The Origin of Rubies from Paranesti, Northern Greece

THE ORIGIN OF RUBIES FROM PARANESTI, NORTHERN GREECE

Kandy Kan, WANG

A thesis presented in fulfilment of the requirements for the degree of Masters of Philosophy in Geoscience

School of Biological, Earth and Environmental Sciences University of New South Wales February 2020

Thesis/Dissertation Sheet

Surname/Family Name : WANG Given Name/s : KANDY KAN Abbreviation for degree as give in the University calendar : M.Phil Faculty : Science School : Biological, Earth and Environmental THE ORIGIN OF RUBIES FROM PARANESTI, Thesis Title : NORTHERN GREECE

Abstract 350 words maximum: (PLEASE TYPE)

Cabochon-quality ruby-bearing occurrences (here termed PAR-1 and PAR-5) located near Paranesti, north-eastern Greece have been systematically studied. Tectonically, the occurrences are located within the Nestos Shear Zone (NSZ) which separates two distinct geological units. The Rhodope Mountain Complex is a heterogeneous unit of gneisses, mafic, ultramafic and meta-sedimentary rocks in the hanging wall. The footwall Pangaion-Pirin Complex consists of marbles and acid gneisses of a Mesozoic carbonate platform on pre- Mesozoic continental basement.

In this thesis, a range of petrographic and geochemical techniques were used to determine: 1. The genesis of this ruby deposit; 2. Distinctive geochemical and/or oxygen isotopic fingerprints (signature) for the Paranesti rubies and; 3. Do these rubies provide any important additional information on the metamorphic/tectonic evolution of the Rhodope Mountain Complex?

The Paranesti rubies occur within small pargasite schist boudins enclosed surrounded by thin clinochlore schist layers. The footwall and hanging wall lithologies mostly comprise amphibolites. Detailed petrographic and whole-rock analyses have found the Paranesti ruby deposit to be of a mafic/ultramafic protolith, with pervasive fracturing of the ruby grains and margarite reaction rims. The surrounding ruby- deficient amphibolites contain anorthite, along with relict sillimanite, kyanite and chlorite/muscovite/epidote overprinting. EMPA major element analysis determined pargasite as the dominant amphibole within the ruby-bearing amphibole schist host. The surrounding non-ruby bearing chlorite schist mainly comprises clinochlore. Detailed LA-ICP-MS trace element analysis of the colour range of ruby from the two occurrences showed it to be mainly of metamorphic origin, though pale rubies from PAR-5 suggest some metasomatic influence. The rubies display distinctive geochemical signatures with an unusually low oxygen isotope value of ~1. The geochemical signature includes a combination of high Cr (average 2300ppm with 15% sample points on core positions >5000ppm and maximum 8600ppm); high Si (average 1400ppm and maximum 2500ppm), low Mg (average 30ppm) and very low V, Ti and Ga. The estimated P-T conditions of ruby formation are P<7 Kbar and T<750°C. The Paranesti ruby deposit most likely formed during regional amphibolite facies metamorphism which created NSZ, along with a lesser influence of more localised metasomatism. Subsequent multiple greenschist facies retrogression resulted in the host amphibole-chlorite schist assemblages.

Thesis/Dissertation Sheet

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents a non-exclusive licence to archive and to make available (including to members of the public) my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known. I acknowledge that I retain all intellectual property rights which subsist in my thesis or dissertation, such as copyright and patent rights, subject to applicable law. I also retain the right to use all or part of my thesis or dissertation in future works (such as articles or books).

14 February 2021 …………………………………………………………… ……….……………………...…….… Signature Date The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years can be made when submitting the final copies of your thesis to the UNSW Library. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

14 February 2021 INCLUSION OF PUBLICATIONS STATEMENT

UNSW is supportive of candidates publishing their research results during their candidature as detailed in the UNSW Thesis Examination Procedure.

Publications can be used in their thesis in lieu of a Chapter if: • The candidate contributed greater than 50% of the content in the publication and is the “primary author”, ie. the candidate was responsible primarily for the planning, execution and preparation of the work for publication • The candidate has approval to include the publication in their thesis in lieu of a Chapter from their supervisor and Postgraduate Coordinator. • The publication is not subject to any obligations or contractual agreements with a third party that would constrain its inclusion in the thesis

Please indicate whether this thesis contains published material or not:

This thesis contains no publications, either published or submitted for publication ☐ (if this box is checked, you may delete all the material on page 2)

Some of the work described in this thesis has been published and it has been ☒ documented in the relevant Chapters with acknowledgement (if this box is checked, you may delete all the material on page 2)

This thesis has publications (either published or submitted for publication) ☐ incorporated into it in lieu of a chapter and the details are presented below

CANDIDATE’S DECLARATION I declare that: • I have complied with the UNSW Thesis Examination Procedure • where I have used a publication in lieu of a Chapter, the listed publication(s) below meet(s) the requirements to be included in the thesis. Candidate’s Name Signature Date (dd/mm/yy) Kandy Kan Wang 14 Feb 2021 COPYRIGHT STATEMENT

‘I hereby grant the University of New South Wales or its agents a non-exclusive licence to archive and to make available (including to members of the public) my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known. I acknowledge that I retain all intellectual property rights which subsist in my thesis or dissertation, such as copyright and patent rights, subject to applicable law. I also retain the right to use all or part of my thesis or dissertation in future works (such as articles or books).’

‘For any substantial portions of copyright material used in this thesis, written permission for use has been obtained, or the copyright material is removed from the final public version of the thesis.’

Signed ……………………………………………......

Date ……………………………………………......

AUTHENTICITY STATEMENT ‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis.’

Signed ……………………………………………......

Date ……………………………………………...... ABSTRACT

Cabochon-quality ruby-bearing occurrences (here termed PAR-1 and PAR-5) located near Paranesti, north-eastern Greece have been systematically studied. Tectonically, the occurrences are located within the Nestos Shear Zone (NSZ) which separates two distinct geological units. The Rhodope Mountain Complex is a heterogeneous unit of gneisses, mafic, ultramafic and meta-sedimentary rocks in the hanging wall. The footwall Pangaion-Pirin Complex consists of marbles and felsic gneisses of a Mesozoic carbonate platform on pre- Mesozoic continental basement.

In this thesis, a range of petrographic and geochemical techniques were used to determine:

1. The genesis of this ruby deposit;

2. Distinctive geochemical and/or oxygen isotopic fingerprints (signature) for the Paranesti rubies and;

3. Do these rubies provide any important additional information on the metamorphic/tectonic evolution of the Rhodope Mountain Complex?

EMPA major element analysis determined pargasite as the dominant amphibole within the ruby-bearing amphibole schist host. The surrounding non-ruby bearing chlorite schist mainly comprises clinochlore.

Detailed LA-ICP-MS trace element analysis of the colour range of ruby from the two occurrences showed it to be mainly of metamorphic origin, though pale rubies from PAR-5 suggest some metasomatic influence. The rubies display distinctive geochemical signatures with an unusually low oxygen isotope value of ~1. The geochemical signature includes a combination of high Cr (average 2300ppm with 15% sample points on core positions >5000ppm and maximum 8600ppm); high Si (average 1400ppm and maximum 2500ppm), low Mg (average 30ppm) and very low V, Ti and Ga.

The estimated P-T conditions of ruby formation are P<7 Kbar and T<750°C. The Paranesti ruby deposit most likely formed during regional amphibolite facies metamorphism which created NSZ, along with a lesser influence of more localised metasomatism. Subsequent multiple greenschist facies retrogression resulted in the host amphibole-chlorite schist assemblages.

This is the first time that SIMS has been used for in situ oxygen isotope analysis of metamorphic corundums.

Acknowledgements

I would like to acknowledge and sincerely thank the following people and organizations that provided assistance, support, and help throughout the four years of my Master study.

My sincere gratitude to Ian Graham, my supervisor, co-author and friend on the project. Thank you for all the patience and guidance over the last 5 years while I manage between full time work, part time research and a wedding.

Special thanks to Panos Voudouris (co-supervisor), my co-authors Gaston Giuliani, Angela Lay, Stephen Harris, Anthony Fallick and my project reviewers Mike Archer, Bryce Kelly, Carol Oliver, David Cohen (co-supervisor) and the late Colin Ward.

I wish to thank Joanne Wilde formerly of the School of Biological, Earth and Environmental Sciences, UNSW Sydney, for the assistance with petrographic sample preparation and the polished mounts required for SIMS analysis. Karen Privat from the Electron Microscopy Unit (EMU) of the Mark Wainwright Analytical Centre (UNSW) for the assistance in EMPA analyses. Funding for this project came from UNSW Faculty of Science Research Grant to Ian Graham (supervisor) and Australian Institute of Nuclear Science and Engineering (AINSE) grants AINGRA07061 and AINGRA08025. Funding for some of the analytical equipment used in this project was provided through UNSW MREII Grants and Australian Research Council (ARC) Large Infrastructure and Equipment Funds (LIEF) grant LE0989067. The analytical data were obtained using instrumentation funded by DEST Systemic Infrastructure Grants, ARC LIEF, NCRIS/AuScope, industry partners, and Macquarie University. This is contribution 982 from the ARC Centre of Excellence for Core to Crust Fluid Systems (http://www.ccfs.mq.edu.au) and 1161 in the GEMOC Key Centre (http://gemoc.mq.edu.au). I would also like to acknowledge the XRF Facility within the Mark Wainwright Analytical Centre at the University of New South Wales, and the Q-ICP-MS unit in the School of Earth Sciences, University of Melbourne, Australia for their analytical support.

I want to thank Laure Martin for the SIMS analyses and acknowledge the facilities and the scientific and technical assistance of the Australian Microscopy & Microanalysis Research Facility at the Centre for Microscopy, Characterisation & Analysis, The University of Western Australia, a facility funded by the University, State and Commonwealth Governments.

I would also like to thank David Turner and Julien Berger for their reviews, and the Editor of the Canadian Mineralogist Lee Groat and assistant editor of Minerals Anker He for their comments, all of which greatly improved the final manuscript.

I would like to extend my appreciation to my managers from AXA Investment Managers (Matthew Spithill, Brett Dillon) and EG Funds Management (Adrian Lee) for the kind understanding and encouragement when I had to take time out of working hours to attend conferences and finish the research project.

To my family, mum, dad and grandma for all the love, care and picking up life’s chores while I re-live my life as a student. It has been a long journey together and you’ve always supported whatever I wanted to achieve in life.

To my dear husband Greg, you shine without fail and always by my side whenever I needed. Thanks for supporting my interest throughout the years and starting a new chapter in life together. If ever two were one, then surely we.

Publications

Wang, K.K., Graham, I.T., Lay, A., Harris, S.J., Cohen, D.R., Voudouris, P., Belousova, E., Giuliani, G., Fallick, A.E. and Greig, A., 2017. The origin of a new pargasite-schist hosted ruby deposit from Paranesti, Northern Greece. The Canadian Mineralogist, 55(4), pp.535-560. (Appendix 1)

Wang, K.K., Graham, I.T., Martin, L., Voudouris, P., Giuliani, G., Lay, A., Harris, S.J. and Fallick, A., 2019. Fingerprinting Paranesti rubies through oxygen isotopes. Minerals, 9(2), p.91. (Appendix 2)

Conference Presentations

35th International Geological Congress 27 August – 4 September 2016, Cape Town, South Africa

Wang, K.K., Graham, I.T., Voudouris, P., Giuliani, G., Fallick, A.E. and Cohen, D.R., (2016), Origin of rubies from the Paranesti region, North-Eastern Greece. 35th International Geological Congress, Cape Town, South Africa, 1766. (Appendix 4)

XXII Meeting of the International Mineralogical Association 13-17 August 2018, Melbourne, Australia

Wang, K.K., Graham, I.T., Martin, L., Voudouris, P., Lay, A., Harris, S.J., Belousova, E., Giuliani, G. and Fallick, A.E, 2018. Geographic typing of gem corundum taken a step further via in-situ oxygen isotope and trace element analysis: the example of Paranesti, Greece. XXII Meeting of the International Mineralogical Association, Melbourne, Australia, 1121 (Appendix 5) TABLE OF CONTENTS

Abstract ...... i Acknowledgements ...... ii Publications and Conference Presentations ...... iii

Chapter 1. Introduction

1.1 Research Significance and Objectives ...... 1 1.1.1 Research Significance ...... 1 1.1.2 Research Objectives ...... 1 1.2 Mineralogy of Gem Corundums ...... 2 1.2.1 Gem corundum chemistry ...... 3 Major elements ...... 3 Trace elements – chromophores ...... 3 Source of trace elements ...... 4 Trace element applications ...... 5 Radiogenic isotopes ...... 5 1.2.2 Inclusions in gem corundums ...... 6 Mineral inclusions ...... 6 Fluid and gas inclusions ...... 9 1.3 Gem corundum deposit types...... 9 1.3.1 Metamorphic ruby deposits ...... 10 1.3.1.1 Metamorphic deposits ...... 10 1.3.1.2 Metasomatic deposits ...... 11 1.3.2 Distribution and age ...... 12 1.3.3 Geological setting ...... 13 1.3.4 Associated mineral assemblages ...... 14 1.3.5 P-T conditions ...... 14 1.3.6 Source of chromium ...... 15 1.4 Worldwide distribution of ruby deposits ...... 15 1.5 Oxygen isotopes ...... 16 1.5.1 Introduction ...... 16 1.5.2 Oxygen isotopes in gem corundums ...... 17 Chapter 2. The Paranesti Ruby Deposit

2.1 Regional geological setting ...... 19

2.1.1 Regional geology ...... 19

2.1.2 Rhodope mountain complex (RMC) ...... 19

2.2 Local geological setting ...... 21

2.2.1 Nestos shear zone (NSZ) ...... 21

2.2.2 Paranesti ruby location ...... 23

2.3 Sample processing ...... 24

2.4 Analytical methods ...... 25

2.4.1 Microscopy...... 25

2.4.2 X-ray diffraction (XRD) ...... 25

2.4.3 X-ray fluorescence (XRF) ...... 25

2.4.4 Inductively coupled plasma–mass spectrometry (ICP-MS) ...... 26

2.4.5 Scanning electron microscopy (SEM) and electron microprobe a nalysis (EMPA) ...... 26

2.4.6 Laser ablation inductively coupled plasma–mass spectrometry (LA-ICP-MS) ...... 27

2.4.7 Oxygen isotope samples and method ...... 27

2.4.7(a) Whole grain dissolution laser-fluorination method ...... 28

2.4.7(b) In situ secondary ionisation mass spectrometry (SIMS) method .. 28

2.5 Results ...... 29

2.5.1 Petrography and mineralogy ...... 30

2.5.1 (a) Ruby-bearing samples ...... 30

2.5.1 (b) Non-ruby-bearing samples ...... 32

2.5.1 (c) Whole-rock XRD ...... 34

2.5.2 Whole-rock geochemistry ...... 34

2.5.2 (a) Whole-rock XRF ...... 34 2.5.2 (b) Whole-rock ICP-MS ...... 36

2.5.3 SEM and EMPA ...... 37

2.5.3 (a) SEM imaging ...... 37

2.5.3 (b) EMPA analysis ...... 37

2.5.4 LA-ICP-MS analysis ...... 42

2.5.5 Oxygen Isotope ...... 46

2.5.5(a) Whole grain dissolution laser-fluorination results ...... 46

2.5.5(b) In-situ secondary ionisation mass spectrometry (SIMS) results ... 47

Chapter 3. Discussion and Conclusions

3. Discussion and conclusions ...... 50 3.1 Distinctive geochemical and oxygen isotope signatures of Paranesti ruby ..... 51 3.1.1 Geochemical signature - Metamorphic vs Magmatic ...... 51 3.1.2 Mafic/ultramafic protolith ...... 53 3.1.3 Oxygen isotope signature - PAR-1 vs. PAR-5 variations ...... 54 3.2 New understanding on the genesis of Paranesti ruby ...... 55 3.2.1 Origin of the Paranesti rubies ...... 55 3.2.2 Possible P-T model for the Paranesti rubies ...... 57 3.3 Comparison with rubies from deposits around the world ...... 60 3.4 Oxygen isotopes as an identifier for gem corundum geological origin ...... 61 3.4.1 Global low to ultra-low oxygen isotope corundum comparison ...... 62 3.4.2 Possible Causes for low oxygen isotope corundum formation ...... 64 3.4.2 (a) Kinetic isotope fractionation ...... 64 3.4.2 (b) Thermal diffusion ...... 64 3.4.2 (c) Other ultra-low δ18O protolith ...... 64 3.4.2 (d) Hydrothermal alteration model ...... 65 3.4.3 Oxygen Isotope standard for rubies ...... 66 3.5 Conclusions ...... 67 3.5.1 Broader tectonic implications ...... 67 3.5.2 Future areas of research ...... 68

List of tables

Chapter 1

Chapter 2

Table 1a. Major element XRF analysis of the ruby-bearing host rock (wt.%) ...... 35

Table 1b. Trace element XRF analysis of the ruby-bearing host rock (ppm) ...... 35 Table 2. Whole-rock ICP-MS trace element analysis of the ruby-bearing host rock (Sample PAR-1) ...... 36 Table 3. Representative EMPA analyses of Paranesti ruby grains. (38 points analysed for PAR-1 rubies and 20 points analysed for PAR-5 rubies) ...... 38 Table 4. Representative electron-microprobe analyses of spinel inclusions within ruby total of 5 points ...... 39 Table 5. Representative electron-microprobe analyses of margarite – reaction rim around ruby grains total of 6 points ...... 40 Table 6. Representative EMPA analyses of amphiboles surrounding ruby grains within the ruby- bearing amphibole schists PAR-1 (37 points) ...... 41 Table 7. Representative EMPA analyses of chlorite from the zone surrounding the ruby-bearing amphibole schist boudins PAR-1 (20 points) ...... 42 Table 8. Chromophore and key trace elements LA-ICP-MS analyses (ppm) of the Paranesti rubies and their ratios. Polished Mounts (Position: C-Core R-Rim; Colour intensity: P-Pale M-Medium D- Dark) ...... 43 Table 9. Other trace element LA-ICP-MS analyses (ppm) of the Paranesti rubies. Polished Mounts (Position: c-core r-rim; Colour intensity: p-pale m-medium d-dark) ...... 44

Table 10. Oxygen isotope results from the 2009 reconnaissance study using the laser-

ﬂuorination method, n=1 ...... 46 Table 11. Oxygen isotope results from the 2018 study using the laser-ﬂuorination method, n=3 ...... 47 Table 12. Oxygen isotope results from 2017 using the SIMS method ...... 47 Table 13. PAR-5 ruby grains core vs rim oxygen isotope results comparison ...... 48

Chapter 3

Table 1. Summary of Paranesti ruby results (Wang et al., 2017) ...... 50 Table 2. Comparison between the Paranesti amphibole schist hosted ruby deposits and African amphibole-hosted ruby deposits (adapted from Mercier et al., 1999a; Mercier et al., 1999b; Schwarz et al., 2008 and Pardieu et al., 2013) ...... 58 Table 3. Global comparison of corundums with low to ultra-low oxygen isotope values ...... 63

List of figures

Figure 1. Gem Corundum Classification Model (adapted from Simonet et al., 2008) ...... 69 Figure 2. Primary Corundum Classification Model – Lithology Based (adapted from Giuliani et al., 2014)...... 69 Figure 3. Reconstruction of Gondwana with the location of gem deposits along the Pan-African – Indian subcontinent Gondwana collision (extract from Giuliani et al., 2007)...... 70 Figure 4. Marble-hosted ruby deposits from central and southeast Asia from Garnier et al., (2006)...... 71 Figure 5. Gem corundum occurrences and associated basaltic fields along the West Pacific continental margin intraplate basaltic fields (from Graham et al., 2008)...... 72 Figure 6. P-T conditions for the formation of gem corundum in metamorphic deposits (From Giuliani et al., 2014)...... 73 Figure 7. Significant ruby deposits from around the world (Shor and Weldon, 2009) ...... 74 Figure 8. Corundum deposit type based on oxygen isotope ratios (generalised from Giuliani et al., 2005 and Vysotskiy et al., 2015) ...... 75 Figure 9. Geological map of the Aegean and Anatolia, with Paranesti shown (red star) (adapted from Bozkurt 2001)...... 76 Figure 10. Geological map of the Rhodope Mountain Complex with Paranesti shown (red star). Adapted from Collings et al., (2016). ER: Eastern Rhodope Mountains; CR: Central Rhodope Mountains; WR: Western Rhodope Mountains; NSZ: Nestos Shear Zone; CSZ: Chepelare Shear Zone...... 77 Figure 11. Topographic map of the Rhodope Mountain Complex showing the Nestos Shear Zone (NSZ) and the Chepelare Shear Zone (CSZ). The two arrows indicate the vergence of Alpine thrusting in the Balkanides (to the north) and the Hellenides (to the southwest) (Gautier et al., 2017)...... 78 Figure 12. Geological map of the Rhodope Mountain Complex showing the main tectonic zones, with Paranesti located within the Nestos Shear Zone (red star) (adapted from Voudouris et al., 2019)...... 79 Figure 13. Location map of the deposits showing their close proximity to each other and distance from the closest town of Paranesti...... 79 Figure 14. PAR-1 main ruby working showing intrusive contact between pargasite schist (green) and granite pegmatite (grey-white, outlined in red)...... 80 Figure 15. Original ruby discovery in road-cutting along the main road – PAR-5...... 80 Figure 16. Kyanite-bearing pargasite-chlorite schist at the road-side deposit PAR-5...... 81 Figure 17. Non-ruby-bearing pargasite schist boudins along the road from PAR-5 towards the Perivlepto village. The white layers above the boudin are thin marble bands...... 81 Figure 18. Stratigraphy schematics of the sample sites...... 82 Figure 19. Deep red-coloured ruby with both tabular and barrel shaped crystals within pargasite schist...... 82 Figure 20. Deep red-coloured ruby with barrel-shaped crystal within pargasite schist ...... 83 Figure 21. Ruby associated with kyanite ...... 83 Figure 22. Numbered ruby grains in grain mounts analysed using LA-ICP-MS ...... 84 Figure 23. Images of ruby samples from Paranesti. Dark red ruby crystals (0.5-1.0 cm) from PAR- 1 in pargasite schist host rock...... 85 Figure 24. Cluster of pale pink ruby crystals (0.5-1.5 cm) from PAR-5 in pargasite schist host rock ...... 85 Figure 25. Clean cut surface of PAR-1 ruby sample free from inclusions used for the 2009 laser fluorination oxygen isotope analysis ...... 86 Figure 26. Cluster of rubies ~ 1cm in diameter, showing the range of colours found ...... 86 Figure 27. Mounted PAR-1 ruby grains used for in situ SIMS analysis ...... 87 Figure 28. Images of SIMS analysed samples showing oxygen isotope values ...... 88 Figure 29. Petrographic images of PAR1A samples ...... 89 Figure 30. Petrographic images of PAR1B sample ...... 91 Figure 31. Petrographic images of PAR5C samples ...... 92 Figure 32. Petrographic images of PAR5D samples ...... 94 Figure 33. Petrographic images of PAR2A samples ...... 95 Figure 34. Petrographic images of PAR3A samples ...... 96 Figure 35. Petrographic images of PAR4A samples ...... 98 Figure 36. Petrographic image – polished thin section PAR4B...... 99 Figure 37. Petrographic images of PAR4C samples ...... 101 Figure 38. Chondrite-normalised rare earth element diagram showing moderate negative Eu anomaly and concave downwards curve for the host pargasite schist (sample PAR-1) (chondrite values used are those of Sun and McDonough, 1989) ...... 102 Figure 39. SEM Images – PAR1A phyllosilicate rims around rubies ...... 103 Figure 40. PAR1B spinel inclusions in rubies (core – 1B20, rim – 1B21, outside – 1B22) ...... 103 Figure 41. Triangular spinel classification diagram (Cr3+-Fe3+-Al3+) (adapted from Gawlick et al, 2015) ...... 104 Figure 42. Binary classification diagram considering the Mg2+-Fe2+ exchange (adapted from Gawlick et al, 2015) ...... 104 Figure 43. Diagram showing the range in composition of host amphiboles from the ruby-bearing amphibole schists. Note that all plot close to the pargasite end-member. Assumption 20% Fe3+ and 80% Fe2+ (adapted from Colville et al, 1966) ...... 105 Figure 44. Classification diagram for chlorite from the Paranesti deposit (adapted from Hey, 1954) ...... 105 Figure 45. Oxygen isotopic compositions from the 2009 Greek corundum study (adapted from Giuliani et al., 2007; Wang et al., 2019; Voudouris et al., 2019) ...... 106 Figure 46. Mg(x100)-Fe-Ti(x10) trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for magmatic and metamorphic corundums. Adapted from Sutherland et al. (2009) ...... 107 Figure 47. Cr(x10)-Fe-Ga(x100) trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for magmatic and metamorphic corundums. Adapted from Peucat et al. (2007) ...... 108 Figure 48. Ga/Mg vs Fe/Mg trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for metamorphic, transitional and magmatic corundums. Adapted from Sutherland et al. (2015) ...... 108 Figure 49. Ga/Mg vs Fe/Mg trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for metamorphic, transitional and magmatic corundums with R2. Adapted from Sutherland et al. (2015) ...... 109 Figure 50. Cr/Ga vs Fe/Ti trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for metamorphic and magmatic corundum. Adapted from Sutherland et al. (2009) ...... 109

Figure 51. FeO+TiO2+Ga2O3 vs FeO-Cr2O3-MgO-V2O3 trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for metamorphic, transitional and magmatic corundums. Adapted from Giuliani et al. (2010, 2014a) ...... 110 Figure 52. (a) P–T diagram (marked in blue) showing mineral equilibria related to the formation of Paranesti rubies and (b) Estimated P-T path (marked in blue) for Paranesti ruby formation compared to other known metamorphic and metasomatic ruby deposits (adapted from Voudouris et al., 2019). P–T conditions for the formation of corundum in metamorphic deposits (modified from Simonet et al., 2008); P–T fields of North Carolina (Tenthorey et al., 1996); Mangare (Mercier et al., 1999); Morogoro (Altherr et al., 1982); southern Kenya (Key and Ochieng 1991; Simonet 2000); Hunza (Okrusch et al., 1976); Sri Lanka (De Maesschalk and Oen 1989); Greenland (Garde and Marker 1988) and Kashmir (Peretti et al., 1990) with three P–T boxes corresponding to the evolution of the fluids in the sapphire crystals from the center (c), to intermediate (i) and outer (o) zones, Urals (Kissin 1994), and Mong Hsu (Peretti et al., 1996)….111 Figure 53. Oxygen isotopic comparison of rubies from Paranesti occurrences compared with low δ18O corundums from Karelia and Soamiakatra in Madagascar (Wang et al., 2019; Vysotskiy et al., 2015; Giuliani et al., 2007; Bindeman and Serebryakov, 2011) ...... 112

References ...... 113

List of appendices

Appendix 1 - Wang, K.K., Graham, I.T., Lay, A., Harris, S.J., Cohen, D.R., Voudouris, P., Belousova, E., Giuliani, G., Fallick, A.E. and Greig, A., 2017. The origin of a new pargasite-schist hosted ruby deposit from Paranesti, Northern Greece. Canadian Mineralogist, 55(4), 535-560 ...... 132 Appendix 2 - Wang, K.K., Graham, I.T., Martin, L., Voudouris, P., Giuliani, G., Lay, A., Harris, S.J. and Fallick, A., 2019. Fingerprinting Paranesti rubies through oxygen isotopes. Minerals, 9(2), 91 ...... 159 Appendix 3 - Voudouris, P., Mavrogonatos, C., Graham, I., Giuliani, G., Melfos, V., Karampelas, S., Karantoni, V., Wang, K., Tarantola, A., Zaw, K. and Meffre, S., 2019. Gem Corundum Deposits of Greece: Geology, Mineralogy and Genesis. Minerals, 9(1), 49 ...... 174 Appendix 4 - Wang, K.K., Graham, I.T., Voudouris, P., Giuliani, G., Fallick, A.E. and Cohen, D.R., (2016), Origin of rubies from the Paranesti region, North-Eastern Greece. 35th International Geological Congress, Cape Town, South Africa, 1766 ...... 216 Appendix 5 - Wang, K.K., Graham, I.T., Martin, L., Voudouris, P., Lay, A., Harris, S.J., Belousova, E., Giuliani, G. and Fallick, A.E, 2018. Geographic typing of gem corundum taken a step further via in-situ oxygen isotope and trace element analysis: the example of Paranesti, Greece. XXII Meeting of the International Mineralogical Association, Melbourne, Australia, 1121 ...... 218 Appendix 6 – Supplementary Table ...... 221 Table 1. Detection limits for major and trace elements of the PANalytical PW2400 Sequential WDXRF Spectrometer ...... 221 Table 2. Detection limit of Agilent 7700x ICP-MS used for trace element analysis ...... 222 Table 3. Analytical setting for EMPA analyses ...... 223 Table 4. Corrected δ18O results (V-SMOW) from SIMS analyses ...... 224 Electronic Appendix 7 – Oxygen isotope data raw (excel) Electronic Appendix 8 – EMPA data raw (excel) Electronic Appendix 9 – LA-ICP-MS data raw (excel) Chapter 1. Introduction

1.1 Research Significance and Objectives

1.1.1 Research Significance

This thesis examines a ruby deposit from north-eastern Greece near the township of Paranesti. It aims to determine the origin and evolution of this deposit and to see if it is possible to geographic-type the rubies in terms of unique geochemical and/or oxygen isotopic characteristics. The significance of this project is not only in it being the first systematic study on rubies from the Paranesti deposit, but the study additionally contributes important mineralogical, petrological, trace element and oxygen isotope data to a larger ongoing study on gem corundum occurrences and deposits worldwide. This study is also the second to use the secondary ion mass spectrometry (SIMS) method for in situ oxygen isotope analysis on corundum and the first on metamorphic corundums. The unique oxygen isotope signature found for the Paranesti rubies, combined with their Cr content and trace element signatures could potentially be used to geographic type rubies from this deposit.

