of  Broken  Fragments  Fractions  Crystal Intactness  F i g . 2.16. Frequency versus degree o f intactness for a l l diamonds i n study. Numbers 1, 2/3, and 1/3 represent the estimated amount o f the original diamond crystal preserved.  fully intact crystals (see section 2.3.9).  F i g . 2.17. Photos and S E M image o f examples from this study for the four classes o f 'intactness'. F r o m left to right: Intact crystal (diamond 3-1); broken crystal (diamond 4-1); fragment (diamond 6-6); and fraction (diamond 6-5).  27  2.3.6 Etching  For the diamonds in this study, trigons, hexagons, tetragons and etch channels are abundant, and found on 80% o f the stones.  2.3.6.1 Trigons Fifty-four  percent  of  the population exhibits trigonal 2.18).  pits  (Fig.  Pits are mostly  small (on the order o f 10-100  um)  however  some are as large as 0.5  mm.  bottomed  Flattrigons  are  Fig. 2.18. SEM photo of trigonal pits. Left image is single trigonal pit (diamond 5-9) and image on right shows several trigonal pits, some super-imposed on one another (diamond 6-4).  predominant but there are several instances o f point-bottomed trigons. Positively oriented trigons were verified on only 3 diamonds while negatively oriented trigons were confirmed on 24 diamonds. Roughly 1/3 o f the diamonds examined lack adequate crystal faces to confirm trigon orientation.  2.3.6.2 Hexagons  Hexagons are found on 54%  of  diamonds  examined (Fig. 2.19). A l l hexagons observed are flat-bottomed  and  Fig. 2.19. SEM photo of hexagonal pits. Left image is from diamond 415 and right image from diamond 5-3.  28  relatively large compared to trigons, some measuring up to 0.75 mm. Although some inclusion pits resemble hexagons, they invariably lack flat bottoms and are thus easily differentiated.  Trigons of both positive and negative orientations are observed in the  bases of hexagons.  2.3.6.3 Tetragons  Thirty-six percent of diamonds  examined  exhibit tetragonal pits (Fig. largest  2.20). pits  The measure  -0.25 mm across while the  average  size  measures less than 20 um.  As discussed in  Fig. 2.20. SEM photo of tetragons (sometimes referred to as quadrons). Photo on left is from diamond 4-3 and photo on right from diamond 3-8.  section 2.3.2, cubic or (100) faces were not observed, thus determining the orientation of the tetragons was difficult. Pit orientations are determined based on their relationship to (111) faces. In the rare case where (111) faces are not detectable, tetragon orientation is not possible to discern. All tetragons observed have positive orientation with the exception of tetragons on two grains. The presence of tetragonal pits is not sufficient evidence for a cubic growth face.  CL examination of polished diamond surfaces (chapter 7.0) as well as  discussion with J. Harris, has led to a more reasonable and consistent interpretation. The abundant brittle fracturing of stones during residence in the mantle (cross-cutting relationships require that fracturing occurred before dissolution in many cases) has created surfaces that are sub-parallel to (100) surfaces; subsequent etching likely formed tetragonal pits on these paleo-fracture surfaces.  29  2.3.6.4 E t c h channels  E t c h channels o c c u r i n 2 8 % o f the d i a m o n d s and are c o m m o n l y associated w i t h i n c l u s i o n pits ( F i g . 2.21).  Fig. 2.21. SEM photo of ruts or etch channels. Photo on left is a more macroscopic view of etch channels emanating from what is likely an inclusion pit (diamond 3-1). The photo on the right is a close-up of two typical sinuous etch channels (diamond 3-1).  2.3.6.5 F r o s t i n g  F r o s t i n g is o b s e r v e d o n 1 9 % o f the d i a m o n d s and is r e c o g n i s e d b y its ' r o u g h ' appearance o n the  diamond  microscope. quantitative  surface  under  R o b i n s o n (1979) divide between  the  optical  provides  coarse  and  no fine  frosting, a n d as s u c h , n o d i s t i n c t i o n w a s m a d e i n this study.  S E M e x a m i n a t i o n o f some frosted  surfaces r e v e a l a v e r y fine ' r o u g h n e s s ' to the d i a m o n d surface (e.g. F i g . 2.22). F r o s t i n g most c o m m o n l y o c c u r s o n r o u n d e d surfaces i - i , . i ^ i , h e a v i l y resorbed d i a m o n d s (classes 1 or 2).  30  of  Fig. 2.22. SEM image of fine frosting (diamond4-13).  2.3.7  Deformation L a m i n a t i o n s  Twenty percent of the diamonds examined in this study display deformation lamellae. In some cases they occur as two sets of pronounced laminations (e.g. Fig. 2.23) while in other cases they are quite obscure. There is ambiguity as some features look like both hillocks (see section 2.3.8) and deformation laminations. There may be an argument for grouping these features into one category.  Fig. 2.23. SEM image of deformation laminations. A) Macroscopic view of laminations on diamond 1-3. B) Close-up of laminations on same diamond.  2.3.8  Hillocks  Ellipsoidal and elongate hillocks were observed on 17% of diamonds examined. However, it was generally difficult to differentiate  between poorly developed  deformation laminations and hillocks. Figures 2.24D-F show three views of successively magnified digital images of the same diamond, illustrating that deformation laminations can be made up of elongate hillocks. Macroscopically, many of the features observed in Fig. 2.24 look  31  40 um Fig. 2.24. SEM images of hillocks. Photos illustrate the association between hillocks and deformation laminations. A) ellipsoidal hillocks are observed in the top half of photo while elongate hillocks are seen in bottom left (diamond 4-8), B) elongate hillocks or deformation laminations? (diamond 1-4), C ) elongate hillocks that grade into ellipsoidal hillocks as the face angle changes (diamond 3-2), D-F) the last three images show successively higher magnifications of the same features (diamond 2-9). In the first image deformation laminations are clearly visible, however, on closer examination, these lamination lines look more like what could be called 'elongate hillocks'.  32  like typical deformation laminations.  F i g . 2.24C illustrates how the morphology o f  hillocks is dependent on which part o f the rounded form o f diamond that it is being viewed on. For planes that closely parallel (111) faces, hillocks take on a more elongate morphology, while on surfaces that do not parallel (111) faces hillocks tend to take on a more rounded or ellipsoidal morphology. There appears to be an association between deformation laminations and both ellipsoidal and elongate hillocks. The morphology the hillock exhibits is controlled by the orientation relative to the (111) face on which the feature develops. Certainly, a case can be made for grouping these two surface features, however, for historical reasons (e.g. Robinson, 1979; Orlov, 1977), they are catalogued here separately.  2.3.9 Fracturing 'Pre-emptive' fracture surfaces are observed on 86% o f the diamonds and late stage conchoidial fractures are observed on 49% o f the diamonds in this study. Fifty-two percent o f stones have both 'pre-eruptive' and late stage fractures.  In  the case o f late stage fractures, it is unclear whether these formed during magma ascent, transport  in  the  surficial  environment,  or  mining.  However, the scarcity o f etching  features on these faces suggests they formed  50 um Fig. 2.25. SEM image of mechanical wear on an edge of a diamond (Diamond 5-9).  sometime after ascent.  the  final  stages o f magma  The high percentage o f fracturing has  resulted i n significant loss o f diamond. The diamonds examined seldom exhibit any signs o f mechanical abrasion.  F i g . 2.25  shows faint mechanical wear along one edge o f a diamond. The relative absence o f this feature is consistent with diamonds that have not been transported far from their source.  33  2.4 Discussion  2.4.1 Summary of the physical characteristics  The morphological characterization o f these diamonds has provided much insight into their history from initial growth tO deposition i n R i o Soriso, A more comprehensive story w i l l emerge when all studies, including studies o f impurities, diamond growth and inclusions are combined.  The morphology o f the diamonds is difficult to describe. Crystals are rarely intact and possess only remnant primary growth faces.  The diamonds are strongly resorbed, with  52% o f stones falling between classes 1 to 3 on the resorption scale o f M c C a l l u m et al. (1994) while 13% exhibit non-uniform resorption. Following Kaminsky et al. (2001a), most diamonds are considered dodecahedroids.  Single crystals make up 87% o f the  population while aggregates (7%) and unknown crystals (6%) make up the remainder o f the suite. Features that are a product o f local etching are observed on most diamonds i n the form o f trigonal, hexagonal, or tetragonal pits, as w e l l as i n the form o f etch channels. Deformation laminations are visible on at least 20% o f the diamonds.  Frosting and  hillocks are present on less than half o f the stones, while signs o f mechanical abrasion are weak and observed on one grain. Based on the studies reported i n this chapter, there is not sufficient evidence to suggest that more than one population o f stones is being represented.  2.4.2 Comparison with other diamond studies from the Juina area  Detailed studies by Hutchison (1997) and Kaminsky et al. (2001a) on Juina diamonds warrant comparison with studies from this chapter. In terms o f morphology, the results from the three studies o f Juina diamonds (including this study) are different (compare F i g . 2.26, previous studies, with F i g . 2.14 from this study).  34  However, the lack o f  Hutchison (1997)  50  Kaminsky et al. (2001a)  50  45  45  40  40  35  35  30  30 B  25  p  25  20  20  -  15  15  -  10  10  5  5  0  0  # /  /  n  * n=82  n=61  Fig. 2.26. Morphology distribution from two previous studies on Juina diamonds. For data from Hutchison (1997), it is unclear if the term made is being used according to its strict definition to describe twins (when two or more crystals share a common crystallographic axis) or whether this category makes no distinction between aggregates (no shared crystallographic axis) and twins.  consistency i n describing diamond morphology (discussed i n detail i n section 2.1.2) is a possible reason for many o f the differences. A l l three studies indicate that the diamonds exhibit complex and irregular forms thus making morphological classification difficult. A l l studies indicate the presence o f a small proportion o f multiple crystal stones, be they aggregates or macles. Another important point is that no cubes or cubo-octahedral stones were found i n any o f the Juina area diamond suites studied.  Hutchison (1997) reports that 77% o f diamonds display plastic deformation laminations while Kaminsky et al. (2001a) also indicates that plastic deformation laminations are abundant. If diamonds from this study containing either lamination lines or hillocks are combined (the reasons and defence o f such a grouping are discussed i n section 2.3.7) then - 3 0 % o f diamonds from this study may exhibit signs o f plastic deformation. this number  is still considerably low, deformation  35  Although  laminations are often  subtle.  Kaminsky et al. (2001a) and Hutchison (1997) record other surface features, such as etch channels, etch pits and frosting, but describe these only i n qualitative terms.  36  3.0 Colour 3.1 Introduction  Colouration in diamond is a result o f full, partial or the lack of, white light absorption (Fritsch, 1998). Factors which control whether diamond w i l l absorb light or not include: impurities, crystal defects, and inclusions (Orlov, 1977).  The colour classification o f  diamond is based on the body colour; any colouration due to mineral inclusions, i f visible, is ignored.  The colour o f gem quality diamond is typically classified using colour charts, such as the Gemological Institute o f A m e r i c a ( G I A ) colour grading scheme.  U s i n g this chart,  diamond colour can be ranked from colourless (D) to yellow (Z). Other 'fancy' colours are classified using different schemes.  Colours o f rough stones, however, are usually  described i n more qualitative terms, such as colourless, brown, or yellow (see Harris et al., 1975; Kaminsky et al., 2001 a; Gurney et al., in print).  3.1.1 Causes of colouration in natural diamond  Colouration i n natural diamond is the result o f impurity defects, dislocations and irradiation. M a n y impurities have been found i n diamond. B i b b y (1982) found  fifty-five  non-substitutional impurities i n diamonds from some South African kimberlite pipes; however, most colouration from impurities is a result o f two elements that substitute for carbon i n the crystal structure: nitrogen and boron (Harris, 1987). Plastic deformation o f diamond creates dislocations at which amorphous carbon precipitates.  This amorphous  carbon results i n a variety o f colours. Colouration due to irradiation is unlike colouration due to impurities and dislocations.  Irradiation typically forms only on the diamond  surface and usually only as round spots. M o r e detailed descriptions for the causes o f colouration i n diamond relevant to this study are discussed below.  37  Colourless diamonds are generally considered the most pure i n terms o f lacking impurities and crystal defects.  In a perfectly colourless diamond, no wavelengths are  absorbed and white light remains unaffected as it passes through the stone, thus emitting a colourless hue (Fritsch, 1998). Some diamonds with nitrogen impurities arranged i n platelet form are also colourless (Brunton, 1978). Although some colourless diamonds have been referred to i n literature as 'white', this term is discouraged here and w i l l refer only to 'cloudy' or ' m i l k y ' stones which are semi-opaque to opaque.  Grey colouration i n diamond is thought to be a result o f microscopic dark inclusions, most likely graphite (Orlov, 1977; Robinson, 1979). Grey colouration is not believed to be a true body colour, however, even microscopic examination o f diamonds sometimes does not resolve these inclusions. Robinson (1979) suggests that grey colouration may be a result o f crystallisation at lower temperatures. Recent work by Titkov et al. (2003) has found that magnetite is the source o f grey colouration i n some Siberian diamonds.  Numerous studies have linked brown colouration i n diamond to plastic deformation (Urusovskaya and Orlov, 1964; Orlov, 1977; Robinson, 1979). Plastic deformation likely occurs i n the mantle where pressures and temperatures are sufficient so that diamond deforms i n a ductile manner. Deformation results i n the destruction o f valency bonds and the creation o f defect centres (Orlov, 1977). Harris (1987) proposes that atomic sized graphite or amorphous carbon precipitates i n these defect centres and imparts a brown colour  Pink colouration has been linked to the same processes that impart brown colouration i n diamond (Harris, 1992; Orlov, 1977).  Harris (1987, 1992) suggests that plastic  deformation produces a continuum o f colours between pink and brown diamonds.  Two types o f yellow stones are commonly recognised i n literature: canary yellow and Cape yellow.  Colouration i n the former is a result o f paramagnetic substitution o f  38  nitrogen into the crystal structure (also referred to as type lb diamond (section 6.1.1)) while colouration i n the latter is a result o f aggregation o f nitrogen into N 3 centres (LRinactive forms comprised o f three aggregated nitrogen atoms) (Harris, 1987; Fritsch, 1998).  Orlov (1977) also recognises two main yellow hues.  The first one he terms  'straw-yellow', which is analogous to Cape yellows, and the second term is 'amber yellow', which is analogous to canary yellows although he considers this colour as being restricted to cubic diamonds.  T w o less common yellow hues are recognised i n the  literature and are interpreted to be the result o f annealing o f green pigmentation patches (Orlov, 1977) and hydrogen defects (Fritsch, 1998).  M i l k y , white or cloudy colouration i n diamond is interpreted by Orlov (1977) to be the result o f microscopic internal defects i n the crystal structure. More specifically, Navon et al. (1988) suggest that numerous small fluid inclusions are responsible for m i l k y colouration in diamond.  Green colouration is caused either by impurities i n the lattice structure or by alphaparticle damage (Fritsch, 1998).  Green colouration due to impurities is a true body  colour, while colouration due to alpha-particle damage manifests itself i n three ways: as green spots on the surface, as a homogenous green coat on diamond (Vance et al., 1973), and as a green halo around a mineral included in diamond (Kopylova et al., 1997). A l p h a particles are emitted from unstable isotopes that, i f near to or i n contact with diamond and in sufficient abundance, may cause green colouration. The length o f time for colouration to occur depends on the radiogenic element, its abundance, and distance from the diamond. Green surface spots arise when an isolated radiogenic element is the source for alpha-particles, while a uniform green coat requires a relatively even distribution o f radiogenic elements surrounding the diamond, most likely as a dissolved component i n ground waters. Green spots on diamond could form i n alluvial deposits or i n kimberlite, however intense green spot colouration is likely to be a product o f irradiation i n an alluvial environment (Vance et al., 1973).  39  Experiments have also demonstrated that  diamonds with green colouration due to alpha-particle damage turn brown on heating to temperatures greater than 500 to 600 °C (Vance and Milledge, 1972). This heat in nature could be generated by a variety o f different sources, such as a later injection o f kimberlite magma or any other proximal intrusive event, or due to intense burial i n a sedimentary basin.  3.2 A n a l y t i c a l techniques  Diamonds were analysed using a Leica M Z FLIII binocular optical microscope with a lOx zoom lens and l x objective lens. Observations were made i n both transmitted and reflected light mode.  Digital images were collected using a Spot Insight Colour 3.2.0  digital camera and enhanced using Adobe Photoshop 6.0.  3.3  Results  The majority o f diamonds were classified as uniform (93%) while the remainder o f grains contain domains o f both coloured and colourless diamond. F i g . 3.1 graphically illustrates the colour distribution i n this diamond suite.  One faint green pigmentation spot was  observed on a colourless diamond. Appendix A contains photos o f every diamond i n this study. cases,  Note that i n several the  apparent  diamond  colours seen i n the photographs  Brown  are a result o f mineral inclusions and/or  secondary  material  in  grooves and pits, and thus do not reflect the true diamond body colour. Every attempt was made to  ignore  effects  obscuring features.  of  these  • (7)  Colourless • (30) Grey  • (14)  Yellow  • (9)  Pink  a  Milky  • (2)  Nonuniform  a (5)  (2)  Fig. 3.1. Distribution of diamond colours. Number in parentheses indicates sample size.  40  3.3.1 U n i f o r m body colours  In order o f decreasing abundance, the six colours observed in this diamond suite are colourless (43%), grey (20%), yellow (13%), brown (10%), pink (3%) and milky (3%). Representative photographs o f each colour are presented in Fig. 3.2  Colourless diamonds make up almost half o f the diamond suite and are generally the most transparent stones. population.  Grey diamonds make up a significant percentage o f this  However, it w i l l be shown i n section 9.1 that the grey diamonds likely  contain graphite i n an otherwise colourless stone. For this reason, 4 3 % is a minimum for the percentage o f colourless diamonds and 20% is a maximum proportion for grey diamonds.  M a n y o f the brown stones have surface features that are consistent with  Fig. 3.2. Photographs of representative diamonds colours observed. A) colourless (diamond 4-1), B) grey (diamond 5-8), C) milky (diamond 1-5), D) brown (diamond 5-4), E) pink (diamond 2-4), and F) yellow (diamond 2-3).  41  plastic deformation.  There are nine yellow stones, all with weak colouration except  diamond 2-3, which exhibits an intense yellow hue (Fig. 3.2F). It is not clear whether these stones represent Cape or canary yellows. Because these yellow hues are a result o f impurity content (section 3.1.1), further classification o f yellow stones w i l l be considered only in conjunction with impurity data obtained from studies o f infrared spectra (section 9.1).  The two m i l k y diamonds observed were the closest grains to being considered  opaque.  3.3.2 N o n - u n i f o r m colours  Five colourless diamonds contained portions o f either brown diamond (3 stones) or grey diamond (2 stones).  One diamond examined displays fairly weak green colouration that occurs as a spot approximately 0.25 m m i n diamond and is a result o f radiation damage. It is uncertain whether colouration occurred in the host rock, in a secondary collector or i n the alluvial environment from where the diamond was recovered.  Fig. 3.3. Photograph of non-uniform colour. On left, non-uniform brown and colourless diamond (diamond 1-4), on right, alpha-particle damage (circled by white dashed line) on diamond surface (diamond 4-18).  42  3.4 Discussion  3.4.1 Comparison with other studies  Other studies on Juina diamonds report similar colour distributions.  Hutchison (1997)  divides the suite into only three categories, colourless (41% o f the population), brown (57%), and cloudy (2%). Kaminsky et al. (2001 a) does not include a detailed description o f diamond colour, but indicates that most stones exhibit shades o f brown colouration (from pale to dark brown) and are semi-transparent with a silky lustre.  This study  includes more colour subdivision: colourless, grey, yellow, brown, pink, cloudy and nonuniform. Faint yellow and brown colourations are similar and there may be evidence for grouping these two colours for the purpose o f comparison.  Grey stones are not  mentioned in either o f the previous work. In section 9.1 it w i l l be shown that many grey stones are likely colourless diamonds with abundant graphite inclusions (or some other dark inclusions).  There is evidence for grouping grey and colourless stones for the  purpose o f comparison.  A l s o for the purpose o f comparison, non-uniformly coloured  diamond can be re-distributed into either brown (3 diamonds) or colourless (2 diamonds) stones. The two pink stones found i n the study are quite remarkable i n colour (Fig. 3.2E) and may not be represented i n the other suites. Under the new grouping described, this suite comprises colourless (64%), brown (30%), cloudy (2%) and pink (3%) diamonds. The sampling bias mentioned i n section 2.3.1 may be responsible for differences i n colour distribution.  43  4.0 Fluorescence of Diamonds  4.1 Introduction  The fluorescence (FL) o f diamond is a property that has been recognised for many decades and is used i n some sorting plants to separate diamonds from other heavy minerals extracted from mines. F L is caused by the excitation o f electrons by exposure to ultraviolet ( U V ) light. U V light 'excites' valence electrons i n optical centers to higher energy states. When these 'excited' valence electrons fall back to their original energy state they emit light, referred to as F L . The F L o f diamond can be used as a preliminary method to determine qualitatively the relative concentrations o f optical centers as w e l l as distinguish between different types o f optical centers.  Its usefulness  in grouping  diamonds when used i n conjunction with other studies, such as diamond morphology and colour (chapters 2 and 3 respectively) or infrared (chapter 6) is greatly improved.  The specific defects responsible for F L colours i n diamond are surprisingly poorly documented.  Blue F L is generally considered to be a result o f nitrogen impurities i n  diamond, with the relative F L intensity reflecting the relative concentration o f nitrogen impurities. Fritch (1998) reports that 1/3 o f gem-quality diamonds fluoresce blue. Clark et al., (1992) finds that diamonds graded as 'brown' that are exposed to U V - l i g h t often fluoresce bright yellow.  Other F L colours have been reported but their specific causes  are generally unknown.  4.2 A n a l y t i c a l methods  Diamond fluorescence was examined using a 100 watt ultraviolet bulb attached to a Leica M Z F L U I optical microscope and powered by an E B Q Netz power source.  A Spot  Insight Colour 3.2.0 digital camera was used to record images o f diamond fluorescence. A steel skeleton frame fitted with a 'skirt' was designed by the in-house machine shop to  44  fit over the microscope stage to block out any contaminating light. T w o images were collected for each diamond; the first one using an exposure time o f twenty seconds while the second one was dependant on the F L intensity o f the particular stone being examined and varied from 1 to 60 seconds.  B y maintaining a constant exposure time for at least  one image for each diamond, the relative F L intensity between stones could be examined. The objective o f the second photo was to match the exposure time to the relative F L intensity so as to best capture the luminescent features for that particular stone.  Through comparison o f F L digital photographs for each diamond, stones are described i n terms o f three variables: colour, homogeneity, and intensity. Each diamond is grouped into one o f the four F L colours (or none), classified as either uniform or non-uniform, and is qualitatively rated i n terms o f relative intensity (strong, moderate, weak or very weak), determined by comparing images o f all diamonds collected using a constant exposure time.  4.3 Results  Sixty-nine  diamonds  were  examined under ultraviolet light. A  summary o f the results  Blue  is .  presented i n Appendix B under the  column  'Fluorescence'. diamond  1  heading  fluorescence  very weak (9)  •  weak (13)  •  moderate (17)  •  urquoisen  Green Brown None  Photographs o f  •  strong (16)  •  weak « r y(3) w«*a)  •  moderate (5)  •  strong (1)  O  weak(l)  •  strong (1)  •  weak (1)  •  no Eborescence (1)  for all  stones, using variable exposure times, are presented i n Appendix  Fig. 4.1. Fluorescence colour and fluorescence colour intensity distribution of Rio Soriso diamonds. (numbers beside colours in legend indicate sample size).  A.  45  Fig. 4.2. Photographs of various FL colours observed for Rio Soriso diamonds. A) Blue (Diamond 4-3), B) Turquoise (Diamond 6-6), C) Green (Diamond 7-1), D) Brown (Diamond 2-4), E) Non-uniform stone with yellow body and speckled blue overcoat (Diamond 4-7), and F) Non-uniform stone with moderately intense blue FL on left and weakly intense FL on right (Diamond 2-9).  Diamond fluorescence colours are mostly shades o f blue (80%) with turquoise (15.5%), green (3%) and brown (1.5%) colours also being observed (Fig. 4.2).  One diamond  (1.5% o f population) did not fluoresce. A photographic compilation o f diamond F L for all stones, using the same exposure time, can be seen i n F i g . 4.3. The distribution o f colours and colour intensity is graphically presented in F i g . 4.1.  Colour intensity,  although a more arbitrary classification scheme than colour hue, is more evenly divided among the four classes than colour hues.  Colour intensity distribution is as follows:  moderate (32%), strong (26.5%), weak (26.5%) and very weak (15%).  Ten diamonds  had non-uniform F L colours (two examples o f which can be seen i n Figs. 4.2E and F).  46  Fig. 4.3. Compilation photograph of FL colours for Rio Soriso diamond suite (using a constant exposure time of 20 seconds). The seven groups are labeled as follows: 1 - strong blue; 2 moderate blue; 3 - weak blue; 4 - very weak blue; 5 - turquoise (strong, moderate, weak and very weak are all grouped in this column); 6 - green (strong and weak); and 7 - brown (weak). One diamond did not fluoresce at all. (note that for comparative purposes, diamonds are scaled to roughly the same size).  47  5.0 Cathodoluminescence of Diamonds  5.1 Introduction  Cathodoluminescence ( C L ) , like fluorescenence, is an induced-luminescence property. The difference between the two is the energy source used; F L is caused b y exposure to ultraviolet light and C L is caused by the bombardment o f electrons. In the same manner that U V - l i g h t induces fluorescence i n diamond, electron-bombardment excites valence electrons i n optical centers to higher energy states and as they fall back to their original energy state, light is emitted.  This induced luminescence is what is referred to as C L .  The main difference between F L and C L is the depth o f penetration o f the energy source. Whereas F L excites most i f not the whole diamond, C L excites only valence electrons i n the uppermost surface o f the diamond (-v5 um depth o f the diamond surface using a 30 k e V electron beam, Hanley et al,  1997).  C L colours o f diamond are mostly related to the presence o f nitrogen impurities which, when i n aggregated forms as is typical for natural diamond (i.e. type la), tend to produce a variable sky-blue C L colour. Some o f the other more commonly observed colours are yellow, yellow-green, green, canary-yellow, pink, brown and white C L . Y e l l o w C L has been linked to hydrogen impurities that form on (100) faces o f diamond (Bulanova, 1995) and to intrinsic defects such as zones around inclusions, along deformation laminations and around radiation spots (Davies,  1998).  Yellow-green C L is believed to be caused b y  N 3 centres (three aggregated nitrogen atoms) while green C L is typical for type lb synthetic diamond (Hutchison, yellow C L i n diamond.  1997).  Radiation damage is thought to induce canary-  Nitrogen-free (type II) diamond containing microscopic CO2  inclusions produces various shades o f pink (including orange and purple) C L and the plastic deformation o f these diamonds produces brown C L (Chinn et al, C L has been attributed to platelets (Woods,  1986  48  and Davies,  1998).  1995).  White  C L intensity is dependant on the concentration o f defects, and for many diamonds, nitrogen is the main defect.  A s such, the blue intensity observed is related to the  concentration o f nitrogen impurities.  However, i f other defects exist in sufficient  quantities, they can alter the colour and produce intermediary C L colours that can be impossible to unambiguously ascribe to a particular defect centre.  A s a result, the  technique o f C L examination o f diamond is not quantitative.  5.2 Analytical techniques  The C L characteristics o f 47 rough diamonds were examined using a Cambridge Instruments Cathode Luminescence ( C I T L 8200 m K 4 ) system attached to an optical microscope with a 2.5 x lens. The accelerating voltage used was 15kV with an electron beam current o f 300 u A arid chamber pressure was maintained using a Varian D S 102 pump. Diamonds were washed with ethanol before being placed on a recessed steel tray specially designed to fit i n the chamber. Typically, 4 to 5 diamonds were loaded i n the tray at a time, being ~1 c m apart. Larger stones were loaded into the machine one at a time. C L images were collected using a N i k o n Coolpix 995 digital camera. times were variable and do not reflect the C L intensity well.  Exposure  Due to an unknown  automatic feature on the camera, some photos may not accurately reflect the true C L colour observed, although they certainly are close.  It is stressed here that optical C L  studies on their own should be used only for non-rigorous, qualitative classification.  5.3 Results  The CL-induced colours o f forty-seven rough, unpolished stones were examined. Photographs o f diamond C L for these rough diamonds are presented i n Appendix A . Diamonds were grouped into seven categories based on observed colours. The colour distribution is graphically illustrated i n F i g . 5.1 and some examples o f C L colours are presented i n F i g . 5.2. There is certainly some ambiguity i n assigning stones to particular  49  colour classes, but i n the context  of  quantitative  this  semi-  study,  Blue  the  • strong (12)  divisions are considered satisfactory.  • moderate (6)  Turquoise • moderate (11) • strong (7)  These  Green  divisions, along with the  • moderate (4) • strong (2)  Other  number o f samples are:  0  (5)  blue 1 ( C L o f moderate blue intensity, n=6); blue 2, (strong blue, n=12); turquoise  1,  Fig. 5.1. Colour distribution for CL colours, (numbers in brackets in legend indicate sample size).  (moderate,  n = l l ) ; turquoise 2, (strong, n=7); green 1, (moderate, n=4); green 2, (strong, n=2); and  Fig. 5.2. Photographs of representative cathodoluminescence colours of diamonds, this study. A) moderate blue (Diamond 5-1), B) strong blue (Diamond 3-9), C) strong turquoise (Diamond 4-13), D) moderate green (Diamond 4-5), E) strong green (Diamond 4-18), and F) other (Diamond 4-21).  50  other, (n=5) (Fig. 5.4). If the subgroups are combined, the C L colour proportions are as follows: blue (38.3%), turquoise (42.6%), green (12.8%) and other (10.6%).  The C L  colours show a gradation from blue to turquoise to green, which is likely an indication that more than one defect centre is responsible for the colouration. Perhaps blue and green C L colours are end-members and they combine i n varying proportions to produce intermediate colours.  Blue C L is likely caused by aggregated nitrogen defects o f various concentrations, but generally in low abundance.  A n adequate explanation for the green C L colouration  observed is not so clear. Green C L is common for synthetic diamond with disaggregated nitrogen, however, diamond with this nitrogen character is extremely rare i n nature (as w i l l be discussed i n section 6.1).  Yellow-green colouration has been attributed to N 3  centres (also an aggregated form o f nitrogen, but is different from the aggregated nitrogen that produces blue C L ) .  Internal growth features are absent for most diamonds, but visible on diamond 4-17 (examined in detail in section 7.3.14) and diamond 3-9 (a macroscopic view o f the C L pattern can be seen in F i g . 5.2B and a greyscale close-up in F i g . 5.3).  The growth patterns  visible on the surface o f diamond 3-9 are typical for strongly resorbed surfaces.  The pattern is similar i n  appearance to 'agate' and cannot be deciphered in terms o f growth habit based on this image. The C L o f internal  growth  structures  is better examined  polished surfaces (Chapter 7.0).  51  on  Fig. 5.3. Close-up greyscale photograph of CL of growth features of resorbed diamond. (Diamond 3-9).  Blue 1. 2 .  Turquoise 1. 2.  Green 1. 2 .  Other  ^^^^  **  1 '-.'1  Fig. 5.4. Photographs of CL images of 47 rough, unpolished diamonds from Rio Soriso. Diamonds have been scaled to the same size for comparative purposes.  52  6.0 Infrared Spectroscopy of rough diamonds 6.1 Introduction  A l l crystals are made up o f atomic bonds which hold atoms together.  Some o f these  bonds can be excited to higher energy levels by infrared (IR) light, which results in the absorption o f this light. The bonds detected by IR spectroscopy (termed IR active) are those which have an asymmetric stretch (Fig. 6.1) and thus produce a change i n the dipole moment.  (IR inactive bonds produce  no change i n dipole moment symmetric stretch vibration).  during the The typical  wavelength o f IR light used i n diamond studies ranges from 900 to 4000 cm" . Each 1  scan over the crystal records the amount o f absorption at each wavelength within this range and absorption is recorded (in arbitrary units), varying from 0, or no absorption, to  symmetric stretch 1340 cm-  asymmetric stretch 2350 cm-  1  1  0=C=0  .... . O = c=o equilibrium 0 = C = 0 - * — .. —*-o = c = o  o =c=o  position  o=c = o  Fig. 6.1. Difference between IR-active and IR-inactive bonds. Bonds which produce a symmetric stretch are IR-inactive and cannot be detected with an FTIR. Bonds which produce an asymmetric stretch are IR-active and can be detected.  infinity, or full absorption.  The recognition and distinction between diamonds based on their IR absorption patterns was first described by Robertson et al. (1934). They found that most diamonds displayed absorption i n the lower frequencies (less than 1500 cm" ) and termed these diamonds type 1  I, and labeled the remaining diamonds that lacked any absorption i n this range as type II. M a n y studies have been performed subsequently, and are summarized b y Clark and Davey (1984) and Clark et al. (1992).  The range 900-4000 cm" used i n most diamond studies can be divided into three regions: 1  one, two and three phonon absorptions (Fig. 6.2).  After the pioneering work by  Robertson et al. (1934), Kaiser and B o n d (1959) demonstrated that much o f the  53  absorption and its intensity i n the onephonon region (900-1333 cm" ) is due 1  to  the  diamond.  presence  of  nitrogen  in  It is now well known that  nitrogen is a common impurity i n diamond and that most one-phonon absorption is due to its presence. "Perfect"  diamond,  lacking  any  3500  3000  2500  2000  1500  1000  Wavenumbers (cm- ) 1  impurities or imperfections, does hot display any one-phonon absorption. The diamonds which Robertson et al. (1934) classified as type I are known  Fig. 6.2. IR spectrum for a type II diamond from 5004000 cm" , illustrating the three regions: the 1-phonon region, where asymmetric bonds produce absorption; the 2-phonon region, which is intrinsic to all diamond; and the 3-phonon region, where most CH complexes produce absorption. 1  as 'nitrogen-bearing' while type II stones are 'nitrogen-free'.  Absorption i n the two-phonon region is characteristic o f all  diamond while absorption o f hydrocarbons occurs i n the three-phonon region.  6.1.1 One-phonon absorption in diamond related to nitrogen impurities  Most studies to date have focused on correlating the absorption i n the one-phonon region to nitrogen concentration and form.  It was observed that diamond displays many  different absorption patterns i n the one-phonon region, and studies focused on attempting to elucidate the cause(s) for these different spectra.  Davies (1972) was the first to  quantitatively decompose the one-phonon spectra o f type I diamond into two distinct components, which he termed A and B . H e discovered that most one-phonon spectra o f natural diamond could be roughly described by linearly combining two end-member spectra ( A and B ) i n different proportions. End-member spectra A and B can be seen i n F i g . 6.3A and 6.3B. Since this time there has been the discovery o f a third component, labelled D (Fig. 6.3C), and some lesser components (Clark and Davey, 1984).  54  Davies (1976) showed that the A spectrum is likely a result o f two bonded nitrogen atoms incorporated into the crystal structure. Evans and Q i (1982) suggest that the B spectrum is a product o f four nitrogen atoms tetrahedrally surrounding a vacancy. Both o f these interpretations are agreed upon in most literature. More controversial is the D component produced in the one-phonon region. Woods (1986) attributes this absorption to planar structures called platelets, which are made up o f carbon atoms. Platelets are discussed in  1500  1350  1200  1050  1500  900  Wavenumbers ( c n r )  1500  1350  1200  1050  1350  1200  1050  900  Wavenumbers ( c n r )  1  1  1500  900  Wavenumbers (cm- )  1350  1200  1050  900  Wavenumbers (cm- )  1  1  Fig. 6.3. IR spectra for common end-member absorption patterns in diamond from 900-1500 cm"'. A) One-phonon spectrum of pure A centre absorption; B) One-phonon spectrum of pure B absorption; C) One-phonon spectrum of pure D absorption and; D) One-phonon spectrum of pure lb absorption. Figures A-C from deconvolution software. D) Reproduced from Evans (1992).  greater detail in the following section. Diamonds with A and B centres are classified as type la (also termed aggregated forms). The cut-off between type I and type II varies from study to study, but is below 50 ppm total nitrogen.  Ninety-eight percent o f natural  diamonds worldwide are classified as type la (Evans, 1992).  Diamonds are further  subdivided into IaA, IaB, and in most cases, i f both A and B centres are observed, as the transitional form IaAB.  55  IR studies o f nitrogen-containing synthetic diamond reveal that they display yet another one-phonon spectrum (Evans, 1992), illustrated i n F i g . 6.3D. The characteristic electron paramagnetic resonance ( E P R ) signal indicates that the nitrogen atoms are singly substituted throughout the crystal structure. These diamonds are classified as type lb and are extremely rare i n nature.  6.1.2 Process of nitrogen aggregation  The history o f IR studies i n diamond and the connection between nitrogen aggregation and the one-phonon absorption has been discussed i n the previous section. After these studies, most work has focused on trying to correlate the intensity o f particular I R frequencies with nitrogen aggregation concentrations (most recently Woods et al., 1990; B o y d et al,  1994 and B o y d et al., 1995) and applying this knowledge to geologically  significant processes (Taylor et al., 1990; Mendelssohn and Milledge, 1995). In order to decompose the IR spectra and to quantify each component, it is necessary to understand the process  o f nitrogen aggregation  and the causes for the varying one-phonon  absorptions.  The degree o f nitrogen aggregation depends on the residence time o f the diamond i n the mantle, the nitrogen content and the temperature history (Evans and Harris, 1989). These parameters have been corroborated by experimental studies o f nitrogen aggregation and concentration in diamond at controlled temperature and with time.  During diamond  crystallisation, nitrogen substitutes for carbon forming point defects and creating type lb diamond.  W i t h time and at the elevated temperatures expected i n the mantle, the  dispersed nitrogen atoms migrate and form A centres (two bonded nitrogen atoms). This conversion, from type lb to IaA, depends o f kinetics, and is thought to proceed quickly i n terms o f geologic time at mantle temperatures and requires only 5 ± 0.3 e V to activate the reaction (Evans and Harris, 1986).  Experiments on radiogenic minerals included i n  diamonds by Richardson et al., (1990) and Deines et al., (1991) show that many  56  diamonds are Archean i n age and much older than the intrusive event which brought them to the surface. Evans and Q i , (1982) show that type lb diamonds o f Archean age would only survive i n the mantle at low temperatures (<800 °C), which is outside out the diamond stability field for cratons with typical geotherms.  These studies illustrate  effectively why type lb diamonds are so rare i n nature; only diamonds with short residence times could preserve type lb character.  M o r e common in nature are diamonds which fall into the I a A B transition. Through longer  residence  times  and  higher  temperatures, A centres are thought to migrate  and  aggregates,  form  more  B centres.  complex  During  <  this  process it is suggested that a carbon atom is displaced to make room for the four nitrogen atoms  (Woods, 1986).  Thus the B centre is interpreted as being four  nitrogen  atoms  tetrahedrally  o 1370 & O  c  3  arranged around a vacancy. M a n y type D  I a A B diamond IR absorption curves  fV  J  H  exhibit several local absorption modes, the most prominent peak occurring near  i  1650  1370 cm" (Fig. 6.4) and is termed B ' 1  (Clark et al,  1992).  absorption (Taylor et al, 1990). Based on  the  strong  positive correlation  between D absorption and the integrated B'  absorption area  (Fig 6.5,  filled  circles), Woods (1986) proposed that  .1  900  Wavenumbers (cm~l)  This peak is  thought to be the result o f platelet  900 1650  Fig. 6.4. IR absorption spectra of the development and subsequent degradation of the B' peak (at 1370 cm" ). The progression from A-H illustrates the changes in impurity character with increasing time and temperature. The progression is as follows: type IaA (A) through 'regular' transitional IaAB (B-D) to pure IaB with only B and D absorption (E), followed by a departure form regularity (F and G) towards pure IaB with no D absorption (H). Reproduced from Woods (1986). 1  57  the D component observed i n the one-phonon range is the result o f platelet absorption. Thus, the relative absorption produced by D centres is interpreted as a measure o f the relative concentration o f platelets per unit area.  Woods (1986) proposes that the physical process occurring in diamond during A to B conversion is that  the  displaced  carbon  atoms  aggregate  themselves, to form platelets which in turn are responsible for one-phonon fR-active bands known as D centres and B' peaks.  Platelets can be  observed by Transmission Electron Microscopy ( T E M ) and sometimes through C L , and range from a few nanometres to several micrometres in size (Evans and Phall, 1962) as seen i n Fig. 6.6.  0.5 B/(B+A)  Matters are complicated by diamonds containing B and D centres which do not correlate positively (open circles, Fig. 6.5). Plots o f B/(B+A) versus D centres i n these diamonds always fall below the  Fig. 6.5. Plot of B/(A+B) centres in diamond versus integrated area under B' absorption peak. Graph illustrates the difference between regular (filled circles) and irregular (open circles) diamonds. Reproduced from Woods (1986).  line, indicating that there are less D centres than should be expected for the number o f B centres.  These diamonds are referred to by  Woods which  (1986) as  irregular  correlate positively  regular.  while  stones  are considered  Woods (1986) proposes that this  non-linearity observed for some diamonds is a product o f platelet degradation and the Fig. 6.6. Transmission electron micrograph showing a cross-section view of platelets. Reproduced from Evans (1992).  subsequent formation o f dislocation loops.  58  Sometimes  when  dislocations  form  edge they  and can  screw produce  closed dislocation loops which form a complete circle around the segment o f the diamond crystal that has slipped. Figs. 6.7A and 6.7B are images o f a dislocation loop encircling numerous voidites. Voidites are another defect in diamond and are commonly found in dislocation  loops.  They are  small  octahedra, 1-10 nanometers i n diameter, bounded by (111) planes (Lang et al., 1992). They produce some one-phonon absorption makes difficult  studies  (Woods,  nitrogen to  concentrations  calculate.  have  1986),  found  Also,  that  which more several  Fig. 6.7. Transmission electron micrograph of A) numerous voidites (small circles) in a dislocation loop (large circle) and B) numerous voidites (small linear features) inside a dislocation loop (one large, discontinuous loop). Reproduced from Field (1992).  