1.1.2 Research Objectives

There are main three questions that this research aims to address.

1. What is the genesis of this north-eastern Greek gem corundum deposit? 2. Is there a unique geochemical and/or isotopic fingerprint (signature) for the rubies from this deposit and what are the scientific and/or commercial implications? 3. Do these rubies provide any important additional information on the metamorphic/tectonic evolution of the Rhodope Mountain Complex?

A number of geochemical, petrographic and isotopic analytical methods have been employed to answer these research questions. The range of analyses undertaken are listed below: 1. Thin sections/ polished mounts – identification of mineral phases, mineral inclusions within rubies, metamorphic and alteration assemblages and overprinting assemblages, paragenesis 2. Electron Microprobe –major and minor element composition of ruby and associated phases 3. XRD - mineral phase identification of the mineral assemblages of the host and surrounding rocks 4. XRF - whole rock major element composition 5. ICP-MS - whole rock trace element composition

6. LA-ICP-MS – detailed in situ trace element composition 7. SIMS - in situ oxygen isotope analysis of the rubies

1.2 Mineralogy of Gem Corundums

The history of ruby mining can be traced as early as 2500 BC. Certain gem corundum locations have consistently produced high-quality stones over time, such as the Mogok field of Myanmar. In this field, archeological finds of pre-historic tools have been made, showing the fascination of people with gemstones has existed for thousands and maybe even tens of thousands of years (Clark, 1998). Gem corundum is one of the most valuable gemstones in the jewelry industry throughout history and across multiple cultures and civilisations. Gem corundum holds two (ruby and sapphire) out of the four (diamond, ruby, sapphire and emerald) precious stone categories. All other gemstones are classified as semi-precious. On a per carat basis, ruby can command a higher price than diamond on the world market (Robertson, 2012). This is primarily attributed to the fact that high-quality commercially viable gem corundum deposits are very rare.

The word corundum is taken from the Sanskrit word ‘Kuruvinda’ (Kucera, 1984). It is the second hardest naturally occurring mineral after diamonds and ranks nine on the Mohs scale. Its habit allows for good polishing facets to be produced. In corundum’s natural state, it is colourless (white). Only the coloured crystals with transparency are classified as gem corundum. Based on their colours, different names have been attributed to the same mineral throughout the history of time.

The most valuable variety of the gem corundum is “Ruby”. The name ruby comes from the Latin word ruber, which means red (Giuliani et al., 2014). In the ancient Indian culture, Ruby was given the name of “Ratnaraj” literally meaning ‘King of Gemstones’ and “Ratnanayaka” meaning “Leader of precious stones” (Clark, 1998). All other colour varieties of the gem corundum are called “Sapphire”. The name sapphire comes from the Greek word sappheiros, meaning “precious blue gem” (Giuliani et al., 2014). The most well-known sapphire colour is blue. Its colour is the standard against which other blue gems—from topaz to tanzanite—are measured. A special orange-pink sapphire colour is called “Padparadscha”. It means “lotus flower” in Sinhalese, the language spoken in Sri Lanka. Given the high value of these corundums with their special features, locality association features dominantly in the ruby and sapphire trades worldwide. For example, the rubies that could command the highest esteem and price are the

Burmese pigeon-blood. Ceylon sapphires from Sri Lanka are viewed as the most valuable for blue sapphires. High quality gem ruby can be as expensive as diamond (Shor and Weldon, 2009).

Traditionally, gemologists have focused their study of the coloured sapphires and rubies based on their locality and artificial treatments. Unlike the Gemological Institute of America (GIA)’s globally standardized system of quality analysis for diamonds, there’s no consistent grading systems developed for gem quality corundums to date.

1.2.1 Gem corundum chemistry

Major elements

Corundum belongs to the hematite group (X2O3) of rhombohedral oxides. It is an aluminums oxide mineral – Al2O3 (52.91 % of Al and 47.09 % of O2) that comes in a variety of colours and forms (Giuliani et al., 2007; Simonet et al., 2008; Dill, 2010). Corundum crystallises in the 32/m holohedral class of the rhombohedral system, though most of its crystalline general forms belong to the hexagonal crystal system.

There are a variety of Al sources. Aluminum could be released during fluid interactions between mafic or ultramafic rocks (e.g. serpentinites) and Al-rich rocks (e.g. granites, pegmatites, paragneisses). Al2O3 remains in excess in the granite rocks where desilication processes have occurred to form corundum, spinel and/or aluminosilicates (Lawson, 1903; Du Toit. 1946; Robb and Robb, 1986). Gabbro is rich in Al and could be the protolith source of releasing the Al to form corundum (Giuliani et al., 2014). Anorthite-plagioclase-rich rocks such as anorthosites are also likely source rocks as they are sufficiently Al-rich to form corundum during partial melting (Palke et al., 2016).

Trace elements – chromophores

Studies have found that the colours are determined by the inclusion of trace amount of impurities (chromophores). The chromophores can influence the colour of corundum in multiple ways.

a. Directly causing colour b. Chemically interacting with one another to cause colour c. Modify the strength of a colour

When the chromophores have the same chemical valence as Al3+ e.g. Cr3+, it is termed isovalent impurity. When the valency is different from Al3+ such as Mg2+ or Ti4+, it is termed as aliovalent impurity (Emmett et al., 2003).

The red in rubies is due to the substitution of chromium (Cr3+) for aluminium (Al3+) while the blue is related to the intervalence charge transfer between iron (Fe2+) and titanium (Ti4+) which can substitute for Al (Fritsch and Rossman, 1988). Some sapphires are yellow or green due to iron atoms (Deer et al., 1992; ferric [Fe3+] and ferrous [Fe2+] iron, respectively). Other common chromophores in the corundum structure include V, Ni, Ga, Ca and Mn. Less than 0.01% of Fe and Ti are necessary to obtain the blue colour of sapphire (Giuliani et al., 2007). Coloured gem corundums ranging from fuchsia to reddish are also classified as sapphires if the concentration of Fe is higher than that of Cr (Caucia and Boiocchi 2005; Andriamamonjy et al., 2013). Therefore, the definition of ruby is not only based on the colour of the stone to the naked eye, but most importantly it is quantified by a chemical definition of Cr2O3 > Fe2O3.

Trace element ratios such as Ga/Al, Cr/Ga and Ga/Mg are useful indicators of the metamorphic vs magmatic origin of the rubies and sapphires. Ga/Al ratios (=10,000 Ga/Al) between 1 – 1.5 are representative of metamorphic environments while those with higher ratios (2.5-5.3) represent magmatic environments (Whalen et al., 1987). Cr/Ga ratios <1 are typical of magmatic corundums and higher values typical of metamorphic corundums (Abduriyim and Kitawaki, 2006). Peucat et al., (2007) found that high Ga/Mg ratios (>10) are indicative of magmatic sapphires while low ratios (<10) are indicative of metamorphic and metasomatic sapphires.

Source of trace elements

Enrichment of Ti may reflect submicroscopic rutile inclusions, resulting in a specific and enlarged Mg–Ti trend. Gallium is known to substitute for Al and the Ga/Al ratio is relatively constant in the continental crust. Whalen et al., (1987) explained that during the circulation of F-rich fluids

− 3 during partial melting of granulites, F and Ga would form GaF6 ions extracted during the formation of alkaline melts (with high Ga/Al ratios), whereas the granulitic restites would be enriched in plagioclase and thus in Al and therefore depleted in Ga (with low Ga/Al ratios). This F-rich fluid circulation process would thus explain the low Ga/Al ratios observed in metamorphic corundum. Mg concentration in corundum appears to be linked to the protolith/host rock. Simonet et al., (2004) found the Mg content to be below detection limits in the syenite hosted Garba Tula sapphires. Metasomatic sapphires related to plumasites were found to contain high

Mg contents such as those from Kashmir, in agreement with the high Mg content of the ultrabasic host rocks (Peucat et al., 2007).

Trace element applications

Several geochemical determination techniques have been developed based on the trace element geochemistry and widely adopted over the last two decades for the identification of corundum genesis between magmatic vs metamorphic. These techniques are especially useful for the understanding of the alluvial gem corundums where the source of the host rock and the corundum genesis is unknown. The main trace element discrimination diagrams are listed below:

a) Cr/Ga vs. Fe/Ti - Metamorphic vs Magmatic (Sutherland et al., 1998) b) Ga/Mg vs. Fe – Metamorphic vs Magmatic (Peucat et al., 2007) c) Fe-Ti-Mg ternary plot – Metamorphic vs Magmatic as well as deposit identification (Peucat et al., 2007)

d) FeO-MgO-V2O3 vs. FeO+TiO2+Ga2O3 plot – protolith composition (Giuliani et al., 2014)

Peucat et al., (2007) found Ti and V to be less of a constraining factor. V has a broad range of concentrations and therefore is not useful as a discriminatory element, although high values are found in metamorphic sapphires only.

For eastern Australian alluvial sapphires and rubies, the distinction between “magmatic” and “metamorphic” sources through trace elemental ratios have already been questioned, due to the finding of a continuous spectrum of trace elemental compositions by Wong et al., (2017).

Radiogenic Isotopes

Radiogenic isotope ratios of 87Sr/86Sr, 143Nd/144Nd and Pb are largely unaffected by processes that can disturb the trace elemental ratios, such as crystal growth from melts (Krebs et al., 2019). Theoretically, these radiogenic isotopes can act as geochronometers to directly date the crystal growth or the recrystallisation of the corundum grains (Krebs et al., 2019). No direct method of dating corundum was developed prior to 2019 due to the difficulties in dissolving the robust crystal lattice in order to release the relevant radiogenic elements for pre-concentration and measurement.

Historically, the time constraints on ruby formation was predominantly restricted to the dating of mineral inclusions within the corundum grains (e.g. U-Pb age dating of zircon inclusions; 5

Garnier et al., 2005; Graham et al., 2008). However, one risk on the accuracy of this approach is that the zircon formation could predate the corundum and become trapped during crystallisation. The U-Pb age of rutile is less ambiguous and could constrain the minimum age of corundum formation (Sorokina et al., 2017). The first direct age determination using the Sr-Pb isotopic system on pink sapphire and rubies from Aappaluttoq deposit in SW Greenland was performed by Krebs et al., (2019) using an off-line laser sampling technique and Thermo Scientific Triton Plus (TIMS).

1.2.2 Inclusions in gem corundums

Dr Eduard J Gübelin – the father of modern gemology and founder of the Gübelin Gem Lab was an early proponent in the study of inclusions and microscopic features in gemstones to determine their origin.

Mineral inclusions

Mineral inclusions in gem corundum are relatively rare but significantly associated with the origin of the corundum host. Syngenetic inclusions within gem corundums can provide important information with regards to their origin (e.g. Guo et al., 1996; Graham et al., 2008).

Rutile is one of the more commonly observed mineral inclusions in corundum. Star sapphire is a gemological term reserved for asteriated sapphire and ruby caused by rutile inclusions (Giuliani et al., 2007; Dill, 2010). Common mineral inclusions in magmatic sapphires include zircon, rutile, spinel, columbite-group and feldspar-group minerals. Metamorphic corundum inclusions are more likely to be sapphirine, spinel, clinopyroxene and garnet. Refer to Table 1 for a summary of mineral inclusions in gem corundums worldwide.

Table. 1 - Summary of mineral inclusions found in gem corundum. 1. Graham et al., 2008; 2. Sutherland et al., 2008a; 3. Pardieu et al., 2014; 4. Sutherland et al., 2008b; 5. Seifert et al., 2014; 6. Uher et al., 2012; 7. Guo et al., 1996; 8. Wang et al., 2017; 9. Yakymchuk and Szilas 2018 and 10. Rakotondrazafy 2008. Inclusions Deposits Corundum Type Ref Anorthite Eger (Ohre) rift, Germany Metasomatic sapphire 5 Oligoclase Weldborough, Tasmania, Magmatic sapphire 1 Australia Nepheline Podgelbanochny, Russia Magmatic sapphire 1 Anorthoclase Cudgegong/Wellington, NSW, Magmatic sapphire 1 Australia Alkali feldspar Kanchanaburi, Thailand Magmatic sapphire 1 Albite Ban Huai Sai, Laos Magmatic sapphire 1

Feldspar and Felspathoid and Feldspar Plagioclase Mercaderes–Rio Mayo area, Metasomatic sapphire 2 Colombia

Biotite Nuuk, Greenland Metasomatic ruby 9 Margarite Dak Nong, Vietnam Magmatic sapphire 1 Mica Phlogopite Soamiakatra, Madagascar Metamorphic ruby 10

4 High Al- Pailin, Cambodia Metamorphic sapphire diopside and ruby Clinopyroxene Chanthaburi-Trat, Thailand Metamorphic ruby 1 Pyroxene

Chlorite Pailin, Cambodia Metamorphic sapphire 4 and ruby Silicate Zircon Sahambano, Madagascar Metasomatic sapphire 10 Beryl Weldborough, Tasmania, Magmatic sapphire 1 Australia 1 Olivine Weldborough, Tasmania, Magmatic sapphire Australia Sapphirine Pailin, Cambodia Metamorphic sapphire 4

and ruby

Pyrope Cudgegong/Wellington, NSW, Metamorphic ruby 1

Other Australia Meionite Cudgegong/Wellington, NSW, Metamorphic ruby 1 Australia Garnet Chanthaburi-Trat, Thailand Metamorphic ruby 1 Sillimanite Zazafotsy, Madagascar Metasomatic sapphire 10 Thorite Kanchanaburi, Thailand Magmatic sapphire 1 Allanite-(Ce) Mercaderes–Rio Mayo, Colombia Metasomatic sapphire 2 and ruby Quartz Weldborough, Tasmania, Magmatic sapphire 1 Australia Hercynite Cerová Highlands, Slovakia Metasomatic sapphire 6 Chromite Weldborough, Tasmania, Magmatic sapphire 1

Australia Spinel Spinel Paranesti, Greece Metamorphic ruby 8 Oxide Cr-spinel Barrington, NSW, Australia Metamorphic ruby 1 Zincian spinel Kedrovka, Russia Magmatic sapphire 1 Magnetite Eger (Ohre) Rift, Germany Metasomatic sapphire 1

Ferrocolumbite Tumbarumba, NSW, Australia Magmatic and 1 Columbite Metasomatic sapphire Columbite Anakie, Queensland Magmatic corundum 7 Pyrochlore Dak Nong, Vietnam Magmatic sapphire 1 Pyrochlore Uranpyrochlore Mambilla Plateau, Nigeria Magmatic sapphire 3 Betafite Anakie, Queensland Magmatic corundum 7 Rutile Mayo Kila, Cameroon Magmatic sapphire 1 Anatase Cudgegong/Wellington, NSW, Metamorphic ruby 1 Australia Hematite Barrington, NSW, Australia Magmatic sapphire 1 Mn-ilmenite Kanchanaburi, Thailand Magmatic sapphire 1 Others Baddeleyite Dak Nong, Vietnam Magmatic sapphire 1 Ilmenite Cerová Highlands, Slovakia Metasomatic sapphire 6 Boehmite Mercaderes–Rio Mayo area, Metasomatic sapphire 2 Colombia and ruby Uraninite Kings Plains, New South Wales Magmatic corundum 7 Euxenite-Y Ban Huai Sai, Laos Magmatic sapphire 1 Samarskite Anakie, Queensland Magmatic corundum 7 Apatite Mercaderes–Rio Mayo area, Metasomatic sapphire 2 Colombia

Th-Monazite Tumbarumba, NSW, Australia Metasomatic sapphire 1 Monazite Eger (Ohre) Rift, Germany Metasomatic sapphire 5

Phosphate Cheralite Ban Huai Sai, Laos Magmatic sapphire 1 Chlorapatite Podgelbanochny, Russia Magmatic sapphire 1 Brockite Kings Plains, New South Wales Magmatic sapphire 7 Molybdenite Weldborough, Tasmania, Magmatic sapphire 1 Sulfide Australia Pyrite Yogo Gulch, Montana Magmatic sapphire 1

Anhydrite Eger (Ohre) Rift, Germany Metasomatic sapphire 5 Sulphate Barite Eger (Ohre) Rift, Germany Metasomatic sapphire 5 Carbonate Calcite Eger (Ohre) Rift, germany Metasomatic sapphire 5

Graphite Graphite Eger (Ohre) Rift, Germany Metasomatic sapphire 5 Others

Halide Fluorite Kedrovka, Russia Magmatic sapphire 1

Fluid and gas inclusions

Fluid and gaseous inclusions are also found in corundum grains, especially along fractures. Glassy melt inclusions have been observed in the Thai/Cambodian rubies (Gübelin and Koivula 2008) and sapphires from Yogo Gulch, Montana (Palke et al., 2016). The highly prized blue sapphires from the Jammu and Kashmir region of India show a velvety lustre due to the presence of layers of microscopic liquid and gas inclusions (Peretti et al., 1990). Primary fluid inclusions trapped within the corundum grain can provide valuable insights into the pressure-temperature conditions of corundum formation. Peretti et al., (1990) were even able to differentiate the variability in P-T conditions between the inner core and outer zone of the crystallisation growth phase.

1.3 Gem Corundum deposit types

Since the 1990s, our understanding on the formation of corundum deposits has vastly improved. However, genetic data on primary gem corundum deposits are still insufficiently available to create genetic models that encompass all deposits. Individual deposits have been studied to various degrees. For example, the basaltic deposits in Asia and Australia have been extensively researched (e.g. Sutherland and Coenraads 1996; Limkatrun et al., 2001; Sutherland and Schwarz 2001; Sutthirat et al., 2001; Garnier et al., 2005; Graham et al., 2008; Sutherland et al., 2009a, 2009b; Sutherland and Abduriyim 2009c; Abduriyim et al., 2012). At the other end of the spectrum, deposits such as those in Azad-Kashmir, Afghanistan and Tajikistan have not been systematically examined mostly due to access difficulties and unstable political environments.

Deposits found around the world formed under different unique circumstances that can be systematically grouped into deposit types. Only an environment of aluminum abundance and silica (Si) deficiency would allow for corundum formation. If silica is present in any significant quantity, aluminosilicate minerals such as feldspar and micas would stabilize and preferentially form first instead of corundum (Giuliani et al., 2007a). Corundum occurs in abundance throughout the world; however, gem-quality corundum is rare. It requires the substitution of Cr, Fe and Ti for Al in the atomic structure (Muhlmeister et al., 1998; Abduriyim and Kitwaki 2006). As rare as gem corundums may be, they are found in a wide range of rock types.

A range of classification models have been proposed over time. The two commonly referred to gem corundum classification schemes are the Simonet et al., (2008) classification system based on the genetic process of corundum formation. Firstly, it is based on whether the deposit is

9 primary vs. secondary. In primary deposits, corundum deposits are further classified into igneous vs metamorphic whereas the secondary deposits are classified between igneous vs sedimentary (Figure 1). There are three further categories under metamorphic – metamorphic, metasomatic and anatetic.

The second widely referred to model is that of Guiliani et al., (2014) where primary gem corundum deposits are further classified based on lithological criteria (Figure 2). The corundums are similarly classified firstly into magmatic vs. metamorphic. The magmatic group is then classified into five mineral system categories: alkali basalt, syenite, lamprophyric dykes, kimberlite dykes and porphyry copper. In the metamorphic category, a further two subcategories based on genetic processes have been established – metamorphism vs. metasomatism. There are five host rock types under metamorphism - migmatite, gneiss, cordieritite, mafic – ultramafic rocks and marble. Under metasomatism, there are three host rock types identified – plumasite, skarn and biotitite. The lithology type classification is further supplemented by the oxygen isotope classifications based on the host rock deposit types.

1.3.1 Metamorphic ruby deposits

This study is focused on metamorphic rubies from northern Greece. Therefore, magmatic ruby deposits will not be discussed in detail. Corundums formed through the metamorphism of Al- rich rocks, are usually formed under high grade amphibolite and granulite-facies conditions (Giuliani et al., 2007; Garnier et al., 2008; Simonet et al., 2008). The formation mechanisms are either regional metamorphism or contact/hydrothermal metasomatic metamorphism. In some cases, multiple episodes of metamorphism have impacted upon the corundum-bearing complexes. The types of metamorphic and metasomatic deposits are listed below based on Simonet et al., (2008).

1.3.1.1 Metamorphic deposits a) Gem corundum-bearing aluminous gneisses and granulites

This type is best represented by the Highland Groups of southern Sri Lankan corundum deposits. They have been subjected to high temperatures and moderate pressures (amphibolite facies — De Maesschalk and Oen, 1989, – to low pressure granulite facies — Dissanayake and Chandrajith, 1999).

10 b) Ruby-bearing meta-limestones

This type is best represented by the marble-hosted “pigeon-blood” ruby deposits of Myanmar (Kane and Kammerling, 1992; Peretti et al., 1996). The chemical composition of rubies from marbles is characterized by a high Cr2O3 content (up to 2.5 wt.%) and a low FeO content (typically less than 0.04 wt.%) (Muhlmeister et al., 1998). Giuliani et al., (2014) argued that the presence of granite in the proximity of the marble does not imply that the deposit is a skarn type related to the emplacement of the granite, as suggested by Kievlenko (2003). c) Ruby-bearing mafic granulites

This deposit type rarely produces facetable quality rubies (e.g., Muhlmeister et al., 1998), except in the Losongonoi deposit (Tanzania) which yielded significant quantities of gem-quality ruby (Simonet, 2000). One notable deposit from this type in an amphibolite is the non-transparent star rubies from the Mysore area, Karnataka, India. These granulites are generally vivid green rocks because of the high Cr content of the rock-forming aluminosilicates (pargasitic amphibole, zoisite) and rich in Cr2O3 (up to 1.7 wt.%) and FeO (up to 0.8 wt.%) (Simonet et al., 2008).

1.3.1.2 Metasomatic deposits a) Metamorphic deposits associated with desilicated pegmatite, skarn related granite or mafic/ultra-mafic rocks

These deposits are formed as the result of hydrothermal metasomatism. This occurs either at the contact of intrusive aluminosilicate plutonic rocks or mafic-ultramafic host rocks depleted in silica, or the contact of a shear zone. Desilicated pegmatite deposits formed at the contact of two contrasting lithologies, i.e. granite or pegmatite against mafic and ultramafic rocks, marble or gneiss (Simonet et al., 2008). Corundum deposits of this type are known from the northeastern Transvaal (Robb & Robb 1986), Yosemite National Park (Rose 1957), Kinyiki Hill and Mangare in Kenya (Simonet 2000), Umba River (Solesbury 1967) and Kalalani (Seifert and Hyrsl 1999) in Tanzania, Sumjam deposit in Indian Kashmir (Atkinson and Kothavala 1983) and the upper Allier Valley in France (Forestier and Lasnier 1969).

11 b) Corundum-bearing skarns

Skarn-related deposits form when pegmatite and granite intruded marble, calcsilicate rocks or Ca-bearing gneiss (Giuliani et al., 2014). The desilication reaction is initiated by the silica- deficient host rock, which is in this case is a meta-carbonate instead of an ultramafic rock. c) Anatexites

Anatexis is a differentiation process. The process occurs when a meta-pelitic rock melts, causing silica to enter the melt first. The remaining rock is enriched in Al and depleted in silica (Simonet et al., 2008). An example of this type of deposit is Morogoro in Tanzania (Altherr et al., 1982).

1.3.2 Distribution and Age

The global distribution of gem corundums are closely associated with tectonic collision, rift and subduction geodynamics. There are three major events in geological history where gem corundum deposits appear to have formed en masse.

1. The Pan-African Orogeny (750-450 Ma)

The ruby and sapphire deposits found in the gemstone belt of India, Sri Lanka, Eastern African countries and Madagascar are linked to the Gondwana collision of the continental plates (Kröner 1984) as illustrated in Figure 3. A Neoproterozoic–early Cambrian gemstone province developed under granulite-facies conditions in East Gondwana was proposed (Menon and Santosh 1995, Dissanayake and Chandrajith 1999). The age was determined by U-Pb dating of zircon inclusions from the John Saul mine in Kenya 612 ± 6 Ma (Simonet 2000), Longido in Tanzania 610 ±0.6 Ma (Le Goff et al., 2010) and the Vohibory deposits in Madagascar 612 ±5 Ma (Jöns and Schenk 2008) which correlates to the age of the East African Orogeny. 40Ar/39Ar ages of syngenetic biotite from Madagascar range between 494 and 487 Ma (Giuliani et al., 2007). These ages constrained the Pan-African gem corundum formation episode to the Kuunga Orogeny (Meert et al., 1995, Meert 2003).

2. Cenozoic Himalayan Orogeny (45 Ma – Quaternary)

40Ar/39Ar dating on syngenetic phlogopite (Garnier et al., 2002; Pêcher et al., 2002; Garnier et al., 2006) and U-Pb analysis of zircon inclusions (Garnier et al., 2005, 2006) have determined that the marble hosted ruby deposits in central to eastern Asia have Oligocene to Pliocene ages (40–5 Ma). These regional extensional tectonic events created a metamorphic ruby-bearing belt from Afghanistan to Vietnam (Figure 4).

3. Cenozoic alkali basalt extrusions (65 Ma – Quaternary)

Xenocrysts/megacrysts gem corundums in xenoliths or basaltic magma ascent have been found along the eastern Australia through to southeast Asia and eastern China/far east Russia (Graham et al., 2008) as shown in Figure 5. Examples of alkali basalt hosted gem corundums have also been found in Nigeria and Cameroon (Wright et al., 1985; Mbih et al., 2016), the French Massif Central (Merle et al., 1998; Gaillou et al., 2010) and Northern and Central Madagascar (Rakotosamizanany et al., 2009). Age determinations were performed using the U-Pb system on syngenetic zircons. Of economic importance is that Eastern Australia is one of the most extensive and prolific basalt-derived gem-corundum provinces along the western Pacific margin (Sutherland et al., 2009b).

1.3.3 Geological Setting

A large number of corundum deposits are concentrated along continental margins and subduction zones (Dissanayake and Chandrajith 1999; Santosh and Collins 2003; Giuliani et al., 2014). There are two major geological environments that are more favourable for the crystallization of gem corundum. 1) amphibolite-medium pressure granulite-facies metamorphic belts and 2) alkaline basaltic volcanism in continental rifting environments.

Under metamorphic conditions in silica- and alumina-depleted rocks, the transportation of alumina by a fluid phase appears to assist in the accumulation of Al and the formation of corundum. Rakotosamizanany et al., (2009) suggested that in metamorphic rocks, corundum crystallises at high temperatures and high-medium pressures such as in the garnet- clinopyroxene assemblages in clinopyroxenite and metagabbro from the lower crust at a temperature of around 1100°C and a pressure around 20 kbar. The gem corundum field is suggested as a domain above 3 kbar and temperature between 500°C - 800°C (Figure 6). Marble is a common host rock for metamorphic corundum (e.g. as in the famous Mogok field of

Myanmar). South-east Asian and Australian ruby and sapphire deposits occur along the West Pacific continental margin intraplate basaltic fields (Figure 5; Graham et al., 2008).

Stern et al., (2013) suggested that metamorphic ruby is a robust indicator of continental collision given the age and proximity of major occurrences to the major tectonic collision events of the East African–Antarctic Orogen and the supercontinent Gondwana. However, this theory did not discuss the other major ruby-forming events of the Cenozoic Himalayan Orogeny and the Cenozoic alkali basaltic extrusions.

1.3.4 Associated mineral assemblages

Common mineral assemblages are found within the deposit types. Skarn-type deposits are associated with Ca-bearing, usually Fe-rich silicate minerals (including amphibole, pyroxene, garnet, epidote, and zoisite), or Mg-rich silicate minerals (such as phlogopite, diopside, pargasite/magnesiohornblende, and forsterite) according to Giuliani et al., (2014). Aluminous gneiss and granulites are associated with aluminous minerals such as garnet, spinel, sapphirine, cordierite and sillimanite (e.g., Cooray and Kumarapeli, 1960). Marble-hosted corundum deposits are associated with minerals such as red or blue gem spinel, and various Al-, Mg- or Ca- silicates, as well as sulfides and oxides. Ruby-bearing granulites are associated with pargasite/magnesiohornblende, gedrite, calcic plagioclase and spinel. Sapphirine is regarded as a higher-grade mineral than corundum when present (Tenthorey et al., 1996). This type formed during the hydration of plagioclase-rich rocks (anorthosites, troctolites, norites) under granulite facies conditions (Lasnier, 1977; Tenthorey et al., 1996).

1.3.5 P-T conditions

A summary of P-T formation conditions for the major metamorphic gem corundum producing mines around the world was compiled by Giuliani et al., (2014; Figure 6). This diagram demonstrates a concentration of mines within the high temperature (>500°C) zone, but under a diverse range of high pressures. Meta-limestones and their host rocks are mostly to be found to have underwent amphibolite (Okrusch et al., 1976; Peretti et al., 1996) to granulite facies (Muhongo and Errera, 1993) metamorphism.

P-T conditions between the corundum and the surrounding lithologies could also point to the origin of the corundums. For example, the contrast between the P-T condition of ruby formation (700–750°C, 8 < P < 10.5 kbar, granulite facies) and the surrounding graphite-sillimanite gneiss

(650°C, 4.5 < P < 6.5 kbar, amphibolite facies) in the John Saul mine, Mozambique led Mercier et al., (1999a) to suggest that the ruby and surrounding ultramafic rocks are exotic. They correspond to fragments of deeper crust brought up by thrust and shear zone tectonics.

The heat source responsible for the high temperature conditions could be explained by either regional or contact metamorphism or thermal anomalies generated by different intrusions (Giuliani et al., 2014). An example of contact metamorphism is where a skarn-type corundum deposit developed around granitic intrusions (Harlow et al., 2006). Multiple phases of metamorphism may affect the corundum deposit and successive recrystallization may erase earlier formed assemblages. Mineral reaction kinetics favor recording of counterclockwise P–T paths or isobaric cooling because the defining parts of such paths occur under conditions of decreasing temperature and reaction rates, favoring the preservation of earlier formed metamorphic phases (Wakabayashi, 2004).

1.3.6 Sources of chromium

The most defining feature in a ruby is its relatively high chromium content. So what is the source of Cr? Chromium is usually considered as an immobile element in marble (Treloar, 1987; Grundman and Morteani, 1989). However, Cr is considered as a relatively mobile trace element in a fluid-rich environment (Giuliani et al., 2014). The mobility of this element can in some cases be high if anions such as F or Cl are present in the fluid phase (Peretti et al., 1996).

The break-down of spinel has been proposed as the Cr source for rubies. Spiridonov (1998) proposed that meta-evaporite and meta-ultrabasite rocks are favourable settings for ruby crystallization given the low Si, K and Na activities and above 400°C alumina is transported within aqueous fluids. This was observed by Kissin (1994) for ruby formed in marble from the Urals and for ruby deposits from southeast and central Asia (Garnier, 2003).

1.4 Worldwide distribution of ruby deposits

The traditional producers of gem ruby and sapphire are found mostly in the Southeast Asian countries of Myanmar, Thailand, Sri Lanka, Pakistan, India and Afghanistan. As the supplies for the most famous Mogok “pigeon-blood” rubies start to dry up, exploration around the world has discovered new deposits in Australia, Greenland, Vietnam and the African countries of Mozambique, Kenya, Tanzania and Madagascar. As this paper focuses on the rubies from Northern Greece, Figure 7 shows a world map of major ruby deposits around the world.

Ruby deposits from Central and Southeast Asia are best represented by the marble-hosted classic rubies from Mogok Myanmar and mines along the Himalayan mountain belt from Hunza Pakistan, Jegdalek Afghanistan to Chumar Nepal and Luc Yen Vietnam (Garnier et al., 2008).

1.5 Oxygen isotopes

1.5.1 Introduction

Oxygen is the most abundant element in the Earth’s crust, mantle, and fluids. Its isotopic composition provides robust constraints on earth material genesis including corundums. The element (O) oxygen has three naturally occurring stable isotopes, 16O, 17O, and 18O. The nucleus of each of these oxygen isotopes contains eight protons and either eight, nine, or ten neutrons, respectively. Of these stable isotopes, 16O is the most abundant on earth, accounting for 99.757 % of atoms, while 17O (0.038%) and 18O (0.205%) occur in far smaller concentrations worldwide (Wright, 2017). Normal saline water (e.g. seawater) is commonly used as the reference benchmark with the activity coefficients for δ18O essentially equal to 1 (Kendall and Caldwell, 1998). A difference in the isotopic ratio between the reactants and the product of a chemical reaction or physical state change is known as the fractionation of the isotopes (Fry, 2006). Oxygen isotope fractionation is a function of the initial Rayleigh evaporation-precipitation cycle, temperature of the system and degree of water-rock interaction and therefore great care must be taken when interpreting oxygen isotope values (Criss and Taylor, 1986; Valley, 2001). Oxygen isotope fractionation is mostly dependent on temperature and there is little or no measurable effect of pressure on fractionation (Valley, 2001). The scale of isotope equilibration and the magnitude of diffusive retrogression is critically dependent on the fluid conditions of a rock. Dry, granulite facies metamorphism has been invoked to explain very slow diffusion and the preservation of high oxygen isotope temperatures in some rocks (Farquhar et al. 1996) while

18 variable f(H2O) during cooling has explained faster diffusion and resetting of δ O in diopsides from granulite facies marbles (Edwards and Valley 1998), and of amphibolites and pelites (Kohn 1999).

Stable isotope ratio mass spectrometers measure the concentrations of the various masses of oxygen isotopes within a sample and compare these to a standard reference of known composition. These ratios are then corrected to Vienna Standard Mean Ocean Water (V-SMOW). V-SMOW is issued by the International Atomic Energy Agency (1968) due to the inconsistencies between the results obtained in different laboratories (Gonfiantini, 1978). The ratio between

16 the stable isotopes of oxygen is reported as a delta value and calculated as follows (Kendall and Caldwell, 1998):

δ18O=[(RSAMPLE/RSTANDARD-1)] 1000

∗ The oxygen isotopic ratios are driven by global variations in the stable isotopic composition of water. Global rainfall largely originates from oceanic water evaporated in the tropics. The “lighter” 16O evaporates more quickly than the “heavier” 18O, so that δ18O of water in clouds and precipitation is low, i.e. negative on the V-SMOW scale. When rain clouds move over land and from warmer regions to cooler regions, the oxygen isotopic ratio of water held in the clouds declines further and is progressively depleted in 18O. Therefore, the isotopic ratio of rain that falls in latitudes far from the tropics contains proportionately less 18O than rain in coastal tropical areas (Rozanski et al., 1993).

Terrestrial rocks and minerals are usually enriched in the 18O isotope relative to seawater with the majority of values lying within the range of +4 to +15‰ relative to Vienna Standard Mean Ocean Water (VSMOW) for most silicate rocks (Faure, 1986). Bindeman (2008) found that nearly all negative δ18O values are occupied by meteoric waters that are isotopically-light as a result of Rayleigh distillation upon vapor transport and precipitation. Most mantle rocks, basaltic magmas and chondritic meteorites plot in a narrow δ18O‰ range of 5.5 to 5.9‰ (Bindeman, 2008).

1.5.2 Oxygen isotopes in gem corundums

Therefore, oxygen isotope ratios are useful for identifying the primary host rock lithology, and to a more limited degree, the specific geographic locality for the gem corundums when this is not known such as facetted stones of unknown provenance (e.g. Giuliani et al., 2005; Graham et al., 2008; Giuliani et al., 2014). Giuliani et al., (2014) suggested that the isotopic composition of corundum is controlled by the isotopic composition of its host rock. The authors therefore proposed that it is the geological environment of formation and not the specific geographic location that restricts the oxygen isotope range of corundums. Giuliani et al., (2005) proposed a framework of specific δ18O‰ values correlated to specific geological environments (Figure 8).

Oxygen isotopic ratios (18O/16O) in the lattice oxygen of ruby and sapphire are often used to determine their possible geological origin, particularly for rubies and sapphires from alluvial and paleo-alluvial deposits, and this is increasingly being used for geographic typing of gem rubies

17 and sapphires (e.g. Vysotskiy et al., 2014; Sutherland et al., 2015; Wang et al., 2019). Although major and trace element geochemistry is commonly used to differentiate between magmatic and metamorphic gem corundums (e.g. Giuliani et al., 2005, 2007; Sutherland et al., 2009; Vysotskiy et al., 2015), oxygen isotopic signatures when combined with geochemical signatures provide a far more powerful tool and can almost exclusively be used to locate the corundum’s geographic source (Vysotskii et al., 2008; Bindeman et al., 2010; Wang et al., 2017).

The worldwide corundum oxygen isotope values cover a wide range from −27‰ (Khitostrov, Russia) to +23‰ (Mong Hsu, Myanmar), though most are in the range of +3‰ to +21‰ (Krylov, 2008; Giuliani et al., 2014). Some oxygen isotopic compositions of corundum are quite unique in that they enable us to ascribe them a specific (or at least regional) geographic location. Such source identification is especially important in the gem trade as it significantly influences the market value of rubies and sapphires. Examples of using oxygen isotope values for geographic source identification includes Karelia, Russia (Krylov, 2008; Vysotskii et al., 2008; Vysotskiy et al., 2014; Bindeman et al., 2010; Bindeman et al., 2014) and Paranesti, Greece (Wang et al., 2018).

Traditionally, the main oxygen isotope analytical method adopted by researchers for corundum analysis is laser-fluorination (Giuliani et al., 2005) which requires the destruction of the corundum grains. More recently, progress has been made on in situ oxygen isotope analysis via the secondary ion mass spectrometer (SIMS) technique (Sutherland et al., 2017; Wang et al., 2018; Graham et al., 2019).

Chapter 2. The Paranesti Ruby Deposit

This chapter is mostly based on the following published papers in Canadian Mineralogist and Minerals with myself as first author

Wang, K.K., Graham, I.T., Lay, A., Harris, S.J., Cohen, D.R., Voudouris, P., Belousova, E., Giuliani, G., Fallick, A.E. and Greig, A., 2017. The origin of a new pargasite-schist hosted ruby deposit from Paranesti, Northern Greece. Canadian Mineralogist, 55(4), 535-560. (Appendix 1)

2.1 Geological Setting 2.1.1 Regional Geology

The in situ ruby deposits of Paranesti are located within the Rhodope Mountain Complex (RMC) in north-eastern Greece (Figure 9), which forms the innermost part of the Hellenides in the Aegean region. The Hellenide Orogen from north to south consists of several continental blocks (Srednogorie and Rhodope-Sakarya, Pelagonian-Lycian and Adria-External Hellenide) and the two oceanic domains of Vardar-Izmir and Pindos Suture Zones (Bonneau 1984; van Hinsbergen et al., 2005; Ring et al., 2010). In the Aegean region, continuous subduction of both oceanic and continental lithosphere beneath the Eurasian Plate since the Early Cretaceous has resulted in the formation of a series of magmatic arcs from the Rhodope massif in the north to the Active South Aegean Volcanic Arc in the south (Fytikas et al., 1984).The Middle Jurassic to Neogene northeast-dipping subduction and convergence of the African-Eurasian plates resulted in the closure of the Tethys Ocean (Bonev et al., 2006; Krenn et al., 2008). The evolution of the Hellenides is marked by a Mesozoic collisional phase, followed by a continuous southward slab retreat in a back-arc setting since the Eocene that triggered large-scale extension with thrusting at the southern part of the Hellenic domain (Jolivet et al., 2013; Kydonakis et al., 2015).

2.1.2 The Rhodope Mountain Complex (RMC)

The RMC occupies a large part of the border regions between north-eastern Greece and southern Bulgaria and extends eastwards into northwest Turkey (Figure 10). It is part of the European continental margin (Kydonakis et al., 2015; Burg 2012) and belongs to the Alpine- Himalayan Orogeny (Burg, 2012). Geological work since the late 1980s has established the affiliation between Rhodope and Tethys (the long recognised oceanic basin from which the Alpine-Himalayan orogenic system arose). The complex northeast-dipping nappe stacking of the RMC developed since the Late Cretaceous with at least 900 km of continental lower crust and lithosphere subducted underneath the Aegean Sea (Burg et al., 1996; Ricou et al., 1998; van Hinsbergen et al., 2005; Robertson et al., 2009; Krenn et al., 2010). This subduction zone was the precursor of the present-day Hellenic subduction zone further south. Formation of sedimentary basins and horizontal extensions began in the Eocene and up to the present-day (Dumurdzanov et al., 2005; Bonev et al., 2006; Burchfiel et al., 2008). This process is associated with 500 km retreat of the Hellenic slab in Cenozoic times (Jolivet and Brun, 2010).

∼ 19

The formation of the RMC is generally accepted as a result of Mesozoic to Cenozoic events related to the Alpine Orogeny and post-orogenic “Aegean” extension. However, there are contrasting views on the spatial distribution of coherent tectono-metamorphic units of the RMC. Various structural models have been proposed (Burg et al., 1996; Ivanov et al., 2000; Krohe and Mposkos, 2002; Turpaud, 2006; Jahn-Awe et al., 2010; Burg, 2012). The tectonic units of the RMC generally comprise rocks of continental or mixed continental-oceanic origin (Dinter 1998; Ricou et al., 1998; Barr et al., 1999). It is largely comprised of two sub-domains: the northern Rhodope domain and the southern Rhodope core complex (both separated by the Nestos thrust fault and the Nestos Suture Zone, Kydonakis et al., 2015).

A generally accepted view is that the central part of the RMC involves at least two major north- dipping shear zones – the Nestos Shear Zone (NSZ) and the Chepelare Shear Zone (CSZ) (Figure 11). In earlier literature, these two shear zones were thought to be distinct, synthetic thrust zones (Ricou et al., 1998) as they display a similar petro-geochronological record (Bosse et al., 2009; Didier et al., 2014). In more recent times, both the NSZ and CSZ have been reinterpreted as two parts of the same shear zone (Turpaud and Reischmann, 2010; Nagel et al., 2011; Burg, 2012). Burg et al., (1996) proposed the subdivision of the Rhodope into lower and upper terranes separated by intermediate ductile mylonitic zones based on major thrust events. This melange zone which contains both high pressure (HP) and ultra-high pressure (UHP) relicts and amphibolites represents the Nestos Shear Zone (Turpaud 2006), in which the ruby-bearing amphibole schists of this present study are located.

The nomenclature varies between literature on the NSZ geological units. Papanikolaou and Panagopoulos (1981) initially defined a lower unit and an upper unit, later named the Pangaeon Unit (PU) and Sidironero Unit (SU). The PU is equivalent to the “Thracia Terrane” (Turpaud and Reischmann, 2010), the “Lower Terrane” (Burg, 2012), and the “Lower Allochthon” (Nagel et al., 2011). Similarly, SU is also referred to as the “Rhodope Terrane” (Turpaud and Reischmann, 2010) and the “Middle Allochthon” (Nagel et al., 2011).

Another debated issue is whether the Alpine Orogeny consisted of a single continuous protracted cycle (Burg, 2012) or involved distinct episodes of subduction and crustal accretion (e.g. Liati et al., 2011; Froitzheim et al., 2014). Four distinct high-pressure metamorphic (HPM) to ultra-high pressure metamorphic (UHPM) events have been reported (Liati and Gebauer, 1999; Liati, 2005; Liati et al., 2011, 2016) as follows: (i) ca. 149 Ma (U)HPM; (ii) ca. 73 Ma (U)HPM;

(iii) ca. 51 Ma HPM; and (iv) ca. 42 Ma HPM based on U–Pb SHRIMP geochronology and REE geochemistry of zircon. The UHP-eclogitic metamorphism was followed by HP granulite-facies, amphibolite-facies metamorphism and finally by greenschist facies metamorphic events starting in the Jurassic (~200–150 Ma) and lasting up to the Oligocene (Liati, 2005; Gautier et al., 2017). The argument stems from the unreconcilable time lapse of at least 60 million years (between ≥ 115 and ≤ 55 Ma) between the two sets of high-temperature metamorphic events observed by a number of researchers (based on U-Pb zircon and U-Th-Pb monazite ages) including Liati (2005), Bosse et al., (2009), Didier et al., (2014) and Wawrzenitz et al., (2015). The existence of a large time gap (≥ 45 Myr) in the geochronological record appears to support the hypothesis of a polycyclic orogenic evolution for the largest part of the RMC. Mposkos et al., (2010) additionally argued that some of the garnet-kyanite metapelites found along the NSZ retain the record of a polycyclic P-T evolution.

There is a lack of consensus on the tectonic evolution of the RMC during the Cenozoic period. There are three main schools of thought. Based on the retrogressed eclogites, the first hypothesis is that a subduction setting existed up to 42–51 Ma (Liati and Gebauer, 1999; Liati,

2005; Liati et al., 2011; Nagel et al., 2011; Kirchenbaur∼ et al., 2012; Froitzheim et al., 2014). The second group of thought is that the synmetamorphic thrusting remained active until 33-35 Ma (Bosse et al., 2009; Gerdjikov et al., 2010) and 37 Ma (Dinter, 1998; Krohe and Mposkos, 2002) without agreeing to the HP conditions interpretation of the first hypothesis. Lastly, for the third hypothesis no significant thrusting event occurred in the RMC during the Cenzoic period and a crustal-scale extension developed from ~40 Ma (Brun and Sokoutis, 2007; Burg, 2012) or earlier. The widespread magmatism and high-temperature metamorphism are interpreted as a result of mantle delamination processes (Jolivet and Brun, 2010; Marchev et al., 2013).

2.2 Local Geological Setting 2.2.1 The Nestos Shear Zone (NSZ)

The in situ ruby deposits of Paranesti are located within the Nestos Shear Zone of the Rhodope Mountain Complex in north-eastern Greece (Figure 12). The Paranesti area is defined by a rugged mountainous geomorphology. The western part is covered by the Rhodope Mountains and is cut by the Nestos River valley and its tributaries (Karageorgiou et al., 2010).

The Nestos Shear Zone (NSZ) is a major high-strain zone separating the Lower and Upper metamorphic terranes. It is traceable for about 100 km from Xanthi in the east to the Bulgarian border in the west (Figure 12). Proposed ages for the NZS range from Jurassic to Eocene (Ricou et al., 1998; Krohe and Mposkos, 2002; Turpaud and Reischmann, 2010; Krenn et al., 2010). The Lower unit is commonly interpreted as a Mesozoic platform on top of Variscan continental crust and displays an Eocene to Oligocene metamorphic overprint at upper greenschist to lower amphibolite facies conditions (Mposkos and Krohe 2000, 2006; Krohe and Mposkos 2002). The Upper unit displays regional upper amphibolite facies metamorphism and several locations contain well-preserved eclogites (Kolceva et al., 1986; Mposkos and Krohe 2000, 2006; Krohe and Mposkos 2002).

Microdiamond-bearing ultrahigh-pressure (UHP) rocks are found in the NSZ and interpreted to be a suture zone with subduction and exhumation of the UHP rocks and terrain accretion during the Mesozoic (Nagel et al., 2011). The NSZ was proposed by Nagel et al., (2011) to represent a mid-crustal mylonite horizon into which brittle and brittle-ductile extensional faults in the hanging wall root. This extension at the surface corresponds to an event of crustal accretion and associated decoupling of the downward slab from the thickened continental crust above. This detachment triggered slab retreat and the rise of asthenospheric mantle into the gap above the slab, thus explaining magmatic events in the Early Oligocene (e.g. Brun and Faccenna 2008; Jolivet and Brun 2010).

Metamorphic zircons from metabasites and metapelites of the Kimi Complex (e.g. rocks similar to those of the Nestos shear zone) indicate that both the eclogite and metapelite underwent (U)HP metamorphism together in the Jurassic, followed by decompression and high- temperature (e.g. granulite) metamorphism at around 170-160 Ma (Bauer et al., 2007). Further metamorphism is recorded in the eclogite possibly at ca. 115 Ma, and at ca. 79 Ma for the amphibolite facies event (Bauer et al., 2007). Ion probe (SHRIMP) zircon dating of sapphirine- bearing kyanite eclogites from the Thermes area by Moulas et al., (2013) indicated an Eocene age (42 ± 2 Ma) for the end of amphibolite facies overprint. The microdiamond-bearing garnet- kyanite-mica schist found at Sidironero, 19.4 km NW of Paranesti, provides evidence that the UHP metamorphic event was experienced in the Upper metamorphic terrane (Schmidt et al., 2010). This high-pressure eclogite assemblage (garnet+ clinopyroxene+ biotite+ quartz +phengite), with phengite being completely retrogressed to biotite in the present rock,

stabilised at 750⁰C and 22 kbar. Retrogressive formation of biotite at the expense of white mica and garnet occurred at conditions of about 700 C/8–10 kbar (Schmidt et al., 2010).

⁰ In addition, meta-ultramafic bodies have been found in the amphibolite-marble series of the Xanthi area (~40 km east of Paranesti) and characterized as metamorphic peridotite (tectonite) and possibly represent dismembered ophiolite fragments (Dimadis et al., 1990).

2.2.2 The Paranesti Ruby Location

The Paranesti ruby occurrence is stratiform and oriented parallel to the main regional foliation. The rubies in the Paranesti area have been found to outcrop on the surface east of Perivlepto village, Paranesti (Figure 13). Rubies are found at two main localities: the main workings approx. 200 metres down a moderately steep hillside (main ruby workings PAR-1 – Figure 14) and the original discovery along the roadside on top of the ridge (PAR-5 – Figure 15). The area is known to individual prospectors and collectors with no commercial mining of the rubies to date.

PAR-1 comprises the main corundum workings approximately 200 meters down a moderately steep hillside. The ovoid-shaped boudin-like amphibole schist lenses are up to five metres in length but generally only 1-1.5 m across and are elongated parallel to the main regional foliation (Figure 14). These are immediately surrounded by a narrow zone (generally < 0.3 m across) of highly schistose clinochlore schist, and rare small boudins of ruby-kyanite-amphibole schists also occur.