some ER-  inactive form(s) o f nitrogen likely occurs in voidites (Barry, 1986; Hirsch et al., 1986). A s such, diamonds with a high density o f dislocation loops and or voidites may contain a greater concentration o f total nitrogen than calculated using ER spectroscopy alone.  The process o f nitrogen aggregation is complex and is not fully understood. However, it is quite clear that nitrogen diffuses through diamond with time and the influence o f heat, creating more complex forms. F i g . 6.8 is a schematic diagram from Mendelssohn and Milledge (1995) illustrating the process o f nitrogen aggregation.  The arrows illustrate  how initially type lb diamond converts to type IaA and then IaB as a function o f time and temperature.  During the conversion o f B centres, platelets may form. The final arrow  indicates that the IaB character may convert to type Ha diamond; however, the question mark indicates that this conversion is uncertain. It is possible that nitrogen migrates to  59  Platelets  Fig. 6.8. The progression of nitrogen aggregation in diamond. From: lb, singly substituted nitrogen; to IaA, pairs of nitrogen; to IaB, four nitrogen atoms about a vacancy; to Ila, low nitrogen diamonds with likely numerous platelets and voidites. Reproduced from Mendelssohn and Milledge (1995).  voidites and dislocation loops and may be i n an IR-inactive form.  B y modelling the  kinetics o f this reaction, we can quantitatively measure nitrogen and its state i n diamond and make predictions about the diamonds temperature and/or time history.  6.1.3 Quantitative calculation of nitrogen concentration  Equations relating the concentration for A and B centres i n type l a diamond have been continually improved over the past twenty years. A s the causes o f ER absorption become better understood, relationships between peak intensity and concentration change. Equations that are commonly used in literature today can be found in Woods et al. (1990), B o y d et al. (1994) and B o y d et al. (1995). D centres are not well understood and no equation exists relating absorption peaks to concentrations.  The equations used to calculate the concentration o f nitrogen are typically expressed i n terms o f absorption units at a particular frequency (LL1282) per unit thickness ( c m ) . The 1  formula used to convert A centre absorption into concentration is:  60  N ( p p m ) = 16.5 ± 1 x A  uA  1 2 8  2  (cm"'), (Boyd et al, 1994)  (6.1)  2  ( c m ) , (Boyd et al., 1995)  (6.2)  and for the conversion o f B centres:  N ( p p m ) = 79.4 ± 8 x B  uB  1 2 8  6.1.4 Time-averaged mantle-residence temperatures  Taylor et al. (1996) show that the conversion o f singly substituted nitrogen (lb) to aggregated A centres (IaA) follows second-order kinetics and can be quantitatively expressed by the Arrhenius rate law. However, o f more interest to the study o f natural diamonds is the conversion o f A to B centres. Unfortunately this reaction occurs at much higher temperatures than lb to IaA conversion and laboratory experiments to accurately model the reaction are difficult.  The limited data collected from experiments indicate  that the conversion o f A to B centres likely follows second-order kinetics as w e l l (Evans and Harris, 1986).  Clark et al. (1992) and others have experimentally studied the effects that time and temperature have on the character o f nitrogen i n diamond by combining the Arrhenius rate law with the second-order reaction rate equation.  The Arrhenius rate law can be expressed by the following equation:  - Ea  K = Ae  L  kT  (6.3)  Where:  61  K = reaction rate (1/ppm • s) A = Arrhenius constant (1/ppm • s) E a = activation energy (J) k = Boltzmann constant (J/K) T = temperature (K)  E a and A are empirical constants which differ from one reaction to another and are calculated through experimentation. B y taking the natural log o f both sides and isolating T, equation 6.3 can be rewritten as:  Ea  T =  In  K_ A  (6.4)  Assuming the conversion o f A to B centres follows second-order kinetics, the reaction rate, K , can be related to concentration and time by the following expression:  ac dt  = -KC  2  (6.5)  B y inverting and integrating equation 6.5 from the initial concentration, C , at time = 0 0  and the final concentration, C , at time = t:  K=  (6.6)  A n d by substituting equation 6.6 into 6.4:  62  T  =  Ea  ,  I C-l  In  C  -.—°-  (6.7)  t.A The  empirical  constants  used  for  calculating  time-averaged  mantle  residence  temperatures i n this study are:  E a = 1.12633xl0" J or 7.03 eV (Taylor et al, 1990) 18  A = 2 . 9 4 1 8 1 x l 0 1/ppm • s (McKenna, 2001), (7.36747xl0 used by Taylor et al 1990) 5  5  And:  C  0  = N(TOT),  concentration o f nitrogen occurring as A and B aggregates (atomic ppm)  C = N ( ) , concentration o f nitrogen occurring as only A aggregates (atomic ppm) A  k = 1.380658xl0" (J/K) 23  In equation 6.7, both C and C can be measured, thus leaving T and t as unknowns. 0  Diamond mantle residence time is well constrained i f two ages can be determined: 1) the age o f the diamond, and 2) the age o f magmatic emplacement.  Mineral inclusions i n  diamond, such as pyrite, garnet and clinopyroxene, may contain significant quantities o f radiogenic isotopes for dating. The difference between the age o f the diamond and the age o f eruption would represent the mantle residence time.  The average mantle residence temperature is a more difficult variable to estimate using other analytical methods.  The main problem is that it is unlikely that the diamonds  remained at the same temperature throughout their residency period in the mantle. Lithospheric diamonds may have been forced to shallower depths in the mantle during slab underplating o f cratons and thus subjected to lower temperatures and  63  'sub-  lithospheric' diamonds are likely to have been brought closer to the base o f the lithosphere through mantle convection or by the action o f a mantle plume before entrainment i n a deep-seated magmatic body.  It is highly probable that diamonds  experience a range o f different residence temperatures, and that the simplest way o f quantifying this is by calculating a time-averaged value. However, mineral inclusions can be used to provide some constraints on equilibrium temperatures (e.g. Ryan et al, 1996; Brey and Kohler, 1990). Temperature constraints can also be estimated based on the pressure-temperature stability field o f some minerals, which w i l l be discussed at length in section 8.1. For example, any mineral that is stable only in the transition zone effectively restricts, at least initially, the temperature range at formation (temperatures estimates at the - 4 1 0 k m and - 6 6 0 k m seismic discontinuities are 1500°C and 1600°C respectively, Ringwood, 1991). However, i n practice, application o f this to modelling a diamonds time-temperature history based on nitrogen character does not work because the process o f aggregation has either gone to completion, or has resulted i n severe platelet degradation.  6.1.5 IR spectra for some other impurities  IR spectroscopy has been used to detect other impurities i n diamond. Hydrogen, water, carbon dioxide and carbonate all produce fairly distinctive absorptions between 500 and 4000 cm" (Navon et al, 1988). 1  6.1.5.1 C H bonds  The region between 2800-3100 cm" is known as the CH-stretch region. Absorption i n 1  this region forms either sharp peaks or one broad absorption band, and may be attributed to either surface complexes (which, for obvious reasons, are not considered as diamond impurities and therefore not o f interest i n this study) or true lattice defects.  True  hydrogen impurities can produce a variety o f absorption bands, depending on the nature  64  of the complex (i.e. C H , CH2 or CH3, among many other possibilities). There is still uncertainty i n understanding the true nature o f absorption i n the C H stretching  region;  published  however,  reports  ascribe  many the  3000  2000  Wavenumbers (cm ) 1  characteristic sharp peak at 3107 cm"  1  Fig. 6.9. IR spectrum of CH absorption in diamond. Characteristic IR pattern produces a strong absorption peak at 3107 cm" and a secondary peak at 1405 cm" .  and the smaller accompanying peak at  1  1405 cm" to C - H complexes (Fig. 6.9). 1  1  CH2 and C H are thought to produce 3  more complex absorption bands that often overlap, thus requiring spectral deconvolution (Dischler et al, 1993).  Furthermore, Sellschop (1992) found no correlation between  relative peak height and hydrogen concentration and thus suggested that there may be significant amounts o f hydrogen i n other bonds, likely i n IR-inactive forms.  6.1.5.2 W a t e r ( O H a n d H O H bonds)  A  broad peak roughly centred  over  3420 cm" is attributed to OH-stretching 1  while a sharper  peak  at 1645 cm"  1  1  The  presence o f water (Fig. 6.10).  \H-O-H I 1645  /  3000  2000  1000  Wavenumbers (cm')  presence o f either o f these absorption bands is considered evidence for the  /O /  1  (1988)) is considered due to H - O - H bending (Koeberl et al, 1997).  j  O-H 3420  (reported at 1630 cm" by Navon et al.,  Fig. 6.10. IR spectrum of water in diamond. H 0 produces a broad absorption peak centred approximately at 3420 cm"' and a narrow peak at 1645 cm" . Reproduced after Koeberl et al. (1997). 2  1  65  6.1.5.3 Carbon dioxide ( C 0 bonds) 2  Carbon dioxide produces a prominent peak at 2383 cm" (reported at 2350 1  cm" by Navon et al., (1988)) and a less 1  obvious peak at 657 cm" , among others 1  (Schrauder  and  Navon,  Absorption - 3 6 0 0 -  1993).  3750 cm"' also  occurs but is relatively weak compared with the 2383 and 657 cm" peaks (Fig. 1  —i  i  3000  i  2000  1000  Wavenumbers (cm ) 1  Fig. 6.11. IR spectrum of carbon dioxide in diamond. C 0 produces strong absorption peaks at 657 and 2383 cm" . C 0 also produces less well defined peaks above 3500 cm" Reproduced after Koeberl et al. (1997). 2  1  2  6.11).  1  6.1.5.4 Carbonate ( C 0  The carbonate C 0 " 3  2  2 3  bonds)  anion produces  absorption bands at 1430 and 876 cm" (Navon et al,  1988) (Fig. 6.12).  1  The  band at 876 cm" occurs i n the same 1  position (±2 cm" ) for calcite impurities 1  and is distinct from that o f dolomite and  2000  1000  Wavenumbers (cm ) 1  magnesite (Navon et al., 1988).  6.2 Analytical techniques  —  1  3000  Fig. 6.12. IR spectrum of carbonate in diamond. Carbonate produces absorption bands at 876 and 1430 cm"'. Reproduced after Navon et al. (1988).  IR spectra were collected over the range 650-400 cm" on a Nicolet Fourier Transform 1  infrared (FTIR) spectrometer with a liquid-N2-cooled detector. Spectra were collected i n transmission mode using a resolution o f 8 cm" by averaging the signal o f 256 scans 1  (similar ranges, resolutions and number o f scans were used by Taylor ei al., 1990 and  66  Mendelssohn and Milledge, 1995). Spectra were automatically converted into absorption units, as required for the future manipulation o f data, using the equation:  absorption = log (100/%transmittance)  (6.8)  Background spectra were collected at the start o f each day and renewed i f the experiment lasted more than 2 hours. Spectra were manipulated using Omnic version 6.0a software and interpretation software supplied by T. Stachel.  Rough diamonds were cleaned i n an ultrasonic bath o f dichloromethane ( D C M ) for 30 minutes and thoroughly washed with ethanol. They were mounted on glass slides with two-sided scotch tape so that only part o f the diamond was on the slide.  Smaller  diamonds could be fixed to the tape on the edge o f the glass slide. If possible, diamonds were mounted i n such a way so that a flat face was perpendicular to the I R ray path. B y doing this, refraction and dispersion through the diamond is minimized and it was usually possible to collect excellent spectra with minimal noise.  Taylor et al. (1990) demonstrate that the effects o f diamond refraction on ER path length increase with diamond thickness. They found that refraction effects are only significant for diamonds >1.5 m m thick. M a n y diamonds in this study were greater than 1.5 m m thick.  For these diamonds, the reported nitrogen concentrations w i l l be less than the  actual nitrogen concentrations.  Mendelssohn and Milledge (1995) find that ER results  become increasingly inaccurate for diamonds thicker than 2.0 m m . Thicker diamonds can be analysed i f they contain little to no nitrogen.  Absorption intensity is a function o f path length, or diamond thickness. T o account for diamond thickness variation, a spectrum must be calibrated against a spectrum o f known IR path length. A n IR spectrum was collected from a 'nitrogen-free' diamond (type II) and calibrated to a thickness o f 1 c m using the conversion factor 11.94 absorption  67  units/cm, measured at 1995 cm" (T. Stachel, personal communication, 2002). The type 1  LI stone (no visible one phonon absorption) was supplied by T. Stachel and is a 0.57 m m thick chip with two reasonably flat, parallel fracture surfaces. The type LI spectrum was corrected by taking a linear base line between - 4 0 0 0 and - 1 5 0 0 cm" . A l l spectra were 1  scaled to match the type U spectrum and base lined following the same procedure as used for the type LI diamond. Ln some cases, more than two points were required to produce an acceptable spectrum for deconvolution. The choice o f base line has an effect on the final calculated nitrogen totals, and is explored i n more detail below (section 6.2.1.2).  Following this, each sample was deconvoluted into three curves, A , B and D , using software supplied by T. Stachel.  In some instances, no curves were fitted to the  deconvoluted data curves. In these cases, files were base lined using more points and rerun through the deconvolution software. The program calculates the absorption at 1282 cm" for the A , B and D curves. Equations relating the absorption measured at 1282 cm" 1  1  to A and B centre concentrations have been developed (see section 6.1.3) but no such equations exist yet for D curves. A s such, the value quoted for D centres is i n absorption units for a diamond o f 1 cm thickness. It is best thought o f as a dimensionless number which describes the relative concentration o f D centres.  The cut-off for type LI diamond is arbitrary but i n qualitative terms, i f absorption attributed to nitrogen defects is visible i n the one-phonon range, the diamond is type I. In quantitative terms, i n this study, the cut-off o f 20 atomic ppm nitrogen as the divide between type I and type LI diamond Kaminsky et al, 2001b), was used. In some cases the deconvolution program yielded concentrations  greater than 20 ppm nitrogen;  however, examination o f the LR spectra did not reveal any detectable nitrogen (any onephonon absorption is likely a result o f noise). In these cases, the deconvolution results are recorded, but the diamond is classified as type LL. The distinction between type I a A B and the end members IaA and IaB was arbitrarily set using the following divisions: any diamond containing >20 ppm total N with less than 10% B centres is considered type  68  IaA; IaB (>20 ppm total N and >90% B centres); and type I a A B (>20 ppm total N and 1 0 - 90 % B centres).  Relative hydrogen concentrations were measured after thickness calibration and base lining and recorded by measuring the absorption difference between the base and peak at 3107 cm" . 1  6.2.1 Examination of error analysis for infrared studies  Error estimates for nitrogen concentrations are difficult to quantify due to the numerous sources o f absorption that may contribute. Most errors in concentrations are produced either during data collection or data processing.  The following section examines these  errors in attempt to quantify each source.  6.2.1.1 Error from deconvolution software 12.0 residual  10.0  The deconvolution software uses IR data between the wavenumbers 900-  8.0  B A  g  1500 cm" that have been calibrated for  spectrum A+B+D  1  a diamond o f 1 c m thickness and base lined.  } 6.0 JS 4.0  The program calculates a 'best2.0  fit' curve (blue line, F i g . 6.13) which is  n n  the sum o f theoretical curves for A (green  line),  B (pink  line)  and  D  (turquoise line) curves to the collected data (red line). The black line indicates the  difference between  theoretical  curves  and  the  1500  sum o f  the collected  1400  1300  1200  1100  1000  900  Wavenumbers  Fig. 6.13. Results from deconvolution software. A,B and D curves represent scaled values for these pure end-member absorption curves (see Fig. 6.3). The other three colours represented are blue (the sum of the three end-members), red (base lined and calibrated data file) and black (the difference between blue and red curves.  69  spectra (red line) and represents the 'degree o f  fitness'.  Based on these curves, the  program calculates the absorption units at 1282 cm" for each o f A , B and D curves. The 1  program also calculates a value for the black line, but it is unknown to the author what this term represents; most likely it quantitatively describes the area under the curve. Because o f the uncertainty associated with the calculation for determining the error value, no component to the overall error can be quantitatively attributed to the process o f spectral deconvolution.  6.2.1.2 Sensitivity to baseline corrections  F T I R spectra i n this study are typically sloped so that there is more absorption at higher wave numbers numbers.  Reasons  numerous,  but  common  causes  scattering,  some  than lower  for  this  of  the  include  inappropriate  Table 6.1. Base line sensitivity study Base line A (ppm) B (ppm) study no.  are more  sample  choice  of  367.5 340.6  11.3 6.9  1 2  B/(B+A)  Total N (ppm)  97.0 98.0  378.8 347.5  Spectra deconvolution results for two curves in Fig. 6.14. Study 1, one line used for base line, study 2, two line segments used.  background and instrument drift (Smith, 1996). straight  Base line corrections may be or  curved,  however,  it  is  important that regardless o f what base line function is used, false peaks are not introduced.  A good way to  avoid  introducing false peaks is to use as few line segments as possible. Mendelssohn and Milledge (1995) suggest fixing the base line at 4000 cm"  1  and at  i  1500  the  minima between 1400-1600 cm" and 1  extrapolating this backward to 900 cm" . 1  i  1400  1  1  1  1300  1200  1100  '  "  1000  900  Wavenumbers (cm ) -1  Fig. 6.14. Examination of base line sensitivity. Curve 1 was base lined using one line segment, curve 2 was base lined using two line segments.  70  In this study most spectra were corrected using one line segment following the approach suggested by Mendelssohn and Milledge (1995). The minima between 1600-1400 cm"  1  was almost always found at 1558 cm" . In some cases the spectra between 1558 and 400 1  cm" increased i n absorption. In these cases two line segments were used, one between 1  4000 and 1558 cm" and the other between 1558 and 400 cm" . 1  1  B y applying base lines o f various slopes to the same raw data file, base line sensitivity was examined. One spectrum was produced by joining a line segment between 4000 and 1558 cm" and extrapolating backwards (curve 1 on F i g . 6.14) while another spectrum 1  was produced by joining line segments from 4000 and 1558cm" and 1558 to 400cm" 1  (curve 2, F i g . 6.14).  1  Nitrogen totals for these two curves are presented i n Table 6.1.  B / ( B + A ) is not very sensitive to the choice o f base line, however, the total nitrogen concentration varies quite considerably. The difference i n total nitrogen between these two spectra is 30 ppm, or 9% relative error.  6.2.1.3 R e p r o d u c i b i l i t y o f I R spectra  Reproducibility  of  LR spectra  was  Table 6.2. Reproducibility study  estimated by collecting ten spectra from  Point  the same point on a crystal with a l l  1 2 3 4 5 6 7 8 9 10 average stan. dev.  other parameters remaining  constant.  Diamond 2-5 was selected for this study because the spectra collected from this crystal did not require any base lining (there was no absorption at 600 and 4000 cm" ) and the diamond contains 1  considerable nitrogen (-300 ppm N ) . By  selecting  reproducibility  this  crystal  study,  any  for  A (ppm) B (ppm) B/(B+A) 36 37 36 36 35 36 35 35 35 34 35 0.8  278 277 275 273 271 270 269 267 266 265 271 4.6  88 88 88 88 88 88 88 89 88 89 88 0.1  Total N (ppm) 315 313 311 309 307 306 304 302 301 299 307 5.4  Columns A (ppm) and B (ppm) are calculated using equations 6.1 and 6.2.  the error  71  introduced during base lining could be removed. Each spectrum collected was multiplied by the same conversion factor i n order to convert the curve to that o f diamond o f one centimetre thickness and then deconvoluted. The results are presented i n Table 6.2.  The average total nitrogen value is 307 ± 5.4 ( l o ) while % B averages 88 ± 0.1 ( l a ) . The apparent reproducibility error is small; however, note that the total nitrogen decreases after each successive analysis. The reason for this is unclear. The relative error between points 1 and 10 at the 95% confidence level is 3.5%.  6.2.2 Precision of I R data  There are other sources o f error to consider than just the three sources explored i n section 6.2.1; however, the overriding contributor to differences i n nitrogen concentration are heterogeneities within the diamond crystal itself. In this study, values within one crystal varied from 30 to 400 ppm nitrogen (1333%).  The error o f the method used i n this study i n estimating nitrogen concentration and aggregation state is a combination o f the analytical precision and the error inherent to spectrum calibration, deconvolution, and conversion o f results into nitrogen centre concentrations.  The analytical precision, calculated through multiple analyses o f the  same grain is estimated at 3.5% (section 6.2.1.3). The remaining errors are as follows: spectrum calibration, mainly affected by the manual selection o f a base line, is estimated at 9% relative (section 6.2.1.2); deconvolution, although certainly a source o f error has no absolute quantitative precision attached to the results (discussed in section 6.2.1.1) and hence cannot be included here; and conversion o f deconvoluted curves into nitrogen concentrations is calculated as 6% for A centres (Boyd et al., 1994) and 10% for B centres (Boyd et al., 1995) (errors on equations 6.1 and 6.2 i n this study). The overall precision o f this method is estimated at 19% for A centres and 23% for B centres, which  72  agrees well with those reported, i.e. 10-20% (Kaminsky et al, 2001b), <25% (Kaminsky et al, 2000), 12-15% (Deines et al, 1991), and 10-20% (Stachel et al, 2002).  Nitrogen concentration errors for this study fall between 19-23%. However, the relative error increases as nitrogen concentrations decrease.  The B / ( B + A ) value (aggregation  state) also becomes increasingly imprecise as nitrogen concentrations approach the minimum detection limits.  The precision o f relative hydrogen concentrations was examined by measuring the peak height at 3107 cm" for the same curve several times. F r o m these measurements, an error 1  o f 2.6% relative was calculated.  6.2.3 M i n i m u m detection limits  M i n i m u m detection limits ( M D L ' s ) were  estimated  examination  of  based  on  LR  curves  Table 6.3. MDL's for IR data  visual  Centre  MDL  A B D H  12 ppm 20 ppm 1 a.u. 0.03 a.u.  and  comparison with calculated A , B and D centre concentrations.  M D L ' s are  a.u. = absorption units for diamond 1 cm thick.  presented i n Table 6.3. Examination o f LR curves i n Appendix D show that only a few diamonds have visible peaks from A centre LR absorption. The lowest calculated A centre concentration was 12 ppm. Visual examination o f deconvoluted curves show that the M D L for B centres is around 20 ppm. D centre concentrations are so small when compared to suites o f regular diamonds (i.e. diamonds with abundant platelets, F i g 6.6) and calculated values produce results which appear meaningless.  Based on data from regular diamonds and diamonds from this  study, the M D L for D centres is estimated at ~1 absorption unit measured at 1282 cm"  1  when calibrated for a diamond o f 1 cm thickness (Fig. 6.17). M D L ' s for hydrogen were calculated based on visual examination o f LR curves.  73  6.3 Results  The results i n this section are summarized i n table format i n Appendix C . A rawspectrum for each diamond, together with deconvolution curves for A , B and D centres, are presented i n Appendix D .  6.3.1 Nitrogen concentration and aggregation state measurements  Total  nitrogen  concentrations  in  the  examined  are  crystals  D. (X  00 300  low, with a mean o f 72 ± 100  72 ( l o ) and mode o f 36  MDL  0 9  ppm (the large standard deviation is due to the skewed  distribution  towards  nitrogen-free  diamonds)  (Fig. 6.15).  13  17  21  25  29  33  37  41  45  49  53  57  61  65  69  No. of diamonds Fig. 6.15. Distribution of total nitrogen concentration (ppm) for Rio Soriso diamonds. Diamonds are sorted in order of increasing total nitrogen concentration. MDL = minimum detection limit for nitrogen (20 ppm total nitrogen).  Nitrogen concentration ranges from 0 to  Type IaAB (7.2%)  541 ppm for total nitrogen, 0 to 116 for  Type IaA (1.5%)  A centres, and 0 to 541 ppm for B centres. nitrogen  There is a gradation i n total concentration  over  the  complete range, providing no indication o f a bimodal character.  The mean  B/(B+A) centre ratio is 95 ± 22 ( l o ) percent, covering the complete range  Fig. 6.16. Diamond type distribution for Rio Soriso suite (n=69). Averages were used for heterogeneous samples with multiple analyses points (chapter 7.0 on growth studies).  form 0 to 100 % B centres. The distribution o f diamond types are type II (37.7%) and type l a (62.3%).  Type l a stones are further sub-divided into type IaA (1.5%), I a A B  74  (7.2%) and IaB (53.6%) (Fig. 6.16). (The cut-off between Type I and II stones used is 20 ppm total nitrogen).  The relative concentration o f D centres is very low in comparison with A and B  9 8  centres (see any deconvoluted curve i n  7 d' 6 — GO 5  Appendix D ) .  is  When compared with  diamonds containing D centres (i.e.,  en  5»  o  Q  regular diamonds), it is clear that the  O n this  figure, the open circles (calculated from unpublished linear  work)  trend;  form  these  a positive  diamonds  are  regular. This trend is not observed i n Rio  Soriso  diamonds  (filled  grey  o This study O  O O  4 3 2  300  MDL ../  1 0 200  O O  O  O O  100  400  500  600  700  B centres (ppm)  measured D centre concentrations are unusually low (Fig. 6.17).  O Regular diamonds  Fig. 6.17. Plot of D centres versus B centres. Small filled circles are from this study and large open circles are from another suite of diamonds that contain roughly equal proportions of A and B centres and are considered regular (after Woods, 1986). A straight line through these points and the origin produces a reasonable fit with R =0.859. a.u. absorption units for diamond of 1 cm thickness. MDL - minimum detection limit. (Compare with Fig. 6.5 after Woods, 1986).  circles) making them irregular. Due to the high aggregation state o f R i o Soriso diamonds, it is not surprising that they are irregular as any platelets (D centres) that existed have likely degraded (see section 6.1.2). Calculated D centres are below the estimated M D L and have not been included i n the results o f this study.  Six diamonds contain measurable quantities o f A centres (diamonds 2-11, 4-10, 4-11, 415, 4-17 and 6-8) and, provided that platelet degradation does not begin until nitrogen is mostly aggregated as B centres (such as observed i n Fig. 6.5), the type I a A B diamonds would be expected to contain platelets. Although calculated D centre concentrations are low for all diamonds, the B ' peak provides another tool for identifying platelets (section 6.1.2, Fig. 6.4). This peak is visible i n the LR spectra for 5 o f the 6 diamonds, occurring at 1363.7 cm" and is absent i n all other diamonds (types IaB and Ha). Absorption values 1  75  measured at peak heights are presented i n Table 6.4. Comparison with data from regular diamonds (diamonds plotted in F i g . 6.17) suggests that all diamonds in this study are irregular,  even those with A centres.  However, note that nitrogen concentrations are  extremely low when compared to other studies (average o f - 5 0 0 ppm total nitrogen for regular diamonds which were compared with R i o Soriso diamonds) and that B ' peaks and D centre absorption curves would likely be small and possibly below detection. The type IaA and I a A B diamonds may be regular, but the low nitrogen concentrations make any classification difficult. Table 6.4. B' absorption and D centre concentrations for type IaA and IaAB diamonds Diamond  B'  uB'  B  A  D  total N  100xB/ (B+A)  89 483 428 55 1.63 5.02 2-ll(point 19) 1363.75 12 21 33 64 0.00 0.30 1363.75 4-10 0 48 0 48 0.01 0.84 1363.75 4-11 16 138 22 116 0.22 0.89 1363.75 4-15 51 53 27 26 0.07 0.29 4-17 (point A) 1363.75 48 52 25 23 0.11 6-8 Multiple points were examined on diamonds 2-11 and 4-17 in Chapter 7.0. Analysis points are indicated on Figs. 7.19 and 7.27 respectively. B' and uB' are measured in cm" . D, A and B centres represent absorption units (cm ) measured at 1282 cm" . Point 19 on diamond 2-11 was not included in Appendix D or Fig. 6.17 because there is considerable noise in the one-phonon range. This nt-data is discussed at greater length in section 7.3.8 (Fig. 7.20). 1  1  6.3.2 Other impurities detected by IR spectroscopy  Section 6.1.6.5 outlines typical absorption spectra obtained from diamond due to impurities other than nitrogen. O f these typical absorption spectra, only those pertaining to hydrogen were observed. Water, carbon dioxide and carbonate were not detected i n these studies.  T w o absorption spectra o f unknown origin were observed i n several  different diamonds.  76  6.3.2.1 C - H bonds  Spectra from seventy-one percent o f the diamonds studied showed a hydrogen peak at 3107 cm" .  The intensity o f the peak varies from 0 to 6.06 absorption units.  1  Only  crystals with large hydrogen peaks at 3107 cm" (~>1.2 a.u.) had a detectable secondary 1  peak at 1405 cm" . Peak intensity difference at 3107 c m ' is recorded i n Appendix C . 1  1  6.3.2.2 C H and C H bonds 2  3  Absorptions i n the CH-stretch region o f spectra from many o f the diamonds occur either as two well defined peaks at 2920 and 2850 cm" , with a shoulder at 2960 cm" (Fig. 1  1  6.18A), or as one broad peak between 2800-3000 cm" (Fig. 6.18B). In some cases, an 1  accompanying smaller peak was observed at 1460 cm" . 1  Several diamonds that exhibit LR absorption between 2700 and 3100 cm" were heated to 1  600 °C for 20 minutes to determine i f the CFL2 and CH3 complexes were forming only on the diamond surface. Prolonged exposure to these elevated temperatures should remove any C H and C H surface complexes. In all cases the diamonds heated did not exhibit 2  3  any absorption i n the CH-stretch region or at 1460 cm" after heating. A n LR spectra o f 1  B)  A)  4000  3000  2000  4000  1000  3000  2000  1000  Wavenumbers (cm- ) 1  Wavenumbers (cm- ) 1  Fig. 6.18. Absorption in the CH stretch region of Rio Soriso diamonds. A) Two sharp peaks at 2850 and 2920 cm" , B) Broad peak between 2800-3000 cm" . 1  1  77  two diamonds before and after heating can be seen i n F i g . 6.19.  Based on  these findings, any absorption observed between  2800  and  3000  cm"'  was  ignored. 4000  6.3.3 U n e x p l a i n e d spectra  1000  2000 1  There are two fR absorption spectra for which the cause is unknown.  3000  Wavenumbers (cm )  To the  Fig. 6.19. IR spectra of diamonds 2-5 and 4-17 before and after heating to ~600°C. 2.5A before heating, and 2.5B after heating, 4.17A, before heating, and 4.17B after heating. Note that in both cases the peaks at 2980, 2850 and 1460 cm" disappear after heating. Absorption in these regions is thought to be a result of the stretching of C H and C H orbitals. In this case, they are only on the surface of the diamond and are liberated by heating. 1  authors knowledge, they have not been mentioned in diamond literature before.  2  3  Both spectra create absorption i n the one-phonon range, the first w i l l  be  referred to as defect ' X ' (confirmed i n three diamonds, 2-5, 3-10 and 4-17), and the other as defect' Y ' (found in five crystals).  Defect ' X ' is characterised by a broad peak with maximum absorption at 1048 cm" , a 1  smaller peak at 867 cm" and a small broad absorption band centred on 1419 cm" (Fig. 1  1  6.20 A and B ) . The defects responsible for this peculiar absorption remain unknown.  Defect ' Y ' is observed in several stones, although the broad absorption spectra fails to match up when overlapped (a case may be made to split up defect ' Y ' into two or more unknown defect patterns). Defect ' Y ' produces the strongest absorption in diamond 2-3. This crystal is unique i n this study because it is the only stone with strong yellow colouration. Another unique feature about this crystal are local patches o f white/cloudy diamond (Fig. 6.21 A ) . The contact between the white and yellow patches, when viewed under a 35x magnification lens, reveals a rainbow effect o f colours. This is interpreted as being a result o f stress or internal fractures.  IR spectra collected through the yellow  diamond, which constitutes the bulk o f the crystal, indicates that this part o f the diamond  78  A)  c o  &  867  14U  co I  C/5  3  #4.17 4000  3000  2000  4000  1000  3000  Wavenumbers (cm')  2000  1000  Wavenumbers (cm')  Fig. 6.20. IR spectra of diamond with unexplained defect 'X'. A) Two IR spectra collected from diamond 4.17 after heating to 600 °C for 20 minutes. The most characteristic aspect of the unknown curve is the strong absorption observed at 1048 cm" . B) IR spectra illustrating the summation of type IaB diamond with impurity "X". Diamond 4.17 contains 'pure' defect 'X' and Diamond 4-18 exhibits pure IaB diamond. Diamond 2-5 (before heating) is interpreted as a combination of B aggregated nitrogen and defect 'X'. 1  contains no detectable IR-active nitrogen (Fig 6.2IB, curve labeled C ) and is thus considered type II.  IR examination o f the white patches contains defect ' Y ' .  The  unexplained curve exhibits a broad asymmetric peak with maximum absorption at 1086 cm" and a shoulder at 1230 cm" . A secondary peak is observed at 815 cm" (Fig. 6.21 1  B).  1  1  J. Milledge (personal communication, 2003) suggests that defect ' Y ' is likely some  form o f clay, most likely as a secondary mineral i n fractures between the white and yellow diamond. N o further studies were done to corroborate or refute this hypothesis.  3000  2000  Wavenumbers (cm') Fig. 6.21. Photograph of diamond 2-3 and IR spectra of defect 'Y'. A) Photograph of IR location points A, B and C on diamond 2-3. B) IR spectra of three points on diamond 2-3. Points A and B are from white patches on diamond while point C is through yellow portion of stone.  79  6.4 Discussion  6.4.1 Relationship between time and temperature  Nitrogen concentration and aggregation state measurements are useful for providing constraints on temperature and residence time the diamonds experienced i n the mantle (equation 6.7 from section 6.1.4)., Without further constraints on either o f these variables, and assuming the diamonds are from one source in the mantle, we can place them along a curve i n time/temperate space (Fig. 6.22). Further constraints for placing these diamonds along this curve come from a variety o f sources.  Constraints can be placed on  the residence time through dating o f the magmatic eruption which brought the diamonds to surface and through dating of  minerals  in  diamond.  Further constraints on the  residence  temperature  included  can be determined  after  examination o f inclusion types, phases  g 1200  3Ga  Io 1 Ga  o  c  o 1400  T3  u SP  ,-1.1 Ma  g  a  1600'  .-' 93,190 years  e  and associations (chapter 8.0).  10  4  Fig.  6.22  illustrates  the  estimated  minimum time diamonds would need to reside at certain temperatures  i n the  mantle to acquire the observed average nitrogen character measured through JR studies.  For example, a diamond at  estimated lower mantle  temperatures  would only need to reside at these depths  for  -93,000  years  before  10  5  •410 km  Transition zone  -oou Km  Transition zone L o w e r mantle  10 10 10 10 Residence time (years) 6  7  s  9  10  10  10  1  Fig. 6.22. Plot of minimum time averaged mantle residence temperatures versus minimum residence times for Rio Soriso diamonds. The residence time curve is specific to the mean total nitrogen concentration (72 ppm) and mean nitrogen aggregation state (95%) for this study. Diamonds would theoretically only have to reside in the lower mantle for 93,190 years to acquire the average nitrogen character measured. Transition zone diamonds should be less than 1.1 Ma. Temperature estimates for transition zone and lower mantle discontinuities are from Ringwood (1991). Time averaged mantle residence temperatures and times are calculated using A = 736747 1/s-ppm and Ea = 7.03 eV, from Taylor et al. (1990).  80  converting 9 5 % o f A centres to B centres.  Conversely, a diamond at temperatures  estimated for lithospheric cratons would require a much longer residence time, o n the order o f 1-3 billion years, to convert 95% o f A centres to B centres.  Note, however, that the there are numerous errors associated with equation 6.7 (the equation used to derive the curve i n F i g . 6.22).  Some o f the main errors are: 1) the  conversion o f A to B centres is still poorly constrained, especially under the effects o f deformation, 2) the constants used i n equation 6.7 (Ea and A ) may not apply to residence temperatures that would be experienced i n the transition zone (1500-1600 °C) or lower mantle (>1600 °C), 3) the variable average residence temperature is considerably abstract as diamonds most certainly existed i n the mantle at varying temperatures, 4) temperature calculations do not work w e l l for irregular diamonds with high aggregation states and may, at best, provide only a minimum estimate, provided the conversion o f A to B centres followed the regular trend to high aggregation states.  6.4.2 N i t r o g e n character o f R i o 10000  Soriso diamonds  1050"C.-  The  nitrogen  concentration  character  o f nitrogen  (total  6 IOOO  and %B  o.  3  c  aggregation) has been shown to vary  <D 00  3  between some diamond suites and is a reasonable  first-order  method  •3  for  20  distinguishing between some diamond  because  the  and temperature, there are many cases where  separate  20  populations  have  40  60  80  100  100xB/(B+A)  nitrogen  character is strongly controlled b y time  13OTC/  I200°C  10  populations (Kaminsky et al., 2001b). However,  100  &  Fig. 6.23. Plot of total nitrogen concentraion versus %B aggregation for Rio Soriso diamonds. Isotherms are calculated using a residence time of 3 Ga. Line at 20 ppm nitrogen marks the MDL for total nitrogen concentraion.  81  experienced similar histories and hence cannot be unambiguously distinguished based on nitrogen character alone.  Nonetheless, the nitrogen character is a useful fingerprint  record for a diamond suite.  to  The character for this suite can be examined i n F i g . 6.23.  From this figure, two separate sub-groups are visible: diamonds that are %100 B aggregated, and diamonds which contain some percentage o f A centres.  A third sub-  group, not depicted, is represented by diamonds which fall below the 20 ppm nitrogen line marking the M D L .  Included i n this figure are isotherms assuming the diamonds  resided i n the mantle for 3 G a . Although there is no data to substantiate this, they are included to illustrate the morphology o f isotherms as a function o f time.  6.4.3 Comparison of relative hydrogen and total nitrogen concentration  Based on observations o f this study and from growth studies (chapter 7.0), there is  a  positive  correlation  hydrogen and nitrogen (Fig. 6.24).  between  concentrations  T w o possible reasons for  this are that: 1) nitrogen and hydrogen impurities in diamond originated as an N - H complex or 2) conditions favorable for  the  incorporation  of  nitrogen 200  impurities are also favorable for the incorporation  of  hydrogen,  or  visa  versa.  400  600  800  Total Nitrogen (ppm)  Fig. 6.24. Plot of relative hydrogen versus total nitrogen concentrations for Rio Soriso diamonds. Relative hydrogen concentration measured as change in absorption units (a.u.) at 3107cm" . Error on total nitrogen concentration is 20% relative, and is 2.6% on relative hydrogen concentraion. A straight line of best through the data and the origin yields R =0.788. 1  The  prevalence  of  hydrogen  in  diamonds from other deposits is not w e l l recorded.  2  A detailed study by  Taylor et al. (1990) on several diamond suites from Australasia shows that hydrogen is  82  not a ubiquitous element in type l a diamond, as one might conclude from the results o f this study. From studies o f several diamond suites, Argyle was the only diamond suite found to contain abundant hydrogen bearing diamonds. Argyle diamonds are similar in terms o f nitrogen character to the suite in this study (see F i g . 6.25).  Total nitrogen  concentrations are generally <100 ppm, the aggregation state is high (close to 100% B centres), and platelet degradation is likely advanced (Taylor et al, 1990). Unfortunately there is no information on the relative hydrogen concentrations for Argyle diamonds.  6.4.4 Comparison of nitrogen characteristics 10000  with other studies  The  nitrogen  characteristics nitrogen and  0  (i.e.  concentration  aggregation  state)  for R i o Soriso diamonds is essentially the same as recorded by Harte and Harris (1994), Kaminsky et al. (2001b) and Araujo  B &  IOOO  a  Koffiefontein (Deines et al., 199I)  ft  Jagerst'ontein (IX'incs el al.. 1991)  ft  DO-27 (Davies*?/al.. I999)  •  Sao l.uiz (Hutchison. 1997) ' 1  mostly in the form o f B centres).  However, this  nitrogen character is not unique  to  diamonds  Juina (Fig.  area 6.25).  Juina (Kaminsky el al.. 200I) Argyle (Taylor el al.. 1990)  ii  I  a  — 8 E3  o  100 20  mot  1200°C  1250°C  I  o i300°C |  10 20  40  60  80  100  100xB/(B+A)  diamonds (i.e. low total concentrations,  ft  c  et al. (2003) for Juina  nitrogen  Rio Soriso (this study)  1t  Fig. 6.25. Plot of total nitrogen concentration versus %B aggregation for selected diamonds and diamond suites worldwide. Isotherms are calculated for a mantle residence time of 3 Ga. All data are for single diamonds except points labelled 1. and 2. Point 1 is the average of 98 diamonds from three rivers in the Juina area: Rio Sao Luiz, Rio Vermelho, and Corrigo Chicora, from Kaminsky et al. (2001b). The error on the data point is unclear, although it is likely similar to the error in this study. Point 2 is the average 13 diamonds from Argyle (Taylor et ah, 1990). The range in %B is from 60 to 95, and for all but one stone, the total nitrogen concentration is less than 100 ppm.  83  The combination o f 'diamond mineral inclusion' and 'nitrogen character' studies has established that the Tower mantle' diamond suite is characterized by type JJ or type IaB diamond with low total nitrogen concentrations (Harte and Harris, 1994; Hutchison, 1999; Davies et al, 1999; and Stachel et al, 2002). There are also isolated occurrences o f diamonds with this character from South Africa and Australia, among other places (also plotted on F i g . 6.25), however these diamonds are rarities within their respective diamond suites.  84  7.0 Growth studies 7.1 Background  Historically, the internal growth morphology o f diamond has been examined using a variety o f techniques, including birefringence studies (e.g. Tolansky, 1966), X-ray topography o f whole diamonds (e.g. Lang, 1964), and etching o f diamond plates (e.g. Seal,  1965).  However,  these  techniques  have  given  way  more  recently  cathodoluminescence ( C L ) studies o f polished diamond sections and plates. background behind the principles o f C L is outlined i n section 3.1.2.  to  A brief  Optical C L is an  inexpensive technique which is extremely sensitive to chemical heterogeneities. A s such, growth layers with even minute differences i n chemistry often exhibit different C L colours or different colour intensities.  C L studies o f diamond plates (two parallel  polished surfaces) and sections (one polished surface) used i n conjunction with infrared (IR) and isotopic studies have allowed scientists to map the variations i n chemistry during diamond growth (eg. B o y d et al., 1987; Milledge et al., 1989; Mendelssohn et al., 1991).  The purpose o f this study is to integrate C L studies o f diamond plates and  sections with IR studies.  7.1.1 Internal structures in diamond  The vast majority o f natural single-crystal diamonds grow as octahedrons (eight (111) faces), with cubes (six (100) faces) and cubo-octahedral forms (eight (111) and six (100) faces) generally being less common (previously introduced in section 2.1.1 on diamond growth habits).  Other primary growth faces are possible on natural diamonds, but are  extremely rare and need not be considered here.  The growth history o f diamond is  inherently more complex than what would be concluded from observation o f external morphology alone.  Once a suitable site for carbon precipitation exists (from herein  referred to as the seed), both (111) and (100) faces may grow.  85  The face that w i l l  dominate the crystal once growth is complete is the face that grows the slowest - the face that grows the quickest grows to a point and to extinction (Clausing, 1997). Matters are complicated because diamonds  often  experience multiple events o f growth and  dissolution; whatever external faces the crystal retains reflects only the last growth event. It is common to observe the 'competition' between the growth o f (111) and (100) faces in the core o f the crystal.  Fig. 7.1 contains images o f opposing sides o f an octahedral diamond (in terms o f external morphology).  In F i g . 7.1 A , clearly visible are the seed, the competition zone (where  octahedral and cubic faces compete for dominance) and the octahedral outer zone.  