This hillside ruby locality is conspicuous by the occurrence of a prominent white-coloured quartz-feldspar-mica-garnet pegmatite (Figure 14 outlined in red) which intruded into the ruby- bearing amphibole schists (subsequent to the ruby genesis). This intrusion event postdates the amphibolite facies metamorphism during which the ruby is suggested to have formed. Hence the pegmatite intrusion bears no relationship to the genesis of the rubies. This intrusive pegmatite has previously been documented in the central RMC as 65–63 Ma pegmatite veins cutting amphibolitised eclogites in the eastern Rhodope (Baziotis et al., 2007). Furthermore, an Rb-Sr isochron age of 65.4 ± 0.7 Ma is given for another cross-cutting pegmatite (Liati et al., 2002). The ruby-bearing zone at the main workings covers an area of approximately 50 (length) x 10 (width) m.

At the roadside discovery (PAR-5), 300 m east of PAR-1, the footwall consists of thinly banded (50-200 mm) amphibolite and a thin shell of clinochlore (<0.3 m across) surrounding boudins of 23

ruby-bearing amphibole schist (Figure 15). The white layer above the boudins are thin marbles. The ruby-bearing boudins are approx. 1m x 0.3 m in size. The hanging-wall consists of kyanite- bearing quartz-plagioclase-amphibolite gneisses and rarer kyanite-amphibole-chlorite schists (Figure 16). Sheared contacts occur between all of these rock types. Additionally, a number of boudins occur within 10 km2 (including some only a hundred metres further north along the same road), though none of these were found to contain any visible ruby (Figure 17). The transition from amphibolites in both the footwall and hanging wall towards the ruby mineralisation is made through a clinochlore-rich schist zone. The ruby-bearing zone at the roadside workings covers an area of approximately 30 (length) x 3 (width) m.

2.3 Sample Processing

Samples were collected in situ from the localities shown in Figures 14-15 and a stratigraphy schematic of the sample collections site is illustrated in Figure 18. The samples comprised rubies enclosed within their host amphibole schists, loose rubies from the surface which had weathered-out of their host amphibole schists, large pure masses of chlorite from the chlorite schist zones, kyanite-bearing amphibole schists and loose kyanite grains which had weathered out of it. Also taken were samples from across the spectrum of rock types found from the footwall and hanging wall of the ruby-bearing amphibole schists, along with samples of similar- looking boudins of amphibole schists (but with no visible ruby) from outcrops a few hundred metres and a few kilometres to the north.

The near-surface rubies are heavily fractured most likely due to weathering. Some samples were collected showing well-developed deep red rubies with both tabular and barrel-shaped crystal habits within pargasite/magnesiohornblende schist (Figures 19 and 20) while rubies associated with kyanite (Figure 21) were found to be rare.

Samples were labelled based on their collection point and all denoted with a PAR (Paranesti) prefix. Prior to XRD and whole-rock geochemical analysis, visible rubies were firstly removed and the remaining raw crushed samples firstly processed into a fine powder (≤ 27 µm grain size). Rock crushing and milling were conducted at the School of Biological, Earth and Environmental Sciences (BEES), UNSW using a combination of a jaw crusher, geological pick and brass plate, and the Rocklabs Tungsten ring mill. Prior to milling, samples were broken into small (< 10 mm) pieces using either the rock jaw crusher or geological pick and brass plate, then any remaining

weathered and oxidised surfaces were removed to ensure that only fresh and unweathered rocks were processed for analysis.

2.4 Analytical Methods

2.4.1 Microscopy

Petrographic analysis on the mineralogy, textural relationships, alteration phases and mineral paragenesis were examined using the Leica DM2700-P research polarizing microscope, which involved both transmitted light and reflected light for thin sections and polished thin sections. Photomicrographs were obtained through the Leica DFC290 digital camera system. Digital photographs of the samples in transmitted and reflected light were taken using the Leica DFC290 digital camera and then modified and annotated with the Leica LAS software.

2.4.2 X-Ray Diffraction (XRD)

X-ray diffraction (XRD) analysis on three PAR-1, four PAR-5, and three host lithology samples was undertaken at the Solid State & Elemental Analysis Unit, UNSW Analytical Centre, using the PANalytical Empyrean II X-Ray Diffraction System with a Co anode X-ray tube. The samples were prepared using an agate mortar and pestle as well as a UNSW Rock Labs standard tungsten carbide ring mill into fine powders. Samples were packed into 20 mm aluminium holders and gently pressed in the front of the holder with a glass slide to reduce the effects of preferred orientation. The operating conditions were 45 kV and 40 mA, with a 2θ scan angle range of 5– 700 2Ɵ and a step-size of 0.013° 2θ. The resulting XRD patterns were then interpreted using the software programs XPLOT 32 and X’pert Highscore.

2.4.3 X-Ray Fluorescence (XRF)

Whole rock X-ray fluorescence (XRF) analysis of the ruby-bearing amphibole schist (sample PAR- 1) was undertaken within the Mark Wainright Analytical Centre at UNSW, Australia. Glass beads and pressed pellets in aluminium caps were prepared for major and trace elements, respectively. Both trace and major elements were analysed using the PANalytical PW2400 Sequential WDXRF Spectrometer that had been calibrated using certified reference materials. The instrument detection limits for each of the major oxides and trace elements are listed in Appendix 6, Table 1.

2.4.4 Inductively Coupled Plasma – Mass Spectrometer (ICP-MS)

ICP-MS analysis was used for detailed trace element analysis. The crushed and milled samples were analysed at the School of Earth Sciences, University of Melbourne using the Agilent 7700x inductively coupled plasma – mass spectrometer (ICP-MS) following the procedure outlined in Eggins et al., (1997). The ICP-MS analytical procedures altered slightly from those of Eggins et al., (1997) through the use of multi-internal standards (6Li, Rh, Re and 235U) for internal drift correction, and two digestions of the USGS standard W-2 (PCC-1 and BIR-1) for instrument calibration.

2.4.5 Scanning electron microscope (SEM) and Electron Microprobe Analyses (EMPA)

Backscattered electron (BSE) images of polished and carbon-coated corundum and amphibole samples were taken using a Hitachi S3400 SEM at the UNSW Electron Microscopy Unit (EMU),

UNSW, Australia using secondary electron (SE) detectors. The microscope was operated under high-vacuum conditions with an accelerating voltage of 20 kV and beam current of 3 nA at working distances from 12.5 mm. SEM images were acquired specifically to highlight any inclusions present within the ruby grains. The column of the instrument employed for the analysis was surrounded by four wavelength dispersive spectrometers (WDS) to detect and measure the X-ray emitted for semi-quantitative analysis of the samples. The detection limit of the trace elements analysed is listed in Appendix 6, Table 2.

Compositions of selected ruby, margarite, amphibole and chlorite grains from polished thin sections were analysed using a JEOL JXA-8500F Hyperprobe within the Electron Microscope Unit, UNSW, Australia. Prior to analysis, thin sections were baked overnight in a vacuum oven and evaporatively carbon coated to ~20nm. The focused beam voltage and operating current were 15kV and 20nA respectively with a beam diameter of 5 μm. Peak and background counting times were 20s and 10s respectively. For quantitative analysis, calibration was achieved using natural mineral standards with similar chemical and structural compositions to the analysed samples (Appendix 6, Table 3). For analyses of all minerals except sulphides, oxygen content was calculated by stoichiometry and results are reported in oxide mass percent. Since H is not directly measurable via WDS, for hydrous minerals (e.g., chlorite), mineral water content was calculated by difference. Accuracy was measured at < 2% relative to known secondary mineral standards, with an achieved precision of 1-2% for major and minor elements.

The quality of the EMPA results was monitored via analysis of additional mineral and synthetic standards of known composition. The standards used were Ga (GaAs), Na (albite), Mg (periclase), Al (MP1 garnet), Si and Mn (rhodonite), P (apatite), Nb (Nb+Sn metal), K (sanidine), Ca (diopside), Ti (rutile), Cr (chromite), Fe (hematite), Ta (Ta metal), F (fluorite), and Ba (benitoite). Grains analysed included the rubies, spinel inclusions within the rubies, phyllosilicates rimming the rubies, surrounding amphiboles and chlorite from the chlorite zone.

2.4.6 LA-ICP-MS

Six crystals from PAR-1 and 5 crystals from PAR-5 (Figures 22 a and b, showing the location of spot analyses), covering the spectrum of colours were embedded in polished mounts and used for the LA-ICP-MS trace element analysis. The LA-ICP-MS analysis was performed within the Department of Earth and Planetary Sciences, Macquarie University, Sydney, Australia using an Excimer 193nm Ar-F gas laser coupled with an Agilent Technologies 7700 Series quadrupole inductively coupled plasma - mass spectrometer (ICP-MS). Ablation was performed in a He atmosphere with a Mass Flow Control (MFC) 1 of 0.465L min-1 and MFC 2 of 0.36L min-1. Ar was used as the carrier gas with MFC of 1.02L min-1. NIST610 and NIST612 were used as primary standards and BCR2G as the secondary standard. The laser was configured to perform two shots of pre-ablation on a 110-micron spot-size at 5 Hz followed by a 30s pause and 59s of background gas, and then pulsed at 5 Hz on an 85-micron spot-size for 2 min with a burst count of 595 shots and fluence of 9.28J/cm2. Cell voltage and pressure were 15 kV and 6490 mbar, respectively, with an overall energy reading of 5.5mJ. The internal Al standard value was set at 98 wt%, based on the EMPA analyses of the same grains to allow for maximum trace element contents. Spot analyses were performed at core and rim positions on each of the grains, with emphasis on colour-zoned regions.

2.4.7 OXYGEN ISOTOPE SAMPLES AND METHODS

Two different oxygen isotopic analytical methods have been used in this study in order to determine the oxygen isotope values of the Paranesti rubies. Prior to both methods, the rubies were mechanically extracted from the pargasite/magnesiohornblende host matrix and carefully cleaned. In many samples, the ruby crystals occur in clusters of platy crystals and the grain sizes generally ranged between 0.5–1.5 cm (Figures 23–26). Importantly, the ruby grains were generally found to be free of inclusions and thus amenable to in situ analysis (Wang et al., 2017).

2.4.7 (a) Whole grain dissolution laser-fluorination method

In 2009, a reconnaissance study was conducted, whereby five individual grains, one each from different corundum localities/geological environments in Greece, were studied for their δ18O composition. These included two colourless to blue sapphires in desilicified granite pegmatite intruding metaperidotites from Naxos, one pink marble-hosted ruby from Kimi and one purple marble-hosted ruby from Xanthi. One medium red intensity ruby in pargasite/magnesiohornblende schist from Paranesti (PAR-1) was also included. Oxygen isotope analyses were performed using a modification of the laser-ﬂuorination technique described by Sharp (1992) that was similar to that applied by Giuliani et al., in 2005.

The method involves the complete reaction of ~1mg of ground corundum. This powder is then heated by a CO2 laser, with ClF3 as the fluorine reagent. There is no evidence of grain-size effect, or a need of any correction factor compared with conventional fluorination using external resistance furnaces (Sutherland et al., 2009). The released oxygen is passed through an in-line

Hg-diffusion pump before conversion to CO2 on platinized graphite. The yield is then measured by a capacitance manometer. The gas-handling vacuum line is connected to the inlet system of a dedicated VG PRISM 3 dual inlet isotope-ratio mass spectrometer. All oxygen isotope ratios are reported in δ18O (‰) relative to Vienna standard mean ocean water (V-SMOW).

2.4.7 (b) In situ secondary ionisation mass spectrometry (SIMS) method

The first known in situ SIMS analysis on rubies was performed on alluvial ruby-sapphires from New England, Australia by Sutherland et al. (2017). Raw oxygen isotope ratios were corrected for instrumental mass fractionation using standard PAR-1 (1.0 ± 0.1‰) based on the 2009 results.

For the Paranesti rubies, oxygen (18O/16O) isotope ratios were analysed using a Cameca IMS 1280 multi-collector ion microprobe hosted by the Centre for Microscopy, Characterisation and Analysis (CMCA), University of Western Australia (UWA). The sample mounts were carefully cleaned with detergent, distilled water and ethanol in an ultrasonic bath and coated with gold (30nm in thickness) prior to SIMS analyses. For O isotopic analyses, secondary ions were sputtered from the sample by bombarding its surface with a Gaussian Cs+ beam and a total impact energy of 20 keV. The surface of the sample was rastered with a 2.5 nA primary beam over a 15 x 15 µm area. Secondary ions were admitted in the double focusing mass spectrometer within a 110 µm entrance slit and were focused in the centre of a 4000 µm field aperture (x 130 magnification). They were energy filtered using a 40 eV band pass with a 5 eV gap toward the 28

high-energy side. 16O and 18O were collected simultaneously in Faraday cup detectors fitted with 1010 Ω (L’2) and 1011 Ω (H’2) resistors, respectively, and operating at a mass resolution of ~2430. The magnetic field was regulated using NMR control. Each analysis includes a pre-sputtering over a 20 x 20 µm area during 30s followed by the automatic centering of the secondary ions in the field aperture, contrast aperture and entrance slit. Each analysis then consists of 20 four- second cycles, which give an average internal precision of 0.16‰ (SE).

External reproducibility during the analytical sessions was evaluated by repeating analyses in one single fragment of PAR-1. External reproducibility in this fragment was 0.3 and 0.4 per mil (2SD) during the two analytical sessions. In total, three large fragments of PAR-1 were analysed for their oxygen isotope composition, altogether yielding an average value of 0.9 ± 0.6 per mil (2SD, n = 57, Table 3.1).

Raw oxygen isotope ratios were corrected for instrumental mass fractionation using standard PAR-1 (PAR-1, 1.0 ± 0.1‰, 2009); embedded in a separate mount and analysed before and after the sample mounts. Figure 27 shows the sample mount of PAR-1 samples used for the SIMS analysis. The spot-to-spot reproducibility was 0.32‰ (2 SD) on PAR1 during the analytical session. Figures 28a and 28b show the spot location for some of the PAR-5 grain samples with their respective oxygen isotope values included. Uncertainty on each δ18O spot was calculated by propagating the errors on instrumental mass fractionation determination, including the standard deviation of the mean oxygen isotope ratio measured on the primary standard during the session and internal error on each sample data point. Corrected δ18O (quoted with respect to Vienna Standard Mean Ocean Water or V-SMOW) are presented in Appendix 6 Table 4.

2.5 Results

2.5.1 CORUNDUM MORPHOLOGY AND COLOUR

Ruby crystals from Paranesti ranged in size from <1 mm to 20 mm (average size 5–10 mm), are of pale pink to deep red colour and are of cabochon quality. Most grains are opaque to translucent with only rare transparent zones. The morphology is mainly flat tabular with basal planes paralleling the orientation of the main regional foliation, and less commonly prismatic and barrel-shaped crystals. Solid mineral inclusions in the rubies are rare (see below). The ruby grain size from PAR-1 (up to 20mm) is generally larger than that for PAR-5 (approx. 10mm). In hand-specimen, more PAR-1 rubies are observed with a deeper red colour than for PAR-5. In

polished thin sections, the ruby grains from PAR-1 appear to be more fractured, finer-grained, anhedral and more commonly twinned (penetration twins) than in PAR-5. Inclusions in rubies were only found to occur in PAR-1 during the present study.

The rubies occur together with pargasite/magnesiohornblende , which is the main amphibole of the assemblage. Some samples also show ruby crystals occurring together in close spatial association to kyanite. A sample of coarse-grained amphibole schist with subhedral rubies to 20 mm is shown in Figure 23. Figure 26 shows a cluster of large coarse ruby grains from PAR-1 of varying colour depth.

2.5.1 Petrography and mineralogy

Petrographic samples of both the ruby-bearing pargasite/magnesiohornblende schists and immediate surrounding country rocks were examined. The ruby-bearing samples are labelled as sample group PAR-1 (main ruby site on hillside) and PAR-5 (roadside cutting). Host lithologies (i.e. non-ruby bearing) samples are grouped into PAR-2 (pod-like chlorite-mica schist), PAR-3 (footwall of sulphide-bearing amphibolite) and PAR-4A-4C (layered kyanite-bearing amphibole gneiss). Images were taken under both plane polarised (PPL) and cross-polarised light (UXP). Images showing the overall texture under both PPL and UXP were taken for all the polished thin sections.

2.5.1(a) Ruby- bearing samples

The ruby crystals exhibited a wide range of habits. Most were heavily fractured or brecciated and were elongate and aligned parallel to the main foliation within the host rock. Some rare ruby grains were euhedral.

The overall texture of PAR1A shows heavily fractured ruby grains (Figures 29a, 29b), commonly with phyllosilicate reaction rims (later found to be margarite) forming distinctive fringes around the ruby grains (Figure 29c). The size of the ruby grains ranged from 2mm-3mm. Margarite and amphibole (later found to be pargasite/magnesiohornblende) infills fractures within the ruby grains (Figure 29d). Evidence of ruby altering to margarite and opaques is relatively widespread (Figure 29e). pargasite/magnesiohornblende most commonly occurs as decussate grains (Figure 29f). Muscovite was observed to fill spaces between the other minerals present (Figure 29g) and Mg-rich chlorite was found to overprint the other phases present (Figure 29h). Subhedral prismatic kyanite intergrown with pargasite/magnesiohornblende was also found in some 30

sections (Figure 29i). There is also relatively moderate alignment of the ruby grains parallel to the main foliation within the pargasite/magnesiohornblende schist.

PAR1B is very similar to PAR1A in terms of its overall texture with intensely fractured rubies (Figure 30a) with their long axis oriented parallel to the main foliation within the host pargasite/magnesiohornblende schist (Figure 30b). The amphibole grains exhibit clear triple junction grain boundaries and granoblastic textures (Figure 30c). What is unique in this sample compared to 1A is the presence of ~20-micron syn-genetic olive-green spinel inclusions within the rubies (Figure 30d). Some rare ruby grains were euhedral with distinctive and unusual colour zonation parallel to their crystallographic axes (Figure 30e), in the form of a more intensely coloured darker pink band near the rim. The spinel and/or fluid inclusions generally occur as clusters of grains within the rubies (Figure 30f).

Although PAR5 is within 500m proximity to the PAR1 ruby occurrence, samples from here show a more complex mineral assemblage and grain relationships. PAR5C is a ruby- -bearing sample from the roadside cutting and once again the ruby grains are similarly fractured and generally anhedral in shape as in the PAR1 samples (Figure 31a). Kyanite was found to be moderately common in the PAR5C samples (Figure 31b) in contrast to the PAR1 samples where it exclusively occurs as subhedral fractured grains rimming the rubies. As in the PAR1 samples, phyllosilicate (margarite) reaction rims around the rubies is common (Figure 31c). Again, as in the PAR1 samples, spinel inclusions within some of the rubies were also found in some of the PAR5C samples (Figure 31d). The transition from amphibolite to greenschist facies assemblages is shown through Mg-rich chlorite overprinting pargasite/magnesiohornblende and kyanite (Figure 31e). Euhedral muscovite grains were observed interstitial to the amphibole grains (Figure 31f) along with rare interstitial plagioclase (Figure 31g). The most notable difference between the PAR5 ruby-bearing samples and the PAR1 ruby-bearing samples is the presence of relatively abundant monazite (Figure 31h). Some of these were found to be distinctly colour zoned/metamictised (Figure 31i) and many syngenetic with the Mg-chlorite aligned along the same foliation (Figure 31j).

PAR5D is also a ruby- bearing lithology from the roadside cutting. Once again, the ruby grains were found to be highly fractured and generally anhedral in shape (Figure 32a), with fractures and rims filled with margarite (Figure 32b). This is very similar to what was observed in the PAR1 samples. There appears to be two different sized amphiboles with the finer-grained amphiboles

overprinting the coarser-grained amphiboles and both muscovite and Mg-rich chlorite once again occur infilling fractures and overprinting the amphiboles and rubies. Rare minute grains of overprinting carbonate were also found (Figure 32c). Under reflected light at high magnification, pyrite was found to occur within the fine-grained amphibole zone (Figure 32d).

2.5.1(b) Non-ruby -bearing host-rock samples

Samples PAR2A were taken from similar looking boudins of the pargasite/magnesiohornblende schists (but with no visible ruby in hand specimen, Figure 17). They show rare remnant ruby grains, largely replaced by margarite (Figure 33a), indicative of a transitional layer towards the non-ruby bearing host rocks. Quartz was also found to occur within these samples, some with monazite inclusions (Figure 33b). Quartz-plagioclase intergrowths were found to occur mostly between the amphibole grains (Figure 33c). Plagioclase was also found to occur as discrete intergrowths with amphiboles (Figure 33d) and with zoisite (Figure 33e). The main foliation is defined by the parallel alignment of the long axes of the amphiboles (Figure 33f).

Samples of PAR3A are from the footwall amphibolite. These show distinct layering (Figure 34a), including amphibole-rich zones and epidote-zoisite dominant layers (Figure 34b). Abundant subhedral monazite was found within the amphiboles (Figure 34c). Shear bands were seen within the amphibole dominant layers (Figure 34d). Granoblastic textures are prevalent in these samples (Figure 34e) and some of the plagioclase grains are partially replaced by epidote-zoisite (Figure 34f). Amphiboles overprint some of the plagioclase (Figure 34g) while some Fe-rich chlorite partially replaces amphiboles (Figure 34h). Micro veinlets (Figure 34i) of epidote and retrogressive muscovite were also found to occur in these samples (Figure 34j).

Samples PAR4A were collected from the hanging wall and these lack rubies. They are distinctly banded with coarser-grained zones (Figure 35a) and finer-grained zones (Figure 35b). Amphibole (later found to be pargasite/magnesiohornblende) is intergrown with kyanite (Figure 35c) and both pargasite/magnesiohornblende and plagioclase were found as inclusions in kyanite (Figure 35d), pointing to co-crystallisation. In the finer-grained zone, sillimanite occurs as distinctive fringes around kyanite (Figure 35e) and much of the kyanite shows well-developed undulose extinction (Figure 35f).

Sample PAR4B is from the hanging wall and once again lacks rubies. In hand-specimen, it appears to be an altered gneiss, dominantly comprising plagioclase, zoisite-clinozoisite, chlorite,

muscovite, titanite and amphibole. Rare compositionally zoned plagioclase was observed (Figure 36a) while epidote and clinozoisite are commonly seen replacing plagioclase (Figure 36b). Euhedral zoisite–clinozoisite are relatively common (Figure 36c) and zoisite-clinozoisite-chlorite is a common overprinting assemblage (Figure 36d). Chlorite occurs as interstitial Mg-chlorite grains (Figure 36e), radial aggregates (Figure 36f) and as more Fe-rich vermicular grains (Figure 36g). Amphiboles are mostly remnant grains commonly partially replaced by chlorite (Figure 36h). Granoblastic textures were also found in the primary amphiboles (Figure 36i). Other features found included euhedral titanite (Figure 36j) and muscovite porphyoblasts in clinochlore (Figure 36k).

Samples PAR4C differ from samples 4A and 4B, mostly seen in the occurrence of relatively large kyanite grains. These kyanite grains are moderately to intensely deformed and show moderate undulose extinction (Figure 37a). The amphiboles exhibit decussate textures and lack any preferred orientation (Figure 37b). Plagioclase and quartz show well-developed granoblastic textures (Figure 37c). Many of the kyanite grains occur as intergrowths with pargasite/magnesiohornblende showing mutually embayed grain boundaries (Figure 37d). Kyanite was also observed as partially resorbed anhedral-shaped grains (Figure 37e) and overprinted by muscovite (Figure 37f). Sillimanite fringes were also seen to overprint some of the kyanite grains (Figure 37g).

Overall, the ruby-bearing rocks mostly comprise pargasite/magnesiohornblende, corundum (var. ruby), chlorite (mostly clinochlore, rarely nimite), margarite, and in some cases, tremolite or monazite. In contrast, the host rocks (which have sheared contacts with the ruby-bearing pargasite/magnesiohornblende schists) comprise mostly pargasite/magnesiohornblende with varying amounts of plagioclase, kyanite, epidote, clinozoisite, zoisite, chlorite, muscovite, quartz and accessory monazite. Most of the ruby grains exhibited a distinctive margarite rim with continuous optical properties. This material was later examined under SEM and EMPA and confirmed to be margarite. No zircon, monazite, or the more common rutile as syngenetic inclusions were identified in the ruby-bearing samples, and hence in situ dating on the age of formation for the Paranesti ruby samples could not be undertaken. How these mineral assemblages and relationships provide evidence for the metamorphic evolution of the ruby occurrences will be discussed in Chapter 3.

2.5.1 (c) Whole-rock XRD

XRD analysis identified the ruby host assemblages of PAR-1 and PAR-5 as consisting of the same mineral phases (pargasite/magnesiohornblende, corundum, clinochlore, and nimite), with pargasite/magnesiohornblende being the dominant amphibole and clinochlore being the dominant chlorite. PAR1A shows the presence of pargasite/magnesiohornblende , corundum and nimite and PAR1B shows pargasite/magnesiohornblende and corundum. PAR1C shows additional tremolite and clinochore in addition to the pargasite/magnesiohornblende and corundum observed in the other PAR-1 samples. PAR5C only shows pargasite/magnesiohornblende and corundum while PAR5D shows corundum, pargasite/magnesiohornblende and clinochlore.

The host rock non-ruby bearing assemblages (i.e., footwall and hanging wall lithologies) also contain pargasite/magnesiohornblende as the main amphibole along with mostly anorthite and clinozoisite. PAR2 shows pargasite/magnesiohornblende and anorthite while PAR3 shows pargasite/magnesiohornblende , anorthite and clinozoisite. However, due to the closeness of the main peaks amongst the Ca-amphibole subgroup, in order to determine the exact amphibole present the XRD results were further verified through EMPA analysis.

2.5.2 Whole-rock geochemistry

2.5.2 (a) Whole-rock XRF

Whole rock XRF major element analysis (Table 1a) shows that the major oxides within the ruby- bearing amphibole schists are ~40% SiO2, 21% Al2O3, 16% MgO and minor amounts of other elements such as CaO, Fe2O3 and LOI etc. Such a low silica and relatively high Mg content would suggest an ultramafic rather than mafic igneous precursor for the ruby-bearing amphibole schists. Al2O3 is high at 20.96 wt.% which is expected due to the inclusions of ruby grains within the crushed sample.

XRF trace element analysis (Table 1b) shows that chromium is prominent at 8,156 to 10,745 ppm while Ni (1,370 ppm and 792 ppm) and Zn (370 ppm and 175ppm) are also relatively high. All other trace elements are noticeably low (Table 1b).