Octahedral growth layers are recognised (in two dimensions) because they typically form straight, parallel lines, while cubic growth layers are typically rounded or hummocky. This is likely the main reason why smooth cubic faces are rare i n natural stones and octahedral  surfaces  are  often flat (provided the stone  has  not  B)  A) (in)  (in)  experienced dissolution). Growth  layers  developing perpendicular to  the  discussed  (111) in  are  literature  using a variety o f names, some  of  common  the  more  terms  tangential (Sunagawa,  are growth  1984a  and  b), flat faces (Harte et al,  1999),  faceted  Fig. 7.1. C L images of central plate of diamond (opposite sides). A ) The central zone reveals the diamond seed (the two black spots). The lighter grey zones surrounding the seed indicates growth of two types: cuboid (forming perpendicular diagonal arms of two brighter shades of grey) and octahedral (homogeneous darker grey zones in the central region). The fine scale laminations on the outer rim formed on octahedral faces. B) The opposing side of the diamond contains lighter grey zones across the short diagonal (indicating cuboid (100) growth) while the remainder of the central light grey zone is comprised of octahedral growth. Note the fine scale, straight character of the outermost layers. Diamond is cut parallel to the ~(110) plane. Magnification x 40. (image reproduced from Bulanova et al., 2002).  86  Fig. 7.2. Cartoon of internal morphology for cuboid and octahedral growth where both faces grow evenly. Left image indicates where diamond slice of right image is cut (parallel to [100] the (010) surface). Black spot in middle is crystal seed. * indicates that the growth [111]" direction is oblique and not perpendicular to the (111) face when viewed on a surface cut parallel to the (010).  growth (Moore and Lang, 1972), layer by layer growth (Moore and Lang, 1972), or octahedral (111) growth (Davies et al., 1999). For the sake of consistency and simplicity, only the terms octahedral and (111) growth/face will be used in this report. Growth bands that develop roughly perpendicular to the (100) direction are typically irregular and may cross-cut one another. These growth bands are sometimes referred to as normal growth (Khachatryan and Kaminsky, 2002), unfaceted growth  (Moore  and  Lang,  1972),  rounded or  hummocky growth (Harte et al., 1999) or cuboid (100) growth (Davies et al., 1999). The more irregular the bands  the  further  the  growing  conditions  are  considered to be from equilibrium. Cuboid growth of fibrous or columnar habit has also been observed (Lang, 1974b; Sunagawa, 1984b). This report will use the terms cuboid and (100) growth/face for simplicity.  Natural diamonds of solely cuboid growth are not very common (particularly in this study) and thus will not be discussed in isolation, rather, combination cuboid and octahedral forms will be examined. As previously mentioned, cuboid and octahedral growth usually develops once a primary nucleation site exists, e.g. Fig.  87  Fig. 7.3. X-ray section image of cubo-octahedral diamond. Note how octahedral (111) sectors are bright while (100) sectors are dark. This is a result of impurity partitioning. Cut parallel to the (010). * indicates that the growth direction is oblique and not perpendicular to the (111) face when viewed on a surface cut parallel to the (010). (image reproduced from Welbourn et al., 1989).  7.1.  In this e x a m p l e , i n the end, octahedral g r o w t h  prevailed.  In  contrast,  Figs.  7.2  and  7.3  show  e x a m p l e s w h e r e b o t h (111) and (100) faces  develop  w i t h o u t one face g r o w i n g to e x t i n c t i o n (both  figures  are o f slices p a r a l l e l to the (100)).  T h e difference i n  C L brightness b e t w e e n c u b o i d a n d octahedral g r o w t h zones i s a result o f sector dependence o f i m p u r i t y uptake (e.g. F i g . 7.3).  F i g . 7.4. C L image illustrating growth-sectorial dependence in synthetic diamond. (image reproduced from Burns et al, 1990).  Sector dependence o f i m p u r i t y i n c o r p o r a t i o n is perhaps best illustrated i n the C L i m a g e o f a synthetic stone i n F i g . 7.4. F o r this d i a m o n d , n i t r o g e n (as disaggregated N centres or type  lb  diamond)  partitions  into  different  growth  zones,  from  highest  to  lowest  concentration, i n the f o l l o w i n g order: (111), (100), (113) and (110) ( B u r n s et al,  1990).  (It i s unclear w h y the m o s t nitrogen r i c h sector e x h i b i t s the faintest C L c o l o u r , as the opposite i s m o r e often true, that i s , zones o f h i g h e r n i t r o g e n e x h i b i t the brightest C L c o l o u r s ( D a v i e s , 1998)). preference  P i o n e e r i n g studies b y L a n g et al. ( 1 9 7 4 ) demonstrated  the  for n i t r o g e n i m p u r i t i e s to f o r m o n (111) faces d u r i n g g r o w t h o f natural  d i a m o n d o v e r (100) faces.  H o w e v e r , i m p u r i t i e s s u c h as o p a q u e m i c r o - i n c l u s i o n s (<1  u m ) or m i c r o - p r e c i p i t a t e s tend to d e v e l o p o n (100) faces ( L a n g et al,  1992) (e.g. F i g .  7.3). B u r n s et al. ( 1 9 9 0 ) f o u n d that sector dependence i s a f u n c t i o n o f i m p u r i t y type and temperature, a m o n g other variables.  S o m e other terms c o m m o n l y used i n literature to describe i n t e r n a l g r o w t h features are 'agate' structure, ' c e l l u l a r ' structure a n d sector z o n i n g . A g a t e ( L a n g , 1 9 7 4 a ; Z e z i n et  al,  1990) and c e l l u l a r ( D a v i e s , 1998) structures are w a v y , often f i n e l y l a m i n a t e d features. T h e y are interpreted as h a v i n g g r o w n r a p i d l y under fluctuating c o n d i t i o n s and i n the presence  o f fluids  (Davies,  1998).  Sector  z o n i n g (sectorial-growth)  consequence o f i m p u r i t y p a r t i t i o n i n g between g r o w t h faces ( D a v i e s et al, al,  1992) (described  i n the p r e c e d i n g paragraph).  88  forms  as  a  1999; L a n g et  Contrasting C L colours  and/or  intensities  for  growth  faces  that  formed  contemporaneously are an indication o f sector zoning. The crystal core in natural diamond is typically rich in impurities and often (Harte et al,  exhibits brighter C L colours  1999; Bulanova, 1995).  It may be  rounded, or in some cases, exhibit a cross or ' X ' morphology (commonly referred to as the 'Maltese cross', e.g. Lang et al (1992) or the 'iron cross', e.g. Shigley et al. (1986)).  (Figs. 7.5 and 7.6A).  The  rounded morphology o f a diamond core is typically interpreted as being an indication o f resorption.  The  centre o f this core is often comprised o f microinclusions o f graphite (Bulanova, 1995) (e.g. black spots seen i n F i g . 3.1 A ) .  Fig. 7.5. SEM-CL image of diamond core. The dark grey sectors indicate growth on the (111) face while the lighter grey, curved and kinked sectors indicate growth developed on the (100) face, (reproduced from Bulanova et al., 2002).  Bulanova et al. (1996)  propose that the impurities found at the genetic core lowers the nucleation barrier, thus facilitating diamond crystallization. The irregular morphology o f diamond surrounding the growth seed may be cited as evidence for rapid growth under less stable conditions than for subsequent crystal growth (Davies, 1998).  The time it takes for natural diamond crystals to growth remains unclear.  However,  evidence suggests that diamond growth is a slow process and that it varies depending on which growth layers are developing.  W e l l formed, parallel, fine scale laminations  indicating octahedral growth are interpreted as signs o f slow growth (Sunagawa, 1984b). Other evidence cited as indicators for slow growth are signs o f both growth and resorption i n the same diamond and hence fluctuating conditions (Davies, 1998) and variation in mineral inclusion chemistries and types from core to r i m , also indicating changing conditions (Bulanova et al, 1995). Cuboid growth, in general, is interpreted as being an indicator o f more rapid growth (Sunagawa, 1984b).  89  It is rather obvious that in order to examine the internal growth structure o f diamond, a polished surface must cut through the crystal, thus intersecting different growth layers. A n d indeed, to examine all stages o f growth, a plane must intersect the 'seed' or primary nucleation site.  A s well, thought should be given to the orientation o f the cut, i.e.,  parallel to the (100), (111), or (110) face. Diamonds that exhibit remnant growth faces i n their external morphology can be cut along particular planes.  Typically, they are cut  parallel to the (110) plane (e.g. F i g . 7.1) or the (100) plane (e.g. Figs. 7.3 and 7.6A-B). However, diamonds that are strongly resorbed (class 1 on the M c C a l l u m et al., 1994 resorption scale) or broken, often cannot be oriented i n terms o f their crystallographic axes. In the case o f broken diamonds, it is merely a guess as to where the crystal core is located or i f the diamond fragment even still contains the core. Diamonds which have experienced strong resorption and/or have been fragmented, are not the most suitable subjects for growth study. Diamond cuts may only intersect a few growth layers, i f any, and w i l l typically not intersect the diamond core.  There is still debate as to whether diamonds are metamorphic or fluid-derived, however, most evidence seems to be i n favour o f a fluid-derived origin. Sunagawa (1984b) uses the supersaturation level o f carbon i n a fluid or melt to describe the crystal habit (see section 2.1.1). The concentric oscillatory nature o f growth pattern zonation i n diamond is considered further evidence for a fluid or melt origin (Bulanova et al., 1995). However, Davies (1998) points out that similar zoning is reported i n literature for regional and contact metamorphic minerals (e.g. Yardley et al, 1991). Brittle and plastic deformation is clear evidence that diamonds are surrounded by a solid medium at some time, either during or between episodes o f growth. Fractured internal cores with euhedral rims and the truncation o f growth structures by other zones indicate brittle fracturing and fine, parallel laminations are evidence for plastic deformation.  A polished surface often reveals the extensive and complex history o f events, including multiple  stages  of  growth,  resorption,  90  fracturing,  deformation  and  changing  environmental conditions. The interpretation of these events in natural diamond is a difficult task. Examination of a polished surface, even when the diamond can be oriented in terms of crystallographic axes, often reveals very little. Even when internal growth morphology can be observed in CL, often it is too complex to describe, let alone interpret. With this in mind, and using Fig. 7.6 as an example, common internal growth features observed in natural diamond and their possible interpretation are as follows: 1) the seed and immediate core, usually the zone of brightest CL from which features radiate, sometimes in the form of a Maltese cross; 2) the rounded 'outer core', which is conventionally cited as evidence for dissolution, but has also been cited as evidence for cuboid growth (eg. Harte et al, 1999) and non-planar growth in rapidly fluctuating conditions (Davies, 1998); and 3) the outer rim, where typically well-formed octahedral faces develop. A)  B)  Fig. 7.6. Digitally enhanced CL images highlighting some typical internal structures observed in diamond. A) These regions are: A, the core (centre cross); B, cuboid growth layers; C and D, dominantly cuboid growth but bounded by octahedral faces; and E, dominantly octahedral growth. B) Growth regions are: A, B, and C, dominantly cuboid growth layers; D, dominantly octahedral habit, (black dots are ion microprobe analysis points), (reproduced from Harte et al., 1999).  91  7.1.2 Infrared spectroscopy of polished diamond  M a n y C L studies o f diamond include I R spectroscopy to measure the abundance o f and distributions o f impurities present (e.g. Davies et al, 2001).  1999a; Davies, 1998; M c K e n n a ,  fR data from the diamond core and subsequent growth bands can be used to say  something about the diamonds residence temperature and time (discussed in detail i n section 6.1.4). For accurate results, it is essential that the IR path through the diamond for each data point only intersect the growth zone o f interest. This is accomplished by polishing two parallel, flat faces that are close together.  The C L o f both sides is  examined to determine i f and how the patterns match up. From these images it may be possible to select points that w i l l only activate specific zones o f interest. However, it is often difficult to polish diamond into thin plates. It is common for IR results to transect more than one growth zone and thus give results that are more qualitative than quantitative.  It is important to consider the effects o f sampling multiple growth layers  when interpreting IR data from diamond plates.  7.2 Analytical techniques  The internal morphology o f fourteen diamonds was examined.  Thirteen large stones  (0.056-0.404 g) were cut and polished and one small stone (0.006 g) with a reasonably flat fracture surface was studied.  The selection process used for choosing which  diamonds to polish essentially followed four steps. excluded because it would be difficult  Firstly, all small grains were  to analyse multiple points on the  FTIR  spectrometer; all diamonds classified as 'large' (Fig. 2.1) became initial candidates for internal growth studies. Secondly, diamonds with large inclusions that could potentially be lost during cutting and polishing were omitted. Thirdly, diamonds with discernable crystal faces were preferentially selected so that polished surfaces could be made parallel to known crystallographic orientations (i.e. polished surfaces parallel to the (100)). However, most grains were strongly resorbed (see section 2.3.3) or broken and irregular  92  in shape, thus making it difficult to instruct polishers how to orient the grains (for this reason most diamonds were polished in a random orientation).  Lastly, a few grains  containing abundant 'micro-inclusions' were selected in hopes that polished faces would expose at least some o f the inclusions, thus making future inclusion analysis possible. Once cut and polished, the thirteen diamonds were examined under C L and seven o f the most interesting grains (in terms o f internal morphology) were returned so that a second parallel face could be polished on the opposite side to make diamond plates.  There was initial concern that the internal structures for diamonds polished at a random orientation would be difficult, i f not impossible, to decipher. However, it was found that most diamonds revealed something about their growth histories and that orientation was not such a concern.  Study o f highly deformed diamonds from Eastern Australia by  Davies et al. (1999b) found that the 'knotting' effect o f the deformation seams made it difficult to cut along a preferred crystallographic face. They found that polishing stones in a random orientation was not critical to deciphering the growth morphology.  The  abundant deformation laminations observed on diamonds i n this study (section 2.3.7) suggests that similar problems would likely be encountered i f grain orientations were specified for all stones.  The 14 diamonds were examined using a Cambridge Instruments Cathode Luminescence ( C I T L 8200 m K 4 ) system described i n greater detail in section 5.2. thoroughly washed with ethanol before mounting in carbon putty.  Diamonds were Grains were  embedded i n putty so that the polished surface was flush with the putty surface. Carbon putty was used so as to minimize internal luminescence and to avoid charging the outside o f the crystal. C L images were collected using a N i k o n Coolpix 995 digital camera.  LR data was collected and processed following the procedures outlined in section 6.2. Precision o f the method is outlined i n section 6.2.2 and M D L ' s are listed in Table 6.3. The main difference between examination o f whole rough stones and polished surfaces is  93  the need to accurately locate the JR points on the diamond. This was accomplished by taking photographs  o f diamonds under both a conventional microscope and C L  microscope. Images were combined, making it possible to locate points o f interest during IR transects.  Aperture size varied but was generally less than 200 urn.  Diamond  thickness (optical path length o f transmitted radiation) was determined using the absorption coefficient calculated by Taylor et al., (1990) measured at 2030 cm" (0.47 ± 1  0.01 units per m m path length).  7.3 Results  The results o f C L and fR examination o f seven diamond plates, six polished diamond surfaces, and one fractured stone are presented below.  Each diamond is discussed  separately i n terms o f growth mechanisms and how nitrogen and hydrogen character varies across different growth zones. A C L photograph o f each diamond is included as well as an IR transect indicating data points. A summary o f fR results (total nitrogen concentration, percent aggregation and relative hydrogen concentration) is also included for each stone (diamond 2-11 only includes hydrogen results).  The results and  interpretations are summarised in Table 7.1 (section 7.4.1).  7.3.1 D i a m o n d 1-2  Diamond 1-2 is a strongly resorbed colourless fragment that possesses weak to moderate blue C L with some evidence for oscillatory zoning (Fig. 7.7). N o particular orientation was specified for polishing. The centre o f the fragment does not appear to have been intersected with polishing and likely exists towards the top o f the image where C L intensity is highest, possibly at depth i n the remaining portion o f the stone (below points 27-29). Fine growth layers are visible i n two places, near the bottom left o f the image and near the top. The fine layers near the bottom left are highly irregular, rounded forms  94  I  2  3  4  5  6  7  X  9  10  II  13 14  15  Id  17  111  Points  III  22  23  24  25  26  27  28  2')  Points  Fig. 7.7. Image of CL of diamond 1-2 with IR data for transects 1 to 18 and 10, 22 to 23. This diamond exhibits weak to moderate blue CL. A fracture surface is visible along the top edge and the crystal core is likely below point 28. Octahedral zoning forms around the bright core while 'agate' zoning forms further from the diamond core (bottom left). Minimum detection limit (MDL) for Total Nitrogen is represented by grey line at 20 ppm. MDL for relative hydrogen concentration is too low to be indicated on graph (listed at 0.03 absorption units in Table 6.3. %B Aggregation for points listed as below MDL for total nitrogen concentration are meaningless.  95  that appears similar i n morphology to 'agate' structure. It is unlikely that these rounded forms be attributed to resorption as dissolution is usually an uneven process which should lead to the formation o f cross-cutting structures.  They are most likely features o f growth,  however, their crystallographic nature is unclear.  Flat oscillatory zoning faintly observed near the top o f the image is brightened and enlarged in F i g . 7.8.  Fig. 7 . 8 . Close-up greyscale CL  of diamond 1-2. Oscillatory zoning with straight boundaries is  Because o f the straight morphology o f the bands, this growth pattern is interpreted as indicative o f octahedral growth.  interpreted as forming on paleooctahedral faces.  The crystal core appears brighter blue, but is too blurred for a detailed  interpretation.  Nitrogen concentration, aggregation state and relative hydrogen peak  concentration are presented at the bottom o f Fig. 7.7 for two traverses, one from points 1 through 18 and the other from 10 and 22 through to point 29. Nitrogen concentrations range from ~0 to 80 ppm while the aggregation state is essentially pure IaB. For the first traverse, nitrogen concentrations generally increase towards the middle, peaking at point 11 (-70 ppm total N ) and then decreasing again to point 18 at the edge o f the fragment. The impossible totals (130-140 % B ) for points 1 and 2 are due to the large error associated with low nitrogen totals. The relative hydrogen concentration peak mimics the total nitrogen curve extremely well, with the exception o f points 3, 4 and 13. The second traverse starts close to where the first traverse peaks in total nitrogen.  C L intensity  increases towards the top o f the image where the second traverse terminates.  Total  nitrogen concentration gradually increases and peaks at point 18 (80 ppm total N ) . The relative hydrogen concentration curve also mimics the total nitrogen curve well (for traverse 10, 22-29), with the exception o f point 26.  The stone is considered type IaB although it gradually becomes void i n nitrogen (type II) towards the fragment edges. Total nitrogen concentration gradually decreases from ~80  96  ppm (type I), likely near the crystal core which is interpreted to be somewhere under the points 27-29, to practically 0 ppm (or type II) at the edges. " A " centres are absent in this crystal, suggesting it experienced either extremely high temperatures or long residence times at moderately high temperatures.  The growth structure is interpreted to be  octahedral (at least near the core) with 'agate' zoning further away from the growth centre. There is a clear correlation between C L colour and nitrogen concentration as C L colour intensity is strongest where nitrogen concentrations are highest and weakest where nitrogen concentrations are low.  7.3.2 D i a m o n d 1-4  Diamond 1-4 is a non-uniformly coloured stone (brown and colourless) with a peculiar shape, having dimensions 6.0 x 5.5 x 1.25 mm.  Both sides were polished (with no  particular orientation specified) on this grain, making the plate ~1.3 m m thick, although very little material was removed i n the process. C L examination o f both sides reveals a very interesting yet complex growth structure (Fig. 7.9).  T o facilitate comparison  between both sides, all images o f side B have been flipped horizontally so they could be overlaid. C L colours vary slightly from side A to B however this is only an artefact o f the digital camera used to capture the images.  The colours are certainly lighter blue  (turquoise) than most C L colours observed for R i o Soriso diamonds.  There is a  remarkable contrast i n C L intensity, with some sections o f the stone being bright turquoise while others show no evidence o f C L . The core o f the crystal is likely contained with the remaining polished fragment where the most intense turquoise C L is observed, which is also where the brown colouration is found, as observed under optical microscope.  Three or four 'arms' with bright turquoise C L radiate outwards from the  crystal core and are separated by diamond that does not C L . There is a r i m o f diamond that possess bright turquoise C L (best observed at the bottom right o f Fig. 7.9 (side B ) . The cross section A - A ' shows a slice orthogonal to the polished surface o f one o f the arms.  97  Fig. 7.9. Photos of CL, FL and body colour for various sides of diamond 14. Clockwise from top left, photo of side A before diamond was polished; fluorescence photo of side B (also before polishing, flipped horizontally for comparison, (note the faint outline of a dark triangle near the bottom of the image, which is clearly visible under CL); CL of side B, (also flipped) with A - A ' transect; CL and side A with A-A' transect. Left, CL image of A-A' cross section (side A face up). The thickness of the cross-section is ~1.3 mm.  98  Fig. 7.10. Image of CL of diamond 1-4, side A, with IR data for transect 1-15 across polished surface. Two triangular shaped arms are visible radiating from a bright core, one radiates to the left and the other towards the bottom. A C L image of side A including transect 1-15 with accompanying LR data can be seen in Fig. 7.10.  Nitrogen concentration varies from - 2 0  to over 400 ppm.  This is a  considerable range in concentration and highlights the potential errors in collecting 'bulk' IR data from one point in a diamond. concentration and C L intensity,  There is a clear correlation with total nitrogen  with high nitrogen totals corresponding with the  triangular arm and the diamond core. Total nitrogen concentration decreases from core to rim although the 'arms' certainly introduce a complex geometry.  When present, nitrogen  is only in the form of B aggregates. There is a remarkable positive correlation between the relative hydrogen and total nitrogen concentrations.  The C L pattern observed is complex. thus  indicating octahedral growth.  There are sections where flat, parallel faces exist, However, most  o f the  stone exhibits  either  homogenous turquoise C L (the majority of the core and radiating arms), or homogenous dark C L (type II) diamond, revealing little in the way of growth type. appearance to sector-dependence  It has a similar  growth for cubo-octahedral diamonds, described in  section 7.1.1. If this was the growth mechanism, then the turquoise arms would indicate  99  octahedral growth while the dark zones would indicate cuboid growth. The dissolution seen in the top portion o f the image (starting just above point 8) appears to truncate the nitrogen-rich arm.  Deformation  laminations  were  observed  on  this  diamond during characterization o f physical features (see S E M image F i g . 2.26B). Plastic deformation is visible in the large triangular shaped arm o f side A , recognized by the three sets o f intersecting parallel lines (Fig. 7.11).  Fig. 7.11. Greyscale photograph of deformation laminations observed under CL. Close-up of Fig. 7.10, side A. (Compare with Fig. 2.24D of SEM image of deformation laminations).  The C L pattern o f side B with two accompanying IR data transects can be seen in F i g . 7.12. Transect 1-14 w i l l be discussed i n reverse order as it is presented, from left to right, in F i g . 7.12.  The transect does not intersect the diamond core and does not reach the  highest vales collected for points 9 (side A ) and 19 (side B ) . The maximum value is recorded at point 7 (-375 ppm total N ) with minimums at points 3 and 13 (<~75 ppm total N ) .  The aggregation state is pure IaB.  The positive correlation between total  nitrogen and relative hydrogen is, again, quite remarkable. Transect 15 to 24 follows one arm from the diamond edge to the core. Total nitrogen concentration increases slightly, peaking at point 19 at - 4 0 0 ppm N . The same positive correlation between hydrogen and total nitrogen concentration is evident.  Growth patterns with straight, parallel morphology (suggesting octahedral growth) are evident in many parts o f the image.  The enclosed type II triangle is quite interesting.  The C L image o f side B suggests sector growth (as did the C L image o f side A ) . There are growth bands on the left side o f one 'arm' (at point 8) that appear to grow into the area o f type 13 diamond. This implies that the diamond was hollow and grew inwards. This is in contrast to the direction o f growth band laminations that are visible i n sector-  100  Side B  14  13 12 II 10 9 II  7 6  5  4  3  2  I  Points  15  16  17  18  19  21)  21  Points  Fig. 7.12. Image of CL of diamond 1-4, side B (flipped horizontally), with IR data from transects 1-14 and 15-21.  101  dependence growth for diamonds in Figs. 7.3-7.4.  Dissolution may be visible near the  outer margin o f the crystal (bottom left and right), truncating a bright turquoise zone. There is a r i m o f diamond with bright turquoise C L (bottom o f Fig. 7.12).  This diamond is considered type  IaB but has  a marked variation in nitrogen  concentration, ranging from 20 to - 4 0 0 ppm. A n y nitrogen occurs as 100% B centres. The core o f the diamond is too blurred and bright to reveal any detailed morphology, however, it is clearly the richest zone in nitrogen and other brownish coloured impurities. Growth development on the core is most likely cubo-octahedral, with octahedral layers soaking up all available nitrogen (the bright turquoise arms) and cuboid layers (dark regions) containing little nitrogen. The enclosed triangle (bottom o f Fig. 7.12, side B ) , formed either cuboid growth as a result o f a sudden cessation o f growth, or the diamond experienced partial resorption, after which only octahedral growth prevailed.  One  unresolved matter is that fine growth bands are visible parallel to one growth arm (below point 8 on side B o f Fig. 7.12).  If cubo-octahedral growth developed, growth bands  should be perpendicular (i.e. concentric from the core) as oppose to radiating. A t some point in time the diamond experienced deformation (syn or post growth) and brittle fracturing.  7.3.3  D i a m o n d 2-1  Diamond 2-1 is a colourless grain interpreted to be an aggregate o f two crystals (Fig. 7.13).  The two  individual crystals are separated by what appeared to be a fracture (crosses transect at point, F i g . 7.14).  It  was believed that the diamond was an aggregate before C L examination because etch pits on either side o f this fracture  were  at  orientations  that  could  not  be  explained i f they belonged to one crystal. The grain  102  Fig. 7.13. Photograph of polished surface of diamond aggregate (Diamond 2-1). Note fracture dividing crystal into two (irregular white line).  was chosen for growth study because o f this hypothesis and was polished into a 2.7 m m thick  diamond  diamond small,  plate.  contains  dark  examination  The  numerous  inclusions. of  the  CL image  reveals that there is a growth centre on the right side o f the 'join' but there is no discernable centre on the left side.  CL  mm  colour is weak to moderate blue (image is brightened for clarity of  features),  typical  stones in this study.  of  most  Brighter  •  an  in  25  green/turquoise C L is visible at the top o f the image and is likely  E2  late diamond crystallization on the aggregate. fractures  There are several  with  intense  CL  observed i n the stone, one being clearly visible below point 5 (Fig. 7.14).  The crystal on the  right exhibits three concentric layers, a bright core, a weak shell and then a brighter outer shell.  The bright core appears  cloudy i n F i g . 7.13 and may be comprised o f numerous  Fig. 7.14. Fig. 7.14. Image of CL of diamond 2-1 with IR data for transect 1 to 11. Diamond is an aggregate of two crystals, core of right crystal is clearly visible above point 9. Growth structures for the crystal on the left are absent. The two crystals are joined along the irregular line that intersects transect near point 7.  small  103  fluid inclusions (Navon et al., 1988). The boundary between the weak and bright outer shell is irregular i n shape.  Growth patterns to the left o f the j o i n are not so clear. The  core o f this grain may be near the edge below point 1 where the most intense C L is observed,  irregular growth lines between brighter and darker zones are also visible,  although not as well developed as on the right half.  It is unknown i f the irregular  boundaries are a result o f growth or dissolution.  fR data for the transect across this aggregate (points 1 through 11), unfortunately, does not intersect the zones o f brightest C L , particularly the bright core on the right half. There is effectively no absorption i n the one-phonon range, although deconvolution calculates -5-20 ppm total nitrogen.  Detailed examination o f IR spectra reveals that  there is possibly some nitrogen i n the form o f B centres, however, this diamond should be classified as type n. It is possible that there are detectable amounts o f nitrogen i n the crystal core on the right half.  This grain is an aggregate o f two crystals, one with probably a more nitrogen-rich core but as a whole, is considered type II and the other crystal, with fewer visible growth structures, is also type II. The mechanism o f growth is uncertain. Certainly the diamond experienced an episode o f resorption to create the rounded overall rounded morphology, however, it is unclear i f the diamond experienced episodic periods o f dissolution during growth.  7.3.4 D i a m o n d 2-2  Diamond 2-2 is a large (6 x 5.5 x 2 mm), strongly resorbed, broken, pink crystal that was polished in a random orientation. C L colours are turquoise blue and somewhat brighter than most diamonds i n this study. There is one localized section with yellow/green C L found near the top middle o f the bright blue C L zone and it is possible that this region is  104  29«  1 mm  30<  105  the core o f the crystal (Fig. 7.15). There are essentially only two zones visible, one with bright C L and the other without C L . The bright C L zone is complex i n shape and is generally irregular and smooth, but i n some places has straight, angular edges (e.g. top right o f F i g . 7.15).  It is possible that there has been some brittle deformation as is  suggested by the 'floating island' o f bright C L above points 16 and 19. The contrast i n bright and dark zones suggests sectorial dependence o f impurity uptake (and thus cubooctahedral growth), however, the morphology is too complex to conclusively draw this conclusion.  Transect 7-19 follows along an arm, likely starting near the crystal core. Attempts were made to analyse points to the left o f 7, but no spectra could be obtained, likely a as result o f the large number o f fractures  i n this part o f the diamond.  Total nitrogen  concentrations range from 25 to -225 ppm and the aggregation state is 100% B centres. Nitrogen concentration remains roughly constant (-175 ppm total N ) until points 16 and 19, where the concentration drops considerably (-25 ppm total nitrogen at point 19). There  is  a  concentrations.  positive  correlation between  relative  hydrogen  and  total  nitrogen  Transect 14, 20-30 starts on the bright blue C L arm and crosses a zone  that does not C L . Nitrogen concentrations range from 50 to - 2 0 0 ppm total nitrogen, all as B centres.  The minimum nitrogen concentration is reached at point 22 and steadily  increases towards point 29 (-150 ppm nitrogen). It is possible that the low C L intensity zone that this transect crosses is pseudo-semispherical i n shape and that the bright C L arms seen i n the middle (along transect 7-19) and the arm seen at the bottom, are joined at depth.  The C L o f the opposing side o f the diamond was examined but was  inconclusive towards elucidating the three dimensional morphology o f the growth structure.  Again, relative hydrogen concentration correlates positively with the total  nitrogen concentration.  This diamond has nitrogen free zones, however, the majority o f the crystal appears to be relatively nitrogen-rich for the suite o f R i o Soriso diamonds. It is a type IaB stone with  106  zones o f type II diamond.  The growth structure may indicate sector dependence, and  hence cubo-octahedral form.  However, there is no clear evidence indicating cubo-  octahedral growth, and as such, the internal structure w i l l be classified as complex. The crystal core may exhibit a yellow/green C L , but this is not certain - this zone is more likely contamination o f a fracture.  The crystal experienced brittle deformation before  diamond growth was complete.  7.3.5  D i a m o n d 2-5  This diamond is a large ( 7 x 6 x 5  mm) grey fragment that has experienced little  resorption and contains numerous small dark inclusions o f an unknown phase.  No  particular grain orientation was specified for polishing. C L colours are weak to moderate blue, although there are many short lines o f brighter C L centred on these small inclusions (Fig. 7.16). These small bright lines are interpreted as stress fractures around inclusions and should be ignored when interpreting growth zones.  The polished surface is  essentially homogenous blue, with one exception near points 10 and 11 where crystal C L is absent.  The boundary between the C L absent and weak blue C L zones is straight,  Fig. 7.16. Image of CL of diamond 2-5 with IR data for transect 1 to 11. Note the numerous bright blue CL lines (interpreted to be stress fractures associated with inclusions). Also note the region by points 10 and 11 that has weak CL.  107  indicating octahedral growth, with orthogonal bends. It is not clear where the crystal core is i n relation to the polished surface.  Nitrogen concentration variation along the transect ranges from 50 to - 1 0 0 ppm while the aggregation state remains constant at 100% B centres. The nitrogen characters measured at points 10 and 11 in the C L absent zone are essentially the same as for all other points, which likely reflects the thickness o f the polished diamond (1.9 mm) and that the results for each data point are more o f a bulk LR character.  The total relative hydrogen  concentration correlates positively with the nitrogen concentration. A peculiar absorption pattern was observed  near 1048 cm" for most o f the spectra collected (described i n 1  section 6.3.3 and referred to as impurity " X " ) .  Diamond 2-5 is a type IaB crystal which, at least i n part, grew by the mechanism o f octahedral growth.  The absence o f numerous growth bands may be a function o f the  plane o f polishing, or, may possibly be an indicator o f more rapid crystallization.  7.3.6 D i a m o n d 2-8  Diamond 2-8 is a grey fragment with a complex external morphology that was described as 'unknown'. It contains abundant graphite inclusions and was polished i n a random orientation. C L colours for this stone are moderate blue with a thin (less than - 1 0 0 um) coat o f bright green/yellow C L (clearly visible on C L image o f opposite side, inset top left, F i g . 7.17).  There are also several localized zones on bright green/yellow C L but  their association is not clear.  The diamond contains abundant  small inclusions  (interpreted as graphite), one large one o f which is visible below point 5. The strange pale green appearance o f the stone on the top right is an artefact o f the C L machine. The only growth zone visible is the thin coat on the stone.  108  Points  1 mm  Fig. 7.17. Image of CL of diamond 2-8 with IR data for transect 5 to 20. Inset is CL photo of opposite side (scale of inset is ~2.0 mm across).  Because the C L colours are relatively homogeneous across transect 5-20 it is not surprising that there is little variation in nitrogen character. Total nitrogen concentrations vary from 40 to 65 ppm while the aggregation state is 100% B centres.  There is a  positive correlation between relative hydrogen and total nitrogen concentrations. This diamond is type IaB with a thin coat o f bright green/yellow diamond that is too thin to analyse with ER but is likely a late stage growth layer on the stone, made up type I a A B or IaA diamond. The growth habit o f this stone is unclear.  7.3.7 Diamond 2-9  Diamond 2-9 is a large (6.5 x 5 x 2.5 mm) broken stone that is strongly resorbed. It was polished in a random orientation. The stone is greyish i n colour and contained several inclusions, one o f which had an orangey-red halo (later revealed during diamond cracking to be an altered inclusion). A fracture crosses the middle o f the grain and crosses the transect near point 15.  The two sides o f the fracture may have different  origins as fluorescence o f the grain reveals that one half is bright blue while the other half is faintly blue (Fig. 7.18).  C L colours are moderate to bright blue with the brightest  109  2 3 4 5 6 7 8 9 10 1112 1314 15 16 17 18 19 2021 22 23 24  Points  Fig. 7.18. Image of CL of diamond 2-9 with IR data for transect 2 to 24. The bright blue cross-cutting linear features are fractures. One fracture transects the stone completely, crossing at point 15. FL of diamond reveals heterogeneous character of halves bisected by fracture, one half possess bright blue FL while the other does not fluorescence.  sections o f the surface being reflections i n the many fractures.  Deformation laminations  occur as numerous parallel lines, seen as sub-vertical lines on Fig. 7.18. Signs o f plastic deformation were visible on the resorbed surfaces o f the crystal. The relative position o f the crystal core to the polished surface is unclear.  IR data collected across transect 2 to 24 was often o f poor quality and contained an absorption o f unknown origin i n the one-phonon range. The abundant fractures are likely the cause for any poor data and perhaps secondary material i n fractures is responsible for the anomalous one-phonon absorption.  Nitrogen concentration across transect 2 to 24  ranges from ~0 to - 2 0 0 ppm with all nitrogen i n the form o f B centres.  There is a  positive correlation between relative hydrogen and total nitrogen concentrations.  This diamond is type IaB with an unknown growth habit that experienced plastic deformation. It is unclear how large the original crystal was before  110  fragmentation.  7.3.8  D i a m o n d 2-11  This heterogeneous grey/colourless crystal fraction is 5.5 x 5 x 3 m m i n size and has an unidentifiable crystal form.  It was quite angular in surface topography and contained  abundant tiny inclusions which turned out to be sulphides (see section 8.3.1.7). The grain was returned for a second parallel polished surface but was found to be too difficult to polish.  A s such, the grain was irregularly shaped and had only one polished surface,  making it extremely difficult for the collection o f decent LR data.  The abundant  inclusions, the almost opaque dark grey regions, and the grain thickness (~2.7 mm) further complicated matters.  A straight transect across the polished surface was not  possible, and data was collected for points wherever possible. The C L o f this diamond is intriguing, with alternating zones o f blue and yellow C L colour (Fig. 7.19).  LR data reveal an unusual absorption pattern i n the one-phonon range that looked similar in shape to type I a A B diamond. The deconvolution program was able to fit curves to some o f the spectra but the residual curves were large (except i n the case o f point 23, which clearly has both A and B centres). A s such, deconvolution results are not included i n F i g . 7.19. The total nitrogen concentration for point 23 is 72 ppm with an aggregation state o f 60% B centres. Other points appear to contain significantly more nitrogen (e.g. point 19, F i g . 7.19), but the spectra for these points were very noisy. The calculated total relative hydrogen concentrations are included i n F i g . 7.19 as their measurement was not obscured by the noisy one-phonon absorption. Hydrogen concentrations seem to increase from points 22 - 20 (note that points are not i n numerical order), but are relatively low.  The growth patterns observed on this polished diamond are complex.  Numerous  alternating bands o f yellow and turquoise C L , generally with straight bounding faces, are cross-cut by both turquoise and yellow C L bands. The straight nature o f many bounding surfaces suggests that growth was octahedral, but numerous cross-cutting features are more typical o f growth i n the (100) direction. Fracturing and re-growth or annealing are  111  ,  — ' 0.2 0  22  26 4  6  20  19  Points Fig. 7.19. Image of C L of diamond 2-11 with IR data for transect points indicated. Results for the nitrogen character are not included as spectra in the one-phonon range are generally noisy.  112  possible mechanisms for the generation o f such a complex morphology. However, the preferred explanation is that the crystal was initially comprised o f diamond with blue C L that was subsequently plastically deformed creating the bands o f yellow C L , which are crystallographically controlled. from  Davies (1998) observed yellow C L i n several diamonds  Eastern Australia and interpreted  its presence as being a result o f plastic  deformation.  Diamond 2-11 is the only diamond o f the R i o Soriso suite to contain abundant D centres (as calculated by the deconvolution program) and B ' absorption at 1360 cm" (Fig. 7.20). 1  This  may be the only regular (as defined i n section 6.1.2) diamond i n this study. However, there is  4000  3000 2000 Wavenumbers (cm ) Fig. 7.20. IR spectrum from point 19 showing large B' peak (Diamond 2-11). Total nitrogen concentration calculated at -500 ppm with -90% B centres. D centre concentration -1.6 a.u. 1  some uncertainty as to the cause o f absorption at 1360 cm" because o f the abundant noise. In 1  all likelihood abundant D centres (platelets) are present i n this diamond (as oppose to the rest o f  the diamonds in this suite that have experienced platelet degradation) making this stone unique.  The presence o f A centres indicates that at least part o f this diamond was not  subjected to the extreme temperatures and/or lengths o f time that are typical for most diamonds o f this suite.  A t least part o f this diamond is type I a A B (~70 ppm N o f which 60% are B centres) with portions that likely contain considerably more nitrogen (i.e. point 19). The reason for the different C L colours is unclear.  They may reflect different chemistries (i.e. the yellow  C L zones may contain nitrogen o f mixed character (type IaAB) while the blue zones may be more highly aggregated (closer to type IaB)) or they may reflect localized plastic deformation.  This crystal has experienced some growth on octahedral faces, but the  details o f its history are certainly more complex.  113  7.3.9 D i a m o n d 3-1  This diamond is the largest stone for  study,  weighing  0.404  grams. It is fully intact, strongly resorbed, exhibits a non-uniform colourless/brown shows  hue,  signs  of  and plastic  deformation. T w o surfaces were polished approximately parallel to the (100) face, making a plate -3.4 m m i n thickness.  It was  possible to orient this diamond for polishing even though  the  grain was so rounded because o f the etching pits visible on the crystal surface. has  The diamond  moderate blue  7.21) , Soriso  typical  of  stones.  growth  CL . (Fig. many  Rio  Oscillatory  bands  are  remarkable,  quite  indicating  fluctuations i n conditions during growth.  The  heterogeneous  colour (seen when view under an  optical  7.22) )  is  microscope caused  by  (Fig.  3  and abundant microinclusions in  5  6  7  8  9  10  II  12  Points  both  secondary material i n fractures  4  Fig. 7.21. Image of CL of diamond 3-1 side A with IR data for transect 3 to 12. Dark circle to left of point 2 is a large inclusion pit.  114  the centre of the crystal.  The cloudy zone with abundant micro-inclusions is also the  region of brightest blue C L (points 2 to 6). The morphology of the core suggests that this grain is twinned. The boundary between the core and the next growth zone (points 1012) is rounded and may be a function of growth or dissolution. After this layer, it appears that growth of the remaining layers continued without disruption, until the final dissolution event responsible for the external morphology occurred. The final growth layer (top left to bottom left) forms on octahedral faces.  LR data (points 3 to 12, Fig. 7.21) reveals that the diamond contains low amounts of nitrogen (<50 ppm). With the exception of point 7, the nitrogen concentration in the core is ~50 ppm and decreases in the next layer, to essentially type LI diamond. Unfortunately, LR spectra could not be collected for the brighter C L layer seen on the left of the image. Any nitrogen detected occurs as 100 % B centres. The relative hydrogen concentration roughly mimics the total nitrogen concentration.  Side B (Fig. 7.22) also illustrates the fine oscillatory nature of this diamond: the moderate blue C L core (points 5 and ~4), the darker C L zone (~3 and ~4), and the well formed, ' octahedral outer zone of variable blue C L (points 6-9). The form of the core and second darker zone are both rounded, likely as a result of dissolution. The nature of growth of these zones is unclear. The final growth zone (points 6-9) is octahedral.  The LR transect, in reverse (points 9 to 2), reveals that the nitrogen concentration increases from rim (type LI diamond) to core (-60 ppm) and that the aggregation state is 100% B centres.  This diamond is both type IaB (core) and type LI (rim), with any nitrogen in the form of 100% B centres. The nature of growth during the early stages of crystallization are unclear; the diamond may have started off as a twin, may have experienced alternating periods of growth and resorption, or may have grown rounded faces. However, the latter  115  Fig. 7.22. Image of C L of diamond 3-1, side B, with IR data for transect 2 to 9 with various photos. Clockwise, from top left, the images are: CL photograph of side B with IR analysis points (note - point 2 is covered with carbon putty in this image); CL photo of side A with IR points for both sides indicated; photo of side B with IR points; photo of polished rough diamond with IR points; and IR data for points 2 to 9 on side B.  116  stages o f growth developed along (111) faces.  Finally, the diamond experienced  extensive resorption, producing a well rounded external morphology.  The crystal  experienced deformation during residence i n the mantle, although evidence o f this is not obvious i n C L .  7.3.10 D i a m o n d 3-5  Diamond 3-5 is a moderately well resorbed (class 3), faintly yellow, broken stone. T w o parallel surfaces were polished parallel to the (100) surface, creating a plate - 3 . 6 m m thick.  The C L colours observed are moderately intense blue, typical to most stones i n  this study (Fig. 7.23). The growth morphology recorded under C L is remarkable, and is comprised o f essentially three zones: the core, a brighter blue C L zone o f possibly two crystals (an aggregate or twin), which has likely experienced resorption (points 5-7 and 13-16); a less intense blue C L zone o f octahedral growth surrounds the core (points 1-4, 8-10 and 17-19); and an outer zone o f weak blue C L with few discernible growth bands (points 11-12 and 20-21). There may also be a r i m o f even darker blue C L on this zone (seen on bottom right o f image). T w o black points are visible i n the crystal core (near points 5 and 16), which turned out to be inclusions o f ferropericlase (section 8.3.1.1). They are not at the genetic centre, and so are not likely seed crystals, but they were undoubtedly trapped during the early stages o f diamond growth. The second growth zone (octahedral layers) warrants further discussion.  The zone appears to have symmetry  (halves are roughly mirrored on a plan draw through points 6-15) which resembles that o f a m a d e twin.  During data collection, I R light most certainly penetrated more than one growth layer due to diamond thickness, therefore, any results likely do not reflect the absolute IR character o f the growth layers observed i n F i g . 7.23, but give a good indication o f the relative fR changes from core to rim. Both transects (points 1-12 and 13-21) show that total nitrogen concentration is greatest i n the core (highest total at point 15 at - 4 0 ppm) and decreases  117  1  2  3  4  5  6 7  8  1  III  II  12  Points  13  14  15  l<)  17  IX  19  21)  21  Points  Fig. 7.23. Image of CL of diamond 3-5 with IR data for transects 1-12 and 13-21. This diamond exhibits weak to moderate blue CL. Growth can be divided into at least three zones: 1) the bright core (highest nitrogen totals); intermediate blue, octahedral growth, outer zone (points 1-4, 8-10 and 17-19) with a morphology that suggests twinning; and the darker blue rim with few discernible features.  118  towards the r i m where the diamond is type U .  The aggregation state calculated is  unrealistic (-150% B centres) because LR spectra for most points was unusual i n that there was no absorption at -1280 cm" (where the prominent absorption for A centres 1  occurs).  The negative A centre totals are certainly meaningless (although they may  indicate the presence o f an unknown impurity). Although many points were noisy i n the one-phonon region, it was still possible to measure the hydrogen peak at 3107 cm" for all 1  spectra collected. This data shows a decrease i n relative hydrogen concentration from core to r i m i n transect 1-12, however, this pattern is not mimicked i n the transect 13-21.  This diamond contains very little nitrogen, although the core is type IaB (-40 ppm) with subsequent growth containing little to no nitrogen (type LL). A n y nitrogen present occurs as 100 % B centres. The growth occurred i n at least 3, and potentially 4 stages. The first two zones possibly grew as twins, the second along octahedral faces.  The nature o f  growth o f the third (and potential fourth) zone is unclear. Finally, the crystal experienced resorption (class 3).  7.3.11 D i a m o n d 3-8  This brown/colourless stone was polished on one side parallel to the (100) plane. The C L colour o f the stone is moderate turquoise (Fig. 7.24), although the image has been brightened for clarity.  N o growth patterns are observed.  The diamond contains  numerous small, dark inclusions, which i n some instances, are highlighted by lines o f brighter C L (likely due to internal stress around the inclusions).  LR data shows that total nitrogen concentration decreases slightly from left to right (-50 to - 2 0 (9-10) and the aggregation state is 100% B centres.  The total relative hydrogen  concentrations decreases on either side o f point 8, but are generally low.  119  Fig. 7.24. Image of CL of diamond 3-8 with IR data for transect 2 to 12. Diamond exhibits moderate blue CL with no visible growth structures. Bright CL lines form around the abundant mineral inclusions found in this diamond.  This diamond is mostly type IaB (although two points are type IT) with 100% B centres. The absence o f visible growth structures may be a result o f the orientation o f polishing (the polished surface does not intersect any growth boundaries) or may indicate that this diamond grew i n one episode o f crystallization from a medium o f homogeneous composition.  7.3.12  Diamond  3-10  This strongly resorbed, colourless diamond fragment was polished i n no particular orientation.  One half o f the stone exhibits moderate blue C L while C L is essentially  absent i n the other half (Fig. 7.25). Bright green/yellow C L is visible i n several places, most notably around a large inclusion pit (below points 3 and 4). The boundary between the two halves is relatively straight and likely a paleo-octahedral face.  IR data across transect 1 to 14 shows that nitrogen concentrations vary from 75-140 ppm for points 1 to 10 and become progressively nitrogen free from points 11 to 14 (IR  120  1 2 3  4  5 6 7 8 9 10 11 12 13 14 Points  Fig. 7 . 2 5 . Image of C L of diamond 3-10 with IR data for transect 1-14. Diamond exhibits two zones, one of moderate blue C L on the left while the right half has weak to absent C L . Yellow/green C L is visible around a large inclusions pit and is likely a result of lattice strain.  spectra for points 13 and 14 show no visible nitrogen absorption). A n y existing nitrogen occurs as B centres. Relative hydrogen concentrations show a similar pattern to nitrogen, varying across points 1 to 10 and then progressively decreasing towards point 14. There are two unusual absorption patterns observed in LR spectra for most points across the transect. A sharp peak at - 1 4 3 0 cm"' is visible i n spectra collected from points 4 to 11. A broader peak in the one-phonon range centred on -1048 c m ' (impurity " X " ) is present 1  in spectra for points 12 to 14. The cause o f this absorption remains unknown (see section 6.3.3).  This diamond is mostly type IaB (-60-80 ppm nitrogen, all as B centres) but with a significant type LI portion.  A s there is no visible core, it is unclear i f nitrogen  concentration decreases from core to rim, as is typical o f most stones i n this study. The growth o f the nitrogen rich zone is likely octahedral.  The conspicuous association  between yellow/green C L and the inclusion pit suggest a genetic link  121  7.3.13 D i a m o n d 3-11  Diamond 3-11 is a w e l l resorbed, broken, grey stone with a frosted surface.  It was not  polished on any particular orientation. The C L pattern o f this polished surface shows bizarre patchy bright blue C L zones on a more homogeneous blue background o f moderate C L intensity (Fig. 7.26). The nature o f these patchy bright blue C L zones i n unclear. 'Fingers' o f weak yellow/green C L can be seen below points 9 to 13 (Fig. 7.26) and are an artefact o f the machine (these 'fingers' were also observed on diamond 2.8). Growth features on this polished surface are not observed.  Transect data for points 2 to 3 reveals a relatively homogeneous distribution (with the exception o f point 11) o f IR character.  Nitrogen concentration is - 6 0 ppm (all as B  centres) and relative hydrogen concentrations are roughly similar at - 0 . 5 absorption units.  This diamond is type IaB (-60 ppm N ) , with moderate blue C L , and with no obvious growth features on the polished surface.  2  3  4  5  6  7  8  9  10  II  12  13  Points  Fig. 7.26. Image of C L of diamond 3-11 with IR data for transect 2 to 13. C L of this diamond is essentially moderate blue (the heterogeneities around points 9 to 13 are an artefact of the C L machine).  122  7.3.14 D i a m o n d 4-17  This colourless diamond fragment is the only stone that was not polished for growth studies. The fracture surface that was examined for growth morphology was reasonably flat and smooth, although a few steps i n the surface are visible i n the image (particularly below the transect).  However, the C L image turned out remarkably well for an  unpolished surface (Fig. 7.27). Most o f the stone has weak blue C L but a r i m on one half o f this fractured diamond exhibits bright yellow/green C L . Three sets o f parallel lines (plastic deformation laminations) are visible i n the outer growth zone.  The boundary  between the two zones is remarkably sharp and indicates octahedral growth.  The diamond is ~1.5 m m and the IR path penetrates zones o f green C L for all data points (examination o f diamond reveals yellow/green C L on most sides, see Appendix A , diamond 4.17). Data from point A likely does not include contamination from the weak blue C L zone. Nitrogen concentration o f the yellow C L zone ranges from -20-50 ppm, possibly decreasing towards the inferred core (within the weak blue C L zone) with an aggregation state o f 50% B centres. The weak blue zone is essentially type n. Hydrogen concentrations are also somewhat unusual for this study i n that the highest totals do not coincide with the highest total nitrogen concentrations (point A ) .  This diamond has an inner zone o f weak blue C L o f type II diamond and a bright yellow/green C L r i m o f type I a A B (-20-50 ppm total nitrogen). The yellow/green r i m grew, on what was likely a flat-faced type II octahedron, by octahedral growth. Subsequent to growth the diamond experienced plastic deformation.  123  Fig. 7.27. Image of C L of diamond 4-17 with IR data for transect A, 1-7. This diamond exhibits a weak blue C L zone with a bright yellow/green C L rim with extensive deformation laminations (numerous lines that are parallel to the boundary between the blue and yellow/green CL).  124  7.4 Discussion  7.4.1 Summary of growth studies of Rio Soriso diamonds  Examination o f the photos i n section 7.3 shows that there are a variety o f internal structures, some that are relatively simple, others that are extremely complex, and others that exhibit no apparent internal structure. A summary o f the results is presented i n Table 7.1.  A s corroborated by the bulk LR studies on whole diamonds (section 6.3.1), nitrogen concentrations are general low (<100 ppm) and nitrogen occurs mostly as B centres. Diamonds 2-11 and 4-17 are the lone exceptions as they both also contain nitrogen i n the form o f A centres (diamond 2-8 may also be included i n this category because o f the bright yellow/green rim).  For diamonds with visible concentric internal  nitrogen concentrations decrease from core to rim.  structures,  Relative hydrogen concentrations  mimic the relative total nitrogen concentrations, as discussed i n section 6.4.3. However, the positive correlation between nitrogen and hydrogen concentrations may be only with nitrogen in the form o f B centres as diamond 4-17, with A centres, does not show a correlation.  The more intense C L visible in the core regions o f several diamonds is a reflection o f the higher  impurity concentrations,  which are  visible i n transmitted  light using  a  conventional microscope (e.g. the heterogeneous brown appearance o f the core i n diamond 1-4 and the cloudy appearance o f the cores o f diamonds 2-1 and 3-1). nature o f growth in the core is unclear as growth features are not visible.  The  Subsequent  growth on the core is often along paleo-octahedral faces (e.g. diamonds 1-2, 3-1 and 3-5) but in most cases growth is unrecognisable. Episodic resorption and growth are visible i n several crystals (e.g. diamonds 1-4, 3-1 and 3-5) and most diamonds have experienced an episode o f resoption before exhumation.  125  T w o diamonds have a yellow C L r i m  Table 7.1. Surnrnary of results for growth studies of Rio Soriso diamonds Range in total nitrogen (ppm)  Range in B/(B+A) (%)  Type(s) of growth  Direction of nitrogen cone, decrease  20-400  100  sector and agate  ?  0-225  100  complex (sector?)  ?  72  60  octahedral arid complex  ?  0-52  50-100  octahedral rim to core  mod. to weak blue  0-60  100  octahedral core to rim  3.6  mod. to weak blue  0-75  100  octahedral  core to rim  random  2.6  mod. blue to none  0-80  100  octahedral and agate  core to rim  1  random  1.9  mod. to weak blue  50-100  100  octahedral and uncertain  ?  2.1  2  random  2.7  mod. to weak blue  0-20  100  uncertain  core to rim  2.8  2  random  2.8  mod. blue with yellow rim  40-65  100  uncertain  ?  3.8  1  100  1.6  mod. turquoise  0-60  100  uncertain  ?  3.11  1  random  2.5  mod. blue  0-80  100  uncertain  ?  2.9  1  random  2.2  strong to mod. blue  0-200  100  uncertain  ?  3.10  1  random  1.1  strong blue  0-150  100  uncertain  ?  Sample No.  No.of sides cut  1.4  2  random  1.3  2.2  1  random  2.4  2.11  2  random  2.7  4.17  0  random  1.5  3.1  2  100  3.4  3.5  2  100  1.2  2  2.5  Orientation Thickness CL colours (mm) of polishing strong turquoise to none strong turquoise to none mod. turquoise and yellow weak blue with yellow rim  Diamond thickness was calculated using the absorption coefficient calculated by Taylor et al, (1990), measured at 2030 cm" (0.47 ± 0.01 units per mm path length). 1  126  (diamonds 2-8 and 4-17), likely o f type IaA-IaAB character.  Plastic deformation is  observed i n some diamonds (e.g. diamonds 1-4, 2-9 and 2-11) and brittle deformation is evident in one crystal (diamond 2-2).  Based on the nitrogen concentration, aggregation state and internal morphology, the 14 diamonds are divided into the following subgroups:  1) Strong turquoise C L with a significant episode o f combined cubo-octahedral growth.  Nitrogen concentrations have a large range (0 to -225 ppm) and  occur only as B centres. Diamond 1-4 (and possibly diamond 2-2) belongs to this subgroup. 2) Alternating bright turquoise and yellow/green C L zones with a complex geometry. Nitrogen concentration likely has a large range (72 to - 5 0 0 ? ppm) and occurs as type I a A B diamond with - 6 0 - 9 0 % B centres. Diamond 2-11 is lone crystal to fall into this category. 3) Diamonds with moderate to weak blue C L that exhibit alternating periods o f growth and resorption. Nitrogen concentrations are generally l o w (0 to - 8 0 ppm), occurring only as B centres, and tend to decrease towards the crystal rim. The majority o f the growth for these diamonds ( i f not all growth) was on octahedral faces. The diamonds that belong to this subgroup are 3-1, 3-5 and possibly 1-2. 4) Weak blue C L (type 13 diamond) with a reasonably thick rim (up to 250 um) o f bright yellow/green C L (-50 ppm total nitrogen o f which - 5 0 % occur as B centres, or type I a A B diamond). Diamond 4-17 is the only diamond i n this subgroup. 5) Diamonds that do not appear to have any internal structures that exhibit moderate to strong blue/turquoise C L . Nitrogen concentrations range from 0 to 150 ppm and occur only as B centres (diamonds 2-1, 2-5, 2-8, 2-9, 3-8 and 3-10).  127  Subgroups 1-3 are likely distinct populations. internal structures and nitrogen character.  They are quite different in terms o f  Subgroup 4 is created for the one diamond  with a reasonably thick bright yellow/green C L r i m (diamond 4-17). R i m s o f this C L colour exist on other diamonds and perhaps are not observed because they have broken off the diamond core o f type Ha or IaB character. Subgroup 5 likely contains diamonds with internal structures that are not visible on account o f the orientation o f the polished face.  It is likely that at least some diamonds i n subgroup 5 could be redistributed  between subgroups 1 to 3 (most likely belonging to subgroup 3) i f the polished surface was on a more favourable orientation.  7.4.2 Comparisons with other studies  The internal growth morphology and IR character for Juina area diamonds have also been examined by Hutchison (1997) and Araujo et al. (2003).  Hutchison (1997) found that most diamonds exhibit weak to moderate blue C L , lack concentric growth features and display internal structures that are truncated at crystal edges, an indication that many stones are broken. The conclusion was that initial growth for at least some stones (as evidenced by step features) was within a reasonably stable growth environment. F o l l o w i n g initial growth, all stones were subjected to a period o f dissolution followed by precipitation o f diamond o f lower C L intensity. M a n y o f these stones also exhibit signs o f plastic deformation, which may be responsible for the realignment o f concentric growth features.  The results from work by Araujo et al. (2003) are similar. C L o f diamonds is typically homogeneous sky-blue o f weak intensity.  The rare internal features observed include  growth on octahedral faces, step-wise growth, truncated growth zones and fine parallel lines. These observations were interpreted as indicators o f octahedral growth, resorption and plastic deformation. Nitrogen concentrations are low (82% o f diamonds classified as  128  type LT), but i f detected, occur mostly as B centres (>90% B centres). Three stones out o f a sample size o f 234 diamonds were classified as type I a A B diamond with <90% B centres (bulk analysis). There is no data on the internal morphology o f the type I a A B diamonds.  The majority o f diamonds found i n the Juina area appear to contain internal morphologies that are either complex or absent, exhibit weak to moderate blue C L and typically show evidence for plastic deformation, brittle fracturing and episodes o f resorption and reprecipitation. Total nitrogen concentrations are low and occur mostly as B centres (90 to 100% B centres), and in some cases, decreases towards the crystal rim.  Other  subgroups are likely mixed i n with the dominant subgroup just described, however, they are poorly represented. Based on internal growth features and LR character, this study highlights two new subgroups: 1) diamonds with high total nitrogen concentrations that grew, at least i n part, as cubo-octahedrons (diamond 1-4 and possibly diamond 2-2), and 2) diamonds with complex internal morphology o f alternating turquoise and yellow/green C L that may contain considerable nitrogen (72 — 5 0 0 ppm) o f mixed (type IaAB) character and may also contain platelets (D centres) (diamond 2-11).  129  8.0 Mineral Inclusions 8.1 Introduction  M a n y diamonds contain mineral inclusions that are syngenetic with their host (Meyer, 1987; Harris, 1992). These inclusions have been trapped by their hosting diamond and have remained in isolation because diamond acts as an impermeable seal around the inclusion, thus preserving a pristine sample o f the mantle. A n extensive list o f minerals which occur as inclusions i n diamond is given by Meyer (1987) and Gurney (1989). From this large group o f diamond inclusion minerals, only a few have been found i n diamonds from the Juina area,  inclusions found i n Juina diamonds are also found i n  diamonds from most continents, but are generally rare occurrences. The Juina area is unique because it has the largest proportion o f diamonds which contain these rare inclusions.  Numerous studies have confirmed that mineral inclusions i n diamond (and thus diamonds themselves) fall into two broad categories in terms o f paragenesis: peridotitic (p-type) and eclogitic (e^type). Inclusions belonging to other paragenetic groups have been found but they are exceedingly rare.  M a n y o f the inclusions found i n Juina diamonds are  considered to represent their own paragenetic group, which has been referred to as both the lower mantle suite (Hutchison, 1997) and the super-deep suite (Kaminsky et al., 2001a).  These terms do not indicate the nature o f the composition but highlight the  extreme depths o f formation.  A distinction should be made when referring to  compositional similarities (i.e. p-type or e-type) and depth o f formation. When possible, the minerals and mineral associations i n this study w i l l be discussed both in terms o f composition (e.g. eclogitic, peridotitic, mafic and ultramafic) and depth o f formation.  A brief review follows on the current models o f the mantle i n terms o f composition, phase changes and pressure-temperature (P-T) constraints.  130  8.1.1 C o m p o s i t i o n o f the mantle  Constraints on models o f the mantle  6 ' 1  come from a variety sources which can be  broadly  sub-divided  into  'categories':  1) methods  o f indirect  examination  (e.g.  three  seismology  Velocity (km/s)  Crust  Mantle  S waves \  \ sphere  .2000 P waves  \  and Depth (km)  interior  (e.g.  and  3)  \  Liquid  -4000  ultramafic -5000  massifs, basalts, xenoliths and diamond inclusions)  \  -3000  materials inferred to be derived from Earth's  1 1 1 1Asthenosphere  -1000  tomography), 2) direct examination on  the  Lithosphere  T3T—1—1——1  examination  Inner Core  of  analogue materials (e.g. lab experiments at high pressures and temperatures on relevant materials, and examination o f  Solid  S waves -6000  Fig. 8.1. Seismic velocities for P and S waves through the Earth (0-6370km). Compositional subdivisions are listed on the left and rheological subdivisions on the right. Reproduced from Winter (2001) after Kearsey and Vine (1990).  non-terrestrial materials).  The first data from the 'deep Earth' came  from  seismic  upper \ , mantle m  studies.  Recognition o f sharp changes i n wave velocities (and in some cases absences) led  to  the  establishment  subdivisions  in the  o f several  Earth's  which when discussed  in a  interior, general  sense, are widely accepted in literature. These  subdivisions are crust,  mantle  (which is subdivided into upper mantle, transition  zone,  and  lower  mantle),  outer core, and inner core (Fig. 8.1). O f  6  "8  9  10  11  Seismic velocity (km/s) Fig. 8.2. Seismic velocities for P and S waves through the Earth (200-800km). Line patterns represent interpretations from different sources. Vs = shear waves, Vp = compressional waves. Modified from Ringwood, 1991.  most interest to this study is the mantle, and in particular, the transition zone (-410-660 km) and lower mantle (-660-2900 km). A more detailed figure o f seismic wave velocity  131  variation with depth (between 200 and 800 km) is presented i n Fig. 8.2. Seismic profile interpretations generally agree on the depth o f the two large discontinuities (-390-420 k m and -650-700 km) and on a less pronounced discontinuity (-500-530 km). For the sake o f any future discussion and simplicity i n figures, this thesis w i l l consider the upper boundaries o f the two major discontinuities at 410 k m (transition zone) and 660 k m (lower mantle) (Ringwood, 1991).  Direct evidence o f mantle minerals comes from a variety o f sources found near the Earth's surface, such as ultramafic massifs, xenoliths, basalts and inclusions recovered from diamond. However, there are significant limitations on the interpretations from the studies o f these materials.  It is important to understand that some materials may be  products o f partial melting (e.g. basalts) or may not be representative o f the bulk mantle (e.g. xenoliths and diamond inclusions). It is unclear i f diamonds crystallise i n a setting that is chemically distinct and isolated from the bulk mantle, or i f diamonds crystallise i n rocks that reflect the bulk chemistry o f the mantle.  A n y study o f diamond inclusions  must consider the possibility that the inclusions i n diamond may be sourced from parental rocks that are minor i n volume relative to the whole mantle.  The integration o f data from geophysics, natural Earth samples and chondrites along with experiments constraints.  at high pressure and temperature have provided several  consistent  Perhaps most notable was the recognition that the depths at which some  large seismic discontinuities occur (-410 and - 6 6 0 km) correspond well with a phase change i n olivine ((Mg,Fe)2Si0 ) at - 4 1 0 k m , and the breakdown o f ( M g , F e ) S i 0 into 4  2  ferropericlase (fPer) and MgSiOyperovskite (-660 km).  4  Based on this observation  among other experimental work, A . E . Ringwood synthesised an analogue material termed 'pyrolite' (composed o f dominantly pyroxene-olivine material) to represent the primary composition o f the mantle.  F i g . 8.3A is a compilation o f high pressure and  temperature data for material o f 'pyrolitic' composition from Ringwood (1991).  This  figure illustrates very succinctly the range o f stability for most mantle phases. Although there is no consensus on the composition o f the mantle, the hypothetical mixture o f pyrolite is widely accepted (Mfune et al., 1998). F i g . 8.3B shows the range o f stability  132  geothermobarometers.  The geothermal gradient below cratonic lithosphere, within the  convecting mantle, is constrained by the adiabat, which is the theoretical geothermal gradient for a system where no heat is lost through conduction. Although heat is lost through conduction, the geothermal gradient must be close to the adiabat i n the convecting mantle. The geotherm and adiabat included i n Figs. 8.4 to 8.7 is an estimated P-T path through the mantle (reproduced from Joswig et al, 1999). There is a large error associated with this path (-200 °C), but i n the context o f this study, the inclusion o f the geotherm is considered useful.  8.1.3 Terminology*  Mineral terminology is an important matter o f discussion because o f the numerous polymorphs for the expected dominant mantle minerals. Nomenclature is often confusing in literature and warrants explanation here.  Firstly, the minerals introduced i n this  chapter contain iron (Fe) i n their chemical formula, although abbreviations tend to drop Fe and only use M g .  For (Mg,Fe)Si03-perovskite,  mg (where mg is defined as  M g / ( M g + F e t ) ) is generally >0.90 and for the sake o f simplicity, the mineral name is 2+  2+  to  abbreviated to M g S i - P r v . Contractions o f this nature are common in literature and w i l l be used i n this thesis.  A l s o worthy o f explanation are terms such as perovskite and  ilmenite used to describe certain isomorphs.  These terms indicate that the mineral i n  question has a structure similar to a well-characterised mineral (e.g. CaSiCVperovskite has the same structure as CaTiC>3 (true perovskite), which is orthorhombic, and M g S i C V ilmenite has the same structure as FeTi03 (true ilmentite)).  Tables are included for each mineral discussed i n section 8.1.4, relating the full mineral name with abbreviations used. When possible, abbreviations are used for minerals from Kretz (1983). However, many o f the higher P-T phases do not have widely accepted abbreviations. Abbreviations for these minerals are collected from a variety o f sources.  134  for material o f initial basaltic composition. In a broad sense, the pyrolite model has withstood many tests, and as such, has widespread appeal. B) MORB  A) Pyrolite  3.50 S o  3.73  a  3.90 4.37  4.16 0.2 0.4 0.6 0.8 Volume fraction  0.2 0.4 0.6 0.8 Volume fraction  1.  Fig. 8.3. Mineral assemblages and (zero-pressure) densities for A) pyrolite, and B) basaltic (MORB) oceanic crust. Constraints used for temperature are T=1400°C at 400 km and 1600°C at 650 km, from mantle geotherm in Brown and Shankland (1981). Modified from Ringwood, 1991.  8.1.2 Geothermal gradient  The geothermal gradient (geotherm) is important to consider i n this study because it provides a possible P - T path that diamonds and their inclusions could follow during exhumation. The number o f likely phases represented by a particular composition (see Figs. 8.4-8.7) can be reduced based on an understanding o f the geotherm. Geotherms are based  on a variety o f lines o f evidence  such as  geophysical observation  and  experimentation as w e l l as high pressure and temperature experimentation on analogue materials. Experiments on the partitioning o f elements between phases have led to the development o f many geothermometers and geobarometers (e.g. Brey and Kohler, 1990), which can be used on coexisting phases found i n xenoliths (e.g. B o y d , 1987) and inclusion pairs i n diamond (e.g. Harris, 1992).  Using the methods listed above, it is  possible to constrain the geotherm i n cratonic lithosphere up to 1400 °C and 60 kbar (Brey and Kohler, 1990), or - 2 0 0 k m . However, the predicted P - T conditions where most o f the Juina area diamonds originate are well outside o f the experimental limits o f  133  8.1.4 Relevant mantle minerals and their stability fields  The minerals considered relevant to this study are essentially the phases depicted i n F i g . 8.3A-B. They are: (Ca,Mg,Fe)Si03 (perovskite and polymorphs), (Mg,Fe) Si04 (olivine 2  and polymorphs), X3Y2Si30i2 (various garnets, where X = M g , Fe and C a and Y=A1 and Cr), (Mg,Fe)0 (ferropericlase) and SiCh (quartz and polymorphs).  Some other phases  found as inclusions i n Juina diamonds are discussed, but most o f these are rare diamond inclusions and likely represent minor mantle phases. In general, the stability fields for these less common phases are poorly constrained.  t h e most compelling evidence for determining what particular polymorph phase is being represented by a diamond inclusion comes from a combination o f chemical and structural analysis.  Chemical studies  crystallographic studies are not.  are  common  practice  on  diamond  inclusions,  but  The reason for this is that inclusions are typically too  small for most analytical equipment used to determine crystal structure, and, even i f analysis is possible, there is a good chance that the inclusion examined has reverted to a more stable, lower P - T , polymorph. Diamond itself is proof that meta-stable minerals can exist at the surface o f the Earth, however, the activation energy required to convert diamond to graphite is extremely high.  Davies and Evans (1972) determined the  activation energy required to induce graphitisation on the (110) surface o f diamond is 728±50 kJ/mol. Many o f the high P - T polymorphs o f inclusions found i n diamond have much lower activation energies.  Without crystallographic data, two main lines o f evidence are used to determine the identity o f the original polymorph before being included i n diamond. The first approach is through study o f associations, as either composite grains (touching phases) or nontouching phases occurring i n the same diamond (that are interpreted as being i n equilibrium when included i n diamond).  The second approach relies on the controls  pressure and temperature has on element partitioning. Because the chemical signature for a monomineralic inclusion survives through time, regardless o f the P - T conditions  135  (provided the inclusion remains isolated inside the diamond), chemical studies are not subjected to the same ambiguities as crystallographic studies.  A comprehensive overview o f the crystal structure, effects o f element substitution and expansion rates is well beyond the scope o f this thesis. Instead, information relevant to deciphering clues from the study o f diamond mineral inclusions for the R i o Soriso suite w i l l be the focus o f the remainder o f this section. Inclusion polymorphs along with their stability fields w i l l be discussed with a particular focus on how to recognise relevant polymorphs based on chemistry alone (i.e., i n the absence o f data on crystal structure). When possible, the inclusion phase and chemistry w i l l be discussed i n terms o f what it reveals about the composition and the P-T stability field o f the parental source rocks.  8.1.4.1 MgSi0  Seven  3  polymorphs  with  MgSiCb  Table  8.1. MgSi0 polymorphs 3  Mineral  Abbreviation  MgSi-perovskite  MgSi-Prv  MgSi-ilmenite  MgSi-Ilm  MgSi-tetragonal garnet  MgSi-TGar  Enstatite  En  High-temperature clinoenstatite  HCen  Low-temperature clinoenstatite  LCen  Protoenstatite  Pen  stoichiometry exist i n the P - T range o f the upper and lower mantle (Table 8.1).  The  stability fields for these polymorphs are presented i n F i g . 8.4.  The polymorphs o f  most interest to this study are M g S i - P r v and enstatite.  However, it  is  noted  that  throughout this report, enstatite inclusions in  diamond  will  be  referred  to  as  orthopyroxene (Opx), to account for the substitution o f Fe.  M g S i - P r v is generally accepted as the dominant phase found in the lower mantle (660 to 2900 k m ' s depth) (Fiquet et al., 1998) and constitutes - 7 0 % o f a lower mantle o f pyrolitic composition (Fig. 8.3A). It is considered to be a highly unstable phase outside of  the lower mantle.  Studies on the activation energy required for the back-  transformation o f M g S i - P r v (mg = 0.90) to enstatite by Knittle and Jeanloz (1987) find that only 70±20 kj/mol are needed, compared to 728±50 kJ/mol required to convert  136  Temperature (°C)  diamond to graphite (Davies and Evans, 0  1972).  Knittle  and Jeanloz  400  800  1200  1600  (1987)  \  low-tempenmlrtti^.  suggest that M g S i - P r v brought to the Earth's  surface  clinoenstatite  2000  X^J^Q/it/  100 eostatite  would likely survive Other authors,  \  ,. . . liquid  \  \  high-temperature  less than 3-100 years.  2400  protocnstatite MgSiOj  200  clinoenstatite 300  such as Sharp et al. (1997), Wang et al. 400  (1992) and M c C a m m o n et al. (1992) have  drawn  Kesson  et  similar al.  (1991)  ^ .  report  the  Prv to E n (as an inclusion in diamond) accompanied by an expansion o f  -20%.  Because o f the relative ease at  which M g S i - P r v reverts to lower P - T phases  and  the  substantial  volume  increase, it is no surprise that M g S i - P r v  1 1  ringwoodiie + stishovite  conclusions. -  -lower mantle  retrogressive transformation o f M g S i -  is  transition zone  <U 16  \ \  tetragonal garnet  ilmcnitc MgSiOj  500  6O0  1 1  perovskite MgSiOj  700  Fig. 8.4. Phase transformations for MgSi0 . Stability fields for liquid, Pen, MgSi-TGrt, MgSi-Ilm, MgSi-Prv, Sti and wadsleyite (p-Ol) and Sti and ringwoodite (y-Ol) are from Fei and Bertka (1999). Stability fields for LCen, En and HCen are from Ulmer and Stadler (2001). Horizontal lines at -14.5 GPa and 23.5 GPa mark the approximate upper limits of the transition zone and lower mantle respectively and the thick bent curve indicates an approximate geotherm (from Jos wig et al., 1999) 3  has not been directly confirmed through crystallographic studies as an inclusion i n diamond.  M g S i - P r v diamond inclusions may be distinguished from Opx inclusions based on A l and N i content. A t shallow levels (~<250 km) garnet is the main host for A l and examination o f Opx inclusions from diamonds o f shallow origin find A l contents <1.00 wt% AI2O3 (Meyer, 1987). However, at depth (-600 k m , see F i g . 8.3) garnet begins to dissolve into M g S i - P r v and CaSi-Prv.  Experiments at high P - T conditions find that M g S i - P r v can  accommodate the complete budget o f AI2O3 predicted in the lower mantle for a pyrolitic composition (-4 mole %) (Kesson et al., 1995; Irifune et al., 1996) and can accommodate as much as - 2 5 m o l % AI2O3 at pressures between 55-70 G P a (Kesson et al., 1995). Inferred former M g S i - P r v inclusions from Kankan, Guinea contain 1.1-1.7 wt% AI2O3 (Stachel et al, 2000b) and inclusions from Juina contain even higher A l contents, up to 12.6 wt% AI2O3 (Hutchison, 1997). Lower A 1 0 contents (-1.1-1.7 wt%) are thought to 2  3  suggest crystallization within the top 10-20 k m o f the lower mantle (Stachel et  137  al,  2000b). A l content i n M g S i 0 (~>1.00 wt% A 1 0 ) is not an indicator o f former M g S i 3  2  3  Prv on its own, indeed A l contents can be considerable for M g S i C h i n both spinel-facies peridotites (1-6 wt% AI2O3) and garnet-facies peridotites (<2 wt% AI2O3). However, low A l content is a signature o f depleted harzburgites, which are the parental rocks for most peridotitic diamonds (Pearson et al, i n print). A s an inclusion i n diamond with AI2O3 >~1.00 wt% is reasonable grounds to suggest a deep origin.  N i O content is another  useful discriminating tool because i n a system o f M g S i - P r v and fPer, N i O is always strongly partitioned into fPer (Kesson et al., 1991). Upper mantle Opx grains typically contain >0.1 N i O wt% (Meyer, 1987) while inferred former M g S i - P r v inclusions typically contain <0.03 wt% N i O (Hutchison, 1997; Stachel et al, 2000b).  M g S i - P r v is stable i n both mafic and ultramafic rocks at lower mantle P-T conditions, however, it is the expected dominant phase i n ultramafic rocks (-70%) and would be the second or third most abundant phase i n mafic rocks after CaSi-Prv and Si02, depending On the C a : M g ratio.  8.1.4.2  CaSi0  3  CaSi03 has three polymorphs in the P-T range of the mantle (wollastonite, CaSi-walstromite, and  Table 8.2. CaSi0 polymorphs 3  CaSi-perovskite) as well as a stability field where it  Mineral  Abbreviation  occurs as two minerals (Ca2Si04 and CaSi 05)  CaSi-perovskite  CaSi-Prv  CaSi-walstromite  CaSi-Wal  CaSi-wollatonite  Wo  2  (Table 8.2 and Fig. 8.5A). MgSi03,  CaSi03 does not  However, unlike have  any stable  polymorphs in an open system in the upper mantle. A s such, the presence o f CaSi03 is a likely indicator o f depths ->580 k m .  Above - 5 8 0 k m , calcium occurs i n either garnet or clinopyroxene. Below - 5 8 0 k m , calcium, i n both eclogitic and peridotititc source rocks, begins to form a new highpressure species, CaSi-Prv (Fig. 8.3A-B). Experiments by hifune and Ringwood (1987) and W o o d (2000) show that CaSi-Prv exsolves from majorite garnet at pressures  138  exceeding 20-21 G P a . In the lower mantle, CaSi-Prv is predicted to be the third most abundant phase in pyrolite (-10 v o l %) and the second most abundant phase in eclogite (-30 v o l %) (Irirune et al., 1993). In terms o f chemistry, there is little substitution o f other elements i n CaSi-Prv, even i n a chemically complex mantle (Gasparik, 1989,1990). A s such, there are no geothermobarometers yet established that can be used for even a crude estimation o f the depth o f formation.  Trace element geochemistry is likely the only method available to determine the nature o f the parental material for CaSi-Prv inclusions (in the absence o f mineral associations) and has been used by several authors (e.g. Stachel et al., 2000b; Harte et al, 1999; Hutchison, 1997).  The E u anomaly observed i n C a S i 0 3 grains by Harte et al. (1999) has been  interpreted as a possible indicator o f a crustal source.  A)  B) Temperature (°C)  Temperature (°C)  Fig. 8.5. Phase transformations for the predicted dominant Ca phases in the mantle. A) Phase transformations for CaSi0 (modified after Gasparik et ah, 1994), and B) Phase transformations for (Ca,Mg)Si 0 , (modified after Koito et al, 2000). 3  2  6  8.1.4.3 Garnets and highly aluminous silicates  Garnets inclusions i n diamond typically fall into one o f two broad compositional divisions: eclogitic and peridotitic.  W i t h the discovery o f non-stoichiometric garnets  included i n diamonds from Sloan, Colorado (Otter and Gurney, 1989), a third category o f  139  garnet was included, termed majorite. mineral,  A fourth  tetragonal-almandine-pyrope  phase  Table 8.3. Aluminous silicates Mineral  Abbreviation  Highly aluminous silicate (general)  Grt  for grouping these fours phases (e-type, p-type,  Tetragonal almandinepyrope phase  TAPP  majoritic and T A P P ) is that they all contain  Eclogitic garnet  eGrt  Peridotitic garnet  pGrt  Majoritic garnet  Maj  ( T A P P ) w i l l be included here, although this mineral i n not really a garnet.  abundant AI2O3 (~>18 wt%). the  absence  The sole reason  In this report, in  o f supporting data,  any  highly  aluminous silicate w i l l be referred to as garnet (Grt).  M o r e specific abbreviations are listed i n  Table 8.3.  The garnet group covers a large number of minerals with the general chemical formula X3Y2Si30i2, where X=Ca, Fe , Mg and Mn, and Y=A1, Cr and Fe . Peridotitic garnets 2+  3+  are characterised by high chromium content (>2.00 wt% Cr203, Gurney, 1989) while eclogitic garnets are typically void of chromium (<2.00 wt % Cr 03, Gurney, 1989). As 2  well, pGrt's are Mg-rich while eGrt's contain more Fe and Ca (Meyer, 1987). Garnets are generally isotropic, however this property tends to change with depth as the garnet structure begins to accommodate elements in non-stoichiometric proportions! Ringwood (1967)  first reported the solubility of pyroxene in garnet at high pressures. Based mainly  on studies by Akaogi and Akimoto (1977 and 1979) and Lui (1977), it was concluded that pyroxenes gradually dissolve into garnet with increasing pressures until the transition zone (-410 km depth), at which point only one 'garnet' phase exists. This 'garnet' phase is called majorite and any garnet containing a dissolved pyroxene component is considered a majoritic garnet. By calculating the number of silica cations (as a ratio of oxygen anions), it is possible to differentiate between majoritic and non-majoritic garnets; any grain with greater than 3.075 S i  4+  cations per 12 oxygens has a majoritic  component (Stachel et al, 2000a). Both eclogitic and peridotitic garnets can contain a majoritic component. Experimental work by Irifune (1987) shows that the amount of S i in majoritic garnets can be used to determine the pressure at formation.  140  4+  A fourth phase is considered here on account o f the large amount o f aluminium accommodated i n the structure. Examination o f mineral inclusions from Juina diamonds by Harris et al. (1997) found a new aluminous silicate which they termed tetragonal almandine-pyrope phase ( T A P P ) .  They concluded that T A P P is a stable phase i n the  uppermost part o f the lower mantle. T A P P grains contain the normal garnet S i : A l ratio without evidence o f a majoritic component. Some characteristics o f T A P P are: unusually low C a O contents (<0.12 wt%); relatively restricted chromium values between 1.39 and 2.80 wt%  and mg between 0.82 and 0.91. The composition o f parental rocks for  &2O3;  T A P P is unclear.  8.1.4.4  Mg Si0 2  4  M g S i 0 4 is the main phase predicted for a 2  pyrolitic upper mantle and transition zone (Fig.  Table 8.4. Mg Si0 polymorphs. 2  4  8.3A) and exists in the form o f three polymorphs  Mineral  Abbreviation  (Table 8.4).  Olivine  a-Ol  Wadsleyite  p-01  Ringwoodite  Y-Ol  The three changes that M g S i 0 2  4  undergoes from the upper mantle to base o f the transition zone are from orthorhombic (referred to  as  a-Ol),  structure  to  a modified  (P-Ol or  spinel Temperature (°C)  Wadsleyite) and  800  — 1  finally to a true spinel structure (y-Ol or  1200  1600  1 1 — 100  Ringwoodite), as has been demonstrated •  200  by numerous experiments (Ringwood, 1991 and references therein) (Fig. 8.6). The  transformation o f a - O l to P-Ol  * CM  Olivine  10 Olivine +  12  3  14  ringwoodite^^  1 300  1 Olivine + wadsleyite  •  400  Q  transition zone  occurs over a depth interval o f 4-35 k m ,  CM" 500  depending on the mantle temperature and  ringwoodit  -  600  -  water content (Frost, 2003) and  -lower mantle  nerovskite +• erropericlase  results i n a - 8 . 0 % decrease i n volume (Lui, 1975).  The next transformation  (to y-Ol) occurs over a depth interval o f  700  Fig. 8.6. Phase transformation for (Mgo.89Feo.n)2Si0 . Modified from Akaogi et al., 1989. 4  141  20 k m (Frost, 2003), which results i n a - 2 . 0 % decrease i n volume ( L u i , 1975). Eventually y - O l breaks down into two phases, M g S i - P r v and ferropericlase (fPer) (Ito and Takahashi, 1998) occurring over a depth interval o f 5-12 k m (Yamazaki et al, 1994).  Olivine inclusions i n diamond (depths o f origin ~<250 km) are typically.Mg-rich (mg = 0.91 - 0.95), contain -0.05 wt% C r 0 2  3  and nickel concentrations o f -0.40 wt% N i O  (Meyer, 1987). The sparse data for M g S i 0 4 o f deeper origin (depths >250 km) suggests 2  that mg is typically lower (i.e. 0.87 to 0.91, Hutchison, 1997), which is i n part corroborated by experimental studies. High-pressure experiments at 1400°C on pyrolite (mg = 0.89) reveal that the mg o f a - O l increases with depth towards the transition zone to mg = -0.925, after which the structural change to P-Ol results i n an immediate decrease i n mg to 0.885, which then steadily increases with greater pressure (Irifune et al, 1998). Although data on the partitioning behaviour o f the three polymorphs o f M g S i 0 2  4  are  sparse, there may be some distinguishing chemical features to determine what initial polymorph formed when the inclusion crystallised. y - O l has the ability to incorporate trivalent cations better than p - O l , which i n turn incorporates trivalent cations better then a - O l (Brey et al, 2003).  The partitioning o f divalent cations also changes between  polymorphs (Brey et al, 2003).  