Table 1a. Major element XRF analysis of the Table 1b. Trace element XRF analysis ruby-bearing host rock (wt.%). of the ruby-bearing host rock (ppm).

Sample DR1 Sample DR1 Rock Pargasite/magnesiohornblende Rock Pargasite/magnesiohornblende type Schist type Schist (ppm) (wt.%) V 55 SiO2 40.12 Cr 10745 TiO2 0.09 Ni 387 Al2O3 20.96 Cu BDL Fe2O3 5.66 Zn 252 MnO 0.15 As 3 MgO 15.92 Rb 16 CaO 9.95 Sr 46 Na2O 1.37 Y 3 K2O 0.46 Zr 12 P2O5 0.01 Mo BDL SO3 0.05 Sb BDL LOI 4.55 Pb 4 Total 99.17

Th BDL

2.5.2 (b) Whole-rock ICP-MS

The whole rock ICP-MS analyses performed on the PAR-1 sample (Table 2) support the XRF trace element analyses with the same high Cr (11,800 ppm), high Ni (450 ppm), and Zn (360ppm) contents. All rare earth elements (REEs) are depleted. The ICP-MS REE results were also used to plot a chondrite-normalised rare earth element diagram (Figure 38). This shows a strongly concave downward curve with HREE depletion.

Table 2. Whole-rock ICP-MS trace element analysis of the ruby-bearing host rock (Sample PAR-1)

Trace Elements ppb Trace Elements ppb

Ca 80,237,522 W 1190 Sc 2376 Tl 98 Ti 366,720 Pb 1985 V5 45381 Th 9 Cr 11,784,253 U 24 Ni 447696 Cu 8365 Zn 378,113 Ga 9264 As 606 Rb 19672 Sr 46021 Y 476 Zr 839 Nb 376 Trace Element Ratios ppb Mo 1212 Nb/Ta 17.3 Cd 36 Zr/Hf 41.5 Sn 465 Y/Ho 37.2 Sb 362 Rb/Sr 0.43 Cs 575 Ba 44409 La 90 Ce 237 Pr 36 Nd 176 Sm 55 Eu 62 Gd 68 Tb 12 Dy 76 Ho 16 Er 48 Tm 8 Yb 55 Lu 9 Hf 20 Ta 16

2.5.3 SEM and EMPA

2.5.3 (a) SEM imaging

SEM imaging was undertaken on the phyllosilicate phases surrounding the corundum grains in PAR1A in order to determine their mineralogy and composition (Figure 39).

Figure 40 shows a cluster of spinel inclusions examined under the SEM as seen during the petrographic studies. These inclusions within PAR1B were analysed with points taken in the core, rim and in the surrounding corundum. Based on these compositions (dominant Al, Cr, Fe, Zn), the inclusions are likely to be zinc-enriched spinels (for quantitative data on these see the EMPA section below).

2.5.3 (b) EMPA

The chemical composition of the Paranesti rubies determined using the EMPA method is shown in Table 3. A total of 38 points from PAR-1 (12 crystals) and 20 points from PAR-5 (6 crystals) were analysed and this shows that the Paranesti rubies are enriched in (in decreasing order of abundance) Cr, Fe and Si and that rubies from the main site (PAR-1) are more enriched in Cr compared to those from the road-cutting (PAR-5). Variations are shown in the concentrations of Cr (Cr2O3 PAR-1 0.11 wt% - 1.68 wt%, PAR-5 0.13 wt% - 0.37 wt%), Fe (FeO PAR-1 0.19 wt% -

0.73 wt%, PAR-5 0.18 wt% - 0.36 wt%) and Si (SiO2 PAR-1 0.01 wt% - 0.37 wt%, PAR-5 0.01 wt% - 0.22 wt%).

Table 3. Representative EMPA analyses of Paranesti ruby grains. (38 points analysed for PAR-1 rubies and 20 points analysed for PAR-5 rubies).

PAR-1 PAR-5 σ² (std μ σ² (std Wt (%) Min Max μ (mean) dev) Min Max (mean) dev)

SiO2 0.01 0.37 0.08 0.07 BDL 0.22 0.07 0.05 TiO2 BDL 0.01 0.02 0.03 BDL 0.06 0.01 0.02 Al2O3 96.83 100.05 98.58 0.90 97.74 99.09 98.36 0.38 FeO 0.19 0.73 0.40 0.15 0.18 0.36 0.36 0.04 MnO BDL 0.05 0.01 0.01 BDL 0.03 0.01 0.01 MgO BDL 0.05 0.01 0.01 BDL 0.01 0.00 BDL CaO BDL 0.05 0.01 0.01 BDL 0.05 0.01 0.01 Na2O BDL 0.03 BDL 0.01 BDL 0.01 BDL BDL K2O BDL 0.01 BDL BDL BDL 0.01 BDL BDL P2O5 BDL 0.04 0.01 0.01 BDL 0.03 0.01 0.01 Cr2O3 BDL 1.68 0.55 0.45 BDL 0.37 0.14 0.07 Ga2O3 BDL 0.04 0.01 0.01 BDL 0.04 0.01 0.01 Nb2O5 BDL 0.07 0.01 0.01 BDL 0.03 0.01 0.01 Ta2O5 BDL 0.10 0.02 0.03 BDL 0.10 0.02 0.03 ZnO BDL 0.07 0.02 0.02 BDL 0.07 0.02 0.03 Total 99.71 0.55 98.90 0.37 Calculated cations assuming stoichiometry

Si 0.00 0.00 Ti 0.00 0.00 Al 3.97 3.96 Cr 0.01 0.00 Fe2+ 0.01 0.01 Mn 0.00 0.00 Mg 0.00 0.00 Ca 0.00 0.00 Na 0.00 0.00 K 0.00 0.00

EMPA analysis (Table 4) of the spinel inclusions within the ruby grains show that they are dominantly composed of Al, Cr, Fe and Mg (with Sp5 also being enriched in Zn - 7.1 wt%) and that their composition varies widely, even though the grains are in close proximity to each other. On the spinel discrimination diagram (Figure 41), the spinel inclusions plot in the fields for Al- chromite (1), picotite (4) and with one result bordering on hercynite. Considering the Mg-Fe exchange, Figure 42 shows that most of the results plot close to pleonaste with one plotting close to hercynite.

Table 4. Representative electron-microprobe analyses of spinel inclusions within ruby total of 5 points.

Oxide (wt%) Sp1 Sp2 Sp3 Sp4 Sp5

SiO2 0.01 0.02 0.04 0.01 0.13

TiO2 0.01 0.11 BDL 0.04 BDL

Al2O3 42.64 58.37 42.92 53.85 37.77 FeO 19.01 14.93 21.80 18.40 25.26

MnO 0.36 0.19 0.46 0.37 1.26 MgO 13.60 8.54 9.89 7.14 1.91 CaO BDL BDL BDL BDL 0.05

Na2O 0.01 BDL BDL BDL BDL

P2O5 BDL 0.02 0.04 BDL 0.02

Cr2O3 20.37 16.22 22.26 18.77 36.94

Ga2O3 BDL BDL 0.01 0.07 0.12

Nb2O5 BDL 0.02 0.01 BDL 0.04 ZnO 1.14 0.79 1.37 0.61 7.09 Total 97.03 99.15 98.80 99.18 98.59 Calculated cations assuming stoichiometry Si 0.00 0.00 0.00 0.00 0.00 Ti 0.00 0.00 0.00 0.00 0.00 Al 1.46 1.92 1.48 1.81 1.22 Cr 0.47 0.36 0.51 0.42 0.80 3+ Fe 0.08 0.00 0.00 0.00 0.00 2+ Fe 0.38 0.37 0.53 0.44 0.69 Mn 0.01 0.00 0.01 0.01 0.03 Mg 0.59 0.36 0.43 0.36 0.09 Zn 0.02 0.02 0.03 0.01 0.17

Some six points on the phyllosilicate reaction rims around the ruby grains from PAR-5 were analysed and all found to be margarite (Table 5). One margarite analysis (Mar4) appears to have lower concentrations of SiO2 (23.4 wt% vs. other 5 points of 37 wt%) and CaO (8.9 wt% vs. 12 wt%) but with higher Al2O3 (60.6 wt% vs. 48 wt%). This contrast in results could be due to the proximity of the analysis point to the corundum grain, resulting in apparently enhanced Al2O3. The consistency of all other five points shows the representative data of the margarite from the Paranesti occurrences.

Table 5. Representative electron-microprobe analyses of margarite – reaction rim around ruby grains total of 6 points.

Oxide (wt%) Mar1 Mar2 Mar3 Mar4 Mar5 Mar6

SiO2 37.09 37.60 37.61 23.42 36.66 37.37

TiO2 0.05 BDL BDL 0.02 0.04 0.05

Al2O3 48.74 48.40 49.45 60.58 49.82 48.46 FeO 0.11 0.11 0.18 0.18 0.16 0.14 MnO BDL 0.01 BDL 0.01 BDL BDL MgO 0.12 0.10 0.11 0.05 0.09 0.10 CaO 11.70 11.59 12.08 8.90 12.26 11.72

Na2O 1.13 0.65 0.50 0.58 0.67 0.71

K2O 0.12 0.84 0.20 0.44 0.25 0.44

P2O5 0.01 0.02 0.01 BDL 0.02 0.01

Cr2O3 0.05 BDL 0.02 0.06 BDL BDL

Ta2O5 0.09 BDL 0.07 BDL 0.02 0.07 ZnO 0.01 BDL BDL 0.07 BDL 0.06 Total 91.22 91.37 92.23 94.37 91.99 91.07

Calculated cations assuming stoichiometry Si 2.02 2.06 2.06 1.63 1.99 2.04 Ti ------Al 3.99 3.96 4.05 4.96 4.08 3.97 Fe2+ 0.01 0.01 0.01 0.01 0.01 0.01 Mn ------Mg 0.01 0.01 - - - - Ca 0.87 0.86 0.90 0.66 0.91 0.87 Na 0.15 0.09 0.07 0.08 0.09 0.09 K 0.01 0.07 0.02 0.04 0.02 0.04

Some 31 points on the amphiboles from PAR-1 were probed. They are clearly Mg-rich (Table 6) with only minor variation in their composition, primarily in Fe, Al, Si, Mg and Ca. The amphiboles are low in Ti, Mn, K and F. On the amphibole classification diagram for the compositional range of pargasite- ferropargasite- hastingsite - magnesiohastingsite (Figure 43), they all plot close to the pargasite/magnesiohornblende end-member.

Table 6. Representative EMPA analyses of amphiboles surrounding ruby grains within the ruby-bearing amphibole schists PAR-1 (37 points).

Pargasite/magnesiohornblende Oxide (wt%) Min Max μ (mean) σ² (std dev)

SiO2 44.81 47.75 46.03 0.78

TiO2 0.00 0.26 0.06 0.06

Al2O3 13.00 15.82 14.64 0.83 FeO 5.24 8.89 5.94 0.99 MnO 0.06 0.24 0.17 0.04 MgO 15.69 17.62 16.87 0.53 CaO 11.07 12.23 11.94 0.37 BaO BDL 0.07 0.01 0.02

Na2O 1.52 1.87 1.71 0.10

K2O 0.23 0.37 0.37 0.03 F BDL 0.17 0.03 0.05

H2O BDL 3.59 2.37 1.05 Total 100.00 0.02

Calculated cations assuming stoichiometry Si 8.00 8.00 8.00 - Ti - 0.03 0.01 0.01 Al 1.08 1.37 1.24 0.08 Fe2+ 0.75 1.37 0.86 0.14 Mn 0.01 0.04 0.02 0.01 Ca 2.92 3.25 3.14 0.10 Ba 0.75 0.85 0.81 0.03 Na 0.51 0.64 0.58 0.04 K 0.05 0.08 0.06 0.01

The chemical composition of the surrounding chlorite at PAR-1 using EMPA is shown in Table 7. These values agree with clinochlore (peninite) as identified in XRD. However, on the chlorite classification diagram of Hey (1954; Figure 44), they mostly plot in the peninite (a discredited former species now defined as clinochlore) field with one analysis plotting in the talc-chlorite field.

Table 7. Representative EMPA analyses of chlorite from the zone surrounding the ruby-bearing amphibole schist boudins PAR-1 (20 points).

Clinochlore μ σ² Oxide (wt%) Min Max (mean) (std dev)

SiO2 36.53 38.44 36.22 2.00

TiO2 0.01 0.39 0.05 0.08

Al2O3 22.03 22.94 22.49 0.37 FeO 6.24 10.69 7.14 1.37 MnO BDL 0.15 0.04 0.04 MgO 19.63 37.44 36.06 2.79 CaO BDL 0.37 0.02 0.07 BaO BDL 0.04 0.01 0.01

Na2O BDL 0.06 0.01 0.01

K2O BDL 1.97 0.12 0.44 F BDL 0.59 0.04 0.13

H2O 8.26 11.10 9.79 0.65

Total 99.98 0.06

Calculated cations assuming stoichiometry

Si 2.71 3.48 2.79 0.17 Ti - 0.03 - 0.01 Al 2.38 2.49 2.45 0.03 Fe3+ - 0.12 0.04 0.03 Fe2+ 0.43 0.84 0.52 0.10 Mn - 0.01 - - Mg 2.65 4.36 4.14 0.39 Ca - 0.03 - 0.01 Na - 0.01 - - K - 0.23 0.01 0.05 F - 0.18 0.01 0.04 OH 7.82 8 7.99 0.04

2.5.4 LA-ICP-MS

The chromophore and trace element LA-ICP-MS results (Table 8) show that the Paranesti ruby samples have high Cr (8637 ppm), Fe (3822 ppm), Si (2,456 ppm) and Ca (max 2119 ppm). There are very low concentrations of V (max 5 ppm), Ga (37 ppm), Ti (184ppm) and Mg (mostly below detection limit but with a few exceptions up to 376 ppm). These results are in line with the EMPA analyses. Results for the other trace elements analysed by LA-ICP-MS are shown in Table 9 and show that other than Mg, Ti, Cr, Fe, Ga, Si and Ca, these rubies are mostly devoid of trace elements.

Table 8. Chromophore and key trace elements LA-ICP-MS analyses (ppm) of the Paranesti rubies and their ratios. Polished Mounts (Position: C-Core R-Rim; Colour intensity: P-Pale M-Medium D-Dark). PAR-1 Mg Ti V Cr Fe Ga Fe/Mg Ga/Mg Cr/Ga Fe/Ti 1M-R1 42 81 2 1,668 1,820 18 43 - 92 23 1M-C1 17 184 2 1,101 1,810 19 104 1 57 10 1M-C2 21 46 2 903 1,841 20 87 1 44 40 1M-R2 12 19 1 1,869 1,689 17 144 1 113 90 2P-R1 12 11 1 384 2,095 18 182 2 21 189 2P-C1 12 6 1 451 2,150 18 184 2 25 379 2P-C2 26 13 1 460 2,485 22 97 1 21 189 2P-R2 13 20 1 460 2,374 17 180 1 37 118 3P-R1 11 19 1 403 2,094 18 185 2 22 108 3P-C1 19 8 1 360 2,065 17 106 1 21 374 3P-C2 8 37 1 448 2,442 20 377 3 23 86 3P-C3 12 37 1 494 2,664 20 224 2 24 80 4M-R1 9 125 1 2,101 1,572 15 167 2 142 13 4M-C1 10 143 1 1,373 1,591 14 152 1 95 11 4M-C2 16 154 1 1,239 1,678 16 104 1 77 11 4M-R2 13 119 2 2,856 2,361 21 179 2 136 20 5M-C1 15 48 3 1,868 2,664 23 172 1 82 55 5M-C2 13 36 3 1,946 2,511 21 190 2 91 69 5M-C3 14 43 2 1,988 2,421 20 178 1 98 57 6M-R1 7 15 2 2,773 1,650 16 225 2 171 108 6M-C1 14 37 3 2,364 1,999 21 146 2 110 61 6M-C2 16 11 2 2,376 1,704 18 107 1 126 159 6M-C3 25 19 3 2,373 1,961 20 79 1 18 102 PAR-5 Mg Ti V Cr Fe Ga Fe/Mg Ga/Mg Cr/Ga Fe/Ti 1D-R1 82 69 5 8,637 3,822 37 46 - 379 56 1D-C1 78 73 5 8,157 3,800 36 49 - 376 52 1D-C2 76 62 5 8,374 3,762 37 50 - 370 60 1D-C3 61 56 4 6,614 3,008 22 50 - 361 54 2P-C1 42 45 3 646 2,437 18 58 - 37 54 2P-R1 23 36 2 524 2,479 18 108 1 36 83 2P-C1 376 46 3 417 2,643 19 7 - 22 58 3D-C1 11 16 2 4,173 1,969 19 175 2 219 124 3D-C2 8 22 2 5,483 1,980 19 246 2 371 88 3D-R1 10 17 3 7,548 1,987 15 195 2 489 116 3D-C3 26 13 3 3,961 2,075 17 78 1 236 164 4M-C1 12 10 2 1,522 2,170 17 188 1 89 237 4M-R1 12 16 2 1,369 2,206 17 179 1 79 142 4M-R2 9 14 2 1,652 2,374 18 246 2 91 165 5M-C1 15 152 2 172 1,837 13 119 1 13 12 5M-C2 44 153 2 4 2,163 14 49 - - 14 5M-R1 13 190 2 49 1,842 13 147 1 4 10 43

Table 9. Other trace element LA-ICP-MS analyses (ppm) of the Paranesti rubies. Polished Mounts (Position: c-core r-rim; Colour intensity: p-pale m-medium d-dark).

PAR-1 Al B Na Si P K Ca Ni Cu Zn 1M-R1 518,665 10 2 2,456 47 6 817 1 <0.087 BDL 1M-C1 518,665 11 3 1,805 21 11 812 10 <0.101 3 1M-C2 518,665 8 <1.02 1,823 26 24 982 1 BDL BDL 1M-R2 518,665 6 2 1,515 19 <1.18 940 0 <0.051 BDL 2P-R1 518,665 7 <1.37 1,743 19 14 926 1 <0.107 BDL 2P-C1 518,665 9 <1.37 1,892 15 6 1,069 2 1 <0.118 2P-C2 518,665 7 2 1,663 19 6 1,484 BDL 1 <0.181 2P-R2 518,665 10 1 2,374 22 <1.58 852 BDL BDL BDL 3P-R1 518,665 7 10 2,045 16 5 1,037 BDL BDL 1 3P-C1 518,665 10 4 1,606 24 4 936 BDL BDL 1 3P-C2 518,665 7 2 1,676 17 <2.70 972 BDL BDL BDL 3P-C3 518,665 7 <1.37 1,536 24 <2.39 1,379 BDL BDL BDL 4M-R1 518,665 6 5 1,689 21 <2.37 769 BDL <0.102 BDL 4M-C1 518,665 8 <1.22 1,587 23 <2.17 965 1 BDL <0.167 4M-C2 518,665 6 <1.09 1,523 24 <1.96 1,170 1 1 3 4M-R2 518,665 4 1 1,113 37 <1.40 1,900 BDL 1 <0.107 5M-C1 518,665 3 4 1,366 20 2 2,119 BDL 2 <0.112 5M-C2 518,665 3 2 956 21 4 1,855 BDL 3 BDL 5M-C3 518,665 3 1 999 19 2 1,815 BDL 2 BDL 6M-R1 518,665 3 <1.18 1,376 18 <1.66 1,017 BDL BDL BDL 6M-C1 518,665 3 1 781 13 <1.18 1,820 BDL BDL BDL 6M-C2 518,665 5 12 1,379 20 <1.42 1,360 BDL BDL BDL 6M-C3 518,665 6 13 1,008 12 1 1,590 1 1 <0.098 PAR-5 Al B Na Si P K Ca Ni Cu Zn 1D-R1 518,665 3 <0.70 837 10 <0.97 1,903 1 BDL 1 1D-C1 518,665 3 <0.73 959 13 <1.02 1,902 1 1 BDL 1D-C2 518,665 4 <0.71 875 10 <0.99 1,756 1 BDL BDL 1D-C3 518,665 3 <0.71 948 12 <0.98 1,147 1 BDL BDL 2P-C1 518,665 6 3 1,136 21 5 1,436 1 <0.37 3 2P-R1 518,665 5 <0.81 993 11 <1.09 1,654 1 1 1 2P-C1 518,665 5 36 2,123 20 177 1,663 2 BDL BDL 3D-C1 518,665 5 <1.54 1,244 36 4 653 1 BDL 1 3D-C2 518,665 5 <1.59 1,258 14 <2.02 909 BDL BDL BDL 3D-R1 518,665 7 <1.65 1,061 15 <2.11 954 1 BDL <0.163 3D-C3 518,665 6 <1.56 1,371 24 <2.02 809 1 <0.139 <0.206 4M-C1 518,665 4 <1.96 1,373 22 <2.55 736 1 <0.119 <0.253 4M-R1 518,665 5 3 1,396 20 <2.22 1,048 1 <0.155 <0.179 4M-R2 518,665 4 3 1,269 25 <2.09 1,394 1 BDL BDL 5M-C1 518,665 3 <1.61 1,362 20 <2.06 737 1 <0.143 <0.152 5M-C2 518,665 5 <1.64 1,193 36 <2.10 765 1 1 2 5M-R1 518,665 4 <1.57 1,370 16 <1.97 1,037 BDL BDL BDL

For the elements that were analysed by both EMPA and LA-ICP-MS techniques, the results show a general consistency. Core/rim contrast on average for all samples combined did not show any significant variation. When colour variation is considered, on average the Fe contents at the rim for the paler rubies are approx. 200 ppm lower than at the core. The paler PAR-1 samples show similar maximum and average Fe contents to medium coloured PAR-1 with less variance in the range. Ti values are marginally higher in medium coloured PAR-1 samples compared to the paler samples (72 ppm vs 18 ppm) with no noticeable difference between core and rim. No noticeable trace element variations were observed between the core/rim results for PAR-5. As expected, however, Cr is significantly higher in the darker coloured rubies compared to the medium and pale coloured varieties. No noticeable variance based on core/rim positions was observed.

PAR-1 shows a higher Cr2O3 content than PAR-5 under EMPA (average 0.55 wt% vs. 0.14 wt% and maximum 1.68 wt% vs. 0.37 wt%, respectively). However, the difference under LA-ICP-MS for Cr shows a higher maximum in PAR-5 (8637ppm) vs. PAR-1 (3656ppm). The difference is most likely due to the individual grains analysed and the colour variance between the samples. The highest Cr sample analysed (8637 ppm) showed a deep red colour as seen in Fig. 5B. PAR-5 shows a higher Fe (3822 ppm) than PAR-1 (3664 ppm). PAR-1 shows a slightly higher Ca content (2119 ppm) than PAR-5 (1903 ppm).

Both Paranesti ruby sets demonstrate low Ti, V, Ga and Mg values, mostly below detection limit of the LA-ICP-MS. The low values are somewhat problematic in the elemental graph comparison against known ratios, as ratios are magnified due to close-to-zero denominators. The sample plots are therefore more skewed towards one axis. PAR-1 shows a higher Si/Ca ratio than PAR- 5 (average 1.45 and 1.16, respectively).

2.5.5 Oxygen Isotopes

Worldwide corundum oxygen isotope values have been found in a wide range from -27‰ (Khitostrov, Russia) to +23‰ (Mong Hsu, Myanmar) with most in the range of +3‰ to +21‰ (Yui et al., 2003; Giuliani et al., 2007; Kyrlov and Glebovitsky, 2007; Vysotskiy et al., 2015). This criterion has often been used to determine the geological origin of coloured corundum and especially the gem corundums, rubies and sapphires. δ18O has been particularly useful in determining the likely primary geological origin of placer corundums where the primary origin is uncertain (Giuliani et al., 2005). As isotopic fractionation is a function of both temperature and geological processes, oxygen isotope data need to be treated with some degree of caution. Thus, there are very few examples where oxygen isotopes have been used to “fingerprint” the geographic location of gem corundums of uncertain location designation (Vysotskiy et al., 2014).

2.5.5 (a) Whole grain dissolution laser-fluorination results

Using the laser-fluorination method, the oxygen isotope ratio for the pargasite/magnesiohornblende schist hosted PAR-1 ruby was found to be δ18O 1.0‰. This analytical run also included a number of rubies and sapphires from different geological environments. Sapphires from desilicified granite pegmatites intruding metaperidotites were found to range from 4.8‰ to 5.0‰ and rubies from marble-hosted deposits were found to range from 20‰ to 22‰ (Table 10). These results confirm the geological typology of the various Greek corundum deposit types. i.e. δ18O 4.8‰ ± 0.2‰ for sapphires in plumasite and 20.5‰ for sapphires in marble in agreement with the range of values found worldwide for gem corundums in a range of geological environments (Figure 45).

Table 10. Oxygen isotope results from the 2009 reconnaissance study using the laser-ﬂuorination method, n=1. Sample Location Sample type Deposit Type δ18O NAX2 Naxos, Greece Colourless sapphire Desilicified pegmatite 4.80 NAX3 Naxos, Greece Colourless to blue sapphire Desilicified pegmatite 5.05 PAR-1 Paranesti, Greece Red ruby Pargasite/magnesiohorn 1.00 blende schist KIM2 Kimi, Greece Pink ruby Marble-hosted 20.50 Xanthi Xanthi, Greece Purple-pink ruby Marble-hosted 22.09

In 2018, PAR-1 samples were again analysed using the laser-fluorination method at the same laboratory as the 2009 study. However, the results obtained were significantly different to the previous laser-fluorination results and are listed in Table 11. Three grains from PAR-1 marked 46

as PAR-1A; PAR-1B; and PAR-1C were sent for analysis together with half a grain of the original SIMS sample still enclosed in araldite. Results obtained were more than double of what both the 2009 laser fluorination and 2017 in situ study had shown at >2.3‰ average. Extrapolating on this relationship, it would place the marble results from the 2009 study well into >40‰ and off the scale for any known ruby and sapphire oxygen isotope values. Calibration results on the original SIMS sample was not received as part of the results and further clarification sought from the lab was not provided. This set of results was therefore deemed as being highly suspect and not used.

Table 11. Oxygen isotope results from the 2018 study using the laser-ﬂuorination method, n=3.

18 Sample δ Ovsmow PAR-1a 2.8 PAR-1a 2.0 PAR-1a 2.5

PAR-1b 1.8 PAR-1b 2.3

PAR-1c 2.3 PAR-1c 2.6

2.5.5 (b) In situ secondary ionisation mass spectrometry (SIMS) results

The oxygen isotope ratios δ18O (V-SMOW) are presented in Table 12. PAR-5 results show values of −0.37‰ to 0.85‰ (0.14 ± 0.24), on average slightly lower compared to PAR-1 results of 0.44‰ to 1.68‰ (1.00 ± 0.42) even though the two occurrences are only 500 m apart.