These authors conclude that M g S i 0 4 with high mg2  values and l o w N i , C o and C r indicates an upper mantle origin (and equilibrium with ferropericlase), whereas M g S i 0 4 with high N i , C o and C r contents at intermediate to low 2  wg-values are indicative o f an origin i n the transition zone or at the - 6 6 0 k m boundary with the lower mantle.  Note that the stability o f ferropericlase i n the upper mantle is  debatable, although traditionally, it is considered stable only in the lower mantle.  8.1.4.5 S i Q  Si0  2  2  is not predicted to be an abundant phase i n  the mantle and is not i n equilibrium with a mantle o f pyrolitic composition o f mg - 0.89. M o r e Fe-rich material, however, is i n equilibrium with S i 0  2  (Fei et al, 1996) as is mafic material  142  Table 8.5. Si0 polymorphs 2  Mineral  Abbreviation  a-quartz  a-Qtz  P-quartz  (3-Qtz  coesite  Coe  stishovite  Sti  (similar to M O R B i n composition) at ~>300 k m depth (Fig. 8.3B).  Si0  exhibits  P-T  four  polymorphs  conditions that are  at  relevant  for  2  the  mantle (Table 8.4 and F i g . 8.7).  The  distinction  between  Si0  2  polymorphs based on chemistry alone is difficult because o f the lack o f element substitution  into  Si0 . 2  Polymorph  determination can be determined based on phase association, or by varying C L  properties (Sobolev et al., 1999), which w i l l be expanded upon i n section 8:3.2.1.  8.1.4.6 Ferropericlase Ferropericlase has the chemical formula ( M g , F e ) 0 and forms a solid-solution mineral i n the lower mantle, with end-members M g O (periclase) and F e O (wustite). Because o f the high mg (0.50<mg<1.00) for most ( M g , F e ) 0 grains, they are most accurately described as ferropericlase (fPer); grains with more Fe (0.00</wg<.50) described as magnesiowustite.  are most  accurately  FPer forms, along with M g S i - P r v , as a result o f the  decomposition o f y - O l (Ito and Takahashi, 1998) and is considered to be the second most abundant phase, - 1 9 v o l %, (16% by weight, Wood, 2000) i n the lower mantle (Ringwood, 1991). A s previously discussed, this transformation is likely responsible for the strong seismic discontinuity observed at ~660 k m and is what distinguishes the upper mantle from the lower mantle (Figs. 8.1 and 8.2). F o r a pyrolitic lower mantle o f mg = 0.90, fPer (mg = 0.84-0.80) and M g S i - P r v (mg = 0.95-0.96) would be i n equilibrium (Wood, 2000). Although fPer is considered to form only in the lower mantle, it is not restricted to lower mantle pressures; it is also stable i n an upper mantle with sufficiently low S i activity (Stachel et al., 2000b).  Ferropericlase does not undergo any phase  changes for P-T conditions experienced i n the upper and lower mantle, although  143  preliminary experimental  work by Dubrovinsky et al., (2003) find that fPer  may  dissociate into M g O and FeO at 85 G P a , corresponding to a depth o f 1900 to 2000 k m . The parental source rocks for fPer are likely pyrolitic or magnesian in bulk composition; fPer is not a predicted phase in a lower mantle o f mafic composition.  8.1.4.7 Stability of other phases  C a T i 0 , magnetite and sulphides are minor phases that have previously been found in 3  Juina diamonds (Hutchison, 1997; Kaminsky et al., 2001a) and warrant consideration here. A s there are little data on the high P-T stability o f these minerals little can be said, particularly on the stability o f these minerals in an open system o f predicted mantle compositions.  The pyrolitic and basaltic models do not predict any o f these three phases  which is likely a result o f them either not being in equilibrium with a pyrolitic or eclogitic mantle, or that they only make a small overall proportion o f the mantle.  Magnetite (Mag) has the chemical formula F e F e 2 0 4 , and has two 2+  3+  polymorphs  between temperatures (0 to 1000°C) and pressures (0 to 30 G P a or - 8 0 0 km depth) (Haavik et al, 2000).  Structural transformation likely takes place somewhere between  200 to 400 k m depths (from extrapolation o f data from Haavik et al., 2000).  Sobolev  (1983) finds M a g in association with eclogitic source rocks, although this mineral is rare as an inclusion i n diamond.  The stability o f M a g within eclogite and peridotite as a  function o f pressure and temperature is unclear.  Perovskite (Prv) has the chemical formula C a T i 0  3  and is a common  mineral in  kimberlites and some other deep-sourced magmatic rocks (Mitchell, 1995).  It is a  relatively rare inclusion in diamond and the stability of this mineral in an open mantle system is unclear.  Sulphides are often cited as being the most common mineral inclusion in diamond, although this may only be true for diamonds from South African kimberlites (Harris and Gurney, 1979) and Yakutian diamonds (Bulanova et al., 1990). Sulphide inclusions are  144  thought to have formed by the trapping o f a primary liquid sulphide melt that crystallises into monosulphide solid solution ( M S S ) during diamond growth (Bulanova et al, 1996). The P-T stability o f sulphides and distribution coefficients as a function o f P - T is unclear. Sulphide inclusions from peridotitic diamonds contain 22-36 wt.% N i , whereas eclogitic diamond inclusion sulphides contain 0-12 wt.% N i (Bulanova et al.,  1996).  An  intermediate class o f sulphides contains 11-18 wt.% N i and may be sourced from pyroxenitic material (Bulanova et al., 1996).  8.2 A n a l y t i c a l Techniques  8.2.1 E x t r a c t i o n a n d mounting o f inclusions  Inclusions  were  extracted  by  mechanical  crushing o f diamond i n an enclosed steel cracker (Fig. 8.8).  \  Diamonds were oriented inside the  cell under microscope, l i d height was adjusted to  I  ensure it rested on the diamond and then the lid was struck with a hammer. The l i d was removed and the contents o f the cell were examined under microscope.  Often the diamond did not break  and the procedure described was repeated, but by applying greater force on the hammer.  The goal  Fig. 8.8. cracker.  Photograph of diamond  o f this method was to use the minimum force required to induce brittle fracturing so as to cause minimum breakage o f the inclusion and to avoid separating the inclusion from its hosting diamond chip. The benefits o f this procedure are twofold: 1) smaller inclusions can be examined because o f the tendency for inclusions to remain partially embedded i n diamond chips, therefore making it possible to pick up and move inclusions o f any size; and 2) it is easier to distinguish lab contaminants from diamond inclusions when inclusions are partially embedded in diamond host (e.g. Fig. 8.9A and B ) .  145  Once  inclusions  were  exposed, they were moved using  fine  tweezers  to  stubs for examination on the S E M . colourless  For the case of inclusions,  it  was often difficult to keep track  of  which  newly  Fig. 8.9. SEM image of inclusions embedded in diamond. A) 4.3 C, D and E. (one Ol and two MgSi0 grains) and B) 4.3A (fPer). 3  produced diamond chips actually  contained  the  colourless mineral. Examination of diamond chips under cross-polarised light did not prove helpful in finding inclusions. In many cases where the inclusion could not be located under the microscope, diamond chips were placed on stubs in hope that the inclusion would be located using the S E M . In several cases this approach proved successful. The material remaining in the cell after prospective diamond fragments were removed and mounted was dumped into a petri dish and the cell was cleaned with compressed air. Once all prospective fragments had been mounted the 'diamond dust' in the petri dish was examined under polarized and cross-polarized light.  Rarely was  anything found as it was uncommon for the 'diamond dust' to be isotropic (likely due to extensive crystal deformation as is inferred from the abundant deformation lamellae observed (section 2.3.7)).  Diamond dust was placed on a clean sheet of paper and  funneled into a vial for reference and future work. Several procedures were followed in order to minimize the possibility of contamination or mixing of samples.  Firstly, the  cracker (including lid) was examined before breaking any new diamond. As well, the cell was cleaned with a sandblaster roughly twice a week, however, the frequency of sandblasting was motivated more by the oxidation of the cracker than by diamond chip contamination. The petri dish in which the 'diamond dust' was collected was cleaned with compressed air and washed twice with ethanol after all the inclusions had been extracted from each diamond. In most cases diamond chips from different diamonds were stored on separate stubs. Photographs of each stub were collected as a 'map' for future reference.  146  8.2.2 Qualitative identification of inclusions (EDS)  Diamond fragments and inclusions were examined i n back scattered electron (BSE) mode on the S E M . A s inclusions are typically made up o f elements with higher atomic number than that o f diamond, they appear bright under B S E and are thus easily identified (Fig. 8.9). Energy dispersion spectrometry ( E D S ) was used to qualitatively identify inclusions and multi-phase inclusions. Prospective inclusions for E P M A were located and marked on the reference maps. Table 8.6. Statistics on oxide analyses 8.2.3 Quantitative identification of Diamond No.  inclusions (EMPA)  The  methodology  employed  in  1.2 1.4 1.5 2.2 2.6 2.7 2.8 2.10 2.11 3.1 3.2 3.4 3.5 3.6 3.7 3.8 3.9 3.10 4.3 4.7 . 4.10 4.11 4.16 5.10 6.1 6.2 6.6 6.8 6.9 7.1 total  this  study to quantitatively examine mineral inclusions was somewhat unusual and warrants  description  here.  The  procedure followed i n most diamond inclusion studies involves: extraction o f individual  inclusions,  mounting  inclusions i n small stubs i n epoxy and then finely polishing inclusions before electron-probe microanalyses ( E P M A ) . This  arduous  because  the  procedure analytical  is  followed equipment  requires a near horizontal surface to produce acceptable results. in  the  case  o f this  However,  study,  it  was  considered too risky or impossible to separate inclusions from their hosting diamond, to mount the separated grains i n glue and then to polish the sample.  147  Total Number of Percent number of analyses between success analyses 98-102 wt% 90 38 52 11 27 45 39 7 6 28 97 21 58 . 41 30 4 11 64 56 28 12 15 6 19 13 11 3 9 11 4 856  29 5 21 3. 3 5 3  8 21 4 17 12  9 15 15 2 8  3 5 2  5 3 198  32.2 13.2 40.4 27.3 11.1 11.1 7.7 0.0 0.0 28.6 21.6 19.0 29.3 29.3 0.0 0.0 81.8 23.4 26.8 7.1 66.7 0.0 0.0 15.8 38.5 18.2 0.0 0.0 45.5 75.0 23.1  Instead, inclusions exposed on cleavage surfaces  were  examined  in  their  unpolished  state.  Some  larger  inclusions  (>100  microns)  were  Table 8.7. Statistics on sulphide analyses Diamond No.  Total number of analyses  Number of analyses between 98-102 wt%  Percent success 25.0  2.11  24  6  separated from their hosting diamond,  4.11  4  2  50.0  6.8  3  3  100.0  however, even these inclusions were  total  31  11  35.5  examined without polishing. Inclusions were mounted on stubs on double-sided black tape.  Because the microprobe has limited vertical range for focusing, it was  necessary to ensure that all inclusions were at roughly the same height on the stub. Although every effort was made to orient inclusions i n such a way as to create a near Table 8.8. List of standards used in electron microprobe analyses Anion  Element  Synthetic/ Natural  Standard  Chemical formula  X-ray lines  Crystal  oxide  Mg  natural  olivine  natural  olivine  Mg,. Feo.2 Si0 Mgi. Fe . SiO  Fe  synthetic  fayalite  Fe Si0  Mn  synthetic  rhodonite  MnSiOa  Na  natural  albite  NaAlSi 0  Al  natural  NaCa(Fe,Mg) TiSi Al 0 (OH)  MgK SiK FeK MnK NaK A\K PK KK CaK TiK CrK  TAP  Si  NiAT  LIF  sulphide  8  8  0  2  2  4  4  4  3  8  TAP LIF LIF TAP TAP  P  natural  kaersutite apatite  K  natural  orthoclase  KAlSi 0  Ca  natural  diopside  CaMgSi 0  Ti  natural  rutile  Ti0  Cr  synthetic  Mg-chromite  MgCr 0  Ni  synthetic  Ni-olivine  Ni Si0  Y  YAG  Y A1 O  Zr  zircon  ZrSi0  Ce La Pr Nd Sm Gd  cerium dioxide La element Pr element Nd element Sm element Gd element  Ce0 Ca-Al-Si-REE glass Ca-AI-Si-REE glass Ca-Al-Si-REE glass Ca-Al-Si-REE glass Ca-Al-Si-REE glass  LaL/3 PrL/5 Ndl Sm/. Gdi  Fe  pyrite  FeS  FeK  LIF  Co  Co metal  CoK  LIF  Ni  Ni metal  Ni/T  LIF  Cu  tetrahedrite  (Cu,Fe) Sb S  sphalerite  ZnS  CuK ZnK  LIF  Zn  FeS  MnK SK  synthetic synthetic synthetic synthetic synthetic  Mn  Mn metal  S  pyrite  4  6  2  22  Ca (P0 ) (OH) 5  4  3  3  8  2  2  2  2  3  6  4  4  5  l2  4  2  2  12  4  2  Ca-Al-Si glass standard is from (Drake and Weill 1972).  148  13  2  PET PET PET PET LIF  YL  TAP  ZTL  PET  CeL  LIF LIF LIF LIF LIF LIF  LIF LIF PET  horizontal surface, many totals collected were poor.  Analyses success rates for each  diamond are presented in Table 8.6 (for silicates) and Table 8.7 for sulphides.  E P M A of inclusions were done on a fully-automated C A M E C A SX-50 microprobe, operating in the wavelength-dispersion mode with the following operating conditions: excitation voltage, 15 kV; beam current, 20 nA; peak count time, 20 s; background count time, 10 s; beam diameter, 1 jura. For elements Y , Zr, Ce, La, Pr, Nd, Sm and Gd, peak count time was 40 s and background count time was 20 s. Data reduction was performed using the 'PAP' <p(pZ) method (Pouchou and Pichoir 1985). Table 8.8 lists the standards, X-ray lines and crystals used for the elements analysed.  M i n i m u m detection limits  (MDL's) for all data listed as weight percent oxides is found in Table 8.9. MDL's for sulphides are listed as element weight percent. Table 8.10 lists the MDL's for all cation calculations. Error limits for E P M A analyses are listed in Table 8.11 (wt% totals) and Table 8.12 (cation totals) at the 95% confidence level (2a).  Table 8.9. Minimum detection limits for weight percent values  po  2  A1 0  0.05 0.06 0.07  0.03 0.32 0.04 0.33 0.42 0.38  0.05 0.06 0.07 0.04 0.04 0.04  0.04 0.22 0.06 0.02 0.06 0.03  Prv  Y 0 0.03  Zr0 0.09  Nb 0 0.06  Sul  S 0.31  Mn 0.03  Fe 0.40  mineral fPer Grt Mag Ol MgSi0 CaSi0  2  SiO  5  0.06 0.09 _  3  3  2  z  4  3  Ti0  2  2  5  La 0 0.06 2  Co 0.05  3  Cr 0 2  3  0.13 0.13 0.10 0.11 0.15 0.09 3  Ce 0 0.07 2  Ni 0.19  3  FeO  MnO  NiO  MgO  CaO  Na 0  K 0  0.65 0.36 1.01 0.28 0.17 0.06  0.07 0.07 0.07 0.06 0.06 0.05  0.10 0.06 0.06 0.09 0.05 0.06  0.38 0.13 0.05 0.33 0.27 0.02  0.04 0.17 0.03 0.03 .0.03 0.56  0.03 0.03 0.03 0.02 0.02 0.04  0.03 0.03 0.04 0.03 0.02 0.03  Sb 0.05  Pb 0.00  Pr 0 0.17 2  Cu 0.05  3  Nd 0 0.10 2  Zn 0.03  3  Sm 0 0.13 2  As 0.02  3  Gd 0 0.07 2  Ag 0.06  3  Th0 0.03  2  2  4  Cd 0.06  MDL's of weight percent results listed for each mineral type identified. Table is subdivided into three sections: major oxides, LREE's for Prv, and sulphides. MDL's for the major oxides in Prv are the same as those for CaSi0 . 3  149  Table 8.10. Minimum detection limits for cation values mineral  p5+  Si  Ti  4+  Al  4 +  Cr  3 +  Fe  3+  Mn  2+  Ni  2+  Mg  2+  Ca  2+  2+  Na  K  +  +  fPer 0.008 Grt 0.007 Mag Ol 0.002 MgSi0 0.001 CaSiOj 0.002  0.005 0.005 0.003 0.002 0.001 0.001  0.003 0.003 0.002 0.001 0.001 0.001  0.006 0.005 0.003 0.001 0.001 0.001  0.010 0.007 0.010 0.005 0.006 0.005 0.003 0.002 0.002 0.001 0.003 0.001  0.006 0.005 0.004 0.002 0.001 0.001  0.007 0.006 0.005 0.002 0.001 0.002  0.007 0.004 0.004 0.002 0.001 0.001  0.004 0.011 0.004 0.003 0.009 0.004 0.002 0.008 0.003 0.001 0.003 0.001 0.001 0.002 0.001 0.001 0.002 0.001  Prv  Y+ 0.001  Zr 0.001  Nb 0.001  La 0.001  Ce 0.001  Pr 0.003  Nd 0.001  Sm 0.002  Gd 0.001  Th 0.000  Sul  S" 0.001  Mn 0.001  Fe 0.001  Co 0.001  Ni 0.001  Cu 0.001  Zn 0.001  As 0.002  Ag 0.001  Cd 0.001  3  3  +4  +2  +S  +2  +3  +2  +3  +2  +3  +2  +3  +  +3  +3  +2  +4  Sb 0.001  Pb 0.001 +  MDL's of cation values listed for each mineral type identified. Table is subdivided into three sections: major oxides, LREE's for Prv, and sulphides. MDL's for the major oxides in Prv are the same as those for CaSi0 . The general formula units with cation totals (R) are as follows: 10RO for fPer; R 0 for Grt; R 0 for Mag and Ol and; R 0 for MgSiQ and CaSiQ . 3  3  8  4  2  3  3  1 2  3  Table 8.11. Precision for weight percent values at 95% confidence level mineral  P2O5  SiO  fPer Grt Mag Ol MgSi0 CaSi0  0.11 0.1.1  2  A1 0  0.11 0.10 0.12  0.06 0.07 0.06 0.07 0.08 0.07  0.05 0.06 0.06 0.05 0.05 0.06  0.06 0.05 0.06 0.05 0.05 0.04  Prv  Y 0 0.05  Zr0 0.12  Nb O 0.08  Sul  S . 0.03  Mn 0.04  Fe 0.05  3  3  2  3  z  4  Ti0  2  FeO  MnO  0.16 0.16 0.15 0.16 0.16 0.18  0.10 0.08 0.11 0.08 0.08 0.08  0.08 0.08 0.09 0.07 0.08 0.08  La 0 0.10  Ce 0 0.09  Pr 0 0.29  Nd 0 0.14  Co 0.05  Ni 0.05  Cu 0.04  Zn 0.08  2  s  2  3  3  Cr 0 2  3  2  3  2  3  2  NiO 0.11 0.10 0.12 0.10 0.10 0.11 3  Sm 0 0.25 2  As 0.14  3  MgO  CaO  Na O  K 0  0.05 0.04 0.05 0.04 0.04 0.04  0.04 0.04 0.04 0.04 0.04 0.05  0.07 0.06 0.08 0.05 0.05 0.06  0.04 0.04 0.04 0.04 0.03 0.04  Sb 0.09  Pb 0.32  Gd 0 0.12 2  Ag 0.10  3  z  2  Th0 0.09 4  Cd 0.10  Precision for weight percent errors listed at the 95% confidence level for each mineral type identified. Table is subdivided into three sections: major oxides, LREE's for Prv, and sulphides. Error for the major oxides found in Prv are the same as those for CaSi0 . 3  150  Table 8.12. Precision for cation calculations at 95% confidence level mineral fPer Grt Mag Ol MgSi0 CaSi0  P  Si  5+  4+  Ti  4 +  Al  3 +  Cr  3+  Fe  2+  Mn  2+  Ni  2+  Mg  2+  Ca  2+  0.004 0.006 0.001 0.001 0.001  0.003 0.024 0.002 0.008 0.007 0.007  0.003 0.003 0.003 0.001 0.001 0.001  0.004 0.019 0.004 0.001 0.001 0.001  0.009 0.008 0.004 0.002 0.002 0.001  0.045 0.023 0.043 0.006 0.003 0.001  0.005 0.005 0.003 0.001 0.001 0.001  0.007 0.003 0.003 0.002 0.001 0.001  0.048 0.015 0.004 0.012 0.007 0.001  0.004 0.013 0.002 0.001 0.001 0.012  Prv  Y+ 0.000  Zr 0.001  Nb 0.001  La 0.001  Ce 0.001  Pr 0.001  Nd 0.001  Sm 0.001  Gd 0.001  Th 0.000  Sul  S" 0.009  Mn 0.001  Fe 0.007  Co 0.001  Ni 0.003  Cu 0.001  Zn 0.000  As 0.000  Ag 0.001  Cd 0.000  3  3  3  +4  +2  +5  +2  +3  +2  +3  +2  +3  +2  +3  +  +3  +3  +2  Na  +  r  0.005 0.005 0.003 0.001 0.001 0.001  0.003 0.003 0.003 0.001 0.000 0.001  Sb 0.000  Pb 0.000  +4  +  Cation errors listed at the 95% confidence level for each mineral type identified. Table is subdivided into three sections: major oxides, LREE's for Prv, and sulphides. Error for the major oxides found in Prv are the same as those for CaSi0 . 3  151  8.3 Results  The  following section contains the results o f E P M A and E D S analysis for mineral  inclusions as w e l l as S E M images for the majority o f grains analysed i n this study. E P M A data was preferred over E D S data, however, due to the small grain size and inclusion heterogeneity, it was often not possible to collect decent E P M A data (see Tables 8.6 and 8.7 for analysis success rates).  When acceptable E P M A data is not  available, E D S spectra are generally included. The mineral phases and frequency o f phases that occur i n each diamond (i.e. mineral associations) that were analysed using E P M A and produced acceptable results (i.e. wt% totals between 98 and 102) are listed i n table format i n Appendix E .  Appendix F lists all the inclusions identified in each  diamond through E D S .  This section is divided into five subsections: inclusions o f primary origin, inclusions o f uncertain origin, touching phases and inclusions o f secondary origin. The fifth section, a somewhat special category, has been created for F e - N i blebs on ferropericlase grains. Assigning inclusions to either a primary or secondary origin is not a trivial matter and is responsible for much debate, and hence the motivation for the 'inclusions o f uncertain origin' category. When resources and time permit, some scientists polish 'windows' on diamonds to determine i f any cracks lead from the diamond surface to the inclusion (e.g. Harris et al, 1997), however, this technique was not used i n this study. Instead, grains were examined using a conventional microscope and any diamond with extensive fractures was noted.  During mechanical fracturing, diamonds with extensive fractures  (previously noted) typically fragmented easily.  A more detailed description o f the  criteria used to distinguish the origin o f inclusions (i.e. primary vs. secondary) is presented in the preamble to each subsection.  152  8.3.1 Inclusions of primary origin  Primary grains are inclusions that have remained isolated inside the diamond since encapsulation during diamond crystallization. They are pristine samples from the mantle. These inclusions are typically competent (i.e. they do not break easily), subhedral to euhedral in crystal form and homogeneous (in terms of chemistry). comprise primary inclusions.  Eight phases  They are: ferropericlase, MgSiC>3, CaSiC^, Mg2Si04,  pyrope-almandine-grossular garnet, tetragonal-almandine-pyrope phase, pyrrhotite and magnetite.  8.3.1.1 Ferropericlase  Ferropericlase (fPer)  is  the most  abundant inclusion found during this  study.  Approximately 100 individual crystals have been extracted from 16 diamonds (36% of diamonds broken or 23% of the total diamonds in this study).  FPer belongs to the  isometric crystal system, with cubes and octahedrons being the most common forms to develop (Nesse, 1991). Grains in this study are typically euhedral and exhibit either an equant cubo-octahedral form (e.g. Figs. 8.10, 3.10C, O and 5.10E) or a cubo-octahedral form with elongation in one or two directions (e.g. Figs. 8.10, 1.2AJ, 1.5G, 3.2A, F, and G, 3.5D, E , and M and 4.3J). In some cases there is minor development of a (110) face (e.g Fig. 8.10, 3.10W).  Because diamond also forms octahedra and cubo-octahedra  (section 2.1.1.1), it is unclear if the euhedral morphology is being imposed by the host diamond or is in fact an expression of the primary growth faces of fPer grains. Regardless of whether or not the cubo-octahedral morphology of fPer is a result of fPer growth or imposed diamond growth, the large percentage of euhedral crystals is mostly uncommon for other phases in this study.  Euhedral crystals may be more common  because the fPer structure is likely stable from the time of entrapment in diamond until extraction in the laboratory, whereas most of the other phases likely undergo a number of structural changes.  153  154  1.2-AI  1.2-AJ  1.5-D  1.5 - F  1.5-G  1.5-K  2.7-E  2.7-0  3.1-B  155  3.2-A  3.2-F  3.2-G  3.2 J  3.5-D  3.5-E  3.5-M  3.5-P  3.6-D  3.6-J  3.6-K  3.6-L  156  157  io um 4.3-Q  4.3-R  5.10-A  5.10-E  6.1-B  6.1-E  6.2-A  6.9-A  6.9-C  Fig. 8.10. SEM images of ferropericlase grains. Number refers to diamond which hosted the inclusion while the letter refers to the inclusion sample code. Major oxide chemistry and cation calculations for each inclusion are found in Tables 8.13 and 8.14 respectively.  Ferropericlase grains are typically black and opaque; however, some grains are pale orange (Diamond 2.7, inclusion A ) while others are pale purple (Diamond 3.2, inclusions F , G , and J). Inclusion colour is likely both a function o f grain thickness and chemistry, although chemical data does not seem to support the latter hypothesis. in size from 20 to 250 microns.  158  Inclusions range  Major oxide chemical data for 57 grains from 15 diamonds are presented i n Table 8.13, with accompanying cation calculations in Table 8.14.  Grains fall along the periclase-  wustite (MgO-FeO) solid-solution series and are skewed towards the more magnesiumrich  end-member.  There  is  considerable  range  in  mg  (herein  defined  as  M g / ( M g + F e ) , where all iron is calculated as F e ) from 45.1 to 88.9 with an average 2+  2+  2+  2+  of 66.9 ± 13.0 (1 a). However, it is important to note that this average is skewed towards diamonds containing more ferropericlase grains.  Analyses for grains 3-5D, M and P are likely a combination o f two phases (fPer and FeN i blebs) and w i l l be removed for the following generalisations (Fe-Ni blebs are discussed in section 8.3.4). M g  2 +  and F e  2 +  fill 92.8 to 99.6% o f the cation site, with the  remainder o f site being filled by N i O (0.11-1.46 with an average 0.72 wt%), C r 0 3 (0.002  1.30 wt%, average 0.47 wf%), M n O (0.12-1.46 wt%, average 0.41 wt%) and N a 0 (0.002  2.31 wt%, average 0.39 wt%).  There is a positive correlation between N i and M g  content.  Fig. 8.11 illustrates the  70  general  60  FeO  similarities  and  MgO  in  wt%  between fPer inclusions liberated from the same diamond.  inclusions ranging  50  mis  mi-i  A 3-1  BJ-2  B3-5  A3-9  • 3-10 ± 4 - 3  • 6-1  06-2  02-7  #3-6 OJ-10J  96-9  40 30  Diamond 3-9  is the main exception, hosting  8°«  01-2  three  fPer  with from  0.74  20 10 30  20  40  0.89. A linear fit to data  70  80  90  MgO  mg to  60  50  Fig. 8.11. Plot of FeO versus MgO of fPer grains by diamond. Legend refers to diamond sample number. Substitution between FeO and MgO is almost 1:1. (Diamond 3-5 (purple squares) is a combination of two phases (fPer and Fe-Ni blebs)).  in F i g . 8.11 yields the equation  FeO  =  -0.9905MgO  +  96.814  159  with  an  R  value  of  0.9813.  Table 8.13. Major oxide chemistry for ferropericlase grains (wt%) Inclusion No. No.  No. of Inclusion assemblage  analyses P2O5 S i 0 2 T i 0 2 A 1 2 0 3 C r 2 0 3  FeO MnO N i O M g O CaO N a 0 K 0 Total 2  2  averaged  1-2H  1  3  na  0.00 0.00 0.00  0.17 59.93 0.31 0.22 37.63 0.00 0.12 0.00 98.37  1-21  2  3  na  0.09 0.00 0.00  0.00 58.38 0.29 0.20 39.94 0.00 0.14 0.00 99.05  1-2J  3  2  na  0.06 0.00 0.00  0.20 58.49 0.30 0.20 40.42 0.00 0.11 0.00 99.80  1-2L  4  1  na  0.08 0.00 0.00  1-2M  5  3  na  0.10 0.00 0.00  0.16 55.67 0.38 0.18 45.19 0.00 0.00 0.00 101.68 0.17 59.51 0.32 0.21 39.38 0.00 0.13 0.00 99.82  1-2P  6  3  na  0.08 0.00 0.00  0.19 59.54 0.27 0.20 39.75 0.00 0.11 0.00 100.13  1-2R  7  1  na  0.12 0.00 0.00  0.17 53.56 0.24 0.21 43.54 0.00 0.11 0.00 97.96  1-2V  8  1  na  0.09 0.00 0.00  1-2W  9  1  na  0.11 0.00 0.07  0.18 59.56 0.38 0.30 37.38 0.00 0.16 0.00 98.04 0.23 58.92 0.33 0.27 37.88 0.00 0.14 0.00 97.94  1-2Y  10  3  na  0.22 0.00 0.00  0.22 57.59 0.43 0.17 40.44 0.00 0.48 0.00 99.55  1-2AE  11  1  na  0.11 0.00 0.00  0.19 59.18 0.37 0.23 38.34 0.00 1.15 0.00 99.56  1-2AF  12  3  na  0.10 0.00 0.00  0.17 59.22 0.37 0.26 38.70 0.00 0.45 0.00 99.28  1-2AI  13  2  na  0.09 0.00 0.08  0.21 59.07 0.35 0.26 37.99 0.00 0.42 0.00 98.46  1-2AJ  14  2  na  0.24 0.00 0.06  1-5D  15  01-MgSi0 -TAPP  3  0.00 0.07 0.00 0.00  0.16 55.28 0.37 0.23 43.70 0.00 0.10 0.00 100.14 0.35 37.56 0.30 1.18 58.97 0.00 0.11 0.00 98.55  3  1-5F  16  01-MgSi0 -TAPP  3  0.00 0.09 0.00 0.00  0.39 36.42 0.32 1.20 60.94 0.00 0.00 0.00 99.35  1-5G  17  Ol-MgSiOj-TAPP  3  0.00 0.08 0.00 0.00  0.41 35.58 0.28 1.14 62.13 0.00 0.49 0.00 100.11  1-5K  18  Ol-MgSiOj-TAPP  . 1  0.00 0.09 0.00 0.00  0.32 3.7.25 0.23 1.13 61.86 0.00 0.23 0.00 101.11  2-2E  19  CaSiOj  2-7A  20  2-7C  3  0.15 0.00 0.11  0.50 33.90 0.26 1.35 64.63 0.00 0.49 0.00 101.40  1  0.00 0.06 0.00 0.00  0.47 28.86 0.36 1.32 66.68 0.00 0.19 0.00 97.95  21  1  0.00 0.12 0.00 0.00  0.42 27.56 0.43 1.07 67.49 0.27 0.41 0.28 98.05  2-7E  22  2  0.00 0.72 0.00 0.50  2-70  23  1  0.00 0.00 0.00 0.00  0.47 25.12 0.41 1.11 69.95 0.26 0.57 0.19 99.30 0.57 28.94 0.38 1.28 68.45 0.00 0.29 0.06 99.97  3  na  3-1B  24  2  na  0.18 0.00 0.18  0.87 31.86 0.49 1.01 64.54 0.00 1.50 0.00 100.63  3-2A  25 01-MgSi0 -CaSi0 -TAPP?  2  na  0.14 0.00 0.00  0.36 25.69 0.25 1.39 71.45 0.00 0.00 0.00 99.28  3-2F  26 01-MgSi0 -CaSi0 -TAPP?  1  na  0.00 0.00 0.00  0.40 25.43 0.24 1.17 73.67 0.00 0.67 0.00 101.58  3-2G  na  0.08 0.00 0.00  0.35 24.67 0.22 1.36 72.41 0.00 0.18 0.00 99.26  3-2J  27 01-MgSi0 -CaSi0 -TAPP? 28 01-MgSi0 -CaSi0 -TAPP?  3-5D  29  01-MgSi0  3-5E  30  3-5M  •  CaSiOj 3  3  3  3  3  3  3  3  3  2  na  0.11 0.00 0.00  0.38 25.27 0.27 1.34 71.30 0.00 0.18 0.00 98.86  3  3  na  0.08 0.00 0.08  0.91 53.24 0.78 2.69 38.96 0.00 2.20 0.00 98.93  01-MgSi0  3  1  na  0.11 0.00 0.06  0.76 48.97 0.79 0.31 50.34 0.00 0.34 0.00 101.68  31  01-MgSi0  3  2  na  0.07 0.00 0.12  0.73 51.02 0.69 6.40 39.34 0.00 0.61 0.00 98.98  3-5P  32  01-MgSi0  3  2  na  0.06 0.00 0.11  0.86 51.01 0.66 2.46 45.48 0.00 0.09 0.00 100.73  3-6D  33  2  na  0.12 0.00 0.00  0.00 53.97 0.23 0.24 46.84 0.00 0.00 0.00 101.41  3-6J  34  3  na  0.00 0.00 0.08  0.00 57.35 0.18 0.30 41.39 0.00 0.00 0.00 99.29  3-6K  35  3  na  0.06 0.00 0.13  0.00 59.38 0.20 0.22 39.17 0.00 0.00 0.00 99.17  3-6L  36  3  na  0.10 0.00 0.10  0.00 59.28 0.19 0.15 41.34 0.00 0.00 0.00 101.15  3-9H  37  3  na  0.11 0.00 0.00  0.35 38.37 0.34 1.01 59.77 0.00 0.14 0.00 100.10  3-91  38  3  na  0.25 0.00 0.08  0.22 18.00 0.16 0.92 81.24 0.04 0.00 0.00 100.91  3-9J  39  3  na  0.20 0.00 0.06  3-1 OC  40  3  na  0.00 0.00 0.18  0.33 37.05 0.30 0.97 60.77 0.00 0.73 0.00 100.41 1.30 64.97 1.46 0.13 30.35 0.00 0.64 0.00 99.02  3-10H  41  2  na  0.13 0.00 0.20  1.27 58.39 1.35 0.11 39.59 0.00 0.56 0.00 101.60  3-10O  42  3  na  0.10 0.00 0.20  1.24 57.71 1.29 0.16 36.68 0.00 1.52 0.00 98.89  3-10Z  43  1  na  0.00 0.00 0.18  1.20 65.00 1.38 0.75 30.42 0.00 1.58 0.00 100.53  3-1OAA 44  2  na  0.00 0.00 0.17  1.24 65.09 1.33 0.12 30.02 0.00 1.20 0.00 99.18  3-1OW  45  3  na  0.14 0.00 0.25  0.95 56.59 1.11 1.31 37.64 0.00 2.31 0.00 100.31  4-3B  46  01-MgSi0  1  na  0.06 0.00 0.11  0.83 24.21 0.22 1.28 73.29 0.00 0.49 0.00 100.50  4-3J  47  Ol-MgSiO,  2  na  0.14 0.00 0.10  0.98 24.71 0.23 1.41 72.82 0.00 0.89 0.00 101.30  4-3K2  48  01-MgSi0  3  2  na  0.26 0.00 0.42  1.01 25.96 0.26 1.35 70.83 0.00 0.00 0.00 100.08  4-3Q  49  01-MgSi0  3  1  na  0.12 0.00 0.10  1.01 26.19 0.20 1.46 68.35 0.00 0.92 0.00 98.35  4-3 R  50  01-MgSi0  3  2  na  0.07 0.00 0.08  1.18 27.57 0.23 1.26 68.98 0.00 0.57 0.00 99.96  3  160  Table 8.13. Major oxide chemistry for ferropericlase grains (wt%) (continued) Inclusion No. No. 5-1 OA 5-10E 6-1B 6-1E 6-2A 6-9A 6-9C  Inclusion assemblage  No. of analyses averaged  P2O5  Si0  1 2 2 3 2 3 2  0.00 0.00 na na na 0.00 na  0.14 0.34 0.13 0.08 0.16 0.08 0.11  51 52 53 54 55 56 57  2  Ti0  2  A1 0 Cr 0 2  3  0.00 0.08 0.00 0.07 0.00 0.22 0.00 0.41 0.00 0.00 0.00 0.00 0.00 0.00  2  2  Total  0.00 0.00 0.00 0.00 0.00 0.00 0.00  98.55 99.02 99.99 100.61 101.69 100.19 98.29  FeO MnO NiO MgO CaO Na 0 K 0 2  3  0.54 6.51 0.44 0.40 0.29 0.25 0.21  28.09 28.94 36.95 37.21 44.39 48.03 46.38  0.28 0.36 0.40 0.38 0.12 0.12 0.12  1.07 1.21 0.91 0.93 0.39 0.26 0.24  68.33 67.52 60.93 60.97 56.34 51.35 51.22  0.00 0.08 0.00 0.23 0.00 0.08 0.00  0.00 0.00 0.00 0.00 0.00 0.00 0.00  Inclusion assemblage - refers to the other confirmed phases in the diamond (blank entries indicate that only fPer was found); No. of analyses averaged - the number of analyses with acceptable results (generally between 98-102) that were averaged, na - not analysed. Values below MDL (see Table 8.9) are replaced by 0.00.  Table 8.14. Cation calculations for ferropericlase Inclusion No. 1-2H 1-21 1-2J 1-2L 1-2M 1-2P 1-2R 1-2V 1-2W 1-2Y 1-2AE 1-2AF 1-2AI 1-2AJ 1-5D 1-5F 1-5G  Inclusion assemblage  Ol-MgSiOj-TAPP Ol-MgSiOj-TAPP OI-MgSiOs-TAPP  1- 5K  01-MgSi0 -TAPP  2-2E 2-7A 2-7C 2-7E  CaSiO,  3  2-70 3- 1B  CaSiOj  3-2A  01-MgSi0 -CaSi0 -TAPP?  3-2F  01-MgSi0 -CaSi0 -TAPP?  3-2G  01-MgSi0 -CaSi0 -TAPP?  3-2J  01-MgSiQ -CaSi0 -TAPP?  3  3  3  3  3  3  3  3  P  5+  Si  4+  Ti  4+  Al  3+  Cr  3+  Fe ": Mn 2  2+  Ni  2+  Mg * C a 2  2+  Na  +  K  +  0.000 0.000 0.000 0.012 4.686 0.024 0.016 5.244 0.000 0.023 0.000 0.008 0.000 0.000 0.000 4.475 0.023 0.015 5.458 0.000 0.025 0.000 0.005 0.000 0.000 0.015 4.444 0.023 0.015 5.475 0.000 0.020 0.000 0.007 0.000 0.000 0.011 4.057 0.028 0.013 5.870 0.000 0.000 0.000 na 0.009 0.000 0.000 0.012 4..548 0.024 0.015 5.364 0.000 0.022 0.000 na 0.007 0.000 0.000 0.014 4..530 0.021 0.014 5.390 0.000 0.019 0.000 na 0.011 0.000 0.000 0.012 4.049 0.019 0.015 5.867 0.000 0.020 0.000 na 0.008 0.000 0.000 0.013 4.671 0.030 0.022 5.225 0.000 0.030 0.000 0.010 0.000 0.007 0.017 4..606 0.027 0.020 5.278 0.000 0.025 0.000 0.020 0.000 0.000 0.015 4..374 0.033 0.012 5.475 0.000 0.085 0.000 0.010 0.000 0.000 0.014 4 .553 0.029 0.017 5.258 0.000 0.204 0.000 0.009 0.000 0.000 0.013 4 .561 0.029 0.020 5.314 0.000 0.080 0.000 0.008 0.000 0.008 0.015 4..596 0.028 0.019 5.268 0.000 0.075 0.000 0.022 0.000 0.006 0.012 4.100 0.028 0.017 5.777 0.000 0.018 0.000 0.000 0.006 0.000 0.000 0.023 2..592 0.021 0.078 7.253 0.000 0.017 0.000 0.000 0.007 0.000 0.000 0.025 2 .473 0.022 0.078 7.375 0.000 0.000 0.000 0.000 0.007 0.000 0.000 0.026 2 .387 0.019 0.074 7.430 0.000 0.076 0.000 0.000 0.007 0.000 0.000 0.020 2 .487 0.016 0.073 7.362 0.000 0.035 0.000 na 0.012 0.000 0.010 0.031 2 .223 0.018 0.085 7.552 0.000 0.074 0.000 0.000 0.005 0.000 0.000 0.029 1 919 0.024 0.085 7.903 0.000 0.029 0.000 0.000 0.010 0.000 0.000 0.026 1 822 0.029 0.068 7.954 0.023 0.062 0.028 0.000 0.056 0.000 0.045 0.029 1 .611 0.026 0.068 7.999 0.022 0.084 0.019 0.000 0.000 0.000 0.000 0.035 1 .882 0.025 0.080 7.935 0.000 0.044 0.006 na 0.014 0.000 0.017 0.054 2 .094 0.033 0.064 7.561 0.000 0.228 0.000 na 0.011 0.000 0.000 0.022 1 ,652 0.016 0.086 8.191 0.000 0.000 0.000 na 0.000 0.000 0.000 0.024 1 .595 0.015 0.071 8.235 0.000 0.098 0.000 na 0.006 0.000 0.000 0.021 1 .580 0.014 0.084 8.265 0.000 0.027 0.000 na 0.008 0.000 0.000 0.023 1.631 0.018 0.083 8.203 0.000 0.028 0.000  161  Total mg 10.005 0.53 10.004 0.55 9.998 0.55 9.987 0.59 9.996 0.54 9.995 0.54 9.993 0.59 10.000 0.53 9.990 0.53 10.015 0.56 10.086 0.54 10.025 0.54 10.018 0.53 9.978 0.58 9.991 0.74 9.980 0.75 10.018 0.76 10.000 0.75 10.005 0.77 9.995 0.80 10.023 0.81 9.959 0.83 10.007 0.81 10.065 0.78 9.978 0.83 10.037 0.84 9.997 0.84 9.994 0.83  Table 8.14. Cation calculations for ferropericlase (continued) Inclusion No.  Ti  4 +  Al  3 +  Cr  3+  Fe  Mn  2+  2+  Ni  2+  Mg"" C a 2  1  2+  Na  +  K  +  Total  mg  3-5D  OI-MgSiOj  na  0.007 0.000 0.008 0.066 4.087 0.061 0.198 5.332 0.000 0.391 0.000 10.151  0.57  3-5E  01-MgSi0  na  0.010 0.000 0.006 0.051 3.457 0.057 0.021 6.334 0.000 0.055 0.000  0.65  3-5M  01-MgSi0  na  0.006 0.000 0.013 0.053 3.920 0.054 0.473 5.387 0.000 0.108 0.000 10.015 0.58  3-5P  01-MgSi0  3  3  9.990  na  0.005 0.000 0.011 0.060 3.728 0.049 0.173 5.925 0.000 0.016 0.000  9.967  0.61  3-6D  na  0.011 0.000 0.000 0.000 3.904 0.017 0.017 6.041 0.000 0.000 0.000  9.990  0.61  3-6J  . na  0.000 0.000 0.009 0.000 4.352 0.013 0.022 5.599 0.000 0.000 0.000  9.996  0.56  3-6K  na  0.006 0.000 0.014 0.000 4.566 0.016 0.016 5.369 0.000 0.000 0.000  9.987  0.54 0.55  3  na  0.009 0.000 0.010 0.000 4.432 0.014 0.011 5.510 0.000 0.000 0.000  9.987  3-9H  3-6L  '  na  0.009 0.000 0.000 0.022 2.607 0.023 0.066 7.240 0.000 0.022 0.000  9.991  0.74  3-91  na  0.018 0.000 0.006 0.012 1.091 0.010 0.054 8.780 0.000 0.000 0.000  9.972  0.89  3-9J  na  0.016 0.000 0.006 0.021 2.495 0.020 0.063 7.293 0.000 0.114 0.000 10.028 0.75  3-10C  na  0.000 0.000 0.020 0.099 5.255 0.120 0.010 4.376 0.000 0.119 0.000 10.000 0.45  3-10H  na  0.012 0.000 0.021 0.090 4.370 0.102 0.008 5.282 0.000 0.097 0.000  3-10O  na  0.009 0.000 0.022 0.091 4.483 0.101 0.012 5.079 0.000 0.274 0.000 10.071  0.53  3- 10Z  na  0.000 0.000 0.021 0.091 5.189 0.112 0.058 4.328 0.000 0.292 0.000 10.091  0.45  3-10AA  na  0.000 0.000 0.019 0.095 5.268 0.109 0.009 4.330 0.000 0.226.0.000 10.056 0.45  3- 10W  na  0.013 0.000 0.027 0.069 4.321 0.086 0.096 5.124 0.000 0.408 0.000 10.144 0.54  4- 3B  01-MgSi0  4-3J  01-MgSi0  4- 3K2  01-MgSi0  4-3Q  01-MgSi0  4- 3R  01-MgSi0  5- 1 OA  0.55  na  0.005 0.000 0.010 0.049 1.528 0.014 0.078 8.245 0.000 0.072 0.000 10.002 0.84  na  0.011 0.000 0.009 0.058 1.554 0.015 0.085 8.158 0.000 0.130 0.000 10.021  0.84  3  na  0.020 0.000 0.038 0.061 1.656 0.017 0.083 8.056 0.000 0.000 0.000  0.83  3  na  0.010 0.000 0.009 0.063 1.716 0.013 0.092 7.982 0.000 0.139 0.000 10.024 0.82  na  0.006 0.000 0.008 0.072 1.782 0.015 0.078 7.950 0.000 0.086 0.000  3  3  3  0.000 0.011 0.000 0.008 0.034 1.842 0.019 0.068 7.987 0.000 0.000.0.000  5- 10E  9.981  0.000 0.026 0.000 0.006 0.031 1.897 0.024 0.076 7.887 0.000 0.012 0.000  9.931 9.997 9.969  0.82 0.81  9.961  0.81 0.75  6- 1B  na  0.011 0.000 0.021 0.028 2.492 0.027 0.059 7.326 0.000 0.000 0.000  9.965  6-1E  na  0.006 0.000 0.039 0.025 2.497 0.026 0.060 7.291 0.000 0.036 0.000  9.980  0.74  6-2A  na  0.013 0.000 0.000 0.019 3.038 0.008 0.026 6.874 0.000 0.000 0.000  9.978  0.69  0.000 0.007 0.000 0.000 0.017 3.417 0.009 0.018 6.511 0.000 0.014 0.000  9.992  0.66  9.983  0.66  6-9A 6-9C  na  0.010 0.000 0.000 0.014 3.346 0.009 0.017 6.587 0.000 0.000 0.000  Cation totals are calculated on the basis of 10 anions. Inclusion assemblage - refers to the other confirmed phases in the diamond (blank entries indicate that only fPer was found); na - not analysed; mg Mg /(Mg +Fe ) with typical errors of 0.010 at the 95% confidence level. Any values below MDL (see Table 8.10) are replaced by 0.00. 2+  2+  2+  8.3.1.2 M g S i 0  3  Fifteen distinct inclusions o f M g S i C h are confirmed i n 5 diamonds (diamonds 1-5, 3-2, 35, 4-3 and 6-8), representing only 11% o f diamonds cracked and 7% o f the diamond population.  They are not referred to as M g S i 0 perovskites as there is no supporting 3  crystallographic data suggesting they are indeed perovskites. However, there is a high  162  probability that these grains at one time possessed (and perhaps still possess) the perovskite structure, as w i l l be discussed i n section 8.4.1.2.  Grains are colourless and break easily once removed from their hosting diamond. M g S i 0 3 inclusions tend not to exhibit crystal faces as commonly as fPer grains.  They  may have an elongated ellipsoid shape (Figs. 8.12, 3 . 2 A D 1 , 3.211, and 4.3E), moderately well-formed euhedral shape but undistinguishable overall form (Figs. 8.12, 1.5A1, 3.5G1, 3.5Q and 4.3D), or no discernable form at all (Figs. 8.12, 1.5J1). There is evidence for imposed octahedral shape on a few grains (Figs. 8.12 1.5A1 and 3.5G1). ranges from 15 to 150 um.  1.5-A1  1.5 - J l  3.2-11  3.2-AD1  3.5-GI  3.5 - Q  163  Grain size  Fig. 8.12. SEM images of MgSi0 grains. Number refers to diamond which hosted the inclusion while the letter refers to the inclusion code. Major oxide chemistry and cation calculations for each inclusion are found in Tables 8.15 and 8.16 respectively. 3  4.3-D  Table 8.15. Major oxide data for MgSi0 grains (wt%) 3  Inclusion No. No.  Inclusion assemblage  No. of analyses averaged  P2O5  Si0  Ti0  2  2  A1 0 2  3  Cr 0 2  3  FeO M n O  NiO  MgO  CaO N a 0 2  K 0 2  Total  1-5A1  1  Ol-fPer-TAPP  3  0.00 57.31 0.16  1.91  0.21  6.34  0.13  0.00 33.91 0.06  0.00  0.00  100.04  1-5J1  2  3  0.00 54.92 0.17  2.22  0.20  6.15  0.13  0.00 35.68 0.04  0.00  0.00  99.52  3-211  3  2  na  58.19 0.20  1.60  0.22  4.21  0.14  0.00 33.43 0.04  0.00  0.00  98.02  2  na  51.87 0.16  1.94  0.29  4.23  0.14  0.00 39.37 0.04  0.00  0.00  98.04  3-2AD1  4  Ol-fPer-TAPP Ol-fPer-CaSiOjTAPP? 01-fPer-CaSi0 TAPP?  3-5G1  5  Ol-fPer  3  na  51.60 0.14  3.37  0.20  6.66  0.27  0.00 35.91 0.00  0.10  0.00  98.26  3-5Q  6  Ol-fPer  3  na  59.88 0.15  2.73  0.19  6.58  0.26  0.00 30.54 0.01  0.00  0.00  100.34  4-3D  7  Ol-fPer  3  0.00 55.36 0.21  2.54  0.20  4.29 0.10  0.00 36.30 0.00  0.07  0.00  99.08  4-3E  8  Ol-fPer  2  2.36  0.20  4.25  0.00 34.41 0.00  0.08  0.00  98.35  3  na  56.73 0.21  0.10  Any values below MDL (see Table 8.9) are replaced by 0.00. Table 8.16. Cation calculations for MgSi0 . 3  Inclusion assemblage  1-5A1  Ol-fPer-TAPP  0.000 0.985 0.002 0.039 0.003 0.091 0.002 0.000 0.869 0.001 0.000 0.000 1.992 0.91  1-5J1  Ol-fPer-TAPP  0.000 0.954 0.002 0.045 0.003 0.089 0.002 0.000 0.924 0.001 0.000 0.000 2.020 0.91  3-211 3-2AD1  0 1  " T ~pp? ° f  ,  p  5  n  5  +  a  s j 4 +  1  0  0  T j 4 +  80  0  0  3  A  0  0  ,  ^  Inclusion No.  3  3  3  +  0  0  0  M  F f i 2 +  3  0  0  6  1 0  0  n  0  2  +  2  N  0  0  j  0  2  0  +  C  0  8  6  3  0  0  0  a  2  1  +  N  0  0  0  a  0  +  0  K  0  0  0  +  ,  T  1  9  7  2  0  9  3  A  ° " ^5 !o TAPP? 1  f  i  r  l03  "  na  0.915 0.002 0.040 0.004 0.062 0.002 0.000 1.035 0.001 0.000 0.000 2.061 0.94  3-5G1  Ol-fPer  na  0.916 0.002 0.071 0.003 0.099 0.004 0.000 0.950 0.000 0.003 0.000 2.047 0.91  3- 5Q  Ol-fPer  na  1.019 0.002 0.055 0.003 0.094 0.004 0.000 0.775 0.000 0.000 0.000 1.950 0.89  4- 3D  Ol-fPer  4-3E  Ol-fPer  0.000 0.957 0.003 0.052 0.003 0.062 0.001 0.000 0.935 0.000 0.002 0.000 2.015 0.94 na  0.983 0.003 0.048 0.003 0.062 0.002 0.000 0.888 0.000 0.003 0.000 1.990 0.94  Cation totals are calculated on the basis of 3 anions, mg = Mg 7(Mg +Fe ) with typical errors of 0.012 at the 95% confidence level. Any values below MDL (see Table 8.10) are replaced by 0.00. 2  164  2+  2+  Major oxide data for eight MgSiC»3 grains from four diamonds are presented in Table 8.15 and cation calculations in Table 8.16. Mg ranges from 89.2 to 94.3 with an average o f 92.1.  The main substitutional elements are AI2O3 (1.60 - 3.37 wt%, average 2.33  wt%), T i 0 (0.14 - 0.21 wt %, average 0.17 wt%), C r 0 (0.19 - 0.29 wt%, average 0.21 2  2  3  wt%) and M n O (0.10 - 0.27 wt%, average 0.16 wt%). N i O is below detection (<0.05 wt%) and C a O contents are l o w (0.00 - 0.06 wt%).  8.3.1.3 CaSiOj  Twenty-seven grains o f C a S i 0 3 are found i n 12 separate diamonds (27% o f diamonds cracked and 17% o f the whole population). In the absence o f crystallographic data, these grains cannot be classified as C a S i 0 3 perovskites; they w i l l be simply referred to as CaSi03 inclusions. However, the presence o f these minerals as inclusions in diamond strongly supports the interpretation that they had, at least initially, the perovskite structure. A n y lower P-T polymorph o f CaSi03, other than CaSi-Prv, is not stable i n the mantle (such as wollastonite).  They are colourless to milky and range in size from 10 to 120 microns. Grains tend to fragment  once removed or exposed (Fig. 8.13, 4.7C), and cleavage tends to control  fragmentation  (Figs. 8.13, 3.2L and 4.7C). About one half o f the crystals are anhedral  (e.g. Figs. 8.13, 2.8L1, 3.1A1 and 7.1 A ) while the other half are euhedral (e.g. Figs. 8.13,  2.8-L1  3.1-Al  3.1-El  165  3.2-L  3.4-E  4.7-C  4.7 - D  4.7-H  4.10-G  3.10-X  4.7 - E  7.1-A  Fig. 8.13. SEM images of CaSi0 grains. 3  3 . 1 E 1 , 3 . 4 E , 4 . 7 D a n d H ) . E u h e d r a l grains m a y e x h i b i t faces that are a negative shape i m p o s e d o n the i n c l u s i o n b y the octahedral d i a m o n d host, h o w e v e r the friable nature o f the i n c l u s i o n s tend to obscure or not preserve p r i m a r y habit.  166  probability that these grains at one time possessed (and perhaps still possess) the perovskite structure, as w i l l be discussed i n section 8.4.1.2.  Grains are colourless and break easily once removed from their hosting diamond. M g S i 0 3 inclusions tend not to exhibit crystal faces as commonly as fPer grains. They may have an elongated ellipsoid shape (Figs. 8.12, 3 . 2 A D 1 , 3.211, and 4.3E), moderately well-formed euhedral shape but undistinguishable overall form (Figs. 8.12, 1.5A1, 3.5G1, 3.5Q and 4.3D), or no discernable form at all (Figs. 8.12, 1.5J1). There is evidence for imposed octahedral shape on a few grains (Figs. 8.12 1.5A1 and 3.5G1). Grain size ranges from 15 to 150 urn.  3.2-AD1  3.5-GI  3.5-Q  163  Fig. 8.12. SEM images of MgSi0 grains. Number refers to diamond which hosted the inclusion while the letter refers to the inclusion code. Major oxide chemistry and cation calculations for each inclusion are found in Tables 8.15 and 8.16 respectively. 3  20  10 um  um  4.3-E  4.3-D  Table 8.15. Major oxide data for MgSi0 grains (wt%) 3  Inclusion No.  Inclusion assemblage  a n a l y s e s  j averaged  p Q  5  Si0  2  Ti0  2  A1 0 2  3  Cr 0 2  FeO M n O N i O M g O CaO N a 0  3  2  K 0  Total  2  1-5A1  1  Ol-ffer-TAPP  3  0.00 57.31 0.16  1.91  0.21  6.34  0.13  0.00 33.91 0.06  0.00  0.00  100.04  1-5J1  2  3  0.00 54.92 0.17  2.22  0.20  6.15  0.13  0.00 35.68 0.04  0.00  0.00  99.52  2  na  58.19 0.20  1.60  0.22  4.21  0.14  0.00 33.43 0.04  0.00  0.00  98.02  2  na  51.87 0.16  1.94  0.29  4.23  0.14  0.00 39.37 0.04  0.00  0.00  98.04  3-211  3  3-2AD1  4  Ol-fPer-TAPP 01-fPer-CaSi0 TAPP? Ol-fPer-CaSiOr TAPP?  3-5G1  5  Ol-fPer  3  na  51.60 0.14  3.37  0.20  6.66  0.27  0.00 35.91 0.00  0.10  0.00  98.26  3-5Q  6  Ol-fPer  3  na  59.88 0.15  2.73  0.19  6.58  0.26  0.00 30.54 0.01  0.00  0.00  100.34  4-3 D  7  Ol-fPer  3  0.00 55.36 0.21  2.54  0.20  4.29 0.10  0.00 36.30 0.00  0.07  0.00  99.08  4-3E  8  Ol-fPer  2  2.36  0.20  4.25  0.00 34.41 0.00  0.08  0.00  98.35  3  na  56.73 0,21  0.10  Any values below MDL (see Table 8.9) are replaced by 0.00. Table 8.16. Cation calculations for MgSi0 . 3  Inclusion No.  Inclusion assemblage  p 5 +  g j 4 +  T j 4 +  A | 3 +  C l  .  3 +  ^  M  n  2  +  N j 2 +  C  a  2  +  N  a  +  K  +  J  M  1-5A1  Ol-fPer-TAPP . 0.000 0.985 0.002 0.039 0.003 0.091 0.002 0.000 0.869 0.001 0.000 0.000 1.992  1-5J1  Ol-fPer-TAPP  3-211  ^ " ^ ^ p f  ^  0.91  0.000 0.954 0.002 0.045 0.003 0.089 0.002 0.000 0.924 0.001 0.000 0.000 2.020 0.91  "  na  1.008 0.003 0.033 0.003 0.061 0.002 0.000 0.863 0.001 0.000 0.000 1.972 0.93  3-2AD1  °'" ^'S™ " TAPP?  na  0.915 0.002 0.040 0.004 0.062 0.002 0.000 1.035 0.001 0.000 0.000 2.061  0.94  3-5G1  Ol-fPer  na  0.916 0.002 0.071 0.003 0.099 0.004 0.000 0.950 0.000 0.003 0.000 2.047  0.91  3- 5Q  Ol-fPer  na  1.019 0.002 0.055 0.003 0.094 0.004 0.000 0.775 0.000 0.000 0.000 1.950  0.89  4- 3D  Ol-fPer  4-3E  Ol-fPer  f  0  3  103  0.000 0.957 0.003 0.052 0.003 0.062 0.001 0.000 0.935 0.000 0.002 0.000 2.015 0.94 na  0.983 0.003 0.048 0.003 0.062 0.002 0.000 0.888 0.000 0.003 0.000 1.990 0.94  Cation totals are calculated on the basis of 3 anions, mg = Mg 7(Mg +Fe ) with typical errors of 0.012 at the 95% confidence level. Any values below MDL (see Table 8.10) are replaced by 0.00. 2  164  2+  2+  Major oxide data for eight M g S i 0  3  grains from four diamonds are presented i n Table  8.15 and cation calculations i n Table 8.16. Mg ranges from 89.2 to 94.3 with an average o f 92.1. The main substitutional elements are AI2O3 (1.60 - 3.37 wt%, average 2.33 wt%), T i 0 (0.14 - 0.21 wt %, average 0.17 wt%), C r 0 (0.19 - 0.29 wt%, average 0.21 2  2  3  wt%) and M n O (0.10 - 0.27 wt%, average 0.16 wt%). N i O is below detection ( O . 0 5 wt%) and C a O contents are low (0.00 - 0.06 wt%).  8.3.1.3 C a S i 0  3  Twenty-seven grains o f C a S i 0  3  are found i n 12 separate diamonds (27% o f diamonds  cracked and 17% o f the whole population). In the absence o f crystallographic data, these grains cannot be classified as C a S i 0 perovskites; they w i l l be simply referred to as 3  C a S i 0 inclusions. However, the presence o f these minerals as inclusions in diamond 3  strongly supports the interpretation that they had, at least initially, the perovskite structure. A n y lower P-T polymorph o f C a S i 0 , other than CaSi-Prv, is not stable i n the 3  mantle (such as wollastonite).  They are colourless to m i l k y and range in size from 10 to 120 microns. Grains tend to fragment  once removed or exposed (Fig. 8.13, 4.7C), and cleavage tends to control  fragmentation (Figs. 8.13, 3.2L and 4.7C). About one half o f the crystals are anhedral (e.g. Figs. 8.13, 2.8L1, 3.1A1 and 7.1A) while the other half are euhedral (e.g. Figs. 8.13,  2.8-LI  3.1-Al  3.1-El  165  3.2-L  3.4-E  4.7-C  4.7-D  4.7-H  3.10-X  4.7-E  4.10-G  7.1-A  Fig. 8.13. SEM images of CaSi0 grains. 3  3.1E1, 3.4E, 4.7D and H ) . Euhedral grains may exhibit faces that are a negative shape imposed on the inclusion by the octahedral diamond host, however the friable nature o f the inclusions tend to obscure or not preserve primary habit.  166  Table 8.17. Major oxide chemistry for CaSi0 grains (wt%) 3  Inclusion No. No.  No.of analyses P 0 averaged  Inclusion assemblage  2  Si0  5  2  Ti0  A1 0  2  2  3  Cr 0, 2  FeO MnO N i O M g O CaO  Na 0 K 0  Total  2  2  2-8L1  1  fPer  3  na  52.59 0.07  0.00  0.00  0.42 0.00 0.00 0.00 46.61 0.16 0.14  99.98  3-1A  2  fPer  3  na  50.31 0.00  0.00  0.00  0.00 0.00 0.00 0.12 47.94 0.00 0.04  98.42  3-1E  3  fPer  3  0.00 52.11 0.21  0.21  0.00  0.28 0.29 0.00 0.08 45.50 0.10 0.00  98.79  3-2L  4  Ol-fPer-MgSiOjTAPP?  3  52.18 0.00  0.00  0.00  0.00 0.00 0.00 0.05 46.97 0.09 0.00  99.30  34E  5  4  0.00 49.15 2.13  0.26  0.00  0.00 0.00 0.00 0.05 47.15 0.00 0.00  98.74  4-7E  6  2  na  52.40 0.00  0.00  0.00  0.12 0.00 0.00 0.00 47.74 0.00 0.00 100.27  4-10G1  7  1  na  52.05 0.10  0.00  0.00  1.14 0.17 0.00 0.35 46.32 0.48 0.08 100.69  7-1A  8  0.00 51.91 0.00  0.00  0.00  0.08 0.00 0.00 0.00 47.94 0.00 0.00  eGrt  na  3  99.93  Any values below MDL (see Table 8.9) are replaced by 0.00.  Table 8.18. Cation calculations for CaSi0 grains 3  Inclusion No.  Inclusion assemblage  p5  2-8L1  fPer  na  1.012  0.001  0.000  0.000  0.007  0.000  0.000  0.000  0.961  0.006  0.005  1.992  3-1A  fPer  na  0.992  0.000  0.000  0.000  0.000  0.000  0.000  0.004  1.012  0.000  0.002  2.009  3-1E  fPer  0.000  1.013  0.003  0.005  0.000  0.005  0.005  0.000  0.002  0.947  0.004  0.000  1.984  3-2L  Ol-fPer-MgSiOjTAPP?  na  1.011  0.000  0.000  0.000  0.000  0.000  0.000  0.001  0.975  0.003  0.000  1.991  3^1E  0.000  0.967  0.032  0.006  0.000  0.000  0.000  0.000  0.001  0.994  0.000  0.000  1.999  4-7E  na  1.007  0.000  0.000  0.000  0.002  0.000  0.000  0.000  0.983 0.000  0.000  1.993  na  1.001  0.001  0.000  0.000  0.018  0.003  0.000  0.010  0.954  0.018  0.003  2.008  0.000  1.003  0.000  0.000  0.000  0.001  0.000  0.000  0.000  0.993 0.000  0.000  1.997  4-10G1 7-1A  eGrt  +  Si  Ti  4 +  4 +  Al  3 +  Cr*  Fe  2+  Mn  2 +  Ni  2 +  Mg  2 +  Ca  2 +  Na  +  r  Total  Cation totals are calculated on the basis of 3 anions. Any values below MDL (see Table 8.10) are replaced by 0.00.  Chemical data o f eight C a S i 0 grains from seven diamonds are presented i n Table 8.17, 3  with accompanying cation calculations i n Table 8.18.  Grains are essentially pure  C a S i 0 , with C a and S i occupying 97.3 - 99.9% o f the cation sites (average o f 99.1%). 3  167  In order o f decreasing average weight percent, substitutional elements are: T i 0  (0.00 -  2  2.13 wt%, average 0.31 wt%), F e O (0.00 - 1.14 wt%, average 0.26 wt%), N a 0 (0.00 2  0.48 wt%, average 0.10 wt%), M g O (0.00 - 0.35 wt%, average 0.08 wt%), M n O (0.00 0.29 wt%, average 0.06 wt%), A l 0 2  3  (0.0 - 0.26 wt%, average 0.06 wt%), and K 0 (0.0 2  0.14 wt%, average 0.03 wt%).  8.3.1.4 M g S i 0 2  4  Ten grains o f M g S i 0 4 have been identified by E D S analysis i n 6 diamonds (13% o f 2  diamonds cracked and 9% o f population).  In the absence o f crystallographic data and  supporting experimental data, these grains are difficult to further subdivide into a - O l , 001 or y - O l based on chemistry alone. For the sake o f conciseness, M g - S i oxides with a catiomanion ratio o f 3:4 w i l l be called 'olivine' (Ol), but the reader is reminded that these grains could have initially crystallised as one o f the three structures mentioned, or be the result o f a retrograde reaction.  Grains are colourless and generally small, ranging i n size from 10-100 microns. Crystal form is evident i n some grains (e.g. Figs. 8.14, 3.2S and 4.3C), but is typically poorly developed. Four grains o f olivine are i n direct contact with M g S i 0 3 inclusions.  Major oxide data are presented for seven olivine grains i n Table 8.19 with cation calculations in Table 8.20 and images i n F i g . 8.14. S i , M g and Fe fill 98.8 to 99.7% o f the cation sites available. The main substitutional elements are: A l 0 3 (0.00 - 0.80 wt%, 2  average 0.26 wt%), N i O (0.00 - 0.40 wt%, average 0.17 wt%), M n O (0.08 - 0.29 wf%, average 0.14 wt%) and C r 0 (0.00 - 0.28 wt%, average 0.04 wt%). Grains are M g - r i c h 2  3  but have a reasonably wide variation in mg, ranging between 0.88 and 0.95 (average o f 0.91).  Five O l grains from three diamonds cluster around mg = -0.89 while two O l  grains liberated from only one diamond have mg = -0.945.  168  1.5 - J 2  1.5-J4  3.2-W  3.2-AD2  4.3-C  4.3-Kl  F i g . 8.14.  3.2-S  3.5-G2  6.8 - A  S E M images o f Ol grains.  All grains but one are in association with fPer and MgSiCb. The mg of olivine does not seem to vary in accord with inclusion associations, rather, it varies between diamonds; olivine grains in diamond 4-3 have an elevated mg (0.95) while the olivine grain in diamond 3-5 has a lower mg (0.89), even though both diamonds have the same MgSiCV fPer-Ol association.  169  Table 8.19. Major oxide data for olivine grains (wt%) . . . inclusion  N  q  a s  i i ^y°"  No. of analyses P 0  e  2  Si0  5  Ti0  2  A1 0  2  2  Cr 0  3  2  FeO MnO N i O M g O CaO N a 0 K 0 Total  3  2  2  averaged 1-5J2  1  fPer-MgSiOj-TAPP  3-2S  2  0  ' " ^ ^ p f  3-2W  3  0  '~  3- 5G2  4  4- 3C  2  0.00 38.03 0.05  0.12  0.00  10.24 0.12 0.00 48.46 0.00 0.00 0.00 97.02  r  3  na  39.42 0.00  0.00  0.00  11.15 0.12 0.31 48.85 0.07  0.00 0.00 99.93  0 : r  3  na  41.38 0.00  0.00  0.00  10.90 0.08 0.40 45.22 0.00  0.06 0.00 98.05  fPer-MgSiO,  3  na  37.45 0.00  0.20  0.00  11.05 0.29 0.00 50.54 0.00  0.07 0.00 99.61  5  fPer-MgSi0  3  2  na  42.74 0.07  0.80  0.00  4.61  0.15 0.00 51.96 0.00  0.00 0.00 100.33  4-3K1  6  fPer-MgSi0  3  1  na  36.97 0.14  0.72  0.28  5.83  0.14 0.19 55.73 0.00  0.05 0.00 100.05  6-8A  7  0.00 43.43 0.00  0.00  0.00  8.85  0.12 0.25 43.29 0.06  0.05 0.00 96.07  l  ^p.f  f f  0  1  Any values below M D L  (see Table 8.9) are replaced by 0.00.  Table 8.20. Cation calculations for olivine grains Inclusion  I  n  c  ]  u  s  J  o  n  a  s  s  e  m  b  l  a  g  e  p  s  +  s j  4  +  T j  4  +  A  )  3  +  C  r  3  +  F  g  2  +  M  n  2  +  N  i  2  +  M  g  2  +  C  a  2  +  N  a  +  K  +  T  m  ]  m  g  1-5J2  fPer-MgSi0 -TAPP  3-2S  01-fPer-CaSi0 -TAPP?  na  0.977 0.000 0.000 0.000 0.231 0.003 0.006 1.805 0.002 0.000 0.000 3.023 0.89  3-2W  01-fPer-CaSi0 -TAPP?  na  1.036 0.000 0.000 0.000 0.228 0.002 0.008 1.688 0.000 0.003 0.000 2.965 0.88  3- 5G2  fPer-MgSi0  3  na  0.935 0.000 0.006 0.000 0.231 0.006 0.000 1.882 0.000 0.003 0.000 3.063 0.89  4- 3C  fPer-MgSi0  3  na  1.015 0.001 0.023 0.000 0.091 0.003 0.000 1.839 0.000 0.000 0.000 2.973 0.95  4-3K.1  fPer-MgSi0  3  na  0.902 0.003 0.021 0.005 0.119 0.003 0.004 2.026 0.000 0.002 0.000 3.084 0.94  6-8A  3  3  3  0.000 0.967 0.001 0.004 0.000 0.218 0.003 0.000 1.837 0.000 0.000 0.000 3.030 0.89  0.000 1.091 0.000 0.000 0.000 0.186 0.003 0.005 1.621 0.002 0.002 0.000 2.910 0.90  Cation totals .are calculated on the basis of 3 anions, mg = Mg /(Mg +Fe ) with typical errors of 0.012 at 2+  the 9 5 % confidence level. Any values below M D L  2+  2+  (see Table 8.10) are replaced by 0.00.  8.3.1.5 G a r n e t a n d T A P P  Six highly aluminous silicates were identified by EDS analysis from four diamonds. Only two of these grains were found before examination on S E M as they exhibited a distinctive pale orange hue (diamond 4-10, inclusions A and B). Two of the grains show evidence of crystal form (Figs. 8.15, 3.212, and 4.16E) (which again, maybe a result of  170  4.I0B  Fig. 8.15. SEM images and accompanying EDS spectra of aluminous silicate grains. Inclusions are likely either: TAPP (1.5A2, 1.5J2 and 3.212) or eclogitic garnet (4.10A, 4.10B and 4.16E).  diamond imposed morphology) while the remaining grains appear anhedral to subhedral. Grains range i n size from 5 to 120 microns. T w o o f the six grains are found i n contact with M g S i 0 3 (Figs. 8.15. 1.5A2 and 3.2 12) and one is i n contact with both MgSi03 and 01 (1.5J3).  Major oxide chemical data for three o f these grains are presented i n Table 8.21 along with cation calculations i n Table 8.22. A s there are several different types o f garnets or Table 8.21. Major oxide data for TAPP and eGrt grains (wt%) Inclusion No. No.  Inclusion assemblage  No. of analyses averaged  P O 2  s  Si0  2  Ti0  2  A1 0 2  3  Cr 0 2  3  FeO  MnO N i O  MgO CaO N a 0 K 0 2  2  Total  1-5A2  1  01-MgSi0 -fPer  3  19.03  2.74  6.87  0.14  0.00 25.75 0.04  0.00  0.00 99.26  4-1 OA  2  CaSiOj  4  na  40.04 0.92 21.52  0.00  17.15 0.32  0.00 10.39 8.87  0.17  0.00 99.38  4-1 OB  3  CaSiO,  2  na  36.98 0.91  0.00  16.65 0.30  0.00 11.30 9.02  0.14  0.00 98.35  3  0.00 39.99 4.71  23.04  Any values below MDL (see Table 8.9) are replaced by 0.00.  172  Table 8.22. Cation calculations for TAPP and eGrt grains Inclusion No.  Inclusion assemblage  1-5A2  01-MgSi0 -fPer  4-1 OA 4-1 OB  3  CaSi0 CaSi0  3  3  p5  +  Si  4 +  Ti  Total  mg  1.600 0.155 0.410 0.009 0.000 2.738 0.003 0.000 0.000 8.018  0.87  Al  4 +  0.000 2.852 0.252  Cr * 3  3 +  Fe  2+  Mn  2 +  Ni  2 +  Mg " 2  Ca  2 +  Na  K  +  +  na  3.016 0.052 1.911  0.000 1.080 0.021 0.000 1.167 0.716 0.025 0.000 7.988  0.52  na  2.829 0.053 2.077 0.000 1.065 0.019 0.000 1.289 0.740 0.020 0.000 8.091  0.55  Cation totals are calculated on the basis of 12 anions, mg = Mg /(Mg +Fe ) with typical errors of 0.02 at the 95% confidence level. Any values below MDL (see Table 8.10) are replaced by 0.00. 2+  2+  2+  aluminum-rich silicates, it is necessary to further classify these grains.  There are a  variety o f mantle garnets and aluminium-rich silicates, each with specific chemistries (previously introduced in section 8.1.4.3). A s well, the possibility that these grains are aluminous pyroxenes must also be considered.  Hutchison (1997) reports several pyroxenes with as much as 12.5 wt% AI2O3 (which he termed type in). The lowest AI2O3 total o f any o f these grains is 19.0 wt%; it is unlikely that these are aluminous pyroxenes.  A Cr203-CaO plot used to differentiate peridotitic from eclogitic garnets (Fig. 8.16).  Inclusion 1.5A2 plots in the  peridotitic, or, more specifically, harzburgitic 4.10A  field,  while  the  inclusions  and B plot i n the eclogitic field.  It should be noted here that mantle garnets found in diamonds are rarely void o f calcium. One highly aluminous 0.00  silicate void in calcium is T A P P , a rare inclusion that so far has only been  5.00  10.00  15.00  CaO  Fig. 8.16. Plot of CaO versus C r 0 (wt%) for aluminous silicates in this study. The lherzolitic field is from Sobolev et al. (1973) the 2% C r 0 cutoff for the eclogitic field is from Gurney (1984). 2  reported i n Juina diamonds (Harris et  3  2  al,  1997; Hutchison, 1997; Kaminsky  173  3  et al, 2001a). Based on E P M A and E D S analyses shared between inclusions without E M P A results, the aluminous silicates likely represent two phases: T A P P (1.5A2, 1.5J3 and 3.212) and eclogitic garnet (4.1 OA, 4.1 OB and 4.16E), or, more specifically, pyropegrossular-almandine garnet. 2.200  4.10B  Pyroxene begins to dissolve into garnet  - 4 1 0 k m (for both a pyrolitic  section  8.1.4.3.  A  4.10A  •  •  and  1.5A2  eclogitic mantle), as discussed i n detail in  •  2.000  at depths - 2 5 0 k m and is complete at  1.600  dissolved 1.400  pyroxene component can be recognised by determining the S i : 0 " ratio; any 4 +  garnet  with  a  ratio  considered majoritic.  >3.075:12  other garnets  1.200  2  is 2.700  F i g . 8.17 shows  that there is no dissolved pyroxene  2.800  2.900  3.000  majoritic garnets 3.100  3.200  3.300  3.400  Fig. 8.17. Plot of Al + Cr versus Si for aluminous silicates in this study. Cations calculated on basis of 12 oxygen atoms. The line at 3.075 cations Si is used to separate majoritic and nonmajoritic garnets. 3+  component i n the grains analysed.  3+  4  4+  The T A P P grain (1.5A2) is deficient i n Si  4 +  (2.852 cations per 12 oxygens) and contains modest amounts o f T i 0 2 (4.71 wt%) and  Cr 0 2  3  (2.74 wt%).  similarities  and  N i O , C a O , N a 0 and K 0 are essentially absent. 2  differences  between  analysed. The main differences are i n A l  2  the 3 +  There are  two pyrope-grossular-almandine  and S i  4 +  garnets  (i.e. Fig. 8.17). The remaining major  oxide constituents are similar: T i 0 (0.91 - 0.92 wt%), M n O (0.30 - 0.32 wt%) and N a 0 2  (0.14 - 0.17 wt%).  &2O3,  2  N i O and K 0 are absent i n both grains. The differences i n S i 2  and A l content between the grains bring into question the quality o f analyses. The large grain size coupled with better data for inclusion 4-1 OA (99.38 versus 99.31 wt% and 7.988 versus 8.091 cations per 12 oxygens) suggests that the data for inclusion B may be of poor quality.  174  8.3.1.6 Magnetite Magnetite grains have been confirmed in two diamonds, 1-4, and 2-6 (Fig. 8.18) and may also occur i n diamond 2-10. Twenty-two individual inclusions were identified by E D S analysis. They are typically euhedral and dark brown/black i n colour and range i n size from 10 to 60 microns. Magnetite belongs to the isometric crystal system and generally forms octahedral crystals.  Thus, the euhedral morphology o f inclusions i n this study  cannot be used to deduce whether or not the euhedral nature is a result o f diamond growth imposition, or is a growth feature.  Major oxide data are presented i n Table 8.23 and cation calculations i n Table 8.24. Grains comprise o f F e 0 2  3  (51.74 - 60.11 wt%, average 56.74 wt%) and F e O (28.33 -  30.80 wt%, average 29.69 wt%). F e / ( F e + F e ) ranges from 0.612 to 0.656 (average 3+  3+  2+  0.632) and was calculated using Formula (Ercit, T.S., 1996). M i n o r constituents include:  1.4-1  1.4-K  1.4 - M 2  1.4 - M 3  1.4-Ml  2.6-K  Fig. 8.18. SEM images of magnetite grains.  175  Table 8.23. Major oxide data for magnetite grains (wt%) Inclusion No.  No.  Inclusion assemblage  No. of analyses averaged  Si0  2  3  2  2  Ti0  MnO N i O  FeO  MgO CaO N a 0 K 0 2  Total  0.00  0.00  99.13  0.00  0.00  98.46  A1 0  0.16  1.50  5.56  0.00  60.11  28.33  0.61  0.00  2.85  0.00  0.16  2.36  4.75  0.00  58.21  30.80  0.58  0.00  1.46  0.15  2  3  Cr 0  Fe 0  2  2  2  3  2  3  2  1-41  1  Si0  1-4K.'  2  Si0  2-6B  3  1  0.17  3.61  6.92  3.03  51.74  29.48  0.62  0.00  3.58  0.00  0.00  0.00  99.15  2-6K  4  4  0.28  2.97  3.44  2.10  56.91  30.16  0.53  0.00  2.04  0.00  0.11  0.00  98.53  Any values below MDL (see Table 8.9) are replaced by 0.00. using Formula (Ercit, T.S., 1996).  Fe 0 and FeO contents calculated using 2  3  Table 8.24. Cation calculations for magnetite grains Inclusion No.  Inclusion assemblage  MI  Si0  2  0.006 0.042 0.241  0.000  1-4K  Si0  2  0.006 0.067 0.210 0.000  2-6B  0.006 0.098 0.295 0.087  2-6K  0.011  Si  4 +  Ti  4 +  Al  3 +  Cr  3 +  0.084 0.152 0.062  Total  mg  1.664  0.872 0.019 0.000 0.156 0.000 0.000 0.000 3.000  0.06  1.645  0.967 0.018 0.000 0.082 0.006 0.000 0.000 3.000  0.03  1.409 0.892 0.019 0.000 0.193 0.000 0.000 0.000 3.000 1.606 0.946 0.017 0.000 0.114 0.000 0.008 0.000 2.999  0.08  Fe  Fe  3+  Mn  2+  Ni  2 +  Mg  2 +  2 +  Ca  2 +  Na  +  0.04  Cations are calculated on the basis of 4 anions, mg = Mg /(Mg +Fe, ) with typical errors of 0.01 at the 95% confidence level. Any totals below MDL (see Table 8.10) are replaced by 0.00. Fe and Fe calculated using Formula (Ercit, T.S., 1996). 2+  2+  ot  2+  Ti0  2  3+  (1.50 - 3.61 wt%, average 2.61 wt%), M g O (1.46 - 3.58 wt%, average 2.48 wt%)  and C r 0 (0.00 - 3.03 wt%, average 1.28 wt%). 2  3  8.3.1.7 Sulphides  Twelve inclusions o f sulphide were found in three diamonds (6.8% o f diamonds cracked or 4.3% o f whole population). Grains are dark/black, very small (<30 microns i n size) and typically anhedral (with the exception o f one inclusion, 2.11G) (Fig. 8.19). Inclusion 6.8A2 was determined to be secondary after S E M examination, which revealed a fracture filled with sulphides leading up to the inclusion. E P M A data for five sulphide grains are presented in Table 8.25 with accompanying cation totals in Table 8.26. However, weight percent totals are generally poor, falling between 93.48 and 95.67. Poor totals are likely a consequence o f the heterogeneous nature o f the grains.  176  2.11 - P  6.8 - A 2 (bright section)  4.11-A  Fig. 8.19. SEM images of sulphide grains.  Grains are comprised o f essentially five elements (Fe, N i , C u , C o and S).  The  heterogeneous character o f the grains makes it difficult to describe them i n terms o f one mineral (they are likely monsulphide solid solution minerals). Nonetheless, a description o f the grains is warranted. For three o f the grains, Fe and S combine to make up between 95.7 and 100.0% o f the total mineral. Based on the Fe:S for these three sulphide grains  Table 8.25. Chemical data for sulphide grains listed as weight percent for each element (wt%). Inclusion type  Inclusion assemblage  No. of analyses averaged  1  pyrrhotite  Si0  2  pyrrhotite?  Si0  pyrrhotite  Si0  Inclusion No. No.  2-11A 2-111 2-1 IP 4-11A 6-8 A 2  3 4 5  pyrrhotite pentlandite  Ni  Cu  Zn  Mn  s  Total  0.23  3.58  0.21  0.00  0.00  35.53  95.42  0.28  14.34  6.18  0.00  0.00  30.93  95.21  56.01  0.19  2.68  1.19  0.00  0.00  35.03  95.10  2  59.90  0.00  0.00  0.01  0.00  0.00  33.58  93.48  3  25.16  0.38  35.06  0.04  0.00  0.00  35.02  95.67  Fe  Co  4  55.87  2  1  43.48  2  1  2  CaSiOj 01  MDL's and precision are listed in Table 8.9.  177  Table 8.26. Cation calculations for sulphide grains Inclusion type  Inclusion assemblage  Fe  Co  Ni  Cu  Zn  Mn  s  2-11A  pyrrhotite  Si0  2  0.903  0.004  0.055  0.003  0.000  0.000  1.000  1.964  2-111  pyrrhofite?  Si0  2  0.807  0.005  0.253  0.101  0.000  0.000  1.000  2.166  2-1 IP  pyrrhotite  Si0  2  0.918  0.003  0.042  0.017  0.000  0.000  1.000  1.980  0.000  0.000  0.000  0.000  0.000  1.000  2.024  0.006  0.547  0.001  0.000  0.000  1.000  1.966  Inclusion No.  4-11A  pyrrhotite  CaSiOj  1.024  6-8 A 2  pentlandite  01  0.412  Total  MDL's and precision are listed in Table 8.10.  (between 0.90 and 1.02), they should be classified as pyrrhotite. The secondary sulphide (6.8A2) contains abundant Fe and N i , where Fe + N i is slightly less than S. This grain should be classified as pentlandite.  The remaining grain, inclusion 2.111, is trickier to  categorise. The Fe:S ratio is too far from unity and contains too much N i (14.34 wt%) and C u (6.18 wt%) to be considered pyrrhotite.  Inclusion 2.111 remains unclassified  although it is likely close to pyrrhotite in composition as two other inclusions from diamond 2.11 (2.1 II and P) are pyrrhotite.  8.3.2 Inclusions of uncertain origin  There are several inclusions that fail to meet the criteria used i n this study for syngenesis. An  amorphous  flaky,  fine-grained  appearance,  anhedral form,  and  heterogeneous  composition are common characteristics used to distinguish inclusions o f secondary origin.  Grains i n this section tend to exhibit one or more o f these characteristics,  however, they also have some features which suggest they are primary. A s well, most o f the grains have been reported as inclusions i n diamonds from other localities worldwide. Inclusions described here are not necessarily secondary, but there is an added element o f uncertainty to their primary origin.  The inclusions are Si02, several Ca-rich minerals,  perovskite (CaTi03) and metallic iron.  178  8.3.2.1 SiQ  2  Twelve grains o f Si02 have been identified, found i n 11% o f diamonds cracked or 7% o f the total diamond population. When examined under S E M , all grains are polycrystalline and fine-grained, which is typical o f secondary minerals (Fig. 8.20).  However, this  appearance may be a result o f crystal expansion due to phase changes from stishovite to any or all o f coesite, P-quartz or a-quartz (e.g. F i g . 8.7).  Another reason for the  uncertainty o f the origin for these grains is the abundance o f SiC>2 found on fracture surfaces o f several cracked diamonds (Fig. 8.20 2.3-C).  Si02 i n fractures is clearly  secondary and demonstrates the existence o f secondary SiC>2 i n at least some o f the diamonds studied.  Although the twelve S i 0  inclusions  no  obvious  with connection  2  grains reported under are discrete  to  fractures or the diamond surface, there certainly is potential  for silica-rich  fluids to alter inclusions through  unrecognised  fractures.  There is no E P M A data for the Si02 grains found in  this study  although  E D S spectra suggest they are essentially pure S1O2. SiC>2  has  polymorphs  several which can  provide constraints on P T.  In the absence o f  1.4 - C  2.3-C  Fig. 8.20. SEM images of SiO . Images 2.11-C, 2.7-K and 1.4-C are discrete inclusions while 2.3-C is secondary Si0 deposited on fracture surface. z  2  179  crystallographic data, cathodoluminescence ( C L ) was found to be an effective tool for discriminating quartz from coesite (Sobolev et al., 1984). They found that quartz (it is unclear i f this is a-quartz or P-quartz) exhibits a pink or greyish yellow C L colour while coesite exhibits a bright blue C L colour.  C L o f inclusion 2.11C shows moderate yellow green colours suggesting the grain is quartz (Fig. 8.21C).  Inclusion 2.7K, however, does not exhibit any C L colours (Fig.  8.2IF) and could not be classified (it likely is not flush with the diamond surface and was thus not excited by electrons).  The reader is reminded that even i f these grains were  shown to be quartz, this does not exclude the possibility that they were coesite or stishovite at one time.  Fig. 8.21. Photographs of Si0 grains under UV light and in the absence of UV light. A-C - Inclusion 2.11C and D-F - Inclusion 2.7K. Photographs A, B, D and E show cloudy colour of Si0 inclusions. Photo C shows the greyish/yellow cathodoluminescence that is typical of quartz while in photo F, Si0 inclusion shows no CL, which is likely a result of the inclusion being somewhat recessed in the diamond host and thus not being activated by electrons. 2  2  2  180  8.3.2.2 Perovskite  Four separate grains o f perovskite (Prv) have been released from two diamonds. Grains are dark grey/black and do not appear to exhibit crystal form (except i n the case o f inclusion 3.7D2). They range i n size from 80 to less than 10 microns. A l l four grains found are touching CaSiCh inclusions (Fig. 8.22).  E D S spectra are included for all four grains (Fig. 8.22) while major oxide data are available for two grains (2.8L2 and 3-7A2, Table 8.27) and light rare earth element chemistry for only one grain (2.8L2).  Cation calculations are included in Table 8.28.  C a O and TiC>2 make up 92-95% o f the total weight percent (although there may be some elements missing from the totals for 3.7A2). Substitutional oxides include SiCh (1.92 2.06 wt%), A 1 0 2  (1.11 - 1.48 wt%) and F e O ( 0 . 1 7 - 1 . 1 3 wt%). Rare earth elements  3  ( R E E ' s ) and high field strength elements ( H F S E ' s ) make up the remainder o f the impurities present.  E D S spectra confirm the substantial amounts o f R E E ' s that are  common i n perovskites.  Peaks indicating the presence o f cerium are present i n three  E D S spectra (Fig. 8.22) (cerium is present although not so clearly discerned in 2.8L2, Table 8.27. Major oxide data for perovskite grains (wt%) Inchon No.  N  Q  .nCusion assemblage  . averaged  0  2- 8L2  1  CaSi0  3- 7A2  2  CaSiOj  3  S,0 T , G A , G C r 0 PeO CaO Y 0 Z r C L a G C e G P . O . N d 0 2  2  2  3  2  3  2  3  2  2  3  3  2.0652.41 1.48  0.00 0.17 38.78 0.13 0.60 0.49  1  1.9251.17 1.11  0.29 1.1336.15 na  na  2  3  2  3  S m 2  0  3  1.33  0.29  0.72  0.00  na  na  na  na  na  G d G Tola, 2  3  0.21 98.67 na  91.78  Any values below MDL (see Table 8.9) are replaced by 0.00. P, Mn, Ni, Mg, Na and K were analysed for both inclusions but were below MDL's and removed. Th, and Nb were below the MDL's in 2.8L2. Table 8.28. Cation calculations for perovskite grains Inclusion No.  Inclusion assemblage  2- 8L2  CaSi0  3  3- 7A2  CaSiQ  3  g  0  .  0  4 +  4  8  T  0  9  .  1  4 +  ]  A ] 3 +  0  0  4  &  0  0  0  3  0  ^  +  0  0  0  0  C a 2 +  3  0  y 2 +  ^  Z f 2 +  C ( ; 2 +  P r 2 +  N ( j 2 +  S m 2 +  G ( j 2 +  T o t a  ,  . 9 5 9 0.002 0.005 0.004 0.011 0.002 0.006 0.000 0.001 1.993  0047 Q 940 p p 2 Q.QQ6 0.023 0.946 3  na  na  na  na  na  na  na  na  1.994  Any values below MDL (see Table 8.10) are replaced by 0.00. P, Mn, Ni, Mg, Na and K were analysed for both inclusions but were below MDL's and removed. Th, and Nb were below the MDL's in 2.8L2.  181  182  which suggests other R E E ' s are also present).  U s i n g E M P A , inclusion 2.8L2 was  analysed and found to contain - 4 . 0 wt% o f various R E E and H F S E ' s .  In decreasing  order o f abundance, the top three oxides present are: Ce 03 (1.33 wt%), Nd203 (0.72) and 2  Zr0  2  (0.60 wt%).  Although 3.7A2 was not analysed for R E E ' s or H F S E ' s , it likely  contains at least the same amount (and likely more, based on the larger Ce peak), which would considerably increase the wt% oxide total from 91.78 (Table 8.27).  8.3.2.3 O t h e r calcium-bearing minerals  Four Ca-Si oxides with 'exotic' elements were found in two diamonds, and can be subdivided into two groups. The first group is Ca-Si-P oxides (two grains i n diamond 3.1 A ) and the second is C a - S i - T i - A l oxides (two grains found i n diamond 3-4, although it is possible that they are two halves o f the same inclusion). It is possible that these gains are altered CaSi03 inclusions, however, their homogenous appearance suggests they may be primary.  Inclusion 3.1E2 is likely secondary.  making E P M A analysis difficult.  Grains are small (<20 microns), thus  S E M images o f all grains, including E D S spectra, are  presented i n F i g . 8.23. Table 8.29. Major oxide data for 'exotic' Ca-Si-O grains (wt%) Inclusion No.  Inclusion assemblage  P 0  3-1A2  fPer-CaSiOj  10.85  9.58  0.02  0.00  0.00  3-1E2  fPer-CaSi0  0.10  27.85  5.35  12.03  0.46  3-4D  CaSi0  0.02  32.14  25.01  8.82  0.04  0.02  3  3  2  5  Si0  2  Ti0  2  Al 0 2  Cr 0  3  2  3  FeO  MnO NiO  MgO  CaO  Na 0  K 0 2  Total  0.00  0.00  0.00  0.00  46.63  0.00  0.00  67.09  4.52  0.69  0.03  5.56  25.51  0.33  0.00  82.44  0.00  0.00  0.00  28.57  0.00  0.01  94.64  r  Total  2  MDL's and precisions were not calculated for these inclusions.  Table 8.30. Cation calculations for 'exotic' Ca-Si-0 grains Inclusion No.  Inclusion assemblage  3-1A2  fPer-CaSi0  3  0.304  0.139  0.000  0.000  0.000  0.000  0.000  0.000 0.000  0.722  0.000  0.000  1.166  3-1E2  fPer-CaSi0  3  0.002  0.221  0.032  0.113  0.003  0.030  0.005  0.000 0.066  0.217  0.005  0.000  0.693  3-4D  CaSi0  0.000  0.217  0.127  0.070 0.000  0.000  0.000  0.000 0.000  0.207  0.000  0.000  0.621  3  P  5+  Si  4 +  Ti  4 +  Al  3 +  Cr  3 +  Cations calculated on basis of 1 oxygen anion.  183  Fe  2+  Mn  2 +  Ni  2 +  Mg * 2  Ca  2 +  Na  +  E P M A data for the four grains are poor (totals ranging from 67-94 wt%). However, the results are included here as diamond inclusions o f this composition have not yet been recorded (Table 8.29). E P M A Results for the two Ca-Si-P minerals are poor and w i l l not be considered further, however, the results for a C a - S i - T i - A l inclusion (3-4D), given the inclusion size, are acceptable and warrant further discussion.  K o i t o et al. (2000)  synthesized minerals i n the CaSi03 - CaTiC>3 solid-solution series and found that intermediate compositions were stable at pressures greater than 12 G P a (-400 km). Stoichiometry for this inclusion is close to the 2:3 catiomanion ratio for minerals o f this solid-solution series (1.864:3) (Table 8.30).  Experimental work mentioned does not  examine the effects o f aluminum, which make up a significant proportion o f inclusion 3.4D (8.82 wt%). Inclusions 3.1A2 and 3.1E2 can only be discussed i n terms o f E D S analysis. These Ca-Si-P oxides also include A l , Ce and Fe. The presence o f C e suggests that other rare earth elements likely exist (as was observed i n perovskites, section 8.3.2.2).  8.3.2.4 M e t a l l i c i r o n  One large grain (150 x 200 um) o f native iron was found in diamond 2-11 (Fig. 8.24). It appears dark black/metallic when viewed under the microscope.  E D S analysis shows  that the grain in comprised o f essentially pure iron, with no complexing anions, such as  3044 IS  2.1 IE  Ie  Fe  hi 1  0  Fe  1  g$  —'—| »0  ~.  _  _U—,  V  4.0  t  o  J L A 1  no  IO.I  keV  Fig. 8.24. SEM image and EDS spectrum of native iron grain (Diamond 2.1 IE). Peaks on EDS spectra on the far left (not labelled) correspond to Li (noise peak) and C (from the carbon coat).  185  oxygen or sulphur ( E D S spectrum i n F i g . 8.24).  S E M examination also shows that the  grain is homogeneous i n composition.  8.3.3 T o u c h i n g phases  The term 'assemblage' as applied to diamond studies is used to indicate a collection o f minerals  that coexisted  in equilibrium under  compositional conditions.  certain pressure, temperature  and  Unless there is evidence to suggest otherwise, primary  inclusions occurring i n the same diamond are considered to represent an assemblage. The assumption that a single diamond traps all o f its inclusions under the same conditions may or may not be valid; there are certainly instances indicating both scenarios occur. The difference i n chemistry from inclusions o f the same phase liberated from one diamond is evidence that dis-equilibrium exits.  A s well, cathodoluminescence images  (e.g. chapter 7.0) show that most diamonds do not grow during a single event.  Primary (pristine) touching inclusions remove the uncertainty o f equilibrium, however, unlike isolated monocrystalline inclusions, touching phases have the opportunity to reequilibrate with changing P - T conditions and to re-distribute elements between touching phases. A s such, the composite grain as a whole represents 'pristine' mantle material, but the individual phases may not reflect the chemical subtleties acquired during initial crystallization.  Table 8.31 summarises the touching phases observed in this study, which  Table 8.31. Touching phases and their associations for Rio Soriso diamonds Touching inclusions .. Assemblage  No. of  CaSiOj-Prv  2  2.8,3.7  2  3.5,4.3  fPer-01-MgSi0  ^  #  3  fPer-01-MgSi0 -TAPP  1  fPer-01-MgSi0 -CaSi0 -TAPP?  1  3  3  Diamond(s)  3  '  M  g  ^  2.8,3.7 3.5,4.3  1.5 3.2  ^  4.3 1.5  3.2  3.2  'Assemblage' refers to all inclusion phases in diamond (touching and non-touching).  186  1.5  _  w i l l be the focus o f the remainder o f this section; non-touching phases w i l l be considered further in the discussion on paragenesis.  Touching inclusions are divided into five subgroups:  1. Large M g S i 0 3 grains hosting small 01 inclusions. (3 occurrences) 2.  Large M g S i 0 3 grains i n contact with likely T A P P . (3 occurrences)  3.  01-fPer (1 occurrence)  4.  CaSi03 inclusions i n contact with exotic calcium-rich minerals. (~2 occurrences)  5.  Large M g S i 0 3 grain i n contact with both likely T A P P and two inclusions o f 01. (1 occurrence)  8.3.3.1 M g S i 0 and M g S i 0 composites 3  2  4  Three occurrences o f small 01 grains hosted i n large M g S i 0  3  inclusions were observed.  The inclusions are 1.5J2 ( O l on left) and 1.5J4 (01 on right) i n 1.5-J1 ( M g S i 0 ) (Fig. 3  8.25, 1.5-J), 3.5G2 (01) i n 3.5G1 ( M g S i 0 ) and 3 . 5 A D 2 (01) i n 3.5AD1 ( M g S i 0 ) . 3  3  Chemical data are available for two olivine inclusions (1.5J2 and 3.5G2) and all three M g S i 0 3 grains and accompanying photos are found i n F i g . 8.25.  Chemical data and  cation totals are presented in Tables 8.32 and 8.33 respectively. The mg's for 01 and  1-5-J  3.2-AD  3.5-G  Fig. 8.25. SEM images of touching olivine and MgSi0 inclusions. 3  187  Table 8.32. Major oxide data for touching inclusions of MgSi0 and olivine (wt%) 3  Inclusion N  o  Inclusion T  Inclusion assembla e  a  n  No. of a  ,  v  s  e  s  p  2°5  S  ° 2  l  T  l  ° 2  -HI /->!/-/-> T*I  .,,-./-,,-, r <-» > , r , ,-./-> C r 0 FeO MnO N i O MgO CaO N a 0 K 0 Total w  2°3  A 1  2  3  2  2  averaged 1-5J1  MgSiOj  1-5J2  Ol  3-5G1  MgSi0 B  Ol-fPerTAPPMgSiOj Ol-fPerTAPPMgSiOj Ol-fPerr." MgSi0  3  3  0.00 54.92 0.17 2.22 0.20 6.15 0.13 0.00 35.68 0.04 0.00 0.00 99.52  2  0.00 38.03 0.05 0.12 0.00 10.24 0.12 0.00 48.46 0.00 0.00 0.00 97.02  3  na 51.60 0.14 3.37 0.20 6.66 0.27 0.00 35.91 0.00 0.10 0.00 98.26  3  na 37.45 0.00 0.20 0.00 11.05 0.29 0.00 50.54 0.00 0.07 0.00 99.61  3  3-5G2  Ol-fPerw OVA MgSt0  Ol  3  Ol-fPer3-2AD1  MgSi0  S^n" TAPP?MgSiQ  3  '  na  2  51.87 0.16 1.94 0.29 4.23 0.14 0.00 39.37 0.04 0.00 0.00 98.04  3  Chemical data for the touching olivine inclusion 3-2AD2 does not exist.  Table 8.33. Cation calculations for touching inclusions of MgSi0 and olivine 3  inclusion No.  Inclusion Type  1-5J1  MgSi0  Inclusion assemblage  p  5  +  4 +  .  T  4 +  A  ]  3  +  &  3  ^  +  *  N f a 2 +  ^  C a 2 +  N  a  +  0  0  0  K  +  "  Ol-fPpr-TAPP-  1-5J2  3  Ol  3-5G1  MgSiOj  3-5G2  Ol  MgSi0  0  Ol-fPer-TAPP-  „ ' MgSi0  M  gS^0  0  0  0  3  °'  9 5 4  0  0  0  2  0  0  4  5  0  0  0  3  0  0  8  9  0  0  0  2  0  0  0  0  0  9  2  4  0  0  0  1 0  0  0  0  0  2  0  2  0  °-  9 1  0.000 0.967 0.001 0.004 0.000 0.218 0.003 0.000 1.837 0.000 0.000 0.000 3.030 0.89  3  "  3  Ol-fPer" IX MgSt0  a  0  ,  9  1  6  0  0  0  2  °'  0 7 1  0  0  0  3  0  0  0  4  0  0  0  0  0  9  5  0  0  0  0  0  0  0  0  3  0  0  0  0  2  0  4  7  °'  9  1  na  0.935 0.000 0.006 0.000 0.231 0.006 0.000 1.882 0.000 0.003 0.000 3.063 0.89  a  0.915 0.002 0.040 0.004 0.062 0.002 0.000 1.035 0.001 0.000 0.000 2.061 0.94  3  Ol-fPer3-2AD1  MgSiOj  TAPW" TAPP?MgSiQ  n  3  MgSi0 cations are calculated on the basis of 3 anions while olivine cations are calculated on basis of four anions. 3  MgSi03 are 0.89 and 0.91 respectively (in both composite inclusions 1.5-J and 3.5-G). The mg for the composite 3 . 2 - A D is only available for M g S i C h , and is considerably higher at 0.94. A l l three M g S i O j grains contain elevated amounts o f AI2O3 (1. 94 - 3.37 wt%), but this is within the range for the results o f other grains analysed i n this study.  188  The mg values reported for these two 'olivines' are the lowest recorded for O l ' s i n this study, but are not unique to grains touching M g S i 0 3 inclusions (the isolated 01 inclusion 3.2S also has wg=0.89). In all other respects, the 01 grains are similar to the five other grains analysed. The prevalence o f 01 inclusions inside M g S i 0 3 grains, coupled with the relative size difference between grains, may be an indication that these O l grains are the products o f a retrograde reaction between fPer and M g S i - P r v .  8.3.3.2 M g S i 0 a n d T A P P composites 3  There are two occurrences o f T A P P in contact with MgSi03 (inclusions 1.5A and 3.21) and one occurrence o f MgSi03 i n contact with T A P P and two 01 grains (inclusion 1.5J) (Fig. 8.26). Major oxide data and cation calculations for these composites are presented in Tables 8.34 and 8.35 respectively.  The mg for both M g S i 0 3 inclusions i n diamond 1-5 (inclusions 1.5A and J) is 0.91, while the T A P P grain (1.5A2) has mg=0.87. There is no E P M A data for T A P P inclusion 1.5J3. The M g S i 0 3 inclusion in diamond 3-2 has a higher mg (=0.93) and there are no E P M A data for the suspected touching T A P P phase. grains are similar in all other respects.  1.5-A  The chemistry data for the three M g S i 0 3  Inclusion 1.5J2 is the only T A P P grain with  1.5 - J  3.2-1  Fig. 8.26. SEM images of touching MgSi0 and TAPP inclusions. 3  189  Table 8.34. Major oxide data for touching MgSi0 and TAPP inclusions (wt%) 3  No. of analyses P20 averaged  Inclusion No.  Inclusion Type  Inclusion assemblage  1-5A1  MgSi0  Ol-MgSiOjfPer-TAPP  3  0.00 57.31 0.16  1-5A2  TAPP  01-MgSi0 fPer-TAPP  3  0.00 39.99 4.71 19.03 2.74 6.87 0.14 0.00 25.75 0.04  I-5J1  MgSi0  01-MgSi0 fPer-TAPP  3  0.00 54.92 0.17  2.22  0.20  Ol-fPerCaSi0 MgSi0 TAPP?  2  58.19 0.20  1.60  0.22 4.21 0.14 0.00 33.43 0.04 0.00 0.00 98.02  3-211  3  3  3  3  MgSiOj  3  5  na  Si0  2  Ti0  A1 0 C r 0 2  2  1.91  2  3  FeO MnO N i O MgO CaO N a 0 K 0 2  3  2  Total  6.34 0.13 0.00 33.91 0.06 0.00 0.00 100.04  0.21  0.00  0.00 99.26  6.15 0.13 0.00 35.68 0.04 0.00 0.00 99.52  3  There is no supporting chemical data for the TAPP inclusions touching grains 1-5J1 and 3-211 Table 8.35. Cation calculations for touching MgSi0 and TAPP inclusions 3  Inclusion No.  Inclusion Type  Inclusion assemblage  1-5A1  MgSi0  ^"^f^pp" 0.000 0.985 0.002 0.039 0.003 0.091 0.002 0.000 0.869 0.001 0.000 0.000 1.992 0.91  1-5A2  TAPP  1-5J1  MgSi0  3-211  MgSiO,  6  3  s j 4 +  T j 4 +  A ] 3 +  &  3  +  ^  M  n  2  +  N j 2 +  M  2 +  Q & 2 +  ^  +  K  +  ^  °  ^fPer-TAPP " | 2r  0.000 2.852 0.252 1.600 0.155 0.410 0.009 0.000 2.738 0.003 0.000 0.000 8.018 0.87  * f'!?'' fPer-TAPP  0.000 0.954 0.002 0.045 0.003 0.089 0.002 0.000 0.924 0.001 0.000 0.000 2.020 0.91  ,  M S 1 A  M  3  p 5 +  Ol-fPer-  S o'S " a  3  MgSi0 TAPP?  na  '-008 0.003 0.033 0.003 0.061 0.002 0.000 0.863 0.001 0.000 0.000 1.972  0.93  3  MgSi0 and TAPP cations are calculated on the basis of three and twelve anions respectively. 3  acceptable E P M A data, thus excluding the possibility o f comparison with other grains i n this study.  8.3.3.3 Mg2Si0 and ferropericlase composite 4  Diamond 4-3 contains one occurrence o f a small composite (-30 microns i n size) o f olivine and ferropericlase (inclusion K ) (Fig. 8.27). Only one acceptable E P M A result was collected from the O l ( K I ) and two were collected for the fPer (K2) grain (Tables 8.36 and 8.37). The mg o f the fPer and O l grains is 0.83 and 0.94 respectively. The O l inclusion is similar i n most respects to another olivine, which was i n isolation, liberated from  diamond 4-3 (inclusion 4-3C).  The main differences being inclusion 4-3K1  190  Table 8.36. MajoT oxide data for touching grains of ferropericlase and olivine (wt%) No. of analyses P 0 averaged  Inclusion No.  Inclusion Type  Inclusion assemblage  4-3K1  01  Ol-fPerMgSiO,  1  na  36.97 0.14  0.72  0.28  4-3K2  fPer  Ol-fPerMgSi0  2  na  0.26  0.42  1.01  2  Si0  5  Ti0  2  A1 0  2  2  0.00  3  Cr 0 2  FeO M n O N i O M g O CaO N a 0 K . 0 Total  3  2  2  0.14 0.19 55.73 0.00 0.05 0.00 100.05  5.83  25.96 0.26 1.35 70.83 0.00 0.00 0.00 100.08  3  Table 8.37. Cation calculations for touching grains of ferropericlase and olivine Inclusion No.  Inclusion Type  Inclusion assemblage  4-3K.1  01  Ol-fPerMgSi0  na  0.902 0.003 0.021 0.005 0.119 0.003 0.004 2.026 0.000 0.002 0.000 3.084 0.94  01-fPerMgSiQ  na  0.020 0.000 0.038 0.061 1.656 0.017 0.083 8.056 0.000 0.000 0.000 9.931  p  5  +  S j  4  +  T j  4+  A  ,  3  +  C  ]  .  3  t  p 2+ e  M  n  2  +  N  j  2  +  M  g  2  +  C  a  2  +  N  a  +  K*  Total  mg  3  4-3K2  fPer  0.83  3  contains 0.19 wt% N i O and 0.28 wt% C r 0 , which are 2  3  absent or below detection i n inclusion 4-3C. Compared to the remaining olivines i n this study, 4-3K1 is the only O l grain to contain detectable chromium and it also contains elevated amounts o f titanium (0.14 wt% Ti02). Compared to the rest o f the population, both olivines liberated from diamond 4-3 contain elevated amount o f aluminium (0.72 - 0.80 wt% A 1 0 ) . The mg o f O l does 2  3  not represent the most extreme value, but is close to the upper limit o f mg's for 'olivines' in this study.  The  Fig. 8.27. SEM image of touching olivine and ferropericlase inclusion (Diamond 4 . 3 - K ) .  chemistry o f the fPer is unremarkable, with mg and all wt% totals falling within the ranges for all fPer grains analysed.  The possibility o f the O l being the result o f a  retrograde reaction between fPer and M g S i - P r v must be considered, but the similarity i n chemistry with the isolated O l in diamond 4-3 makes this conclusion less likely.  191  8.3.3.4 M g S i 0 , M g S i 0 3  2  4  a n d T A P P composite  Inclusion 1.5J is the only composite with more than two touching phases found. This composite is composed o f T A P P (J3), M g S i 0 ( J l ) and two 01 grains (J2 and J4) 3  (Fig. 8.28).  There appears to be a triple junction  between one o f the 01 grains and the M g S i 0 3 and T A P P . Tables  8.38  and  8.39  contain  chemical data  for  inclusions J l (MgSiOa) and J2 (01), as well as A l ( M g S i 0 ) and A 2 ( T A P P ) .  Chemical data are available  3  only for M g S i 0 3 and the 01 grain (J2) not in contact with T A P P . M g ' s for M g S i 0 respectively.  Both  3  and 01 are 0.91 and 0.89  01 (J2) and M g S i 0  (Jl)  3  Fig. 8.28. SEM images of threephase composite (Inclusion 1.5-J) Individual phases are: Ol (J2 on left, J4 on right), MgSi0 (Jl) and TAPP (J3). 3  are  unremarkable in terms o f chemistry, with all E P M A wt% totals falling within the ranges measured for the complete sets o f inclusions. Although there is chemical data for another T A P P grain hosted in the same diamond (inclusion 1.5A2), its chemistry may not reflect that o f 1.5J3. A s both T A P P grains are in contact with MgSi03, a comparison between Table 8.38. Major oxide data for inclusions in diamond 1-5 (wt%) Inclusion Inclusion No. Type  1-5A2  TAPP  1-5A1  MgSi0  Inclusion assemblage Ol-fPerTAPPMgSiO,  3  Ol-fPerTAPPMgSi0 Ol-fPerTAPP-  No. of analyses P2O5 averaged  Si0  2  Ti0  A1 0 2  2  3  Cr 0 2  3  19.03 2.74  FeO  MnO  N i O M g O CaO N a 0 K 0 2  Total  6.87  0.14  0.00 25.75 0.04  0.00  0.00  99.26  2  3  0.00 39.99 4.71  3  0.00 57.31 0.16  1.91  0.21  6.34  0.13  0.00 33.91 0.06  0.00  0.00 100.04  3  0.00 54.92 0.17  2.22  0.20  6.15  0.13  0.00 35.68 0.04  0.00  0.00  99.52  2  0.00 38.03 0.05  0.12  0.00  10.24 0.12  0.00 48.46 0.00  0.00  0.00  97.02  3  1-5J1*  MgSiOj  MgSiO, 1-5J2*  01  01-fPerTAPPMgSi0 3  The three-phase touching inclusion 1.5J contains Ol (2 grains), TAPP and MgSi0 . Only chemical data for the MgSi0 and one Ol exist (1-5J1 and 1.5J2 respectively). Chemical data for the two-phase touching inclusion 1.5 A of MgSi0 and TAPP that is found in the same diamond is included for comparison. 3  3  3  192  Table 8.39. Cation calculations for inclusions in diamond 1-5 Inclusion Inclusion No. Type  1-5A2  Inclusion assemblage 01-fPerTAPPMgSi0 01-fPerTAPPMgSi0 Ol-fPerTAPPMgSi0 01-fPerTAPPMgSi0  TAPP  P  5 +  Si  4 +  Ti  4 +  Al  Cr  3 +  3 +  Fe  2+  Mn  2 +  Ni  2 +  Mg  2 +  Ca  2 +  Na  +  K  +  Total  mg  0.000 2.852 0.252 1.600 0.155 0.410 0.009 0.000 2.738 0.003 0.000 0.000 8.018 0.87  3  1-5A1  MgSiOj  0.000 0.985 0.002 0.039 0.003 0.091 0.002 0.000 0.869 0.001 0.000 0.000 1.992 0.91  3  1-5J1  MgSiOj  0.000 0.954 0.002 0.045 0.003 0.089 0.002 0.000 0.924 0.001 0.000 0.000 2.020 0.91  3  1-5J2  Ol  0.000 0.967 0.001 0.004 0.000 0.218 0.003 0.000 1.837 0.000 0.000 0.000 3.030 0.89  3  these two M g S i 0 3 inclusions is warranted. less S i 0  2  1.5J1 contains more AI2O3 (2.22 vs. 1.91),  (54.92 vs. 57.31) and more M g O (35.68 vs. 33.91) than inclusion 1.5A1.  Because o f these differences i n chemistry, the chemical data for T A P P inclusion 1.5A2 may be different from the chemical data for the T A P P inclusion 1.5J3 despite the similarities i n E D S spectra (Fig. 8.15). The mg o f T A P P 1.5A2 is 0.87.  8.3.3.5  CaSK>3 a n d C a - r i c h m i n e r a l composites  There are six occurrences o f composite grains i n contact with C a S i 0 3 .  They can be  subdivided into three groups:  1. C a S i 0 and C a T i 0 . (4 occurrences, 2.8L, 3.7A, D and E) 3  3  2.  CaSi03 and a C a , S i , A l , P Ce, Fe oxide. (1 occurrence, 3.IE)  3.  C a S i 0 3 and a C a , P, S i , Ce oxide. (1 occurrence, 3.1 A )  These composite grains have a high degree o f uncertainty attached to their pristine nature. They are small, i n some case heterogeneous, and typically anhedral (and therefore fail to meet some o f the criteria for primary inclusions).  However, they are included here  because some pass all the criteria and there is merit i n presenting them together. Group 1 occurs i n two diamonds and appears most likely to be primary. T w o grains are quite  193  2.8-L  3.7-A  3.7-D  3.7-E  3.1-A  3.1-E  Fig. 8.29. SEM images of composite grains of CaSi0 and 'exotic' Ca-rich phases. Inclusions 3.1A2 and 3.1E2 were analysed using EDS and are a Ca,P,Si oxide and a Ca,Si,P,Al oxide respectively. 3  large and each phase is homogeneous (2.8L and 3.7A). In the case o f inclusion 3.7A, the grains did not separate easily as is common for grains which have experienced alteration. Group 2 is most likely secondary, as E P M A results were poor for the 'exotic' phase and the S E M image shows that it is heterogeneous (Fig. 8.29 3.IE). Table 8.40 is a summary of E D S and E P M A results.  194  Table 8.40. EMPA and EDS data for composite grains of CaSi0 and other Ca phases 3  averaged 2-8 LI  CaSiOs  3  na  52.59 0.07  2-8L2  CaTi0  3  3  na  2.06 52.41 1.48  0.07 0.17 38.78 0.13 0.60  3-1 A l  CaSi0  3  3  na  50.31 0.03  0.09  0.02 0.08 47.94  3-1A2  Ca,P,Si,0?  1  10.85 9.58  0.02  0.00  0.00 0.00 46.63  3-1 E l  CaSiOj  3  0.14 52.11 0.21  0.21  0.01 0.28 45.50  3-1E2  Ca,Si,PAl?  1  0.05 47.58 1.43  3.80  0.10 2.35 38.66  3-7A1  CaSiOj  1  0.00 49.19 0.05 0.01  0.03 0.50 36.15  3-7A2  CaTiOs  1  3-7D1  CaSi0  3  3-7 D2  CaTi0  3  3-7E1  CaSiOj  3-7 E2  CaTiC-3  na  0.01  1.92 51.17 1.11  0.49  1.33  0.29  0.72  0.11  0.21  98.98 98.47  +  67.08 98.47  +  93.97 85.92  +  0.29 1.13 36.15  +  91.78  + +  1  99.70  0.00 0.42 46.61  0.02 45.41 0.08 0.02  +  +  +  +  0.00 0.42 44.41  +  90.37  +  +  '+' indicates the element was detected using EDS. Blank entries for major oxides (P O through CaO) indicate phase was not analysed using EPMA. Blank entries for oxides Y O through G d 0 indicate they were not detected using EDS. 2  2  8.3.4  s  s  2  3  F e - N i blebs on ferropericlase  Blebs o f F e - N i alloy were observed on the surfaces o f several ferropericlase grains released from two diamonds (3-5 and 3-10). The blebs are small, generally less than 2 microns i n size, have an elliptical morphology, and tend to form i n lines. E D S analysis shows that they consist essentially o f iron and nickel alloy. However, most analyses are likely contaminated to varying degrees by their ferropericlase hosts, thus introducing M g and O, among other elements, into the results. The E D S analysis for inclusion 3-10W2 (Fig. 8.30) is considered to be the most representative o f the blebs.  