Table 12. Oxygen isotope results from 2017 using the SIMS method. δ18O Mean Number of Sample Grain δ18O Min δ18O Max Analyses PAR-1a 0.80 1.20 1.01 19 PAR-1b 0.65 1.37 0.96 15 PAR-1c 0.75 1.20 1.03 15 PAR-1 Total 0.65 1.37 1.00 49 PAR-5central -0.04 0.51 0.37 12 PAR-5a -0.14 0.85 0.25 10 PAR-5b -0.37 0.42 0.03 8 PAR-5c -0.22 0.16 -0.06 9 PAR-5d 0.08 0.37 0.17 5 PAR-5 Total -0.37 0.85 0.14 44 Combined PAR- -0.37 1.37 0.60 93 1 and PAR-5 47

Based on the in situ analysis spot positions (Figure 28, Table 13) on the individual PAR-5 ruby grains A, grains B and C show average negative (-0.12‰ and -0.08‰ respectively) values for the cores and close to zero values for the rims (0.19‰ and 0.16‰ respectively). However, grain A appears to show slightly higher values for both the core (0.13‰) and rim (0.43‰). All three grains clearly show lower δ18O values from the cores to the rims. PAR-1 fragments were too small in size to do any meaningful core vs. rim analysis (refer Figure 27). Although the δ18O variance between the PAR-1 and PAR-5 grains are close, there is enough difference to distinguish between ruby from the two localities. Corundums are extremely slow diffusors of oxygen and hence grain to grain variability in δ18O is to be expected, as found by Bindeman et al. 2010, 2014) in their detailed study on metamorphic-hosted corundums from Karelia in northwest Russia.

Table 13. PAR-5 ruby grains core vs rim oxygen isotope results comparison.

Sample Delta Sample Delta Sample Delta points values points values points values PAR-5 δ18O Position PAR-5 δ18O Position PAR-5 δ18O Position Grain- Grain- Grain- A@10 0.13 core B@3 -0.14 core C@2 0.13 core Grain-A Grain- Grain- @3 0.37 core B@4 0.23 core C@3 0.03 core Grain-A Grain- Grain- @4 0.19 core B@6 -0.37 core C@4 -0.12 core Grain-A Grain- Grain- @7 0.24 core B@7 -0.37 core C@5 -0.18 core Grain-A Grain- Grain- @8 -0.14 core B@1 0.01 rim C@6 -0.22 core Grain-A Grain- Grain- @9 -0.02 core B@2 0.15 rim C@7 -0.10 core Grain-A Grain- Grain- @1 0.85 rim B@5 0.42 rim C@8 -0.18 core Grain-A Grain- Grain- @2 0.51 rim B@8 0.18 rim C@9 -0.03 core Grain-A Grain- @5 0.37 rim C@1 0.16 rim Grain-A @6 0.10 rim Core - average 0.13 -0.12 -0.08 Rim - average 0.43 0.19 0.16

The spot-to-spot reproducibility of the PAR-1 result was further validated by the 2017 study of Sutherland et al., (2017) where it was calculated as 0.36‰ (2 SD) during the analytical session. The PAR-1 sample was used as the standard for the study of oxygen isotopes within unusual

alluvial sapphires from the Orosmayo region of northwest Argentina by Graham et al. (2019). In this study, the spot-to-spot reproducibility for PAR-1 was calculated as 0.42% (2 SD) (Graham pers. comm, 2020). The consistency of the results with minimum variances for the PAR-1 sample demonstrates the accuracy of the results obtained in this study.

The ruby in pargasite/magnesiohornblende schists from Paranesti has a very low δ18O-values of 1‰ or lower and can be interpreted in different ways which will be discussed in detail in Chapter 3. (i) inherited pre-metamorphic reactions between sea-water and hot basic/ultrabasic rocks before subduction and metamorphism (Bindeman et al., 2010); (ii) syn-metamorphism depletion in 18O related to hydration during amphibolite facies metamorphic conditions (Krylov and Glebovitsky, 2007); (iii) post-amphibolite facies metamorphism with recrystallization under the effect of metasomatism of the metamorphosed mafic/ultramafic rocks with a high depletion in 18O (Wilson and Baksi, 1983); and (iv) metamorphic/metasomatic conditions involving deeply penetrating meteoric waters along major crustal structures (Wilson and Baksi, 1983).

Chapter 3. Discussion and Conclusions

This chapter is mostly based on published papers in the Canadian Mineralogist and Minerals with myself as first author.

Wang, K.K., Graham, I.T., Lay, A., Harris, S.J., Cohen, D.R., Voudouris, P., Belousova, E., Giuliani, G., Fallick, A.E. and Greig, A., 2017. The origin of a new pargasite-schist hosted ruby deposit from Paranesti, Northern Greece. Canadian Mineralogist, 55(4), 535-560. (Appendix 1)

3. Discussion and conclusions

PAR-1 and PAR-5 are hillside and roadside surface ruby outcrops in north-eastern Greece studied systematically for the first time. Ruby crystals from Paranesti range in size from <1 mm to 20 mm (average size 5–10 mm), are of pale pink to deep red colour and are of cabochon quality. The ruby grains are found to be mostly free of inclusions with only one crystal found to have Mg- and Cr-bearing hercynite (spinel) inclusions. The rubies are found within pargasite/magnesiohornblende schist boudins surrounded by clinochlore schist. The main mineral assemblage associated with rubies comprises pargasite/magnesiohornblende +spinel+kyanite +margarite +muscovite +monazite +clinochlore. Plagioclase is found within the non-ruby bearing host rock but not found within the ruby-bearing assemblage. Both EMPA and LA-ICP-MS results have indicated high Cr, Si and Ca, and low Ti, V and Ga contents. A summary comparison of the main findings from the petrographic and geochemical studies on the Paranesti rubies is presented in Table 1.

Table 1. Summary of Paranesti ruby results (Wang et al., 2017). Attributes PAR-1 PAR-5 Physical characteristics Site Hillside surface outcrop Roadside surface outcrop – 300m east of PAR1 Grain-size 10mm-20mm 5mm-10mm Colour Deeper red than PAR-5 (generally) Medium red Inclusions Picotite (Mg- and Cr-bearing None hercynite -Spinels) Textural More fractured, finer-grained - characteristics Host rock pargasite/magnesiohornblende pargasite/magnesiohornblende schist schist EMPA – whole-rock analyses (wt.%) Cr2O3 0.11-1.68 0.13-0.29 FeO 0.19-0.73 0.18-0.36 TiO2 0-0.01 0-0.06 Ga2O3 0-0.04 0-0.04 LA-ICP-MS – trace element analysis (ppm) Cr 360-2856 4-8627 Fe 1572-2664 1833-3822 V 1-3 2-5 Mg 7-42 8-376 Ti 6-184 10-190 Ga 14-23 13-29 Si 781-2456 837-2123 Ca 769-2119 653-1903 50

3.1 Distinctive geochemical and oxygen isotope signatures of Paranesti ruby

3.1.1 Geochemical signatures - Metamorphic vs. Magmatic

A range of studies have used chromophore elements (Fe, Cr, Ti, V) and elemental genetic indicators (Ga and Mg) to distinguish corundum from different primary sources, using a range of elemental diagrams (e.g. Peucat et al., 2007; Sutherland et al., 2009; Sutherland et al., 2015). These have been utilised to help understand the origin and evolution of the Paranesti rubies and are shown below.

3.1.1 (a) Ga/Al fractionation

Gallium is close in size to aluminium (Ga+3 = 0.62Å; Al+3 = 0.57Å) and the Ga/Al ratio is relatively constant in the continental crust (Skinner, 1979). Ga contents are found to be lower in metamorphic corundums than in magmatic ones, both of which have a constant Al content (Peucat et al., 2007). It has been proposed that Ga/Al ratios (=10,000 Ga/Al) falling in the range between 1 – 1.5 are representative of metamorphic environments while those with higher ratios (2.5-5.3) represent magmatic environments of crystallisation (Whalen et al., 1987). This differentiation has been explained as due to circulation of Fe-rich fluids during partial melting of granulites. F and Ga would form GaF6−3 ions extracted during the formation of alkaline melts (with high Ga/Al ratios), whereas the granulitic restites, enriched in plagioclase and thus in Al, would be depleted in Ga (with low Ga/Al ratios). The Paranesti rubies have a Ga/Al range of 0.25- 0.56, therefore indicating a metamorphic environment of crystallisation.

3.1.1 (b) Mg fractionation

Ga/Mg was originally used to differentiate between blue sapphires of magmatic versus metamorphic origin (Peucat et al., 2007) and was extended to ruby suites by Sutherland et al., (2017). The Mg content is found to be 5-20ppm for magmatic sapphires and 60-130ppm for metamorphic sapphires and it has been proposed that relatively high Mg contents in corundum are due to fluid circulation during a metamorphic stage where metasomatic exchange occurs between mafic and adjacent Si-rich acidic rocks (Peucat et al., 2007). According to Saminpanya et al., (2003) Ga contents >100 ppm and Cr2O3/Ga2O3 ratios <1 typify corundum of magmatic origin, while Ga contents <100 ppm and Cr2O3/Ga2O3 ratios >1 are more characteristic of corundums of metamorphic origin. Peucat et al., (2007) found that high Ga/Mg ratios (>10) are

indicative of magmatic sapphires while low ratios (<10) are indicative of metamorphic and metasomatic sapphires. Cr/Ga ratios <1 are typical of magmatic corundums and higher values typical of metamorphic corundums (Abduriyim, 2006). The Paranesti rubies have Ga contents between 13-29ppm, Ga/Mg ratios of 0.05 – 2.6, Cr/Ga ratios of 3.8 – 488.6 and average

Cr2O3/Ga2O3 ratios of 123, all indicating a metamorphic origin, as expected from the field relationships and petrographic analysis.

3.1.1 (c) Ti, Cr, V

Titanium concentration has been found to be highly variable within sapphires with the potential to differentiate between sapphires of magmatic and metamorphic origins (Peucat et al., 2007). This may reflect the presence of submicroscopic rutile inclusions, resulting in enlarged Ti trends. The use of triangular plots of Mg(x100)-Fe-Ti(x10) helps to categorise Fe-rich magmatic and Mg- rich metamorphic fields (Sutherland et al., 2009; Figure 46). As Ti and Mg are both relatively low in the Paranesti rubies, the Mg(x100)-Fe-Ti(x10) triangular plot did not show a clear distinction between magmatic and metamorphic origin for the Paranesti rubies with most points falling close to the Fe-Mg*100 line and across either side of the defining line.

Although vanadium has also been found to exhibit a broad range of concentrations in both magmatic and metamorphic sapphires, only metamorphic sapphires show high values of V and therefore V alone was not found to be a discriminating element by Peucat et al., (2007).

The triangular plot of Cr(x10)-Fe-Ga(x100) of Peucat et al., (2007) helps to categorise Cr/Fe-rich metamorphic from Ga-rich magmatic fields (Figure 47). The Paranesti rubies are found to contain very low levels of Ga and thus all plot in the metamorphic region on this diagram. The plot also shows an almost straight linear trend from just to the left of the centre of the Fe- Ga*100 baseline to the Cr*10 apex indicating an almost continuous range in Cr at almost constant Fe and Ga.

Both PAR-1 and PAR-5 corundums demonstrates a linear trend in their chromophore trace elements. The elemental diagrams Mg(x100)-Fe-Ti(x10) (Figure 46), Cr(x10)-Fe-Ga(x100) (Figure 47), Ga/Mg vs. Fe/Mg (Figure 48 and 49) and Fe/Ti vs. Cr/Ga (Figure 50) based on LA-ICP-MS results, show that the Paranesti corundums fall mostly within the metamorphic fields on all of these diagrams. Transitional values across this magmatic/metamorphic relationship may represent metasomatic processes (Sutherland et al., 2015). There is one outlier noted from

Figures 48 to 50, most likely due to the analysis hitting a minute inclusion within the analysed grain.

The high R2 value based on the Fe/Mg vs. Ga/Mg elemental discrimination diagram shows that both PAR-1 and PAR-5 corundums contain highly consistent trace element compositions (Figure 49).

3.1.2 Mafic/ultramafic protolith

Further differentiation of the metamorphic field using the FeO+TiO2+Ga2O3 vs. FeO-Cr2O3-MgO-

V2O3 discrimination diagram of Giuliani et al., (2010, 2014a) shows that the dark-red rubies from PAR-5(A) clearly fall in the field for rubies with a mafic-ultramafic origin, while the paler-coloured PAR-5(B) and most of the PAR-1 rubies fall in the overlapping region between ruby in mafic- ultramafic rocks and metasomatic corundum (Figure 51). Most of the rubies can clearly be attributed to a mafic-ultramafic protolith. The low SiO2 (40.1 wt%) and high MgO (15.9 wt%) content shown by the whole-rock XRF analysis (Chapter 2, Table 1) also clearly points to a mafic- ultramafic protolith.

Whole-rock ICP-MS results are also indicative of a mafic-ultramafic precursor. Zr-Hf, Nb-Ta, Y/Ho pairs have similar electronegativity and ionic radii, and therefore should not fractionate from each other during charge-and-radius-controlled (CHARAC) magmatic processes (Bau, 1996). It has been generally accepted that the Zr/Hf, Nb/Ta and Y/Ho ratios in all earth materials should be constant and should be the same as the most undifferentiated Type 1 carbonaceous chondrite with Zr/Hf = 37.1, Nb/Ta = 17.6 and Y/Ho = 28.8 (McDonough and Sun, 1995; Sun and McDonough, 1989). In continental settings, highly evolved magmatic rocks such as high-silica granites display variations to the standard values above. This has been attributed to the involvement of volatile fluids such as H2O, P, F and others (Bau, 1996). The Paranesti ICP-MS results (Chapter 2, Table 2) have Zr/Hf=36.5, Nb/Ta=17.3 and Y/Ho=29 in line with a mafic- ultramafic source. Rb/Sr ratios of <0.2 also indicate an ultramafic source rock (Bau, 1996) and the Paranesti sample has Rb/Sr of 0.009, indicating this. The moderate negative Eu anomaly on the chondrite-normalised REE diagram (Figure 38) for the host pargasite/magnesiohornblende schist (sample PAR-1) is the result of Eu being preferentially incorporated into plagioclase leaving the melt depleted in Eu (Bau, 1991). It also shows a well-defined concave downwards curve with depletion in HREE relative to LREE. This provides further evidence for a mafic/ultramafic source rock.

Overall, the geochemistry of PAR-1 and PAR-5 rubies show distinctive ranges for each despite the sampling site proximity. The PAR-1 occurrence has overall larger ruby crystals which are also generally darker red in colour than those from PAR-5 (albeit the deepest-coloured rubies found within this study came from PAR-5). This suggests that the two occurrences formed under the same conditions but with variable access to Fe and Cr during their formation potentially as a result of varying degrees of metasomatic influence (discussed in later sections).

3.1.3 Oxygen isotope signature - PAR-1 vs. PAR-5 variations

The oxygen isotope values obtained using SIMS indicates that the Paranesti rubies have a narrow defined band of oxygen isotope signatures centred on +1‰ (ranging from -0.31‰ to 1.31‰). This is lower than any ratios based on the existing framework for rubies. There are further distinctive values between PAR-1 (+0.65‰ to 1.31‰) and PAR-5 (-0.31‰ to 0.85‰). Although there is a slight overlap in terms of the individual highest value for PAR-5 and the lowest value for PAR-1 the average for PAR-1 is +1‰ whilst the average for PAR-5 is +0.14‰.

There may be some differences between core-rim oxygen isotope values observed in the PAR-5 SIMS results where the core average (-0.02) is lower than the rim average (0.29). However, this is within the range of uncertainty when the errors are taken into account. As the traditional laser fluorination method consumes the entire grain, such subtle zoning would not be seen using this technique. Thus, the greater spatial resolution of the SIMS technique enables us to analyse discrete isotopic domains (i.e. rims, cores, sectors) within single corundum crystals.

Based on the spot analysis, the results suggest that δ18O values are lower towards the core of the ruby grains when compared to rim positions. For PAR-1, Fe content is observed to be on average higher at core than rim (183ppm) and minimal variance for PAR-5. Cr content had a few anomalies in PAR-5 spot analysis potentially due to hitting submicroscopic inclusions and two values at core (4ppm and 172ppm) and one value at rim (49ppm) should be adjusted to understand the core vs. rim analysis. On average, the values in the core are higher than in the rim of PAR-5 by 417ppm. In contrast, PAR-1 cores show on average 260ppm lower concentrations than their rims. The geochemical results with opposite core vs. rim values are somewhat surprising and more analyses would be required to confirm that rubies from the two locations can truly be differentiated based on the core-rim analysis.

3.2 New understanding for the genesis of Paranesti ruby

3.2.1 Origin of the Paranesti rubies

Margarite occurs as discontinuous narrow alteration rims around some, but not all, of the rubies from both Paranesti occurrences in quartz and calcite free assemblages. Margarite in quartz/calcite-free rocks (higher metamorphic grade margarite assemblage) from the Central Alps was found to decompose at about 620⁰C and 7 kbar (Bucher-Nurminen et al., 1983). The prograde metamorphism is supported by changes in the margarite-bearing assemblages where:

1 margarite  1 anorthite + 1 corundum + 1 H2O (1)

Margarite is found to be no longer stable after this P-T constraint under prograde conditions. Petrographic observations of the Paranesti rubies suggests that the margarite formed reaction rims around ruby grains, suggesting that the corundums did not form due to the break-down of margarite. Rather, the rubies formed first at a higher temperature and the margarite formed subsequently at its lower stability temperature, due to retrogression. This indicates that the rubies formed at a temperature in excess of 620⁰C. In a corundum-bearing xenolith from the Qôrqut Granite Complex, Godthåbsfjord, Greenland, Rosing et al., (1987) found plagioclase and corundum to react with water to form margarite at ~ 5 kbar and ~ 580⁰C. Margarite from the current study consists of reaction rims around ruby grains. Based on the P-T constraints discussed above, margarite developed during the retrogressive phase of metamorphism, subsequently at the expense of ruby.

Corundum-kyanite-sapphirine amphibolites (comprising an assemblage of pargasite/magnesiohornblende/tschermakite, anorthite, corundum, kyanite, sapphirine and spinel) from the French Massif Central - FMC (Berger et al., 2010) bear similarities to the 55

Paranesti occurrences. The FMC was found to have formed from subduction of a plagioclase- rich troctolite, part of the Limousin ophiolite. The difference between Paranesti and the FMC is the complete lack of anorthite (or any other plagioclase) and sapphirine and lack of any direct evidence of ophiolitic indicators in the Paranesti corundum assemblages.

Based on the results presented above and the regional geological and tectonic framework studies of other authors, the Paranesti rubies are inferred to have formed within an ultramafic precursor. Metamorphic events at either ca. 149 Ma or ca. 73 Ma, prior to the pegmatite intrusion at ca. 65 Ma, resulted in the formation of the Paranesti rubies during amphibolite facies metamorphism. The rubies are likely to have formed during the retrograde metamorphic path of high-temperature/medium-pressure metamorphism of platform carbonates and amphibolites during the Cenozoic collision that resulted in the formation of the Nestos Suture Zone (Voudouris et al., 2019).

Neither feldspars nor quartz have been observed in the ruby-bearing assemblages from Paranesti in contrast to the surrounding amphibolites. Based on the elevated Ca content (up to 10 wt%; Chapter 2. Table 1), the igneous protolith most likely had large proportions of clinopyroxene and was most likely an aluminous clinopyroxenite.

Due to the lack of mineral inclusions available for dating, it is difficult to ascertain whether both PAR-1 and PAR-5 formed during the same metamorphic event, though both contain similar mineral assemblages and overall geochemistry. The trace element diagrams display a metasomatic influence on the paler coloured PAR-5 corundums. This is supported by the location of the occurrences within the melange suture zone between two distinct tectonic units. The Paranesti ruby-bearing assemblages went through retrogression from amphibolite facies to greenschist facies based on the observed mineral assemblage of remnant sillimanite to pargasite/magnesiohornblende/kyanite /margarite/ plagioclase and lastly replacement by clinochlore/muscovite and clinozoisite within the surrounding corundum-free amphibolites (Figures 29(c), 34(f), 34(j), 35(e), 36(h), 37(g)).

There are no convincing field or petrographic relations to suggest the original protolith rock type for the ruby-bearing pargasitic schists from Paranesti. These metamorphic rocks have been affected by intense tectonic processes and no primary phases and/or textures have been preserved.

3.2.2 Possible P-T model for the formation of the Paranesti rubies

The pressure-temperature (P-T) evolution for ruby genesis is of geodynamic interest because it elucidates the processes of burial and uplift in convergent tectonic settings. The secondary formation of margarite around the ruby rims as well as the existence of sillimanite indicates the retrogressive metamorphic history at Paranesti.

Previous studies have shown that the polymetamorphic terrains of the Rhodope Mountain Complex show a complex multi-stage development. Overprinting structural relationships were originally interpreted to represent three separate metamorphic events in the central Greek Rhodope. First, an early high-pressure metamorphic record of pre-tectonic eclogite for which metamorphic conditions of 19 kbar and 700⁰C have been estimated (Liati and Mposkos 1990; Liati and Seidel 1996). Based on studies of relict mineral assemblages from the high-grade gneisses, the pressure is estimated at > 15 kbar, compatible with > 45km burial depth (Mposkos and Liati 1993). This eclogite-facies event was followed by amphibolite facies metamorphism which overprinted the high-pressure metamorphism (P = 0.8 - 1.1 GPa, T = 580–690°C, Liati and Seidel 1996).

The traditional view of kyanite in eclogites being replaced by sapphirine-bearing symplectites is attributed to granulite facies metamorphic events (Liati and Seidel 1994, 1996). This view was challenged by Moulas et al., (2013) where the metamorphic grade was found to be at the much lower amphibolite facies (0.4 GPa < P < 0.7 GPa, 580°C < T <800°C). Through the various analytical methods deployed in this study, no granulite facies minerals (or relicts of these minerals) were identified in the samples. Kyanite is found within the ruby-bearing rock assemblages (though not always) as well as in the surrounding wall rocks that are devoid of rubies. As these kyanite grains are more euhedral in shape and only weakly fractured compared to the anhedral to subhedral highly fractured ruby grains, this suggests earlier formation of the ruby. Remnant sillimanite was found in the hanging wall, providing evidence for retrogression from amphibolite facies to the current greenschist facies metamorphism in line with the findings of Moulas et al., (2013).

Simonet et al., (2008) discussed the existence of a “gem corundum domain” in his corundum classification model, with P-T conditions ranging from P > 3 kbars and T > 450⁰C and P > 7 kbars and T > 550⁰C. Figure 52(a) illustrates the hypothetical P–T diagram showing mineral equilibria related to the formation of Paranesti rubies. Figure 52(b) illustrates the theoretical P-T path for

the Paranesti rubies compared to some other metamorphic deposits from around the world. Based on the P-T constraints discussed above, the Paranesti rubies are thought to have initially formed under amphibolite facies conditions at around 4 kbar < P <7 kbar and 580°C < T < 750°C, similar to the African amphibolite-hosted ruby deposits summarised in Table 2. The Paranesti ruby-bearing assemblages followed a nearly isothermal decompression within the amphibolite facies and then a further evolution towards the greenschist facies, along the path that was proposed by Krenn et al., (2008) for the Nestos Suture Zone. This path records a transition from the kyanite to the sillimanite stability field during retrogression, as observed by former kyanite surrounded by fibrolitic sillimanite (Krenn et al., 2008; Mposkos et al., 2010; Voudouris et al., 2019 and this study).

Table 2. Comparison between the Paranesti amphibole schist hosted ruby deposits and African amphibole-hosted ruby deposits (adapted from Mercier et al., 1999a; Mercier et al., 1999b; Schwarz et al., 2008 and Pardieu et al., 2013). Deposits Geological Main Paragenesis Estimated Ruby Occurrences P-T Chemical Conditions Characteristics

Malagasy - Amphibolites forming -hornblende T: 730- Cr203: 0.2 - Madagascar diffuse zones within (tschermakitic)+ruby + 870°C 1.2% basic/ultrabasic plagioclase (An88-91)+ Cr- P: 9-11 Fe2O3: 0.3- complexes spinel + phlogopite kbar 0.7% Ejeda- in anorthosite layers -plagioclase (An94-96)+ Ti02: 0.0 - Fotadrevo and veins within ruby+garnet+gedrite+ 0.03% Madagascar basic/ultrabasic sapphirine + zoisite+ V2O3: 0.01- complexes clinopyroxene+ 0.07% hornblende (pargasitic) +Cr-zoisite + ruby + Cr- spinel+ margarite Tanzania In veins of zoisite- -hornblende+Cr- T: 750- Cr203: 0.5 - Longido amphibolites zoisite+ruby Cr- 870°C 2.0% Lossogonoi (anyolites) and spinel+garnet(Prp-Alm) P: 8.5-11 Fe2O3: 0.2- anorthosites cutting -plagioclase( An65-93)+ kbar 0.5% through ultrabasic ruby hornblende+ Cr- Ti02: 0.0 - bodies spinel 0.03% V2O3: 0.01- 0.05% Tanzania Mafic migmatitic and -amphibole + garnet T: 800 ± Cr203: 0.07 ± Winza well-foliated gneisses +plagioclase + kyanite 50°C 0.02% +apatite +spinel P: 8–10 Fe2O3: 0.36 ± inclusions kbar 0.06% Ti02: 0.02 ± 0.02% V2O3: 0.005– 0.015% 58

Deposits Geological Main Paragenesis Estimated Ruby Occurrences P-T Chemical Conditions Characteristics

Mangari - SE Kyanite-plagioclasites -ruby + plagioclase + T: 700- Cr203: 0.23- Kenya and/or sapphirinites sapphirine 750°C 1.4% in the contact zones + spinel + clinochlore + P: 8-10.5 Fe2O3: 0-0.7% between ultrabasic phlogopite + graphite kbar Ti02: 0.01- bodies and country 0.85% rocks V2O3: 0-0.05% or veins of kyanite/sapphirine - bearing desilicated pegmatites cutting through ultrabasic bodies Montepuez Green actinolite with -ruby + actinolite + ? Cr203: 0.06 - Mozambique small lenses rich in margarite + 0.15% anorthite Chalcopyrite Fe2O3: 0.003 - 0.02% Ti02: 0.002 – 0.005% Paranesti - pargasite/ pargasite/ T: 580- Cr203: 0.01- Greece magnesiohornblende magnesiohornblende 800°C 1.3% schist boudin +ruby+plagioclase P: 4-7 kbar Fe2O3: 0.2- surrounded by +spinel+kyanite 0.5% clinochlore schist -margarite+muscovite Ti02: 0-0.03% +clinochlore V2O3: 0-0%

3.3 Comparison with rubies from deposits around the world

The Paranesti rubies show relatively high Si, Ca and very low Ga, Ti and V contents. Si values of up to about 500 ppm were considered within the solubility limits for the corundum structure, with higher values signalling the presence of submicroscopic inclusions (Harlow and Bender 2013). Nevertheless, the Paranesti Si range from 780 ppm - 2450 ppm is well above the accepted solubility limits for the corundum structure. However, no Si-bearing primary inclusions were found within the corundums during this study even during the LA-ICP-MS experiment. The possible source for this Si could potentially be attributed to extremely fine-grained and overprinting margarite (reaction rim post ruby formation). This overprinting of margarite would explain both the high Si and Ca content detected and is in line with the Si/Ca ratio from the LA- ICP-MS results. All the evidence shown above suggests that the rubies formed in situ and are either pre- or just syn- to the major tectonometamorphic event that affected the region.