Fractured  ferropericlase grains in diamonds 3-5 and 3-10 show that the blebs only develop on the crystal surface (it is also possible that they are too small to resolve on fractured surfaces, however, i f they do exist on fracture surfaces, they are proportionally minor compared to surface blebs).  Ferropericlase grains released from several other diamonds appear to  have small (<1 urn) blebs as well, however they were too small to analyse with E D S (e.g.  195  Fig. 8.30. SEM images and EDS spectra for Fe-Ni blebs on ferropericlase gains. Blebs are Fe-Ni-rich and occur as moderately aligned elliptical blebs, only on the inclusion surface.  Figs. 8.31 1-2H, 1-5F and 6-2A). A l l blebs were too small for E P M A , however, several results from ferropericalse grains released from diamond 3-5 contain anomalously high iron and nickel contents as a result o f overlap with F e - N i blebs (Tables 8.13 and 8.14).  196  3.10-H  6.2-A  1.2-H  1.5-F  Fig. 8.31. SEM images of Fe-Ni blebs and linear features on ferropericlase grains. Linear feature on inclusion 1.5F were too small for EDS analysis.  8.3.5 Inclusions o f secondary origin  Inclusions were classified as secondary based on several criteria. The possibility o f a secondary origin was suspected during inclusion extraction and mounting when the inclusion appeared amorphous when released from the diamond host, was flaky, had a tendency to crumble, or was reddish i n colour.  There is certainly uncertainty as to  whether an inclusions' amorphous nature is a result o f crystal structure inversion (as was hypothesized for SiCh inclusions, section 8.3.2.1) or as a result o f alteration. However, after further examination using the S E M , it was usually possible to resolve this issue. Under the S E M , secondary inclusions typically appear heterogeneous (with the exception o f primary sulphides, section 8.3.1.7), have rough crystal faces with irregular 'holes' or pits or have no crystal faces at all.  197  8.3.5.1 A l t e r e d ferropericlase grains  Altered ferropericlase grains were observed in a few diamonds.  Once  inclusions  were  released  from  diamond, they were easy to recognize as they were difficult  to move without destroying and they were  typically reddish i n colour (Fig. 8.32).  Diamond 3-10  contained a number o f altered fPer grains (Fig. 8.33). From this figure, the general progression from primary to secondary can be seen.  E P M A data is not available  for the highly altered grain in Fig. 8.33, but based on the  8- - hoto of altered ferropericlase grain (Inclusion 3 9A). Fi  8  3 2  p  E D S spectra (and E P M A data for the pristine fPer's from diamond 3-10) it appears that alteration increases the Fe, S i , A l and C r content o f fPer gains while lowering the M g content.  Fig. 8.33. S E M images of weathered ferropericlase grains. These images illustrate the progression of alteration of fPer grains, from least altered on the left, to most altered on the right, (inclusions released from diamond 3-10).  8.3.5.2 L o c a l oxidation o f ferropericlase grains Twelve grains o f fPer were found in diamond 3-6, o f which several grains contained small (<5 um) spots that appeared bright under S E M and were restricted to particular domains on the inclusion surface (Fig. 8.34). E D S analysis o f these spots indicate that  198  they are mainly Fe-oxide with minor M g . E P M A o f several  fPer grains  from  this diamond yielded mg values ranging from 0.540.61.  One Fe-rich value  was poor (90.92 wt%) and was recalculated assuming  3.6-K  3.6 - L 3.6P2  the presence o f both ferric and ferrous 8.41  iron.  lists the  Table  oxide  values  and  Table  8.42  lists  the  for  cation  this analysis.  Fe  oxygen  e  Fe  1 JYSL*  values  2,»  11.11  t.u  1 11  When  anions,  LA  M  1U 1  keV  Fig. 8.34. SEM images and EDS spectrum of secondary magnesioferrite spots on ferropericlase (from Diamond 3.6). EDS spectra of spot from inclusion 3-6P.  calculated on the basis o f four  IS  reassigned  major  calculated  iiJ  F  the  sum o f calculated cations is  three.  The iron-rich spots are thus interpreted  to  magnesioferrite solid solution series in the spinel group.  fall  along the  magnetite-  The formula o f these blebs  would be: (Feo.43Mgo.49Cao.06Nao.01) (Fei.aoAlo.oiXV These features w i l l be referred to as magnesioferrite spots.  Magnesioferrite spots are different from the primary inclusions o f magnetite identified i n section 8.3.1.6 as they contain significantly more magnesium (8.96 wt% M g O compared to 1.46-3.58 wt% M g O for the primary magnetite grains). A s well, these secondary spots contain little or no T i 0 , A 1 0 , C r 0 , or M n O . 2  2  3  2  3  199  Table 8.41. Major oxide data for secondary magnesioferrite spots on ferropericlase grain 3-6P (wt%) Inclusion No.  Inclusion assemblage  No. of analyses y averaged  3-6P  fPer  1  a n a |  SL  s e s  °2 T1O2  0.18  A1 0  0.00  2  Cr 0  3  2  0.24  Fe 0  3  2  O.OQ  FeO  3  72.68  MnO NiO MgO CaO N a 0 K 0 2  14.17  0.12  0.00  8.96  1.55  2  0.18  0.00  Total 98.08  Fe 0 and FeO values determined using Formula (Ercit, T.S., 1996). 2  3  Table 8.42. Cation calculations for secondary magnesioferrite spots on ferropericlase grain 3-6P Inclusion No.  Inclusion assemblage  3-6P  Ti * 4  fPer  0.007  Al * 3  Cr  3 +  0.000 0.010 0.000  Fe  3+  1.989  Fe  21  0.431  Mn  Ni  2 +  0.004  2 +  Mg * 2  0.000 0.486  Ca  2 +  Na  +  0.060 0.013  K  +  0.000  Total  mg  3.000  0.53  „3+ 3+  Fe and Fe values determined using Formula (Ercit, T.S., 1996). 2+  8.3.5.3 A l t e r e d C a - r i c h grains  Heterogeneous grains are generally  interpreted  as  being secondary in origin, although  this  necessarily sulphides 4.3.1.7). two their  is  not  true (see  for section  S E M images o f  grains clearly reveal secondary  nature  Fig. 8.35. SEM images of weathered Ca-rich grains. Inclusion 3.4A (left) and inclusions 3.4B (right).  (Fig. 8.35). Although little can be said about these grains, they are included here for the purpose o f illustration.  They were likely C a S i 0  3  grains (one grain o f CaSi03 was  identified i n diamond 3-4) before they were exposed to external fluids, likely through a small fracture in the diamond. The relative timing o f alteration is unknown, it could have occurred in the mantle, during transport to the surface, or i n the secondary collector at surface. Because o f the uncertainty surrounding secondary inclusions (i.e. what they say about their primary compositions and the relative timing o f alteration), they are generally ignored i n diamond studies.  200  8.4 Discussion  8.4.1 Comparison with diamond inclusions from other studies  In order to properly evaluate the results from examination o f mineral inclusions i n this study, an i n depth comparison with other diamond inclusions is warranted.  More  specifically, comparison with other diamond inclusions should further help differentiate 'typical' upper mantle peridotitic and eclogitic phases from inclusions with a  deeper  origin.  8.4.1.1 Ferropericlase  There is an increasing amount o f published data on ferropericlase inclusions from localities all over the world (Table 8.43). From this table, it is striking that fPer grains from Juina, Central South America (including Sao Luiz) have a distinctly different mg range and average (Fig. 8.36). Juina fPer grains cover essentially the complete range i n mg from 0.36-0.89 with an average o f 0.68. The range o f mg's from all other localities is 0.75-0.94 with an average o f 0.86 (although it is important to note that the vast majority o f data for all other localities is from Guinea, Western Africa). spectroscopy, M c C a m m o n et al. (1997) found that F e  3 +  Using Mossbauer  is low ( F e / I F e t < 7%) for 3+  to  ferropericlase grains extracted from Juina diamonds. Although no quantitative analytical technique to specifically analyse F e  3 +  has been used on ferropericlase i n this study,  weight percent oxide values are close to 100.00 (97.94-101.69 wt%, average 99.75 wt%), thus suggesting iron occurs mostly as F e . 2 +  Stachel et al. (2000b) observed a positive  correlation between N a 0 and CrC>3 and explained this correlation as being due to 2  2  coupled substitution. This coupled substitution is not observed i n this study as there is considerable scatter when N a 0 wt% is plotted against Cr 03 wt%. Aside from the few 2  2  analyses i n this study that include F e - N i contamination (section 8.3.5), the chemistry o f fPer grains recovered i n this study is similar to that previously reported.  201  Table 8.43. Published data on ferropericlase diamond inclusions G e n e r a l region Central South A m e r i c a  Northern South A m e r i c a Certral N o r t h A m e r i c a  Locality  No. of grains  Range o f mg  H u t c h i s o n (1997)  36  0.36-0.85  0.71  Juina, B r a z i l  28  0.49-0.83  0.66  this study  R i o Soriso, B r a z i l  57  0.45-0.89  0.67  121  0.36-0.89  0.68  K a m i n s k y et ai,  (2000)  D a v i e s et ai,  (1999)  G u a n i a m o , Venezuela  1  0.88  Sloan, U S A  1  0.88  .  D O - 2 7 , Canada  7  0.80 - 0.87  0.85  H u t c h i s o n (1997)  Guinea  2  0.87  0.87  S t a c h e l e r al., (2000b)  Guinea  44  0.75-0.94  0.86  46  0.75-0.94  0.86  Stachel etal., (1998)  Southern A f r i c a  M c D a d e and H a r r i s (1999) Scott-Smith et ai,  (1984)  K o p y l o v a et al., (1997)  Scott-Smith et al., (1984)  M w a d u i , Tanzania,  1  Letseng-la-Teria, Lesotho  1  Koffiefontein, South A f r i c a  2  River Ranch, Zimbabwe  1  Orroroo, A u s t r a l i a  total  8.4.1.2 M g S i 0  mg  Sao L u i z , B r a z i l  East A f r i c a  Australia  Average  K a m i n s k y et al., (2001 a)  Otter and G u r n e y (1989)  Northern N o r t h A m e r i c a  West A f r i c a  Author  0.86 0.89 0.86-0.87  0.87 0.85  4  0.85-0.89  0.87  2  0.86-0.87  0.87  183  0.36-0.94  0.74  3  Table 8.44 is a compilation o f sources for data on M g S i 0 3 diamond inclusions that are interpreted as having initially crystallized as M g S i - P r v .  Other data may exist, but  because grains can be confused for upper mantle Opx, they may have been overlooked (e.g. a grain from M w a d u i , Tanzania, as discussed i n Stachel et al., 2000b).  A s previously introduced i n section 8.1.4.1, there are usually three chemical differences between M g S i 0  3  grains that crystallized i n the lower and upper mantle: N i and A l  contents, and mg.  202  Fig. 8.36. Plot of Fe versus M g for ferropericlase grains from Juina, Brazil and Guinea, West Africa. Cation totals are calculated on basis of ten oxygens. Fe calculated all as Fe (cation totals were recalculated for all published data for consistency). Numbers 0.40 to 0.90 on the upper right side of the data indicate mg. Rare data from other locales not included (Table 8.43) plot between mg 0.80 and 0.89. 2+  2+  2+  Table 8.44. Published data on MgSi0 grains with a probable deep origin (>660 km) 3  General region Central South America  Author  Locality  Hutchison (1997)  Sao Luiz, Brazil  Kaminsky et al., (2001 a)  Juina, Brazil  this study  Rio Soriso, Brazil  No. of grains  Northern North America  Davies et al, (1999)  DO-27, Canada  1  West Africa  Stachel et al., (2000b)  Guinea  7  Southern Africa  Scott-Smith et al., (1984)  Koffiefontein  1  28  203  0.86-0.94  0.90  0.89-0.94  0.92  0.86-0.94  0.91  0.87 19  total  Range of mg Average mg  0.94 0.93-0.95  0.94 0.95  0.86-0.95  0.92  N i c k e l content for all M g S i 0 grains listed i n Table 8.44 is less than 0.06 wt%. M g S i 0 3  3  inclusions i n diamonds from more shallow sources typically contain ~>0.1 wt% N i O (Meyer, 1987) and M g S i 0  grains i n mantle xenoliths contain 0.03-0.11 wt% N i O  3  (Pearson et al, i n print).  Although the aluminium content i n mantle xenolith M g S i 0 grains may be relatively high 3  (up to 6 wt%, Pearson et al., i n print), typical M g S i 0 diamond inclusions (of lithospheric 3  origins at depths ~<250 km) contain little aluminium (<1.00 wt% A l 2 0 , Meyer, 1987). 3  A1 0 2  3  is generally >1.00 wt% for lower mantle MgSi-Prv's, with a few exceptions (two  inclusions from Guinea (Stachel et al., 2000b) and one each from South Africa (ScottSmith et al,  1984) and Australia (Scott-Smith et al, 1984)). Hutchison (1997) found  several grains o f M g S i 0  3  with significant A l 2 0 , which he termed type II (-10 wt% 3  A 1 0 ) and type m (-10 wt% A 1 0 and - 5 wt% C a O and 6 wt% N a 0 ) . M g S i 0 2  3  2  3  2  grains  3  liberated from Juina area diamonds all have A 1 0 values above 1 wt%. Thirteen type I 2  3  inclusions have been found i n Juina (with A 1 0 wt% range o f 1.23-3.37 and average o f 2  3  2.19). The seven grains identified in Hutchison (1997) and classified as either type IT or DT contain an average o f 10.46 wt% A 1 0 with a range o f 8.34-12.58 wt% A 1 0 . 2  3  2  3  The  sole grain found by Davies et al, (1999) falls into the type I classification. Stachel et al, (2000b) and Scott-Smith et al,  (1984) found seven grains i n association with  ferropericlase. These M g S i 0 grains have a more restricted range (0.55-1.68 wt% A 1 0 ) 3  2  3  and lower average (1.15 wt% A 1 0 ) . The reason for the gap i n A 1 0 wt% between type 2  3  2  3  I and II M g S i 0 grains is unclear. 3  The mg distribution is different from upper mantle O p x ' s only for Juina area inclusions (Fig. 8.37), although data from other locales is sparse.  M g ' s for upper mantle diamond  inclusion O p x ' s are generally restricted between 0.91-0.95 (with very few exceptions) and are skewed to the more Mg-rich end o f the range.  The Orapa mine i n Botswana  appears to be an exception i n that diamond inclusion O p x ' s found there are more Fe-rich, ranging from 0.77 to 0.93 (Meyer, 1987). M g S i 0  3  grains found from Junia tend to be  more Fe-rich, although not nearly as extreme as those found in Orapa. (1997) type II and HI M g S i 0  3  Hutchison's  grains average 0.89 mg and range from 0.87 to 0.91 mg.  204  The thirteen type I grains from Hutchison (1997), Kaminsky  et  al,  (2001a), and this study average  0.92 mg and  range from 0.86 to 0.94 mg.  The lone MgSi03  grains  found  in  association with fPer by  87  88  89  90  91  92  93  94  95  96  mg Fig. 8.37. Plot of mg vs. frequency for MgSi0 inclusions worldwide. U.M. Opx = upper mantle Opx from Meyer (1987) (plotted against relative percent, n=82). Juina, Guinea and Other* are plotted against frequency. Other* are MgSi0 data from locales other than Juina and Guinea listed in Table 8.44. 3  Davies et al. (1999) and Scott-Smith are  al, (1984)  more Mg-rich, at  3  0.94 and 0.95 mg respectively. There are seven grains from Stachel et al, (2000b), of which all but one occur in association with fPer. They have a more restricted range closer to the Mg-rich end-member, lying between 0.93 to 0.94 mg and averaging 0.94 mg. In terms of mg, there is considerable overlap between upper mantle Opx and possible lower mantle type I MgSi03 perovskite.  It appears that upper mantle M g S i 0 3 grains can be distinguished from lower mantle M g S i 0 grains based on chemistry alone, however, all criteria should be used in unison. 3  Low N i content appears to be the most unambiguous indicator of a lower mantle origin while mg has a considerable overlap between the two fields, and may not be such a useful indicator on its own. Provided diamonds are only stable in a depleted harzburgite source, then the aluminium content is also a useful indicator. However, i f diamond can remain stable in a less depleted harzburgite or lherzolite, there is no reason why an aluminous M g S i 0 3 (>1.00 wt% AI2O3) could not be included in diamond in the upper mantle.  Chemistry of the M g S i 0 3 inclusions is consistent with grains of a lower mantle origin. They are similar in composition with all other lower mantle MgSi03 grains, with the exception of the exceptionally aluminium-rich grains (types II and III) reported by Hutchison (1997).  205  8.4.1.3 C a S i 0  3  Roughly 20 grains o f CaSiCh have been reported as inclusions i n diamond prior to this study from three localities: Juina, B r a z i l (Hutchison, 1997 and Kaminsky et al., 2001a), Kankan, Guinea (Stachel et al, 2000b) and Sloan, Colorado, U S A (Otter and Gurney, 1989) (Table 8.45). A l l grains are essentially pure CaSi03 which makes any comparison between major oxide chemistry data difficult.  U n l i k e M g S i O s , there is no C a S i 0  3  polymorph found i n equilibrium with typical inferred mantle compositions at shallow levels and hence no need to weed out lower P-T polymorphs; they are all deep (~>580 k m ' s depth).  T a b l e 8.45.  P u b l i s h e d data o n  General region Central South A m e r i c a  CaSi0  3  diamond inclusions Author  Locality  N o . o f grains  H u t c h i s o n (1997)  Sao L u i z , B r a z i l  3  K a m i n s k y et al., (2001 a)  Juina, Brazil  6  this study  R i o Soriso, B r a z i l  8 17  Central N o r t h A m e r i c a  Otter and G u r n e y (1989)  Sloan, U S A  1  West Africa  Stachel et al., (2000b)  Guinea  6*  total  24  * Stachel et al. (2000b) found some Ca-silicates that do not have CaSi0 stoichiometry. They interpreted these grains as retrograde phases of CaSi0 . 3  3  8.4.1.4 'Olivine'  Because olivine is the major constituent o f the upper mantle, it is one o f the most common inclusions found i n peridotitic upper mantle diamonds, as reported by Harris and Gurney (1979) and Meyer (1987). This is not the case, however, for diamonds which originate from a deeper source, such as those found i n Juina and Guinea.  Olivine  inclusions are relatively rare i n Juina and most o f those which are reported from Guinea (Stachel et al., 2000b) are likely sourced from shallower depths (~<410 km) in the upper mantle.  Table 8.46 is a compilation o f all 'olivine' grains reported from Guinea and  Juina that are interpreted as having a deep origin.  206  For Guinea, where shallow upper  Table 8.46. Published data on olivine grains with a probable deep origin (>400 km) General region Central South A m e r i c a  West A f r i c a  Author  Locality  No. of grains  R a n g e o f mg  Average mg  H u t c h i s o n (1997)  Sao L u i z , B r a z i l  3  0.87-0.91  K a m i n s k y et ai, (2001a)  Juina, B r a z i l  3  0.87-0.89  0.88  this study  R i o Soriso, B r a z i l  7  0.88-0.95  0.91  13  0.87-0.95  0.90  4  0.94-0.97  0.95  17  0.87-0.97  0.91  Stachel et al, (2000b)  Kankan, Guinea  total  0.89  mantle diamonds are also abundant, the only 'olivines' reported are those that occur in association with ferropericalse. Data from Juina, where there is no evidence for a large source of shallow contamination, contain all 'olivine' grains reported from Hutchison (1997), Kaminsky et al, (2001a) and this study.  The separation of 'olivines' into a-Ol, P-Ol, y-Ol and retrograde olivine from MgSi-Prv + fPer based on chemistry alone is difficult. Although Brey et al, (2003) suggest that y-Ol incorporates trivalent cations more easily than either a-Ol or P-Ol, there are no known published works on laboratory experiments corroborating this hypothesis.  'Olivine'  inclusions in this study contain little or not trivalent cations (Ol inclusions in diamond 4-3 are the lone exception). The  only  striking  difference between upper mantle  Ol (a-Ol) and  'olivines'  associated  with fPer is the variation in  mg  (Fig  .  8.38).  'Olivines' in association with fPer are either mgrich (mg=0.94-0.97) or mg-poor  (wg=0.87-  0.91), whereas the mg  87  88  89  |  90  91  92  93  94  95  96  97  ™g  Fig. 8.38. Plot of mg vs. frequency for 'olivine' inclusions worldwide. U.M. Ol = upper mantle olivines from Meyer (1987) (plotted against relative percent on right, n=154) Guinea and Juina inclusions are plotted against frequency on left.  207  for upper mantle olivine inclusions is restricted between 0.90-0.95 (Meyer, 1987). The mg for olivines from off-craton xenoliths range from 0.88-0.92 and from 0.91-0.94 for on craton xenoliths (Pearson et al., in print). The low mg for diamond inclusion 'olivines' in this study is consistent for a non-cratonic source, such as transition zone.  There is no  apparent explanation for the small population of 'high-mg olivines' in Juina, which is similar to the 'olivine' in association with fPer form Kankan.  At  present, there is not enough experimental work demonstrating the partitioning  behavior of M g 2 S i 0 polymorphs. At the very least, the low-mg 'olivines' recovered 4  indicates an off craton source. Although this says nothing about their depth of formation, it is consistent for a deep source, such as the transition zone.  8.4.1.5 Garnet and T A P P  Two  different highly aluminous silicates were recovered from diamonds in this study,  eclogitic garnet and TAPP. Because no majoritic garnets were found in this study, these deep inclusions (>250 km) will not be considered further in this section, however, it is Table 8.47.  Aluminous silicates and T A P P data from select localities with a deep origin No. of TAPP grains  N  ° ' .°f arnet  No. of Eclogitic garnets  No. of Peridotitic garnets  General region  Author  Locality  Central South America  Hutchison (1997)  Sao Luiz, Brazil  9  6  0  1  Wilding (1990)  Sao Luiz, Brazil  0*  14*  20*  0*  Kaminsky era/., (2001a)  Juina, Brazil  1  3  0  0  this study  Rio Soriso, Brazil  1  0  2  0  11  23  22  1  0  1  11  1  Northern North America  Davies et al, (1999)  West Africa  Stachel et at, (2000a)  Kankan, Guinea  0  6  6  6  Hutchison (1997)  Kankan, Guinea  0  1  -  -  Moore & Gurney (1985) Monastery, S. A.  0 35  39  Southern Africa  DO-27, Canada  total :  11  data is referenced through Hutchison (1997).  208  important to point out that numerous majoritic garnets have been recovered from the Juina area (Table 8.47).  It appears that it may be 2.200  9: A  *s «  I  A • x o o • A,  %t  i  2.000 TAPP  .'  i  :  ^BF H A  as ;  1.800  u +  e-type p-type majorite (e-type) majorite (p-type) TAPP TAPP - this study e-type - this study  possible to discriminate between T A P P and other aluminous  silicate  minerals based solely on  w  * **  * 1.600  - /  '  7--  /  chemical data, although  :  /  r  *  X  criteria are based on a  'typical'diamond inclusion garnets :  very limited number o f 4  1.400  k  V * *  T A P P inclusions and i n  \V  the  1.200  i  1.000  :  silica  3.300  Fig. 8.39. Plot of Al + Cr versus Si for aluminous silicates worldwide. Cations calculated on the basis of 12 oxygen anions. The line at 3.075 Si is used to sepate majoritic from non-majoritic garnets (from Stachel et al., 2000a). Data compiled from Stachel et al. (2000a), Stachel et al. (2000b), Mayer (1987), Hutchison (1997), Kaminsky et al., (2001a) and Moore and Gurney (1985). Question mark (?) indicates anomalous value from this study which may be a result of poor quality data. Arrow indicating increasing depth only applies to majoritic garnets. J+  experimental  and  the  low  calcium content are the  Si 3+  of  work. The deficiency i n  1  3.100  2.900  2.700  supporting  majoritic garnet  n o n - m a j o r i t i c garnet  absence  4+  4+  two  characteristic  features o f T A P P . 8.39  is  highlights between  a  plot  the  Fig. that  division  majoritic  non-majoritic  and  garnets.  However, it may also prove useful in separating T A P P from other mantle garnets as T A P P grains contain less S i  4 +  per 12 oxygens than typical mantle garnets (both eclogitic  and peridotitic). Based on empirical observation, a TAPP/garnet divide may be placed at -2.95 S i  4 +  per 12 oxygens. Silica deficiency may be a reflection o f elevated F e  3 +  and T i  4 +  contents, however, elevated amounts o f titanium are only found i n three T A P P grains (titanium is absent in the remaining eight), while F e for the complete database.  3 +  has not been accurately determined  T A P P grains also are markedly different from all other  garnets (majoritic included) in that they contain no calcium (Fig. 8.40). U s i n g these two  209  criteria, the current data seems to  be  relatively  A e-type  three  types:  TAPP.  e  x majorite (e-type) O majonte (p-type)  normal  oTAPP • TAPP - this study  garnets, majoritic garnets and  P-tyP  n  easy to separate into the  A e-type - this study  The  conventional discriminating used  to  factor distinguish  peridotitic from eclogitic garnets using  CX2O3 wt%  may not apply to T A P P grains.  The  Eclogite field 0.00  ,—A-Axjkx-x„i-X-x—, x*x*A—A-xr  0.00  5.00  of  TAPP  grains is unclear (Harris et al, has  1997), although it been  15.00  CaO  compositional paragenesis  10.00  found  in  Fig. 8.40. CaO versus C r 0 (wt%) in garnets from Juina, Guinea, and around the world. The outline of the lherzolitic field is taken from Sobolev et al., (1973) and the cutoff at 2.00 wt% C r 0 for the eclogitic paragenesisis from Gurney (1984). Data compiled from Stachel et al. (2000a), Stachel et al. (2000b), Mayer (1987), Hutchison (1997), Karninsky et al., (2001a) and Moore and Gurney (1985). 2  3  2  3  association with material that is equilibrium with a peridotitic bulk composition.  Grain 4-10A plots i n the same field as 'normal garnets' (Fig. 8.39) and is likely sourced from no deeper than 250 k m . Although inclusion 4-10B had the same orange colour and was recovered from the same diamond as 4.1 OA, its chemistry is markedly different.  It  plots i n a field o f its o w n (Fig. 8.39) and thus brings into question the reliability o f the chemical data for this small (20-40um) grain. The chemistry for eGrt (inclusion 4-1 OA) is typical for eclogitic garnets with an origin less than 8 G P a (~<250 km).  8.4.1.6 Perovskite  To the authors' knowledge, there have only been three published reports o f perovskite occurring as an inclusion i n diamond: one from Juina (Kaminsky et al, 2001a), another  210  from River Ranch, Zimbabwe (Kopylova et al., 1997), and a third from Sloan, Colorado, U S A (Meyer and M c C a l l u m , 1986). The grain reported by K a m i n s k y et al. (2001a) is found i n association with ilmenite, majorite and an unidentified S i - M g phase, thus indicating a deep origin (>~250 km). The grain reported by Meyer and M c C a l l u m (1986) occurs as a polyphase inclusion o f perovskite, ilmenite and phlogopite and the grain reported by Kopylova et al. (1997) occurs i n association with chromite.  The four  perovskites i n this study occur as composite grains with CaSiC>3.  Perovskites reported are divided here into two groups, those containing R E E and H F S E and those void i n these elements.  The grain reported by K a m i n s k y et al. (2001a) is the  lone example o f a diamond inclusion perovskite void i n R E E ' s and H F S E ' s .  It has  almost a stoichiometric composition with minor Si02 (1.05 wt%), AI2O3 (0.64 wt%), and N a 0 (0.22 wt%). 2  The grains from River Ranch and this study contain appreciable  amount o f R E E ' s and H F S E ' s (7.00 wt% and ~>4.00 wt% respectively), and while these elements were not analysed for i n the grain reported from Colorado, the low values reported (93.3 wt%) suggest appreciable amounts are also present. R E E ' s were analysed, Ce contents are greater than L a .  In all cases where  There are some similarities  between grains i n this study and those reported i n K a m i n s k y et al. (2001a), particularly i n the abundance o f silica (~2.00 wt% S i 0 ) and aluminium (-1.30 wt% A1 C»3), although 2  2  these is no detectable sodium.  Perovskites inclusions found i n diamond are traditionally considered secondary because perovskite is not considered stable i n the deep upper mantle i n the diamond stability field and perovskite is a common primary phase i n kimberlite. Elevated content o f Ce over L a is a distinguishing feature o f kimberlitic perovskites (Mitchell et al., 1988). Because o f the similarities in chemistry between diamond inclusion perovskites and kimberlitic perovskites, grains reported from Sloan and River Ranch were interpreted as secondary i n origin.  Experiments have shown that CaTiC<3 (in isolation) is stable at high pressures and temperatures, and that complete solid-solution between C a S i 0 3 perovskite is possible  211  (Koito et al, 2000), however, its stability in mafic and ultramafic rocks at high pressures and temperatures is poorly constrained. It is possible that in the absence of pyroxene (such as in the lower mantle), CaTiC»3 may be the main reservoir for titanium, REE's and HFSE's. The association of CaTi03 with CaSiCh in this study suggests a deep (~>580 km) origin, possibly in the lower mantle. High Si and Al may be indicators of deep (lower mantle) pero vskites.  8.4.1.7 Si0  2  Si0 is a relatively common inclusion in diamond (Table 8.48). Although this table is 2  not a comprehensive list of all localities where Si0 has been recovered, it does illustrate 2  that Si0 has been found in most localities where fPer has also been reported. There are 2  three instances where it has been reported in association with fPer (two from Juina and Table 8.48.  Si0 data from select localities worldwide 2  General region Central South America  Northern South America  Author  Locality  No. of grains  phases in association  Hutchison (1997)  Sao Luiz, Brazil  5  (1) moissanite?, (1) fPer, biotite and plagioclase  Kaminsky et al., (2001a)  Juina, Brazil  3  (l)fPer  this study  Rio Soriso, Brazil  12  Guaniamo, Venezuela  27  (6) eGrt, (8) Cpx, (3) eGrt and Cpx, (1) eGrt and sanadine, (1) Cpx and rutile, (1) Cpx and Mag, (1) rutile and ilmenite, (1) eGrt, Cpx and Ti-Mag, (1) eGrt, rutile and Mag  Sobolev  a/., (1998)  West Africa  Stachel et al., (2000b)  Kankan, Guinea  2  (1) fPer  Eastern Africa  Stachel etal., (1998)  Mwadui, Tanzania  1  Harzburgitic garnet  1  (Mg,Fe)Si04-pGrt (not majoritic)  Southern Africa  Russia  Australia  Kopylova et al., (1997) River Ranch, Zimbabwe Gurney et al., (1984)  Orapa, Botswana  ?  McKenna (2001)  Helam, S.A.  12  eGrt, (2) corundum  ?  Mag, metallic Fe, eGrt, Cpx and rutile  Sobolev (1984)  Siberia platform  Sobolev et al, (1976)  Siberia platform  Sobolev et al, (1984)  Southeastern Aus.  ?  Hall and Smith (1984)  lamproite in W. Aus.  9  Jaques etal., (1989)  Argyle  Meyer etal., (1997)  New South Wales  212  eGrt and omphacite  12  one from Kankan). There are several reports o f S i C ^ and magnetite i n association (e.g. Siberian platform, Sobolev, 1984; and Guaniamo, Sobolev et al, 1998). The uncertainty  Si02 grains has already been discussed (section 8.3.2.1).  in the primary origin o f  However, there are two grains which are reasonable candidates for a primary origin, occurring i n diamond 2-7  (Si02 and fPer) and diamond 2-11 (Si02 and eclogitic  pyrrhotite).  Si02 and fPer are not i n equilibrium in a peridotitic mantle o f mg=0.89, however, they are stable in a more iron-rich peridotite, where mg<~0.78, depending on the specific P - T conditions (Ito and Takahashi, 1989). The mg's for fPer grains i n association with  Si02  are 0.69 (Hutchison, 1997), 0.78 (Kaminsky et al., 2001a), 0.86 (Stachel et al., 2000b) and 0.81 (this study).  8.4.1.8 Pyrrhotite, magnetite and native iron  Pyrrhotite, magnetite and native iron were also recovered from diamonds i n this study. The chemical subtleties for these inclusions are unremarkable i n comparison to grains recovered from other locations.  In the case o f pyrrhotite, there are abundant data from localities all over the world and the four grains reported i n this study are similar to all pyrrhotites o f eclogitic paragenesis. Six grains have been previously reported from Juina, one o f which was i n association with a majoritic eGrt (Hutchison, 1997). Other locations where both pyrrhotite and fPer have been reported (but not in the same diamond) include D O - 2 7 , N W T (Davies et al., 1999) and Guaniamo (Kaminsky et al., 2000). N o grains have yet been reported from Kankan, Guinea.  2"T"  3"1"  Magnetite ( F e W 2O4) is a rare inclusion i n diamond. It has previously been observed z  in diamonds from Juina (Hutchison, 1997), Guaniamo, Venezuela (Sobolev at al., 1998), Sloan, U S A (Meyer and M c C a l l u m , 1986) and several localities on the Siberan plateau (Sobolev et al., 1981).  It has been found in association with S i 0 , among other 2  213  inclusions, i n Guaniamo (Sobolev et al, 1984).  1998) and the Siberian platform (Sobolev,  It is mostly found in association with inclusions o f eclogitic paragenesis.  Magnetite has never been found i n cratonic peridotite, but can occur i n cratonic eclogite (Pearson et al, i n print).  To the author's knowledge, there are only a few published occurrences o f native iron as an inclusion i n diamond.  One has been recovered from Sloan (Meyer and M c C a l l u m ,  1986) and two from Siberia (Sobolev et al, 1981). Native iron from Siberia was found i n association with pGrt, chromite and sulphides. The inclusion i n this study was found i n association with pyrrhotite (eclogitic) and Si02, which is i n contrast to the peridotitic association found for Siberian diamonds containing native iron.  8.4.1.9 Fe-Ni blebs and magnesioferrite spots on ferropericlase  Hutchison (1997) distinguishes between two types of blebs found on several fPer grains. Transmission electron microscopy ( T E M ) o f one type o f bleb finds they are comprised o f magnesioferrite ( M g F e ) 2 0 , while E D S o f the other type o f bleb finds that they are 2+  3+  4  F e - N i alloy. Both types o f features were observed in this study. F e - N i blebs were best developed  on  fPer  grains  from  diamonds  3-5  and  3-10  (section  8.3.5.1)  and  magnesioferrite was observed only on fPer grains from diamond 3-6 (section 8.3.5.2).  Characterisation and interpretation o f these features was more difficult i n previous studies because o f the differences between the procedures followed. In previous studies, inclusions were mounted i n epoxy and polished before examination, whereas in this study, inclusions were mounted on stubs and not polished. S E M images o f both types o f features on fPer show that they develop on the grain surface and are not pervasive with depth.  A n y polishing o f fPer inclusions would result i n the reduction o f bleb size  available for examination.  214  The  observation that magnesioferrite spots are restricted to particular domains on  ferropericlase inclusions was interpreted as an indication that these features  are  secondary. A s , such , they are o f less interest than the F e - N i blebs.  In contrast, the F e - N i blebs are found covering the complete surface o f the ferropericlase inclusions.  The regularity to their formation suggests that they are likely a product o f  exsolution.  The mg's for fPer grains with blebs i n this study are low, but not unique  (mg=0.65 for fPers from diamonds 3-5, and wg=0.45-0.55 for fPers from diamond 3-10). Several grains released from other diamonds have similar mg's without F e - N i blebs. Hutchison (1997) proposed a possible link between blebs and the core-mantle-boundary ( C M B ) . A t present, little can be concluded regarding these peculiar blebs. Experiments on the ability o f fPer grains (of varying mg) to incorporate varying amounts o f nickel at P-T conditions up to the C M B are likely required to better understand this phenomena.  8.4.2  Inclusion paragenesis  The separation o f diamonds into subgroups based on the predicted composition o f the parental material is common practice i n any diamond study (e.g. Davies et al., 1999; Stachel et al., 2000a and b). The stability o f relevant minerals has been introduced i n section 8.1. U s i n g the results from section 8.3, this discussion w i l l divide the inclusionbearing diamonds into subgroups based on paragenetic associations.  Subgroups w i l l  include diamonds o f similar parental material, as well as diamonds o f similar origin o f depth.  A number o f problems and uncertainties were encountered when trying to subdivide the diamond suite into paragenetic groups. 1) A s highlighted throughout section 8.3, it is difficult to unambiguously classify an inclusion as primary.  Chemical homogeneity  (except i n the case o f sulphides) and euhedral form were considered the most convincing evidence for a primary origin. 2) It is also important to bear i n mind that material from a variety o f regions has the potential o f being included i n diamond, from depths where diamond began crystallization to depths where the diamond was (presumably) entrained  215  in a magmatic eruption. Disequilibrium between phases is certainly a possibility. 3) As inclusions are accidental material in diamond, it is unlikely that all phases from one source rock will be represented in one diamond. 4) The variation in phase stability between different inclusions, along with variations in partitioning, make it easier to place some phases in P-T space (e.g. majoritic garnet and T A P P ) while others are difficult to place (e.g. olivine). When combined with the scenario highlighted in point 3, this becomes a more serious problem. For example, in this study, the association Ol-fPerMgSiCh  is considered to have formed at the lower mantle/upper mantle boundary.  However, when only Ol is found, it is difficult to determine if it has a similar paragenesis (i.e. lower mantle/upper mantle), or whether it truly represents its own subgroup. In order to avoid any assumptions, diamonds will only be placed in a paragenetic group that the chemistry of the extracted inclusions permit. For example, diamonds containing fPer-CaSi03 are considered as belonging to lower mantle paragenesis, however, CaSi03 in the absence of fPer (and MgSi03 and TAPP) will be considered as having formed at depths ~>580 km. In all likelihood, many diamonds containing CaSi03 in the absence of fPer have a lower mantle paragenesis, but such a classification requires more evidence.  8.4.2.1 L o w e r mantle  Although the ferropericlase stability field is not restricted to the lower mantle, it is generally used as an indicator of the lower mantle paragesesis.  In general, the  transformation of fPer in an open system (i.e. in the presence of silicates) is considered to take place at the 660 km discontinuity (Akaogi et al., 1998). Many argue that this seismic discontinuity is a result of the breakdown of y-Ol to fPer and MgSi-Prv. Twelve diamonds, or 27% of the diamonds cracked, contain fPer in the absence of 01 and are thus grouped in the lower mantle paragenesis (the importance of this will become clear in section 8.4.1.2). Of these 12 diamonds, the number of grains in the particular association, along with the average mg for fPer grains in that association are: fPer as the only phase (found in 7 diamonds, with average mg=69), fPer + CaSi03 (3 diamonds,  216  average mg=6$), fPer + Si02 (1 diamond, average mg=S\), and fPer + eGrt? (1 diamond, no E P M A data for fPer grain).  There is a wide spread i n mg for fPer grains i n this  paragenetic group (0.45-0.83).  In terms o f bulk composition o f the mantle, the inclusion associations are mostly i n equilibrium with a peridotitic or pyrolitic lower mantle. C a S i 0 3 is found i n association with fPer i n three diamonds, which is consistent for a mantle o f pyrolitic composition at lower-mantle P - T conditions.  Phase associations in two diamonds (2-7 and 4-16) are  either not i n equilibrium, or have different protoliths.  Diamond 2-7 also contains an*  inclusion o f Si02 (Fig 8.21, 2.7K), w h i c h i f indeed primary and syngenetic, would suggest the source region was iron-rich peridotite (Ito and Takahashi, 1989). Diamond 416 contains two euhedral grains that leave little doubt i n their primary origin, however, both grains are too small for E P M A analysis (<10 um).  The garnet is unlikely to be  T A P P as it contains too much calcium (see section 8.3.1.5), and as it is found with fPer, it should either have a large majoritic component, or not be i n equilibrium with fPer. The similarities i n E D S spectra between the aluminous silicate i n diamond 4-16 and a nonmajoritic garnet (confirmed by E P M A ) is a first approximation that the garnet does not have a majoritic component, and is thus sourced from depths <250 k m .  8.4.2.2 L o w e r m a n t l e / u p p e r  mantle  M a n y authors have demonstrated experimentally that olivine is not stable i n the lower mantle (e.g. Ito and Takahashi, 1998; Yamazaki et al., 1994). Therefore, its presence i n association with fPer has implications on the origin o f the diamond.  Four different  scenarios where diamond could include fPer and 01 are considered here:  1. fPer formed at depths <660 k m where silica activity was low enough for it to form in the presence o f y - O l . 2.  01 is a retrograde phase o f M g S i - P r v plus fPer.  3.  fPer and 01 were not trapped at the same time and are thus not i n equilibrium.  217  4. Inclusions were trapped at the lower mantle/upper mantle ( L M / U M ) boundary at the predicted conditions where fPer, 01 and M g S i - P r v are i n equilibrium.  There are four grains containing fPer and O l (diamonds 1-5, 3-2, 3-5 and 4-3), or 9% o f diamonds cracked.  A l l four grains also contain M g S i 0 3 , one o f which also contains  T A P P while another contains both T A P P (based only on E D S analysis) and CaSi03. The M g S i 0 3 inclusions contain elevated amount o f AI2O3, consistent with that o f grains which initially crystallized with the perovskite structure and formed i n the upper -10-20 k m o f the lower mantle (Stachel et al,  2000b), thus excluding scenario #1 for all  diamonds (provided that the fPer and M g S i 0 3 inclusions are i n equilibrium). Further corroborating an origin o f depth below 660 k m is the presence o f T A P P , which has been interpreted as being sourced from the uppermost part o f the lower mantle (Harris et al, 1997; Gasparik et al, 2000). The similarity i n chemistry o f the touching (4-3K2) and the non-touching (4-3C) O l inclusions excludes the possibility o f a retrograde origin from touching fPer and M g S i - P r v inclusions for diamond 4-3.  In any case, scenario #2 is  unlikely for olivine inclusions that are not i n direct contact with other phases because it requires diamond to trap the perfect proportion o f fPer and M g S i - P r v to make O l , without any leftover components. A strong case for a retrograde origin for O l grains i n diamonds 1-5 (inclusions J2 and J4) and 3-5 (inclusion G2) could be made.  In the absence o f  convincing experimental data demonstrating the chemical differences between a-Ol, p - O l and y - O l , scenario #3 is difficult to corroborate or refute. The variation i n mg for O l i n this study does suggest there are some differences i n parental material for diamond 4-3. It is difficult to exclude the first three scenarios for all four diamonds, however, scenario four is the most probable origin for the associations found i n diamonds 3-2 and 4-3. Olivine inclusions i n diamonds 1-5 and 3-5 may be retrograde phases and the diamonds may have origins deeper than - 6 6 0 k m .  The associations i n these diamonds provide a rare opportunity to examine the partitioning o f elements between various phases. The variation i n mg for fPer, O l ,  MgSi03 and T A P P  grain(s) is recorded i n Table 8.49 and depicted graphically i n F i g . 8.41. Diamond 3-2 is the only sample where O l , fPer and  MgSi03 fall along a straight line (Fig 8.41); the  218  Table 8.49. Mg for olivine-ferropericlase-MgSi0 associations (this study) 3  mg Diamond  fPer  01  MgSi0  TAPP  3  range  ave  range  ave  range  ave  ave  range  1.5  0.74-0.76(4)  0.75  0.89(1)  0.89  0.91(2)  0.91  0.87(1)  0.87  3.2  0.83-0.84(4)  0.84  0.93-0.94(2)  -  0.60  0.94-0.95(2)  0.89-0.91(2) 0.94(2)  -  0.83  0.89 0.95  0.94 0.90 0.94  -  0.57-0.65(4) 0.82-0.84(5)  0.88-0.89(2) 0.89(1)  0.89  3.5  -  4.3  -  Mg calculated as Mg /(Mg +Fe ) where all Fe was calculated as Fe . Numbers in brackets represent the numbers of inclusions. 2+  2+  2+  2+  remaining three diamonds contain O l with mg that is shifted to elevated magnesium contents. This suggests that the parental material is likely not the same for the four grains. Experimental work on olivine (mg=0.90) at high P-T conditions  find  that  the  MgSi03  (mg=0.93±0.006) is i n equilibrium with fPer (/wg=0.82±0.009) at predicted P-T conditions at the upper mantle-lower  Fig. 8.41. Plot of M g versus Fe versus S i for MgSi0 , Ol and fPer grains in association in this study. 2+  2+  4+  3  mantle MgSi0  boundary 3  (~660 km) and  (mg=0.919±0.004)  is  in  equilibrium with fPer (/wg=0.84.9±0.008) at predicted P-T conditions at the upper coremantle boundary (-2650 km) (Fig 8.42A, Kesson et al, 2002). These results are plotted with data from this study i n Fig. 8.42B. The fPer grains i n diamonds 3-2 and 4-3 contain magnesium contents that are predicted for a mantle o f mg=Q.9G, however, the olivine data from diamond 4-3 is curious. The fPer inclusions i n diamonds 1-5 and 3-5 contain l o w magnesium contents, suggesting the parental material also contains considerably less magnesium.  FPer grains from diamond 3-5 were also anomalous i n that they contain  small blebs o f F e - N i alloy on their surfaces (section 8.3.4).  219  j y j g 2 » 100  90  80  70  60  50  ] ^ j g 2 » 100  40  90  80  70  60  50  40 p 2 + e  Fig. 8.42. Plot of M g versus Fe versus Si for MgSi0 and fPer grains from experimental studies and associations in this study. A) Experimental data on olivine (/ng=0.90) at predicted P-T conditions at the upper mantle-lower mantle boundary (~660 km) and the core-mantle boundary (~2650 km) (Kesson et al., 2002). B) Comparison between four diamonds from this study with fPer and MgSi0 in association with resultsfromexperimental data. 2+  2+  4+  3  3  8.4.2.3 Deep transition zone/lower mantle (>~580 k m )  Although CaSi03 is considered the dominant calcium host in the lower mantle, it is likely the second most dominant calcium-bearing phase i n the deeper reaches o f the transition zone where it begins to exsolve from garnet. mineral inclusion, i n the absence o f fPer,  A s such, the presence o f CaSiC«3 as a  TAPP or aluminous MgSiCh, cannot be  considered an indicator o f the lower mantle paragenetic association without neglecting the possibility o f a deep transition zone origin. A s well, unlike fPer, C a S i 0 3 is stable i n both eclogitic and peridotitic rocks. For these reasons, C a S i C ^ in the absence o f the three minerals discussed, w i l l not be considered as an indicator o f a lower mantle paragenesis; it w i l l be considered an indicator o f the deep transition zone/lower mantle ( D T Z / L M ) , or depths  >~580 k m .  There are seven grains (16% o f cracked stones) that belong to this paragenesis (diamonds 2-8, 3-4, 3-7, 4-7, 6-6, 7-1, 4-10). E P M A data exists for four C a S i 0 diamonds and for three  3  grains from four  CaSi0 grains from two diamonds that also contain fPer. There 3  does not appear to be any chemical distinction between  220  CaSiCh grains whether or not  they are i n association with fPer. However, the purity o f all CaSi03 grains makes any such comparison difficult.  CaSiC>3 grains from two diamonds are in contact with CaTiC»3 (diamonds 2-8 and 3-7) and another is i n association with non-majoritic eclogitic garnet (diamond 4-10). composition o f the parental material which crystallised C a T i 0 majoritic garnet in association with C a S i 0  3  3  is unclear.  The  The non-  cannot be in equilibrium. A s such, this  diamond either belongs to two different paragenetic associations, or one or both phases are secondary.  The parental material for most diamonds i n this paragenetic group could be sourced from either eclogitic or peridotitic rocks.  Diamond 4-10 has an eclogitic inclusion, but it  cannot be in equilibrium with CaSiC>3.  8.4.2.4 Peridotitic Diamonds 4-18 and 6-8 both contain olivine i n the absence o f fPer (comprising 5% o f the diamonds cracked). Unfortunately, the E P M A data for the olivine inclusion i n 6-8 are poor (wt% value o f 96.07) and there are no chemical data for the accompanying MgSiC»3 (identified by E D S analysis) nor are there chemical data for the olivine grain (identified by both Raman and E D S analysis) extracted from diamond 4-18 (grain was lost during polishing). The mg for the olivine inclusion 6.8A is 0.90, which falls slightly below the typical range for iithospheric' olivines (0.91-0.95, Meyer, 1987). T o suggest that these inclusions belong to a different paragenesis than the olivines found i n association with ferropericlase based solely on the inability to find fPer i n the diamond is tenuous: it is certainly possible to not include fPer even i f these diamonds did originate at the U M / L M boundary, as w e l l , it is possible that fPer was included but not found. The chemistry is poor or non-existent and therefore  assigning these inclusions to the lower mantle  paragenesis can not be justified. The depth o f formation o f these diamonds is unclear, however, the composition o f the parental rocks certainly was peridotitic.  221  8.4.2.5 Eclogitic  Six diamonds have been classified as belonging to the eclogitic paragenesis: diamond 211, 4-10, 4-11, 2-6, 1-4 and 2-10. Diamonds i n this paragenesis are recognised by the presence o f l o w - N i sulphides, Ca-rich garnets or magnetite.  Diamond 4-10 contains two garnet grains with abundant calcium and no chromium, as well as a small inclusion o f CaSiCh (previously discussed i n section 8.4.1.3). The garnets plot i n the eclogitic field delineated by Gurney (1984) and are considered pyropegrossular-almandine i n composition. They do not have a majoritic component and thus likely formed at depths shallower than - 2 5 0 k m . C a S i C ^ and non-majoritic garnet is a 'forbidden' assemblage i n both pyrolite and eclogite. The chemistry o f the CaSiC«3 grain (Fig. 8.13 and Table 8.17) is markedly different from the seven other grains with chemical data; it contains more F e O (1.14 wt%, average o f 7 other grains is 0.13 wt%), M g O (0.35 wt%, average o f 7 other grains is 0.04 wt%) and N a 0 (0.48 wt%, average o f 2  7 other grains is 0.05 wt%) than the other grains studied. There appears to be something unique about the CaSi03 grain found i n this diamond. This diamond may have began crystallization at depths >580 k m where the CaSi0 grain was included, and later, after 3  the diamond had been exhumed to shallower levels (<250 km), included eGrt.  This  diamond belongs to two paragenetic associations, one deep (>580 km) o f either eclogitic or peridotitic material, and one that is relatively shallow from eclogite source material.  Diamond 2-11 contains abundant Si02 (7 grains) and sulphides (9 grains), as well as an inclusion o f native iron.  Out o f all the Si02 grains observed i n diamonds from R i o  Soriso, the only Si02 grain with crystal faces (Fig. 8.20, 2.11-C) was found in diamond 211. Si02 is also found i n several fractures which places some doubt on the primary origin o f inclusions recovered (certainly i n the case o f Si02). sulphide grains analysed ranges from 2.68 to 14.34.  The wt% o f N i O i n the three  T w o o f the sulphides would be  considered to be i n equilibrium with eclogite (Gurney, 1989; Bulanova et al., 1996) while the third is i n equilibrium with pyroxenitic source material (Bulanova et al., 1996). This diamond w i l l be classified as eclogitic based on the abundance o f Si02 and the low N i O  222  wt% values o f two out o f three sulphide inclusions, but it is recognized that the there is secondary material present i n the diamond which could have overprinted an eclogitic paragenesis. The depth o f origin o f this grain is unclear.  Three diamonds were found to contain euhedral crystals o f magnetite. Chemical data are only available for grains from two o f the diamonds (2-6 and 1-4) and it is possible that the grains originally identified as magnetite by E D S i n diamond 2-10 are some other FeO phase or are secondary in origin. Diamond 1-4 also contains an inclusion o f SiC^ (Fig. The primary nature o f this Si02 inclusion is certainly  8.20 i n section 8.3.2.1).  questionable, although primary silica is common in eclogitic rocks. Magnetite is stable i n cratonic eclogite and has not been reported from any cratonic peridodite (Pearson et al., in print). However, the stability o f this inclusion at higher P-T conclusions is unclear. A s there are no other inclusions i n association with magnetite, the depth o f formation is unknown.  8.4.2.6 Paragenesis s u m m a r y  Although the sample size is small (44 diamonds cracked, 30 diamonds containing identifiable inclusions) comments about the number o f different source regions where diamonds grew as well as the frequency n=44  o f sampling o f these different sources are  warranted.  Three  • lower mantle (>660 km)  distinct  • lower mantle/upper mantle (-660 km)  associations are recognized ( L M / U M ,  • >580 km depth  lower mantle and eclogitic) as well as  • eclogitic  two associations which may represent  ta peridotitic  their own distinct groups, or may appear  • no inclusions recovered  unique only on account o f other phases not being included i n diamond (>580 km  and peridotitic).  proportions  of  each  The relative subgroup  are  Fig. 8.43. Distribution of diamond paragenetic groupings based on mineral inclusion data. Diamond 4-10 is in equilibrium with two groups (>580 km and eclogitic) but has been grouped with the eclogitic paragenesis.  presented i n Fig. 8.43.  223  9.0 Discussion 9.1 Correlations between d i a m o n d body colour, F L , C L a n d IR  One goal o f this thesis was to identify correlations between diamond body colour, fluorescence, cathodoluminescence and its infrared absorption, with a particular emphasis on understanding the effects that impurities (revealed through IR analysis) have o n diamond body colour, F L and C L .  Fluorescence colours for most diamonds worldwide are various shades o f blue and are generally interpreted as being a result o f nitrogen impurities. The defects responsible for other F L colours, however, are poorly understood, or are a result o f several combined defects.  In this study, where diamond body colours are compared w i t h F L colours, it  becomes apparent that most o f the F L colours other than blue are found i n diamonds that are coloured (with the exception o f grey diamonds) (Fig. 9.1). This is strong evidence for a correlation between diamond body colour and diamond F L colour. Therefore, centres that are responsible for variations  i n diamond  body colour also have a strong  control  diamond F L . also  on  F i g . 9.1  highlights  the  similarities i n F L colour distribution grey  between  and  diamond.  distribution patterns are  interpretation  (30)  for that  the grey  £  r e v  (16)  yellow  brown  milky  pink  (9)  (10)  (2)  (2)  Diamond colours  colourless  These similar  evidence  colourless  Fluorescence colours • blue(s)  • blue^)  •blue(w)  Oblue (vw)  • turquoise (s)  • turquoise (m)  • turquoise (w)  • turquoise (vw)  • green (s)  • green (w)  • brown (w)  Dnone  Fig. 9.1. Comparison of F L colour distribution by diamond body colour. Number in brackets below each diamond body colour represents sample size.  224  diamond is not a true body colour, but results from  incorporation  of  numerous  dark  inclusions  (section  3.3.1).  Although both cathodeinduced and ultravioletinduced are  luminescence  caused  defects their and  by crystal  and impurities, resultant  patterns  colours may be  quite different (Fig. 9.2). Most diamonds  exhibit  C L and F L colours o f various shades o f blue, but  there  are  marked which  some  differences are  likely  a  mm  combination o f both the volume o f diamond and the optical centre that is activated technique.  by  each  A s illustrated  in F i g . 9.1, diamond F L is strongly controlled b y the same defect centres  Fig. 9.2. Photos comparing diamond CL (on left) and diamond FL (on right). From top to bottom, 4-17, 2-11, 4-21, and 4-3.  225  responsible for diamond body colour, and as such, is a bulk crystal property.  CL,  however, only activates a small volume o f diamond (provided certain precautions are taken, such as those outlined i n sections 5.2 and 7.2), and thus any resultant colours reflect only local defects.  Whereas diamond F L generally appears  homogeneous,  diamond C L can be extremely heterogeneous (e.g. images o f internal structures i n diamonds i n section 7.3).  Comparisons  between  nitrogen 600  character  and diamond F L , and C L  reveal several correlations. intensity  of  generally  diamond reflects  o median 500  Firstly, the  FL the  and  1 range  CL  ba  tend to  have  This correlation is  There  so strong  does  correlation  not  200  J3 none very weak weak moderate strong n=l n=l() n=l« n=22 n=18  particularly true for C L colour intensity, but not  300  100  brighter F L (Fig 9.3) and C L (e.g. Figs. 7.12 and 7.15).  400  o o c o u a a oo o  nitrogen  concentration; diamonds with high total nitrogen concentrations  0 mean  Fluorescence intensity  for F L intensity.  appear  between  to  be  Fig. 9.3. Plot of nitrogen concentration versus fluorescence intensity.  any  nitrogen  aggregation state and F L intensity or colour.  There may be a correlation between C L  colour and aggregation state i n some instances, but there are not enough transitional (type IaAB) diamonds to test this hypothesis. Diamond 4-17 has a rim with yellow-green C L (type IaAB), which becomes more aggregated towards the core, but also decreases i n nitrogen concentration to below detection.  Not all type I a A B diamonds have yellow-  green C L , and not all yellow-green C L colours should be considered type I a A B diamond. Yellow-green C L colours observed i n some diamonds (i.e. diamond 2-11, section 7.3.8) are interpreted as being a result o f plastic deformation.  226  Because plastic deformation  cannot focus slip on only the diamond rim, this mechanism o f formation was ruled out for diamond 4-17.  The three main causes o f body colouration i n diamond (impurity defects, dislocations and irradiation) have been discussed in section 3.1.1. The defect centres revealed through IR spectroscopy observed.  should reflect diamond colouration, however,  few correlations  were  Y e l l o w colouration i n diamond is usually found i n type lb diamond or i n  diamonds with N 3 centres.  N o diamonds  in this study contain nitrogen i n a  disaggregated form (type lb) and N 3 centres are not IR-active and thus could not be detected.  It is possible that the yellow stones contain N 3 centres.  B r o w n and pink  colouration is a result o f plastic deformation, which does not produce IR-active defects. The two m i l k y diamonds contain high concentrations o f both nitrogen (all as B centres) and hydrogen. W i t h the exception o f m i l k y diamonds, bulk diamond body colour reveals nothing about the nitrogen character.  The internal morphology o f several diamonds  examined in chapter 7.0 reveals that some crystal cores may be recognised, without the aid o f C L or F L , by either a local brownish body colour (i.e. diamond 1-4) or by a local cloudy aggregate (i.e. diamond 2-1).  These recognisable cores contain the highest  nitrogen concentrations.  9.2 D i a m o n d subpopulations  Diamond paragenesis has been discussed based solely on inclusion chemistry i n detail i n section 8.4.1. In this section, diamond paragenesis w i l l be discussed using an integrated approach b y including all methods o f diamond fingerprinting described i n Chapters 2 through 7.  227  9.2.1 U p p e r mantle diamonds  The term 'upper mantle' diamonds, as used here, applies to the typical assemblage o f mineral inclusions for both peridotitic and eclogitic sourced diamonds, as described by Meyer (1987) and Gurney (1989). Diamonds are assigned to this subgroup based mainly on two upper-mantle 'indicators': 1) diamond type (either IaA or IaAB) and 2) mineral inclusion chemistry and associations.  Six diamonds i n this study still contain nitrogen in the form o f A centres and are considered either type IaA or I a A B .  M o s t diamonds i n Juina contain relatively low  nitrogen concentrations, with all nitrogen i n the form o f B centres. N o diamond assigned to the lower mantle paragenesis (in this study or i n any other) has ever been found to contain A centres. This observation has been cited as evidence that these diamonds have been exposed to high mantle residence temperatures and/or long mantle residence times. A s well, diamonds from lithospheric sources with only B centres are relatively rare.  Certainly there is a high probability that these diamonds are old as most studies on the age o f diamond find that they are >1 G a (e.g. Pearson et al,  1999), however, without  further constraints, we cannot rule out the possibility that type IaA and I a A B diamonds are young and that although they were subjected to high mantle residence temperatures, they were not subject to long residency periods, thus preserving A centres.  F i g . 6.22  demonstrates the short duration o f time required to achieve 9 5 % B centres with total nitrogen concentrations less than 100 ppm. The short duration o f time to achieve 95% aggregation (1.1 M a ) for temperatures predicted at - 4 1 0 k m makes a deep but young origin highly improbable.  A s well, there are no known processes that slow down  aggregation, only those that enhance the rate o f aggregation, such as plastic deformation, which would result i n shifting the curve in F i g . 6.22 to the left.  228  Constraints on the age and equilibrium  9(111  temperatures o f peridotitic and eclogitic  1000  xenoliths recovered from kimbelites in  111(1  Juina could  explain the  aggregation  states observed in four o f the type IaA and  I a A B diamonds.  1100 y a  i_  ^  2003),  are  brought  by  95  c  §  Ma  160(1  1700  indicating a mantle residence time o f  I son 100E+04  -1100  to 1800 M a . F i v e different types  of xenoliths recovered are plotted i n time-temperature space, along with the aggregation containing  pathways 10  and  for 100  diamonds ppm  Centres  / \ /\ \  1500  kimberlites (Heaman et al, 1998), thus  \  5"„n  14011  "j  al,  10 ppm N 100 ppm  1300  u•— 3 a  The xenoliths,  dated at 1166-1884 M a (Costa et  granular peridotites sheared peridotites • eclogites • opx-rutile eclogites Mf sanidine-cocsitc eclogites  95% B  10 ppm N  100 ppm N  Centres  1 1 00E+05  I 00E+06  1.1  Ma  1 00E+07  l.OOE+OS  Cia  1.8 lia  1.00E+09  1 00E+10|  Time (years)  Fig. 9.4. Plot of mantle residence time versus mantle residence temperature for xenoliths and diamond nitrogen character. Xenoliths are recovered from kimberlites in Juina (Costa et al, 2003) and nitrogen character curves are calculated using equation 6.7. Note that the time scale is logarithmic.  total  nitrogen with 5 and 9 5 % B centres. In 1.1 G a , 95% o f A centres w i l l be converted to B centres at a temperature o f -1325 °C (Fig. 9.4). In other words, diamonds with 10-100 ppm total nitrogen exposed to temperatures greater than -1325 °C would have >95% B centres.  Assuming these diamonds have similar ages (or greater) than the xenoliths  recovered, they likely have origins i n the upper mantle, possibly i n the cratonic lithosphere.  9.2.1.1 Upper mantle peridotitic diamonds  This subpopulation includes diamonds that have crystallised i n a peridotitic source in the upper mantle.  Two diamonds were classified into this subgroup based on mineral inclusions as they contained O l in the absence o f fPer (diamonds 4-18 and 6-8). E P M A chemical data is  229  only available for the O l inclusion in diamond 6-8 and is dissimilar to both U M / L M 01' s and shallow olivines (a-Ol) (Meyer, 1987; Gurney, 1989).  In this case,  however, comparisons based on chemistry are subject to errors as the analysis is poor. Diamond 6-8 contains the only sulphide with abundant  nickel (inclusion  6.8A2, section 8.3.1.7), and is consistent for sulphides o f peridotitic origin o f lithosperic depths (although the sulphide is clearly secondary).  These two diamonds  have many shared characteristics and are different in several respects from the majority o f the population.  Fig. 9.5. Photos of diamond body colour (left), FL (middle) and CL (right) for upper mantle diamonds of peridotitic paragenesis. From top to bottom, diamonds are: 4-17, 4-18 and 6-8.  The most obvious difference is the bright green C L colour, which has not been found on eclogitic or deep R i o Soriso diamonds (Fig. 9.5). Based on the green C L colour and low aggregation state (-75% B centres), a third diamond (4-17) is tentatively assigned to this paragenesis, although it contained no inclusions.  Comparison o f morphology, F L , C L and body colour o f the three diamonds reveals more similarities than differences. The diamonds are generally intact (relatively small amounts o f diamond have been lost from brittle fracturing) and two grains show no signs o f resorption while one shows strong resorption (class 2).  One o f the diamonds is an  aggregate o f many octahedral diamonds. T w o diamonds exhibit strong blue F L and one exhibits moderate turquoise F L (4-18).  A l l three stones exhibit moderate to strong  yellow-green C L and clearly stand out among the whole population o f diamonds. Diamonds 4-17 and 4-18 have the brightest green C L colours (diamonds i n F i g . 5.3 i n category 'Green 2') while diamond 6-8 is the next brightest green C L diamond in the suite. T w o diamonds are colourless while one is classified as grey.  230  Two diamonds are type I a A B (53 and  Temperature (°C) 700  75% B centres with ~50 total ppm  20  nitrogen) while one diamond (4-18)  30  contains 144 ppm total N all as B  40  centres. have  Assuming  these  diamonds  recovered  from  t/1 V  Juina  diamonds can be plotted on F i g . 9.6 to determine their time-averaged  mantle  90  °C).  100  type  'amoiJfj  D  temperatures  (-1200  Estimated  temperatures  of  nitrogen  aggregation are consistent with those calculated for both sheared peridotite and granular peridotite xenoliths (Costa et al., 2003) (Fig. 9.6). I f diamonds are  —  i  1500 64 96 128 160  granular peridotite - xenoliths •  residence  1  -_i^j|Ptote  V)  80  two  1300  6-8 — -4-17  |^ 60  £ 70  the  1  £50  IaAB  kimberlitres,  UOO  g,  similar ages to the peridotitic  xenoliths  900  192 sheared peridotite 224 xenoliths 256 288 320  Fig. 9.6. Plot of temperature versus pressure for peridotitic xenoliths and diamonds. Grey P-T fields are estimates for xenoliths from Juina kimberlites (Costa et al., 2003). The lower pressure estimates for sheared peridotites are unusual in the woldwide context. Diamond/graphite stability is from Kennedy et al. (1976). The estimated temperatures for diamonds 6-8 and 4-17 represent time-averaged mantle residence temperatures for ~ 1.1 Ga diamonds.  sourced from granular peridotite, they are likely lithospheric i n origin.  The low mg (0.90) o f olivine in diamond 6-8 is  consistent with composition o f olivines from sheared asthenospheric peridotite (Boyd, 1989) and may indicate that the diamond is sourced from the sublithopheric mantle. This is highly unusual as diamonds have never been reported i n sheared peridotites or having olivine inclusions with fertile asthenospheric mantle compositions (Meyer, 1987; Gurney, 1989).  Information on the internal structure is available from two diamonds, 4-17 and 6-8. Diamond 4-17 has a r i m o f bright green C L (type IaAB) that forms on a core o f weaker blue C L (type Ha) (section 7.3.14). The aggregation state o f the r i m increases towards the crystal centre, but is below detection in the blue C L interior. Although the crosssection o f diamond 4-17 does not reveal the complete growth history, the growth o f the  231  outer rim likely formed on a flat-faced octahedron. It is certainly possible that the type I a A B diamond formed on a core that initially crystallised in the deep mantle (the type II core may attest to such a history). Chips from diamond 6-8 (produced from cracking o f diamond for mineral inclusion extraction) also reveal a blue C L interior, however the core appears bright turquoise and is surrounded by a zone o f weaker blue C L and a thin rim o f bright yellow-green C L diamond. There are no IR data for specific C L zones for this diamond.  The bright green C L r i m is perhaps the hallmark o f this subgroup. The relatively intact nature o f crystals, the general absence o f resorption, the tendency for grains to be I a A B , and the presence o f 01 i n the absence o f fPer are other distinguishing features.  9.2.1.2 U p p e r mantle eclogitic diamonds (type I a A a n d I a A B )  This  subpopulation includes type  IaA and  IaAB  diamonds that have crystallised in an eclogitic source in the upper mantle.  However, i f the diamonds  experienced short mantle residence times (<1 M a ) , than a deeper origin for some grains cannot be ruled out.  Three grains were classified as having an eclogitic origin based on inclusion studies alone (diamonds 211, 4-10 and 4-11).  &  ;#  They are considered eclogitic  based on the low nickel content o f sulphides (2-11 and 4-11) and the calcium-rich, chromium-poor garnets (with no majoritic component) found in diamond 4-10 (indicating a depth o f formation o f less than 250 km). Diamond 4-10 contains two forbidden phases, CaSi03  232  Fig. 9.7. Photos of diamond body colour (left), FL (middle) and CL (right) for upper mantle eclogitic diamonds. From top to bottom, diamonds are: 2-11, 4-10, 4-11 and 4-15.  and non-majoritic garnet, and as such, can be placed i n two paragenetic subgroups. S i C ^ was found in diamond 2-11 and is likely primary on account o f its euhedral form (it occurs i n association with l o w - N i O pyrrhotite). A fourth grain is tentatively included (diamond 4-15) based o f its nitrogen character (type I a A B ) and similarities i n F L , C L and crystal form to diamond 4-10.  L i k e the peridotitic diamonds o f upper mantle origin, this subpopulation tends to be less resorbed than most diamonds, although they are slightly more resorbed than the p-type diamonds (resorption scale 2,3 and 4 and one stone is classified as 'unknown'). Diamonds are mostly intact crystals o f various shades o f colourless to grey. Two o f the stones are aggregates.  The F L o f all diamonds is blue, and the intensity is strong in all  but one (Fig. 9.7). C L colour o f the exterior o f three diamonds occurs as shades o f blue o f various intensities (Fig. 9.7).  Temperature (°C) 500  700  900  1100  20  '  A l l four diamonds contain A centres, 25  one is type IaA (48 ppm N ) while the 30  other three are type I a A B (16-64 % B , 33-138 ppm total N ) .  Assuming that  these diamonds have similar ages as  r 35 8  kimberlites,  we  can  place  some  constraints  on  temperatures  and  50  pressures o f formation for the diamonds in  this  subgroup.  Three  types  of  eclogitic xenoliths have been recovered, they are: orthopyroxene-rutile eclogites (1648 M a ) , sanidine-coesite eclogites (1166 M a ) and coarse-grained eclogites (1593 M a ) (Costa et al,  2003).  The  40 45  eclogitic xenoliths recovered from Junia  4-II  opx-rutilc eclogitic xenoliths  1300  r  -4-^5 —2-11  sanidine|I coesite ! 'eclogitic xenoliths  1500 64 80 96 112 128 144 160  55 -  176  60  192  Fig. 9.8. Plot of temperature versus pressure for eclogitic xenoliths and diamonds. Grey P-T fields are estimates for xenoliths from Juina kimberlites (Costa et al., 2003). Diamond/graphite stability field is from Kennedy et al. (1976). The estimated temperatures for diamonds 2-11, 4-10, 4-11 and 4-15 represent time-averaged mantle residence temperatures for ~ 1.1 Ga diamonds. Diamond 4-11 contains 100% A centres, which is outside the limits of equation 6.7.  233  time-averaged mantle residence temperatures for minimum residence times o f 1.1 G a for each diamond are shown i n F i g . 9.8. Note that the estimated temperature for diamond 411 falls somewhere to the left o f the line indicated. Although the estimated temperatures are consistent with those predicted for the sariidine-coesite xenoliths, the estimated pressures fall outside o f the diamond stability field (Kennedy et al,  1976).  Only  diamond 4-11 could be sourced from the opx-rutile eclogite for these residence times. Temperatures o f formation for the course-grained eclogites were estimated between 1182 - 1287 °C, however no pressure estimates were determined (Costa et al, 2003). These temperatures are consistent for two eclogitic diamonds recovered, however it is unknown whether or not these xenoliths are sourced within the diamond stability field.  Only one diamond (2-11) from this subgroup was polished for growth studies, which revealed an internal morphology unique to this study (section 7.3.8). C L studies reveal a complex, mostly alternating pattern o f turquoise blue and yellow-green C L bands. The yellow-green zones were interpreted as being the result o f plastic deformation. Growth, at least i n part, was on octahedral faces.  Eclogitic diamonds from the upper mantle are characterised by their lower aggregation states (<100% B centres) and by the presence o f mineral inclusions that are commonly found in eclogitic diamonds from cratonic sources (e.g. non-majoritic pyrope-almandinegrossular garnets, l o w - N i pyrrhotites and possibly Si02). They tend to be colourless to grey and exhibit F L and C L colours or various shades o f blue o f moderate to strong intensity. Aggregated forms may be more common than i n other subgroups identified.  9.2.2 Eclogitic diamonds (type I a B )  Diamonds i n this subgroup are type IaB and contain inclusions o f eclogitic paragenesis. Three diamonds were grouped into the eclogitic paragenesis in section 8.4.3.2 based on the presence o f inclusions o f magnetite (diamonds 1-4, 2-6 and 2-10).  234  Three other  diamonds  were  also  grouped  into  the  eclogitic  paragenesis based on inclusions, however examination o f nitrogen character separated these six diamonds into two subgroups within the eclogitic paragenesis, those with A centres, types IaA and I a A B (which defines the subgroup o f likely upper mantle origins described i n section 9.2.1.2) and those with only B centres, or type IaB (the three diamonds in this subgroup). Diamond 1-4 also contains an inclusion o f Si02 (the primary nature o f which is questionable) and the magnetite i n  Fig. 9.9. Photos of diamond body colour (left), FL (middle) and CL (right) for eclogitic type IaB diamonds. From top to bottom, diamonds are: 1-4, 2-6 and 2-10.  diamond 2-10 was determined based only on E D S analysis.  The three diamonds are mostly intact, appear strongly resorbed and are yellow, m i l k y and non-uniform brown/colourless.  F L and C L colours range from blue to turquoise o f  moderate to strong intensity (Fig. 9.9).  A l l three grains are classified as type IaB.  T w o grains contain unusually high  concentrations o f nitrogen for this study (maximum recorded values o f 233 and 541 ppm) while diamond 2-10 contains low concentrations o f total nitrogen (24 ppm). It should be noted that IR analysis o f this diamond was performed on a diamond chip, which may contain unusually low amounts o f nitrogen. Examination o f Figs. 7.9 - 7.12 illustrates that that nitrogen concentrations are highly variable. One IR analysis o f a diamond chip w i l l clearly give no indication o f a heterogeneous nitrogen character. The less intense F L colours o f diamond 2-10, compared to the colours i n diamonds 1-4 and 2-6, suggest that the total nitrogen concentrations are low, but likely higher than the suite average o f 72 ppm N . The IR results for diamond 2-10 are questionable, and a large range in nitrogen concentration may a characteristic o f this subgroup. The high aggregation state indicates  235  that the diamonds have likely resided i n the mantle at temperatures higher than is typical for cratons.  Internal growth studies reveal that diamond 1-4 has a complex and unique pattern (although there are similarities with the internal structure o f diamond 2-2). There is a core o f bright turquoise C L , which has what appears to be combined cubic and octahedral faces growing on the core (Figs. 7.10 and 7.12). Nitrogen concentrations range from 20 to 400 ppm and occur only as B centres.  This paragenesis is characterised by the presence o f magneite inclusions, type IaB character, and possibly high nitrogen concentrations. Diamonds exhibit variable F L , C L and body colours. Growth o f diamond, i n part, may occur on both cubic and octahedral faces, which is unique to this study.  9.2.3 Eclogitic and/or peridotitic diamonds from depths greater than ~580 km  Six diamonds have been included i n this subgroup based on the presence o f CaSi03 grains i n the absence o f fPer (diamonds 2-8, 3-4, 3-7, 4-7, 6-6, 7-1). This subdivision is based purely on inclusion assemblage.  T w o diamonds contain touching inclusions o f  CaSi03 and CaTi03 (although the perovskite may be secondary). Diamond 4-10 could also be included i n this subgroup because o f its CaSi03-eGrt disequilibrium association.  The prevalence o f CaSi03 compared to MgSi03 inclusions i n this study is somewhat curious considering that a lower mantle o f peridotitic bulk composition should contain, by weight, 79% M g S i 0 and only 5% C a S i 0 (Wood, 2000). The M g S i 0 / C a S i 0 ratio 3  3  3  3  in mafic material at lower mantle depths should be much closer to unity. Particularly striking are the number o f CaSi03 inclusions occurring i n the absence o f fPer, while no diamonds are found with MgSi03 i n the absence o f fPer. The most plausible explanation for this observation is that CaSi03 inclusions are sourced from mafic material from the  236  transition zone where CaSiCh is stable but M g S i O j and fPer are not (e.g. Figs. 8.3A and B). It is suggested that this paragenetic group is sourced from eclogitic rocks from the relatively restricted depth interval between 580 and 660 km. A n alternate model for the origin of these diamonds would place them in the lower mantle, where CaSi03 is also stable. The observed high C a S i C V M g S i C h ratio is unlikely to occur by chance alone, unless a process can be proposed whereby MgSi03 is selectively removed over C a S i C V One such process could be a preferential fracturing of diamond that contains MgSi03 because of its higher thermal expansion rate compared to CaSi03. To account for both models, diamonds with CaSi03 inclusions in the absence of fPer are classified as having either a mafic or ultramafic source, from depths greater than 580 km.  Out of the six diamonds, two have moderately resorbed forms and one is strongly resorbed. Diamonds occur as either fragments or broken stones. F L colours and intensity are variable, from blue (3 diamonds), turquoise (2) and green (1), while C L colours are either yellow or turquoise (Fig. 9.10). Four grains are type IaB and two are type Ha.  Nitrogen concentration  ranges from -40-225 ppm and averages 105 ppm. The internal morphology of one diamond was examined (28, section 7.3.6), however results show either that the diamond lacked any internal structure, or that the polished surface does not intersect any growth zones. This diamond may have a green C L rim (Fig. 7.17) similar to those seen on upper mantle peridotitic diamonds (Fig. 9.5).  This subgroup is characterised by the presence of CaSi03 in the absence of fPer. Diamonds are either  237  Fig. 9.10. Photos of diamond body colour (left), FL (middle) and CL (right) for diamonds classified to the >580 km paragenesis. From top to bottom, diamonds are: 2-8, 3.4,  3 - 7 , 4 - 7 ,  6-6  and  7-1.  type Ila or IaB with nitrogen concentrations less than 225 ppm. They exhibit a variety o f F L and C L colours and tend to show moderate signs o f resorption and tend not to occur as intact crystals.  9.2.4 U p p e r mantle/lower mantle (-660 k m ) diamonds  Four diamonds have been grouped into the upper mantle/lower mantle ( L M / U M ) subgroup based on the association o f O l and fPer, which can only be i n equilibrium at a narrow depth range straddling the upper mantle/lower mantle boundary (diamonds 1-5, 32, 3-5 and 4-3). Olivines i n diamonds 1-5 and 3-5 may be retrograde, which would place these two diamonds in the lower mantle paragenesis. A l l diamonds contain fPer-01-MgSi03, while two also contain T A P P and one  contains  CaSi03.  In  comparison to  other  inclusions o f the same phases i n this study, the chemistries are unremarkable.  The MgSi03 grains  contain AI2O3 contents that are consistent for M g S i -  Fig. 9.11. Photos of diamond body colour (left), FL (middle) and CL (right) for diamonds classified to the UM/LM peridotitic paragenesis. From top to bottom, diamonds are: 1-5,3-2, 3-5 and 4-3.  Prv near the top o f the lower mantle and are not consistent for diamond inclusion MgSi03 grains from cratonic sources. The average mg of the fPer grains is somewhat higher i n comparison to most grains i n this study for three diamonds (75, 83 and 84 - the average in this study 66.9 ± 1 3 ( l o ) ) and is low for the third (average mg = 60 for fPer's i n diamond 3-5).  Diamonds are moderately resorbed and occur as fragments, broken stones or fractions o f stones.  The F L colours o f this subgroup are variable, from shades o f blue to turquoise  and green.  F L intensity is also variable.  C L colours were only examined for two  diamonds (3-5 and 4-3) and appear moderate blue and turquoise (Fig. 9.11).  238  A l l four diamonds are type IaB, however there is considerable range i n the nitrogen concentrations. Diamond 3-5 contains 32 ppm N , 4-3 (82 ppm N ) , 3-2 (228 ppm total N ) and 1-5 (311 ppm N ) . The high aggregation state is consistent with diamonds that have resided i n the mantle at high temperatures, such as temperatures predicted near the 660 km seismic discontinuity.  The internal growth habit o f diamond 3-5 was examined.  It was one o f only two  diamonds i n this study to exhibit a reasonably simple pattern o f concentric growth layers with some intermittent episodes o f resorption. It has a brighter blue C L core and nitrogen concentration (all as B centres) decreases from core to rim.  This subgroup is characterised by the presence o f fPer and O l inclusions. Diamonds are type IaB with concentrations ranging from 32-211 ppm total nitrogen.  Diamonds  fluoresce blue and are generally w e l l resorbed.  9.2.5 L o w e r mantle diamonds  Twelve diamonds were classified into the lower mantle ( L M ) subgroup based on mineral inclusions alone (diamonds 1-2, 2-2, 2-7, 3-1, 3-6, 3-9, 3-10, 4-16, 5-1, 6-1, 6-2 and 6-9). A l l diamonds contain fPer i n the absence o f 01 (and are thus not restricted to the upper mantle/lower mantle boundary). Four diamonds also contain C a S i 0 3 (2-2, 3-1, 3-10 and 4-16), one o f which may also contain an eclogitic garnet (diamond 4-16), which was determined based on E D S alone (in the absence o f supporting E P M A data and considering the small grain size, the identification o f eGrt cannot be considered conclusive). origin).  Another diamond (2-7) also contains Si02 (although it is o f questionable  The variation i n mg for fPer does not appear to be controlled by inclusion  association. The Mg-number for fPer grains that occur alone are mg = 55, 57, 66, 69, 75, 79, 81 (seven diamonds), for fPer grains i n association with C a S i 0 3 , mg = 50, 77, 78  239  (three diamonds), fPer in association with eGrt (no E P M A results for the fPer) and fPer in association with Si02 (mg  4  = 81).  Diamonds are more resorbed than in any o f the previous subgroups  (mostly resorption categories  tetrahexahedroids  and dodecahedroids).  1 and 2, or Diamonds are  mostly colourless (7 grains) with yellow (3), pink (1), and non-uniform remainder  brown/colourless of  the  subgroup.  (1)  making  Plastic  up  the  I MB ' A • -  deformation  laminations are observed on half o f the diamonds, which is significantly higher than the percentage observed on the population as a whole (20%, section 2.3.7). F L colours for all twelve diamonds are blue and moderate to very weak i n intensity (Fig. 9.12).  Although these F L colours are not  unique to this subgroup, a preliminary subdivision o f diamonds based on this criteria alone may be valid.  Diamonds are either  type  f \  Jv  •  •  2  Ha or IaB, with nitrogen  concentrations ranging from 0 to 400 and averaging ~60 ppm.  The internal structures were examined for four o f the diamonds i n this subgroup. CL  O  ^  *  Three diamonds exhibit blue  o f variable intensity while one diamond exhibits  turquoise C L (diamond 2-2).  The internal structure i n  diamond 2-2 was different from the others (it is similar i n internal morphology to diamond 1-4) and was interpreted as being complex, but likely showing signs o f sector  240  Fig. 9.12. Photos of diamond body colour (left), FL (middle) and CL (right) for diamonds classified to the L M peridotitic paragenesis. From top to bottom, diamonds are: 1-2, 2-2, 2-7, 3-1, 3-6, 3-9, 3-10, 4-16, 51,6-1, 6-2 and 6-9.  dependence o f impurity incorporation.  Some domains o f the diamond were rich i n  nitrogen (225 ppm total nitrogen) while others were essentially type Ha diamond. Diamonds 1-2 and 3-1 exhibit typical diamond growth features, i.e. concentric pattern o f growth with episodic events o f resorption.  Nitrogen concentration for both o f these  diamonds decreases towards the crystal rim.  This subgroup is characterised by the presence o f fPer inclusions i n the absence o f olivine. CaSi03 and T A P P (as well as MgSiC>3) may also be present. Diamonds are type Ha or IaB with generally low total nitrogen concentrations (~<60 ppm N ) . F L colours are blue o f weak to moderate intensity. C L colours are more variable, exhibiting a variety o f shades o f blue o f weak to strong intensity.  Diamonds often exhibit signs o f plastic  deformation, are colourless to various shades o f brown and pink, and are strongly resorbed (classes 1 or 2).  9.2.6 D i a m o n d s of u n k n o w n paragenesis  There are 37 diamonds that cannot be assigned to a source rock subgroup, however the high aggregation state suggests that these diamonds are not sourced from lithospheric mantle. A few diamonds contain elevated concentrations o f total nitrogen (e.g. diamond 4-4, 335 ppm N ; diamond 1-3, 243 ppm N ; and diamond 3-3, 226 ppm N ) , however, the nitrogen concentrations for the 37 diamonds cover the complete spectrum from <20 to 336 ppm N , with no indication o f bimodal character. Sixteen diamonds contain <20 ppm nitrogen, thus making impossible the time/temperature estimates based on nitrogen aggregation.  However, low nitrogen concentrations are uncommon for most diamond  suites and are considered to be an indicator o f a deep source (Kaminsky et al., 2001b).  In terms o f morphology, colour, F L and C L characteristics, these diamonds exhibit a range o f features compatible with those for all other paragenetic groups defined.  There  are a couple o f diamonds that are peculiar and may represent their o w n paragenetic  241  groups. Diamond 2-3 is unique i n that it possesses the most intense yellow body colour and is the only diamond which does not fluorescence. N o inclusions were recovered after cracking this type Ila diamond.  Diamond 2-4 has an intense pink colour (one other  diamond was classified as pink), contains moderate amounts o f nitrogen (94 ppm), all as B centres.  9.3 D i s t r i b u t i o n o f paragenetic groups  The  distribution  of  n=69  paragenetic groups (Fig. 9.13) does not vary much from  the  distribution  determined  in  8.4.1.7.  section  • lower mantle (>660 km) • lower mantle/upper mantle (-660 km) • ~>580km • eclogitic (type IaA and IaAB) • eclogitic (type IaB)  However,  integration o f all studies helps divisions  to  based  on  mineral  inclusion  studies.  In particular,  nitrogen  • peridotitic (upper mantle) • unknown  reinforce  aggregation  Fig. 9.13. Distribution of paragenetic groups for Rio Soriso diamonds based on diamond morphology, colour, FL, CL, internal morphology, nitrogen character and mineral inclusions, n - represents sample size.  state helps to separate diamonds o f upper mantle origin from deeper sourced stones. S i x diamonds contained at least some A centres (-9% o f the population) and one diamond was added to this subgroup, increasing the proportion o f diamonds that likely resided i n the upper mantle to 7 diamonds (10% o f the population). The remaining 62 diamonds (90%) are likely sourced from greater depths, as suggested b y either the 100% aggregation to B centres or the type Ila character. N o diamonds can be placed, with any certainty, to depths between - 2 5 0 and 660 k m . However, any diamonds containing CaSiCh i n the absence o f fPer, T A P P or MgSi03, could be sourced from depths below 580 k m . Four diamonds o f peridotitic origin (6% o f population), are likely sourced from  242  somewhere near the upper mantle/lower mantle divide at - 6 6 0 k m (assuming all phases are i n equilibrium, which may not be justified i n all cases). A t least twelve diamonds (17%) are likely sourced from the lower mantle (660 - 2900 k m depth), however, as o f yet, there are no constraints to place the diamonds within a more restricted depth interval in the lower mantle.  Three diamonds containing magnetite are sourced from eclogitic  material, likely o f deeper origins than the type IaA and I a A B eclogitic diamonds.  The 37 diamonds o f unknown origin likely belong to either the lower mantle, the L M A J M or the ~>580 k m subgroup based on similar nitrogen character (they are all either type Ha and IaB) and they are generally w e l l resorbed, broken crystals with variable F L colours and C L colours o f various shades o f blue and turquoise. However, i n the absence o f supporting data, they w i l l remain unclassified.  9.4 P l u m e origin of R i o Soriso diamonds  The discovery o f several subpopulations i n a diamond suite is typical for most studies (e.g. M c K e n n a , 2001; Gurney et al, in print). However, i n the case o f most studies, the variation i n depth o f origin between subgroups is not so markedly different; most diamond suites represent subgroups sourced from only the cratonic mantle between - 1 5 0 - 250 k m depths, whereas the diamonds i n this study are interpreted as having origins spanning depths from - 2 0 0 to >660 k m . Certainly an explanation is required describing how material spanning such a large depth interval in the mantle could be sampled.  The simplest explanation invokes entrainment i n a mantle plume as it rises from depths within the lower mantle to the base o f the craton.  This model has been proposed by  several authors to explain the occurrence o f fPer as an inclusion in diamond (e.g. Haggerty, 1994; Hutchison, 1997; Griffin et al, 1999; Kaminsky et al, 2001a).  243  Based on evidence from xenoliths and the presence o f lower mantle diamonds, Griffin et al. (1999) conclude that the Slave craton is comprised o f two layers: a shallow ultradepleted upper layer and a less depleted lower layer, with a boundary at 140-150 k m . They concluded that the less depleted lower layer represents the head o f a plume or diapir that incorporated both moderately depleted mantle and subducted crustal material during ascent from >660 k m depth and eventually accreted onto the base o f the craton.  The theory o f superplumes was first proposed by Larson (1991) to explain the voluminous mid-Cretaceous basaltic lavas at Ontong Java. Haggery (1994; 1999) used this superplume model to relate the periodicity o f kimberlite eruptions with superchron events. It has been shown that the Earth's polarity reverses every few m i l l i o n years, with the exception o f several extended periods where no reversals are observed. These periods are referred to as superchrons and occur every - 2 0 0 M a (Haggery, 1999).  The  recognition that kimberlite ages are restricted to time intervals that tend to match those o f superchrons is strong support that there is a link between the outer core and the generation and eruption o f kimberlite magma. Superplumes originate at the core-mantle boundary and have potential to sample material at any depth during, ascent, including lower mantle diamonds.  The age o f emplacement for Juina kimberlites is 95 M a  (Heaman et al., 1998), which places them i n the middle o f the 80-120 superchron. It is suggested that the R i o Soriso suite was entrained i n an ascending superplume from the mid-Cretaceous superchron and were subsequently exhumed to shallow crustal levels i n a kimberlitic magma generated from this superplume.  This model explains how any sublithospheric material can be brought to upper mantle levels.  The advected material was picked up by Group 1 Juina kimberlites ( H .  Cookenboo,  personal  asthenospheric mantle.  communication)  at  -250-300  km  depths  in  the  upper  Group 1 kimberlitic magmas must have asthenospheric origins  based on the Sr-Nd isotopic compositions similar to the B u l k Silicate Earth (Mitchell, 1995). The maximum depth where GrOup 1 kimberlites can form is not constrained, but  244  general consensus is that it becomes increasingly difficult to generate and propagate magma through mantle at greater depths.  During magma ascent to the surface, the  kimberlite must have sampled the cratonic lithosphere containing diamondiferous peridotite and eclogite.  In the proposed scenario, the plume material from the lower  mantle (superplume) was neither underplated to the Amazonian craton nor became part o f the cratonic root. Equally possible, however, is the alternate scenario, where the plume was frozen to the cratonic keel before being sampled by a kimberlite.  9.5 Origin and distribution of eclogitic diamonds  A significant proportion o f R i o Soriso diamonds may have been sourced from rocks o f mafic composition from sub-lithospheric depths up to ~660 k m .  Eclogites i n the cratonic mantle can be produced either by hig