The Paranesti occurrences share similar characteristics with a number of amphibolite-hosted mafic/ultramafic ruby deposits from Africa, including the Ejeda-Fotadrevo deposit of southwest Madagascar which shows a prominent internal growth structure, the Longido deposit of Tanzania (veins of anyolite zoisite-amphibole) and the Mangari deposit of Kenya (Mercier et al., 1999). Table 2 provides a comparison of the geological occurrence, estimated P-T conditions, main paragenesis and ruby chemical characteristics for selected African deposits and the Paranesti occurrences. The similarities in the geochemistry of the rubies provide further support for an ultramafic origin for the Paranesti rubies as well as guidelines on the possible P-T conditions.

The closest ruby deposit type in the worldwide literature database to date is the Tanzanian Winza ruby deposit. The main rock types here are migmatitic and well-foliated gneisses, indicative of upper amphibolite to granulite facies conditions (Schwarz et al., 2008). The corundum crystals are embedded within dark-coloured amphibolite with accessory Cr-spinel, mica, kyanite, and allanite. The possible lithologies for the protolith include high-alumina layered gabbro or leucogabbro. Spinel that overgrew, or is included in, the corundum is Fe- and Mg-rich. Similar spinel inclusions have been found within the Paranesti rubies (Figure 30(f). The average Cr content of Paranesti rubies from PAR-1 and PAR-5b are comparable to other world corundum occurrences. However, PAR-5a (range 4300 ppm – 8600ppm), is very high compared to other known occurrences. Winza rubies have ~ 0.1-0.6 wt% Fe which is comparable to that found for the Paranesti rubies (~ 0.2-0.8 wt% Fe). Also, the low V content found in Paranesti (2 60

± 1 ppm) and Ti (6-190ppm) is also observed in Winza rubies V (2.5 ± 2 ppm) and Ti (2.5 ± 2 ppm). Rubies from the Montepuez Area, Mozambique are also hosted in metamorphic green amphibolitic (actinolite) lenses with margarite associations; however, the Cr content is lower at < 2,600 ppm (Pardieu et al., 2013).

The high chromium (Cr) content based on the LA-ICP-MS and EMPA studies found in the Paranesti rubies (peak 0.9 wt% LA-ICP-MS and > 1 wt% EMPA) is rare compared to other in situ rubies (~ 0.3 wt%) worldwide. Most rubies of metamorphic/metasomatic origin have significantly lower values. One other ruby deposit with high Cr (up to 14,000 ppm) is the Aappaluttoq deposit, Fiskenæsset Greenland (Keulen and Kalvig., 2013). The Fiskenæsset rubies are found within an amphibolite-anorthosite contact zone in an assemblage of corundum, sapphirine, pargasite/magnesiohornblende, spinel, hornblende, gedrite, biotite, plagioclase, phlogopite ±cordierite and anthophyllite (Appel and Ghisler 2014; Keulen and Kalvig 2013, Smith et al., 2016). The Fiskenæsset rubies are relatively rich in Fe and Si, but relatively poor in Ti and Ga (Keulen and Kalvig 2013). The host rock anorthite is distinctly different to the Paranesti pargasite/magnesiohornblende schist and the mineral assemblage shows higher metamorphic grade and the occurrence of both plagioclase and sapphirine. The rubies from both Paranesti and Aappaluttoq share a similar chemical signature of high Cr, Si and Fe but low Ti and Ga. In addition, Paranesti rubies are also low in Mg.

The Paranesti rubies’ oxygen isotope results are rather unique when compared to the other global corundum deposits. Detailed comparisons will be discussed in the following section.

3.4 Corundum oxygen isotopes as an identifier for geological origin

A framework on the interpretation of the geological origin of gem corundums using the δ18O ratio proposed by Giuliani et al., (2007a) is now widely adopted. Based on this framework, rubies and pink sapphires can be classified into 5 categories based on their δ18O value range:

1. Mafic gneiss hosted from 2.9‰ to 3.8‰;

2. Mafic-ultramafic rocks (amphibolite, serpentinite) from 3.2‰ to 6.8‰; 3. Desilicated pegmatites intruding ultramafic rocks from 4.2‰ to 7.5‰; 4. Shear zones cross-cutting ultramafic lenses and pegmatites within sillimanite gneisses 11.9‰ to 13.1‰; 5. Marble-hosted rubies 16.3‰–23‰.

This framework has been further validated by numerous subsequent corundum oxygen isotope studies (e.g. Garnier et al., 2008; Graham et al., 2008; Sutherland et al., 2008; Zaw et al., 2015; Graham et al., 2019; Wang et al., 2019).

The reconnaissance 2009 laser ﬂuorination results (see also Voudouris et al., 2019) on the sapphires and rubies from different geological environments in Greece very closely fits the oxygen isotope value ranges from the model of Giuliani et al., (2005), where over 200 corundum samples were analysed under the same method. Critically, the oxygen isotope values for both the marble-hosted rubies (20‰ to 22‰) and sapphires from desilicified pegmatites (4.8‰ to 5‰) from Greece fit well within the ranges for their respective geological environments using the framework of Giuliani et al. (2005). Therefore, the δ18O results obtained using the laser- fluorination method in 2009 are further validated as accurate measurements.

3.4.1 Global low to ultra-low oxygen isotope corundum comparison

δ18O (SMOW) values for gem corundums below 1‰ are very rare and not shown on the original systematic framework of Giuliani et al., (2005). Other than the Paranesti ruby results shown and discussed in Chapter 3, the only negative values for corundums are from Karelia in northwestern Russia and sapphire from a secondary deposit in Madagascar. Table 3 lists the global low to ultra- low oxygen isotope analyses for corundums.

The Madagascar sapphire deposit of Ilakaka with δ18O of -0.3‰ to 16.5‰ is a consolidated placer formed in a sandstone environment. The geological origin of the different ranges of isotopic values found for the sapphires corresponds to at least five different geological environments (Giuliani et al., 2007b). The low- δ18O delta values for some sapphires correspond up to an unknown geological sapphire type.

The PAR-1 result of +1.0‰ is lower than most known primary corundum oxygen isotope values other than the unique ultra-low values of corundums from Karelia (Bindeman and Serebryakov, 2011; Krylov, 2008) and one instance of ruby from the Soamiakatra area of Madagascar (Giuliani et al., 2007b). The Karelia corundum formed under unique circumstances (see discussion below) and can be easily distinguished from the Paranesti rubies. The Madagascar rubies show much higher average δ18O values and the minimum value obtained corresponds to the maximum value from Paranesti. Therefore, oxygen isotope analysis is a valuable tool that can be used to fingerprint the Paranesti rubies from other worldwide occurrences (Figure 53).

Table 3. Global comparison of corundums with low to ultra-low oxygen isotope values.

δ18O‰ δ18O‰ Primary vs. Corundum Type Country District Host rock (Min) (Max) Secondary Greece1 Paranesti-1* 0.65 1.31 pargasite/ magnesiohornblende Primary Ruby Greece1 Paranesti-5* -0.31 0.85 schist Primary Ruby Madagascar2 Soamiakatra* 1.25 4.70 Pyroxenitic enclaves in basalt Primary Ruby Madagascar2 Ilakaka* -0.30 16.5 Placer in sandstone Secondary Sapphire Madagascar2 Andilamena* 0.50 3.9 Placer in basalt Secondary Ruby Russia3,4,7 Khitostrov^ -26.3 -17.7 plagiogneiss Primary Corundum Russia4,7 Khitostrov * -26 - Crn-St-Gt-Bi-Prg-Pl rocks with Primary Corundum coarse grained Crn Russia7 Khitostrov * -18.6 - Ky-Crn-Pl, leucocratic Primary Corundum Russia3 Varastskoye^ -19.2 -11.3 plagiogneiss Primary Corundum Russia4 Varastskoye # -17.3 - Crn-Cam rock, coarse grained Primary Corundum Varastskoye# -19.2 - Crn and Crn-St-Pl substituting Ky Primary Corundum Russia4 Dyadina# 0.49 - Inclusion of Cam-Crn in giant Gt Inclusion Corundum Russia4 Dyadina# 0.10 - Crn-Cam rock, coarse grained Primary Corundum Russia5 Dyadina* 0.4 0.8 Corundum amphibolite Primary Corundum

Russia4 Kulezhma# 0.31 - Cam-Crn rock Primary Corundum Russia4 Pulonga# 0.67 - Crn-Gt-Ged rock Primary Corundum Russia4 Perusel’ka* 0.26 3.45 Crn-Cam rock, coarse grained Primary Corundum Russia5 Perusel’ka# 0.6 - Corundum-kyanite amphibolite Primary Corundum Russia5 Perusel’ka# 1.5 - Corundum amphibolite Primary Corundum Russia5 Notozero* -1.7 -1.5 Ged-Gt rocks with Crn and St Primary Corundum Russia4 Mironova Guba^ (2.34) - Cam-Crn rock Primary Corundum Thailand6 Bo Rai* 1.30 4.20 Placer in basalt Secondary Ruby

* Individual grain analysis ˆ Whole-rock analysis # Only one analysis result, no range. Bi—biotite, Cam—Ca-amphibole, Crn—corundum, Ged—gedrite amphibole, Gt—garnet, Ky—kyanite, Pl—plagioclase, Prg—pargasitic amphibole, St—staurolite. 1 - Wang et al., (2017); 2 - Giuliani et al., (2007b); 3 Vysotskiy et al., (2015); 4 - Bindeman and Serebryakov (2011); 5 - Vysotskiy et al., (2014); 6 - Yui et al., (2006); 7 - Bindeman et al., (2010). 63

3.4.2 Possible causes for low oxygen isotope corundum formation

There are several current theories on how corundums can form with low δ18O isotope ratios. These range from hydrothermal alteration of deeply penetrating surface meteoric waters to isotope separation by thermal diffusion during endogenous fluid flow (Bindeman et al., 2013; Akimova and Lokhov, 2015).

3.4.2 (a) Kinetic isotope fractionation

Kinetic isotope fractionation occurs when rapid thermal decomposition of hydrous phases results in isotope disproportionation into a high-δ18O residue and a low-δ18O fluid (Clayton and Mayeda, 2009; Mendybayev et al., 2010). However, the high- δ18O residue material should also be found within proximity of the studied samples for this theory to apply. The water-rock interaction is kinetically restricted to supracrustal rocks, and isotope fractionation factors are large at low temperatures, favoring higher-δ18O solids (Bindeman et al., 2014). In contrast, isotopic exchange is more rapid within a hydrothermal system. As the Paranesti rubies formed under amphibolite-facies conditions above 600 ᴼC (Wang et al., 2017), significant kinetic isotope fractionation is highly unlikely and therefore effectively rules out this hypothesis.

3.4.2 (b) Thermal diffusion

For thermal diffusion, the oxygen in a temperature gradient is redistributed with low δ18O at the hotter end and high δ18O at the colder end of the melt or hydrous solution (Bindeman et al., 2013). Akimova and Lokhov (2015) proposed a model of cascading thermo-diffusion within shear zones to explain the Karelian ultra-low δ18O corundums. This scenario would require several individual thermal cells to align in the correct position simultaneously with a similar convection rate and timing. Given that only two locations have shown ruby-bearing pargasite/magnesiohornblende schist with other similar pargasite/magnesiohornblende boudins nearby being devoid of ruby, this model appears to be unlikely.

3.4.2 (c) Other ultra-low δ18O protoliths

Ultra-low δ18O protoliths could potentially provide the low δ18O during corundum syn- metamorphic formation (Bindeman et al., 2014). However, a source for the ultra-low δ18O protolith would be needed under this scenario such as a low δ18O mantle reservoir or previously surface-exposed and then rapidly buried metamorphic rocks. Neither were observed at

Paranesti. However, as oxygen isotope analyses were not performed on the whole-rock and associated mineral phases for the Paranesti occurrences, this hypothesis cannot be ruled out.

3.4.2 (d) Hydrothermal alteration model

This theory involves the conservation of the initial isotopic ratios of the protolith in the corundum-bearing rocks and then isotopic exchange between these rocks and meteoric waters before metamorphism (Krylov and Glebovitsky, 2007). According to Sharp (2017), volatilisation reactions in metamorphic rocks cannot change the δ18O values of a lithology by more than several per mil under any conditions. Larger shifts in δ18O values in geological environments require the infiltration of a fluid or melt. Wang et al., (2017) demonstrated that the rubies from Paranesti were most likely syn-metamorphic and largely free of inclusions. Therefore, it is unlikely that the low isotopic ratios observed within the Paranesti rubies were due to preservation of the initial ratios within the protolith.

For granulite facies metamorphism, Wilson and Baksi (1983) proposed three processes that could produce a low oxygen isotope value. These are (1) pre-granulite reaction between heated seawater and hot basic intrusives, or an initial protolith such as a palaeosol for the sapphirine– spinel–(cordierite) assemblages; (2) syn-granulite depletion in 18O related to dehydration during granulite metamorphism and removal of the resultant products of partial melting with a depletion in 18O by up to 2 or 3‰ for the restite; and (3) post-granulite facies metamorphism with recrystallization under the effect of biotite and/or amphibole-metasomatism with depletion in δ18O up to 4‰. Based on the prior discussions within this current study, the Paranesti rubies were found to have formed under amphibolite facies conditions and there is no evidence that they ever reached granulite facies within the specified zone. Therefore, this theory is inconclusive for explaining the low δ18O values for the Paranesti rubies. However, there are other locations within the Rhodope Mountain Complex (RMC) where regional metamorphism reached granulite facies conditions, though these are some distance away from Paranesti and no rubies are known from these locations.

The glacial meltwater influence during formation of corundums was proposed to explain the ultra-low δ18O isotopic ratios observed for corundums from Karelia in northwestern Russia (Bindeman et al., 2011, 2014; Bindeman and Serebryakov, 2011). However, there is no evidence suggesting the existence of glaciers in the Mediterranean region based on the reconstruction of the tectonic evolution of the East Mediterranean region since the late Cretaceous (Menant et

al., 2016). Could the RMC be a higher mountain with glaciers that have melted during ruby genesis? There is no evidence in the literature to suggest such and this would only be a remote possibility. Therefore, it is unlikely that a glacial meltwater source played a role during ruby formation at Paranesti. However, it is likely that meteoric water (but not glacial melt) interaction caused by downward flow of surface waters along deep crustal fractures/structures during the formation of the corundum would contribute in producing low δ18O values for the Paranesti rubies.

3.4.3 Oxygen Isotope standard for rubies

The in situ average oxygen isotope SIMS analysis of +1‰ on PAR-1 exactly matches the 2009 laser fluorination results. This +1‰ value is significant in the context where before this current study, there have been no corundum standards for in situ SIMS oxygen isotope analysis. The reproducibility of the results for PAR-1 has been verified by two other studies to date. These include Sutherland et al., (2017) for the Australian New England magmatic rubies and sapphires and Graham et al., (2019) for Argentinian sapphires. Both studies returned a 2SD of 0.32‰ and 0.42‰ respectively.

There are a number of issues to consider for geological standards. The variation noted on the 2018 laser-fluorination analysis emphasised the requirement of multilateral calibration between different laboratories around the world and agree on the value with minimal deviations. There should be an abundant supply of the Paranesti rubies that are inclusion free and readily available to other organisations for it to be considered as a possible standard. A ruby standard should ideally come from a high-grade terrain with the oxygen completely equilibrated. Paranesti rubies are formed under amphibolite facies metamorphism which is high grade only below the granulite facies. However, no granulite facies rubies to date have been proposed as a suitable oxygen isotope standard. The three SIMS studies that have used PAR-1 were all conducted within the Centre for Microscopy, Characterisation and Analysis (CMCA), University of Western Australia (UWA) and not by other laboratories. Consistency in results obtained by multiple international laboratories would be a prerequisite for a standard (Hut, 1987; Coplen, 1988; Valley et al., 1995).

Thus, PAR-1 rubies are a suitable first pass standard for in situ oxygen isotope analysis, but significantly more calibrations and reproductive analyses from a number of laboratories would

be required to recognise it as an internationally accepted primary SIMS oxygen isotope corundum standard.

3.5 Conclusions and suggested further work

This is the first systematic research on the origin and genesis of the Paranesti ruby occurrences. The range of analytical results provides a framework for the P-T conditions of ruby formation during amphibolite facies conditions estimated at 4 Kbar < P <7 Kbar and 580°C < T < 750°C and with subsequent retrogression.

Geochemical signatures of the Paranesti rubies display a distinctive high Cr, low V, Ti, Mg and Ga trace element fingerprint as a geographical locality marker. This distinctive geolocation signature for corundum profiling could provide benefits for future commercial exploration and marketing of the rubies from the area.

The in situ SIMS oxygen isotope analyses on the Paranesti rubies is the first time that a primary (and exclusively) ruby occurrence was found to have ~ +1‰ for its δ18O-isotope composition. Based on the low δ18O value and the local geology, it is most likely that the Paranesti rubies formed under metamorphic/metasomatic conditions involving deeply penetrating meteoric waters along major crustal structures related to the Nestos Shear Zone. PAR-5 is potentially closer to the source of the hydrothermal influence during ruby formation compared to PAR-1, and thus has a lower δ18O.

Importantly, this study shows that in situ gem corundum oxygen isotope analysis using the SIMS method together with distinctive geochemical signatures may be used to determine the likely geographic origin for corundums lacking any provenance details. The SIMS method being only minimally destructive, with the analysis spot (15 x 15 µm) amenable to repolishing, a wider adoption of this technique has important applications/implications for the international gem and jewellery industry.

3.5.1 Broader tectonic implications

Dissanayake and Chandrajith (1999) discussed the East African metamorphic belt, termed the Metamorphic Mozambique Belt in the context of the existence of a wide gem corundum province covering East Africa, Madagascar, south India, Sri Lanka and east Antarctica. The formation of this belt was due to continental collision between the eastern and western

Gondwana blocks with associated high-grade metamorphism and fluid circulation (metasomatism) at amphibolite to granulite facies conditions. Rubies and sapphires have been reported to occur in various rock types in the metamorphic units of Greece (Voudouris et al., 2010; Voudouris et al., 2019). This study on Paranesti rubies contributes to a potentially wider gem corundum province covering the Greek Hellenides, and eastwards towards the border with Turkey.

Rubies of metamorphic origin are associated with continental collision zones and have been proposed as plate tectonic indicators by Stern et al., (2013). The Paranesti occurrence is located within the NSZ (Alpine suture zone) in which remnants of a partially subducted magmatic arc, foundered on the attenuated margin of Europe is preserved. Subduction and collision of the arc with the incoming Lower Terrane continent on one side of the NSZ produced decoupling within the arc and subduction of its deeper parts along with the frontal parts of the Lower Terrane, one of the intra-Tethys continental blocks derived from Gondwana (Burg, 2012). Brun et al., (2016) now consider the Nestos Suture Zone to correspond to the Vardar Suture Zone (e.g. subduction of the Vardar ocean beneath Eurasia). The Paranesti corundum occurrence thus adds support to the suggestion that metamorphic rubies hosted in mafic-ultramafic rocks can be used as plate tectonic suture zone indicators. This broad tectonic framework can thus be used to assist in future exploration for this valuable gemstone, both in northern Greece and other collisional zones of the world.

3.5.2 Future areas of research

A detailed cathodoluminescence analysis to determine the homogeneity or heterogeneity of the ruby grains is suggested for future studies. Further studies in quantitative thermobarometry, oxygen isotope signatures and age-dating would complement the research data on these occurrences. The non-destructive in situ SIMS methodology could be applied to gem mineral varieties other than corundum and emeralds in the future if well-characterised standards are available. The aim of such future work would be to determine if in situ SIMS oxygen isotope analysis can be used to both better understand gem formation and to see if it can be used to clearly separate the same gem mineral from different geographic locations.

Figures

Chapter 1

Figure 1. Gem Corundum Classification Model (adapted from Simonet et al., 2008).

Figure 2. Primary Corundum Classification Model – Lithology Based (adapted from Giuliani et al., 2014).

Figure 3. Reconstruction of Gondwana with the location of gem deposits along the Pan-African – Indian subcontinent Gondwana collision (extract from Giuliani et al., 2007).

Figure 4. Marble-hosted ruby deposits from central and southeast Asia from Garnier et al., (2006).

Figure 5. Gem corundum occurrences and associated basaltic fields along the West Pacific continental margin intraplate basaltic fields (from Graham et al., 2008 ).

Figure 6. P-T conditions for the formation of gem corundum in metamorphic deposits (From Giuliani et al., 2014).

Figure 7. Significant ruby deposits from around the world (Shor and Weldon, 2009) .

Figure 8. Corundum deposit type based on oxygen isotope ratios (generalised from Giuliani et al., 2005 and Vysotskiy et al., 2015).

Chapter 2

Figure 9. Geological map of the Aegean and Anatolia, with Paranesti shown (red star) (adapted from Bozkurt 2001).

Figure 10. Geological map of the Rhodope Mountain Complex with Paranesti shown (red star). Adapted from Collings et al., (2016). ER: Eastern Rhodope Mountains; CR: Central Rhodope Mountains; WR: Western Rhodope Mountains; NSZ: Nestos Shear Zone; CSZ: Chepelare Shear Zone.

Figure 11. Topographic map of the Rhodope Mountain Complex showing the Nestos Shear Zone (NSZ) and the Chepelare Shear Zone (CSZ). The two arrows indicate the vergence of Alpine thrusting in the Balkanides (to the north) and the Hellenides (to the southwest) (Gautier et al., 2017).

Figure 12. Geological map of the Rhodope Mountain Complex showing the main tectonic zones, with Paranesti located within the Nestos Shear Zone (red star) (adapted from Voudouris et al., 2019).

Figure 13. Location map of the deposits showing their close proximity to each other and distance from the closest town of Paranesti.

Figure 14. PAR-1 main ruby working showing intrusive contact between pargasite schist (green) and granite pegmatite (grey-white, outlined in red).

Figure 15. Original ruby discovery in road-cutting along the main road – PAR-5.

Figure 16. Kyanite-bearing pargasite-chlorite schist at the road-side deposit PAR-5.

Figure 17. Non-ruby-bearing pargasite schist boudins along the road from PAR-5 towards the Perivlepto village. The white layers above the boudin are thin marble bands.

Figure 18. Stratigraphy schematics of the sample sites.

Figure 19. Deep red-coloured ruby with both tabular and barrel shaped crystals within pargasite schist.

Figure 20. Deep red-coloured ruby with barrel-shaped crystal within pargasite schist.

Figure 21. Ruby associated with kyanite.

a. PAR-1 (6 grains). Figure 22. Numbered ruby grains in grain mounts analyzed using LA-ICP-MS.

b. PAR-5 (5 grains). Figure 22. Numbered ruby grains in grain mounts analysed using LA-ICP-MS.

Figure 23. Images of ruby samples from Paranesti. Dark red ruby crystals (0.5- 1.0 cm) from PAR-1 in pargasite schist host rock.

Figure 24. Cluster of pale pink ruby crystals (0.5-1.5 cm) from PAR-5 in pargasite schist host rock.

Figure 25. Clean cut surface of PAR-1 ruby sample free from inclusions used for the 2009 laser fluorination oxygen isotope analysis.

Figure 26. Cluster of rubies ~ 1cm in diameter, showing the range of colours found.

Figure 27. Mounted PAR-1 ruby grains used for in situ SIMS analysis.

a. SIMS in situ analysis spot location with values individually marked on the ruby grain – PAR-5 Grain A.

b. SIMS in situ analysis spot location with values marked on the ruby grain – PAR-5 Grain C.

Figure 28. Images of SIMS analysed samples showing oxygen isotope values.

29a. Overall texture of PAR1A showing highly 29b. Overall texture of PAR1A showing highly fractured ruby grains under plane polarised light fractured ruby grains under cross polarised light (PPL). (UXP).

29c. Margarite (Mrg) fringes around ruby (PPL). 29d. Margarite (Mrg) and pargasite (Amp) filling fractures within the ruby grain (PPL).

29e. Ruby altering to margarite and opaques 29f. Decussate pargasite grains (UXP). (PPL).

Figure 29. Petrographic images of PAR1A samples.

29g. Muscovite filling space between amphiboles 29h. Overprinting Mg-rich chlorite (Chl)(UXP). (UXP).

29i. Subhedral prismatic kyanite (Ky)(UXP).

Figure 29. Petrographic images of PAR1A samples.

30a. Overall texture of PAR1B (PPL). 30b. Ruby parallel to main foliation (PPL).

30c. Triple junction grain boundaries in 30d. Olive-green spinel inclusions in ruby (PPL). amphibole (UXP).

30e. Euhedral ruby gain with colour zoning on 30f. Olive-green spinel inclusions in the centre of the grain the ruby (PPL). boundary (PPL).

Figure 30. Petrographic images of PAR1B samples.

31a. Overall texture of PAR5C showing 31b. Kyanite around ruby grain (PPL). anhedral fractured rubies (PPL).

31c. Margarite rims around ruby grain (UXP). 31d. Brown spinel inclusions in rubies (PPL).

31e. Chlorite (Chl) overprinting kyanite 31f. Euhedral muscovite space-fill grain (UXP). (Ky)(UXP).

Figure 31. Petrographic images of PAR5C samples.

31g. Interstitial plagioclase (UXP). 31h. Abundant angular monazite (PPL).

31i. Zoned monazite in Mg-chlorite (PPL). 31j. Monazite along foliation in Mg-chlorite (PPL).

Figure 31. Petrographic images of PAR5C samples.

32a. Anhedral-shaped rubies (PPL). 32b. Margarite (Mrg) rimming rubies (ruby) (UXP).

32c. Rare overprinting carbonate (UXP). 32d. Pyrite in fine-grained amphibole (RL).

Figure 32. Petrographic images of PAR5D samples.

33a. Remnant rubies breaking down into 33b. Monazite inclusion in quartz (Qtz)(PPL). margarite (Mrg) (UXP).

33c. Intergranular quartz and plagioclase 33d. Plagioclase intergrown with amphiboles (UXP). feldspar (UXP).

33e. Zoisite (Zois) and amphiboles (Amp) (UXP). 33f. Main foliation defined by parallel aligned amphiboles (UXP).

Figure 33. Petrographic images of PAR2A samples.

34a. Distinctive planar layering (UXP). 34b. Epidote-zoisite dominant layer (UXP).

34c. Abundant subhedral monazite in pargasite 34d. Shear bands in amphiboles (UXP). (PPL).

34e. Prominent granoblastic texture (UXP). 34f. Epidote-zoisite (Zois) replacing plagioclase (Plag) (UXP). Figure 34. Petrographic images of PAR3A samples.

34g. Amphiboles (Amp) overprinting plagioclase 34h. Fe-rich chlorite replacing amphiboles (Plag)(UXP). (UXP).

34i. Micro-veinlets of epidote (UXP). 34j. Retrogressive muscovite around amphiboles (UXP).

Figure 34. Petrographic images of PAR3A samples.

35a. Overall texture of coarse-grained zone (PPL). 35b. Overall texture of fine-grained zone (PPL).

35c. Kyanite (Ky) intergrown with pargasite 35d. Pargasite (Amp) and plagioclase inclusions in (Amp)(UXP). Kyanite (Ky)(UXP).

35e. Sillimanite fringes around kyanite (Ky)(PPL). 35f. Undulose extinction in kyanite (UXP).

Figure 35. Petrographic images of PAR4A samples.

36a. Compositionally zoned plagioclase (UXP). 36b. Epidote and clinozoisite (Ep) replacing plagioclase (Plag)(UXP).

36c. Euhedral zoisite-clinozoisite (UXP). 36d. Zoisite-clinozoisite-chlorite assemblage (UXP).

36e. Interstitial Mg-chlorite (UXP). 36f. Radial chlorite aggregates (UXP).

Figure 36. Petrographic image – polished thin section PAR4B.

36g. Vermicular chlorite (PPL). 36h. Chloritisation of primary amphibole (UXP).

36i. Granoblastic texture of primary amphibole 36j. Euhedral titanite (UXP). (UXP).

36k. Muscovite porphyroblasts in clinochlore (UXP).

Figure 36. Petrographic image – polished thin section PAR4B.

100

37a Well-developed undulose extinction in kyanite 37b. Decussate texture in amphiboles (UXP). (UXP).

37c. Granoblastic plagioclase and quartz (UXP). 37d. Plagioclase-amphibole intergrowth (UXP).

37e. Partially resorbed anhedral-shaped kyanite 37f. Muscovite overprinting kyanite (UXP). (Ky)(UXP).

Figure 37. Petrographic images of PAR4C samples.

101

37g. Sillimanite (Si) overprinting kyanite (Ky)(UXP).

Figure 37. Petrographic images of PAR4C samples.

100.000 REE diagram - Eu Anomaly

10.000

1.000 La Ce Pr Nd Pm Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Figure 38. Chondrite-normalised rare earth element diagram showing moderate negative Eu anomaly and concave downwards curve for the host pargasite schist (sample PAR-1) (chondrite values used are those of Sun and McDonough, 1989).

102

Figure 39. SEM Images – PAR1A phyllosilicate rims around rubies.

(a) petrographic image of euhedral tabular ruby (b) SEM BSE image of the same grain showing distribution of grain showing unusual colour zonation and minute syngenetic spinel inclusions spinel inclusions (PPL)

(c) close-up view of central aggregate of spinel inclusions from (b). Figure 40. PAR1B spinel inclusions in rubies (core – 1B20, rim – 1B21, outside – 1B22).

103

Figure 41. Triangular spinel classification diagram (Cr3+-Fe3+-Al3+) (adapted from Gawlick et al, 2015).

Figure 42. Binary classification diagram considering the Mg2+-Fe2+ exchange (adapted from Gawlick et al, 2015). 104

Figure 43. Diagram showing the range in composition of host amphiboles from the ruby-bearing amphibole schists. Note that all plot close to the pargasite end-member. Assumption 20% Fe3+ and 80% Fe2+ (adapted from Colville et al, 1966).

Figure 44. Classification diagram for chlorite from the Paranesti deposit (adapted from Hey, 1954). 105

Figure 45. Oxygen isotopic compositions from the 2009 Greek corundum study (adapted from Giuliani et al., 2007; Wang et al., 2019; Voudouris et al., 2019).

106

Chapter 3

Figure 46. Mg(x100)-Fe-Ti(x10) trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for magmatic and metamorphic corundums. Adapted from Sutherland et al. (2009).

107

Figure 47. Cr(x10)-Fe-Ga(x100) trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for magmatic and metamorphic corundums. Adapted from Peucat et al. (2007).

Figure 48. Ga/Mg vs Fe/Mg trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for metamorphic, transitional and magmatic corundums. Adapted from Sutherland et al. (2015).

108

Figure 49. Ga/Mg vs Fe/Mg trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for metamorphic, transitional and magmatic corundums with R2. Adapted from Sutherland et al. (2015).

Figure 50. Cr/Ga vs Fe/Ti trace element discrimination diagram showing the Paranesti ruby analyses relative to the fields for metamorphic and magmatic corundum. Adapted from Sutherland et al. (2009). 109

110

Figure 52. (a) P–T diagram (marked in blue) showing mineral equilibria related to the formation of Paranesti rubies and (b) Estimated P-T path (marked in blue) for Paranesti ruby formation compared to other known metamorphic and metasomatic ruby deposits (adapted from Voudouris et al., 2019). P–T conditions for the formation of corundum in metamorphic deposits (modified from Simonet et al., 2008); P–T fields of North Carolina (Tenthorey et al., 1996); Mangare (Mercier et al., 1999); Morogoro (Altherr et al., 1982); southern Kenya (Key and Ochieng 1991; Simonet 2000); Hunza (Okrusch et al., 1976); Sri Lanka (De Maesschalk and Oen 1989); Greenland (Garde and Marker 1988) and Kashmir (Peretti et al., 1990) with three P–T boxes corresponding to the evolution of the fluids in the sapphire crystals from the center (c), to intermediate (i) and outer (o) zones, Urals (Kissin 1994), and Mong Hsu (Peretti et al., 1996).

Abbreviations: An = anorthite; And = andalusite; Cc = calcite; Cld = chloritoid; Clin = clinochlore; Co = corundum; Do = dolomite; Dsp = diaspore; Ky = kyanite; Ma = margarite; Mag = magnetite; Qz = quartz; Sill = sillimanite; Spr = sapphirine; Sp = spinel; St = staurolite;

111

Figure 53. Oxygen isotopic comparison of rubies from Paranesti occurrences compared with low δ18O corundums from Karelia and Soamiakatra in Madagascar (Wang et al., 2019; Vysotskiy et al., 2015; Giuliani et al., 2007; Bindeman and Serebryakov, 2011).

112

References

Abduriyim, A., and Kitawaki, H. 2006. Applications of laser ablation-inductively coupled plasma- mass spectrometry (LA-ICP-MS) to gemology. Gems and Gemology 42, 98–118. Abduriyim, A.A., Sutherland, F.L. and Coldham, T., 2012. Past, present and future of Australian gem corundum. Australian Gemmologist, 24, 234-242.

Akimova, E.Y. and Lokhov, K.I., 2015. Ultralight oxygen in corundum-bearing rocks of North Karelia, Russia, as a result of isotope separation by thermal diffusion (Soret effect) in endogenous fluid flow. Journal of Materials Science and Chemical Engineering, 3(12), 42.

Altherr, R., Okrusch, M., Bank, H., 1982. Corundum and kyanite bearing anatexites from the Precambrian of Tanzania. Lithos, 15, 191–197.

Andriamamonjy, A., Giuliani, G., Ohnenstetter, D., Rakotondrazafy, A.F.M., Ralantoarison, T., Ranatsenho, M., Rakotosamizanany, S. and Fallick, A.E. 2013. Les corindons d’âge néoprotérozoïque de Madagascar, 2, 109-117.

Appel, P.W.U. and Ghisler, M., 2014. Ruby- and sapphirine bearing mineral occurrences in the Fiskenaesset, Nuuk and Maniitsoq Regions, West Greenland. Danmarks og Grønlands Geologiske Undersøgelse, 2014/72, 71.

Atkinson, D. & Kothavala, R.Z. 1983. Kashmir sapphire. Gems and Gemology 19, 64–76.

Barr, S.R., Temperley, S. and Tarney, J., 1999. Lateral growth of the continental crust through deep level subduction–accretion: a re-evaluation of central Greek Rhodope. Lithos, 46(1), 69- 94. Bau, M., 1991. Rare-earth element mobility during hydrothermal and metamorphic fluid-rock interaction and the significance of the oxidation state of europium. Chemical Geology, 93(3- 4), 219-230.

Bau, M., 1996. Controls on the fractionation of isovalent trace elements in magmatic and aqueous systems: evidence from Y/Ho, Zr/Hf, and lanthanide tetrad effect. Contributions to Mineralogy and Petrology, 123(3), 323-333.

Bauer, C., Rubatto, D., Krenn, K., Proyer, A. and Hoinkes, G., 2007. A zircon study from the Rhodope metamorphic complex, N-Greece: time record of a multistage evolution. Lithos, 99(3-4), 207-236. Baziotis I., Mposkos, E. and Perdikatsis, V., 2007. Geochemistry of amphibolitized eclogites and cross-cutting tonalitic–trondhjemitic dykes in the Metamorphic Kimi Complex in East

113

Rhodope (N.E. Greece): implications for partial melting at the base of a thickened crust. International Journal of Earth Sciences, 97(3), 459–477. Berger, J., Femenias, O., Ohnenstetter, D., Plissart, G. and Mercier, J.C., 2010. Origin and tectonic significance of corundum–kyanite–sapphirine amphibolites from the Variscan French Massif Central. Journal of Metamorphic Geology, 28(3), 341-360.

Bindeman, I., 2008. Oxygen isotopes in mantle and crustal magmas as revealed by single crystal analysis. Reviews in Mineralogy and Geochemistry, 69(1), 445-478.

Bindeman, I.N., Schmitt, A.K., Evans, D.A.D., 2010. Limits of hydrosphere–lithosphere interaction: origin of the lowest-known δ18O silicate rock on Earth in the Paleoproterozoic Karelian rift. Geology 38 (7), 631–634.

Bindeman, I.N. and Serebryakov, N.S., 2011. Geology, petrology and O and H isotope geochemistry of remarkably 18O depleted Paleoproterozoic rocks of the Belomorian Belt, Karelia, Russia, attributed to global glaciation 2.4 Ga. Earth and Planetary Science Letters, 306(3-4), 163-174.

Bindeman, I.N., Lundstrom, C.C., Bopp, C. and Huang, F., 2013. Stable isotope fractionation by thermal diffusion through partially molten wet and dry silicate rocks. Earth and Planetary Science Letters, 365, 51-62.

Bindeman, I.N., Serebryakov, N.S., Schmitt, A.K., Vazquez, J.A., Guan, Y., Azimov, P.Ya., Astafi ev, B.Yu., Palandri, J., Dobrzhinetskaya, L., 2014. Field and microanalytical isotopic investigation of ultradepleted in 18O Paleoproterozoic “Slushball Earth” rocks from Karelia, Russia. Geosphere 10(2), 368-379.

Bonev, N., Burg, J.P. and Ivanov, Z., 2006. Mesozoic–Tertiary structural evolution of an extensional gneiss dome—the Kesebir–Kardamos dome, eastern Rhodope (Bulgaria–Greece). International Journal of Earth Sciences, 95(2), 378-360. Bonneau, M., 1984. Correlation of the Hellenide nappes in the south-east Aegean and their tectonic reconstruction. Geological Society, London, Special Publication, 17(1), 517-537. Bosse, V., Boulvais, P., Gautier, P., Tiepolo, M., Ruffet, G., Devidal, J.L., Cherneva, Z., Gerdjikov, I. and Paquette, J.L., 2009. Fluid-induced disturbance of the monazite Th-Pb chronometer: in situ dating and element mapping in pegmatites from the Rhodope (Greece, Bulgaria). Chemical Geology, 361, 366–362. Bozkurt, E., 2001. Neotectonics of Turkey–a synthesis. Geodinamica Acta, 14(1-3), 3-36. Brun, J.P. and Faccenna, C., 2008. Exhumation of high-pressure rocks driven by slab rollback. Earth and Planetary Science Letters, 374(1-2), 1-7. 114

Brun, J.P. and Sokoutis, D., 2007. Kinematics of the southern Rhodope core complex (North Greece). International Journal of Earth Sciences, 96(6), 1079-1099. Brun, J.P., Faccenna, C., Gueydan, F., Sokoutis, D., Philippon, M., Kydonakis, K. and Gorini, C., 2016. The two-stage Aegean extension, from localized to distributed, a result of slab rollback acceleration. Canadian Journal of Earth Sciences, 53(11), 1142-1157.

Bucher-Nurminen, K. Frank, E. and Frey, M., 1983. A model for the progressive regional metamorphism of margarite-bearing rocks in the Central Alps. American Journal of Science, 283(A), 370-395.

Burchfiel, B.C., Nakov, R., Dumurdzanov, N., Papanikolaou, D., Tzankov, T., Serafimovski, T., King, R.W., Kotzev, V., Todosov, A. and Nurce, B., 2008. Evolution and dynamics of the Cenozoic tectonics of the South Balkan extensional system. Geosphere, 4(6), 919-938. Burg, J.P., Ricou, L.E., Ivano, Z., Godfriaux, I., Dimov, D. and Klain, L., 1996. Syn‐metamorphic

nappe complex in the Rhodope Massif. Structure and kinematics. Terra Nova, 8(1), 6-15. Burg, J.P., 2012. Rhodope: From Mesozoic convergence to Cenozoic extension: Review of petro- structural data in the geochronological frame. Journal of the Virtual Explorer, 42(1), 1-44. Caucia, F. and Boiocchi, M., 2005. Corindone. I cromofori nei cristalli policromi di Amboarohy, Ihosy, Madagascar. Risultati degli studi microanalitici. Rivista Mineralogica Italiana, 2, 126- 129.

Clark, C., 1998. Tropical Gemstones. Periplus Editions, Hong Kong, 12.

Clayton, R.N. and Mayeda, T.K., 2009. Kinetic isotope effects in oxygen in the laboratory dehydration of magnesian minerals. Journal of Physical Chemistry A, 113(10), 2212-2217.

Collings, D., Savov, I., Maneiro, K., Baxter, E., Harvey, J. and Dimitrov, I., 2016. Late Cretaceous UHP metamorphism recorded in kyanite–garnet schists from the Central Rhodope Mountains, Bulgaria. Lithos, 246, 165-181. Colville, P.A., Ernst, W.G. and Gilbert, M.C., 1966. Relationships between cell parameters and chemical compositions of monoclinic amphiboles. American Mineralogist, 51(11-12), 1737- 1754. Cooray, P., Kumarapeli, P., 1960. Corundum in biotite sillimanite gneiss from near Polgahawela, Ceylon. Geological Magazine 97, 480–487.

Coplen, T.B., 1988. Normalization of oxygen and hydrogen isotope data. Chemical Geology: Isotope Geoscience Section, 72(4), 293-297.

115

Criss, R.E. and Taylor Jr, H.P., 1986. Meteoric-hydrothermal systems. Reviews in Mineralogy, 16, 373-424.

Deer W.A., Howie R.A. and Zussman J. 1992. Corundum. In Rock-Forming Minerals Non-Silicates: Oxides, Hydroxides and Sulphides, The Geological Society, London, 5A, 47-91.

De Maesschalk, A., Oen, I., 1989. Fluid and mineral inclusions in corundum from gem gravels in Sri Lanka. Mineralogical Magazine 53, 539–545.

Didier, A., Bosse, V., Cherneva, Z., Gautier, P., Georgieva, M., Paquette, J.L. and Gerdjikov, I., 2014. Syn-deformation fluid-assisted growth of monazite during renewed high-grade metamorphism in metapelites of the Central Rhodope (Bulgaria, Greece). Chemical Geology, 381, 206–222. Dill, H.G., 2010, The “chessboard” classification scheme of mineral deposits: Mineralogy and geology from aluminium to zirconium. Earth-Science Reviews, 100, 1-420.

Dimadis, E., Kassoli-Fournaraki, A., Kolcheva, K.,Filippidis, A., Tsirambidis, A. 1990. Ultramafic body in Gorgona area, North of Xanthi (Central Rhodope Massif, Greece). Geologica Rhodopica, 2, 117–136. Dinter, D.A., 1998. Late Cenozoic extension of the Alpine collisional orogen, northeastern Greece: Origin of the north Aegean basin. Geological Society of America Bulletin, 110(9), 1208-1236. Dissanayake, C., Chandrajith, R., 1999. Sri Lanka–Madagascar Gondwana linkage, evidence for a Pan-African mineral belt. Journal of Geology 107, 223–235.

Du Toit, A.L., 1946. Discussion on “corundum indicator basic rocks and associated pegmatites in the Northern Transvaal by J.W. Brandt”. Transactions of the Geological Society of South Africa 49, 51–102.

Dumurdzanov, N., Serafimovski, T. and Burchfiel, B.C., 2005. Cenozoic tectonics of Macedonia and its relation to the South Balkan extensional regime. Geosphere, 1(1), 1-22. Edwards, K.J. and Valley, J.W., 1998. Oxygen isotope diffusion and zoning in diopside: The importance of water fugacity during cooling. Geochimica et Cosmochimica Acta, 62(13), 2265-2277.

Eggins, S.M., Woodhead, J.D., Kinsley, L.P.J., Mortimer, G.E., Sylvester, P., McCulloch, M.T., Hergt, J.M. and Handler, M.R. 1997, A simple method for the precise determination of≥ 40 trace elements in geological samples by ICPMS using enriched isotope internal standardisation. Chemical Geology, 136(4), 371-370.

116

Emmett, J. L., Scarratt, K., McClure, S.F., Moses, T., Douthit, T. R., Hughes, R., Novak, S., Shigley, J. E., Wang, W., Bordelon, O., and Kane, R.E., 2003. Beryllium Diffusion of Ruby and Sapphire. Gems and Gemology, 39, 84-135.

Faure, G., 1986. Principles of Isotope Geology. 2nd Edition John Wiley & Sons, New York.

Farquhar, J., Chacko, T. and Ellis, D.J., 1996. Preservation of oxygen isotope compositions in granulites from Northwestern Canada and Enderby Land, Antarctica: implications for high- temperature isotopic thermometry. Contributions to Mineralogy and Petrology, 125(2-3), 213-224.

Forestier, F.H. and Lasnier, B., 1969. Découverte de niveaux d'amphibolites à pargasite, anorthite, corindon et saphirine dans les schistes cristallins de la vallée du Haut-Allier. Contributions to Mineralogy and Petrology, 23(3), 194-235.

Fritsch, E. and Rossman, G.R., 1988. An update on color in gems. Part 2: Colors involving multiple atoms and color centers. Gems and Gemology, 24(1), 3-15.

Froitzheim, N., Jahn-Awe, S., Frei, D., Wainwright, A.N., Maas, R., Georgiev, N., Nagel, T.J. and Pleuger, J., 2014. Age and composition of meta-ophiolite from the Rhodope Middle Allochthon (Satovcha, Bulgaria): A test for the maximum-allochthony hypothesis of the Hellenides. Tectonics, 37(8), 1477-1500. Fry, B., 2006. Stable isotope ecology, New York: Springer, 521.

Fytikas, M., Innocenti, F., Manetti, P., Peccerillo, A., Mazzuoli, R. and Villari, L., 1984. Tertiary to Quaternary evolution of volcanism in the Aegean region. Geological Society, London, Special Publication 17(1), 687-699. Gaillou, E., Devouard, B., Vielzeuf D., Boivin, P., Rochault, J. Valley, J. & Harris, C. 2010. Le Mont Coupet, Menet et le Sioulot: trois gisements de saphirs du Massif central. Le Règne Minéral 93, 28–36.

Garde, A.A. and Marker, M., 1988. Corundum crystals with blue-red colour zoning near Kangerdluarssuk, Sukkertoppen district, West Greenland. Rapport Grønlands Geologiske

Undersøkelse 140, 46–49.

Garnier, V., Giuliani, G., Maluski, H., Ohnenstetter, D., Trong, T.P., Quang, V.H., Van, L.P., Van, T.V. and Schwarz, D., 2002. Ar–Ar ages in phlogopites from marble-hosted ruby deposits in northern Vietnam: evidence for Cenozoic ruby formation. Chemical Geology, 188(1-2), 33-49.

117

Garnier, V. 2003. Les gisements de rubis associés aux marbres de l’Asie Centrale et du Sud-est: genèse et caractérisation isotopique. Thèse de Doctorat INPL, Nancy, France.

Garnier, V., Ohnenstetter, D., Giuliani, G., Fallick, A.E., Phan Trong, T., Hoang Quang, V. and Pham Van L, S.D., 2005. Age and genesis of sapphires from the basaltic province of Dak Nong, Southern Vietnam. Mineralogical Magazine, 69(1), 21-38.

Garnier, V., Ohnenstetter, D., Giuliani, G., Maluski, H., Deloule, E., Trong, T.P., Pham Van, L. and Hoàng Quang, V., 2005. Age and significance of ruby-bearing marble from the Red River Shear Zone, northern Vietnam. Canadian Mineralogist, 43(4), 1315-1329.

Garnier, V., Maluski, H., Giuliani, G., Ohnenstetter, D. and Schwarz, D., 2006. Ar Ar and U Pb ages of marble-hosted ruby deposits from central and southeast Asia. Canadian Journal of Earth Sciences, 43(4), 509-532.

Garnier, V., Giuliani, G., Ohnenstetter, D., Fallick, A.E., Dubessy, J., Banks, D., Vinh, H.Q., Lhomme, T., Maluski, H., Pêcher, A. and Bakhsh, K.A., 2008. Marble-hosted ruby deposits from Central and Southeast Asia: Towards a new genetic model. Ore Geology Reviews, 34(1-2), 169-191.

Gautier, P., Bosse, V., Cherneva, Z., Didier, A., Gerdjikov, I. and Tiepolo, M., 2017. Polycyclic alpine orogeny in the Rhodope metamorphic complex: the record in migmatites from the Nestos shear zone (N. Greece). Bulletin de la Société Géologique de France, 188(6), 36. Gawlick, H.J., Aubrecht, R., Schlagintweit, F., Missoni, S. and Plašienka, D., 2015. Ophiolitic detritus in Kimmeridgian resedimented limestones and its provenance from an eroded obducted ophiolitic nappe stack south of the Northern Calcareous Alps (Austria). Geologica Carpathica, 66(6), 473-487. Gerdjikov, I., Gautier, P., Cherneva, Z., Bosse, V. and Ruffet, G., 2010, September. Late Eocene synmetamorphic thrusting and syn-orogenic extension across the metamorphic pile of the Bulgarian Central Rhodope. In 19th Congress of the Carpathian-Balkan Geological Association, Thessaloniki, Greece. Geologica Balcanica, 39, 136-137. Giuliani, G., Fallick, A. E., Garnier, V., France-Lanord, C., Ohnenstetter, D., and Schwarz, D., 2005. Oxygen isotope composition as a tracer for the origins of rubies and sapphires, Geology, 33,249-252.

Giuliani, G., Ohnenstetter, D., Garnier, V., Fallick, A.E., Rakotondrazafy, M., Schwarz, D. and Groat, L.A., 2007a. The geology and genesis of gem corundum deposits. In Geology of Gem Deposits, 1st Edition (L. Groat, ed.). Mineralogical Association of Canada Short Course Series 37, 23-78.

118

Giuliani, G., Fallick, A., Rakotondrazafy, M., Ohnenstetter, D., Andriamamonjy, A., Ralantoarison, T., Rakotosamizanany, S., Razanatseheno, M., Offant, Y., Garnier, V. and Dunaigre, C., 2007b. Oxygen isotope systematics of gem corundum deposits in Madagascar: relevance for their geological origin. Mineralium Deposita, 42(3), 251.

Giuliani, G., Lasnier, B., Ohnenstetter, D., Fallick, A.E. and Pégère, G., 2010. Les gisements de corindon de France. Le Regne Minérale, 93, 5-23.

Giuliani, G., Caumon, G., Rakotosamizanany, S., Ohnenstetter, D. and Rakotondrazafy, M., 2014a. Classification chimique des corindons par analyse factorielle discriminante: application à la typologie des gisements de rubis et saphirs. Revue de Gemmologie, 188, 14-22.

Giuliani, G., Ohnenstetter, D., Fallick, A., Groat, L. and Fagan, A. 2014b. The geology and genesis of gem corundum deposits. In Geology of Gem Deposits. 2nd. Edition (Groat, L., ed.) Mineralogical Association of Canada, Short Course Series 44, Tucson AZ, 29-112.

Gonfiantini, R., 1978. Standards for stable isotope measurements in natural compounds. Nature, 271(5645), 534.

Graham, I., Sutherland, L., Zaw, K., Nechaev, V. and Khanchuk, A., 2008. Advances in our understanding of the gem corundum deposits of the West Pacific continental margins intraplate basaltic fields. Ore Geology Reviews, 34(1-2), 200-215.

Graham, I.T., Harris, S.J., Martin, L., Lay, A. and Zappettini, E., 2019. Enigmatic Alluvial Sapphires from the Orosmayo Region, Jujuy Province, Northwest Argentina: Insights into Their Origin from in situ Oxygen Isotopes. Minerals, 9(7), 390.

Grundmann, G., Morteani, G., 1989. Emerald mineralization during regional metamorphism, the Habachtal (Austria) and Leydsdorp (Transvaal, South Africa) deposits. Economic Geology, 84, 1835–1849.

Gübelin, E.J. and Koivula, J.I., 2008. Photoatlas of Inclusions in Gemstones, Volume 3.

Guo, J., O'reilly, S.Y. and Griffin, W.L., 1996. Zircon inclusions in corundum megacrysts: I. Trace element geochemistry and clues to the origin of corundum megacrysts in alkali basalts. Geochimica et Cosmochimica Acta, 60(13), 2347-2363.

Harlow, G.E., Pamucku, A., Naung,U. S. and Thu, U.K., 2006. Mineral assemblages and the origin of ruby in the Mogok Stone Tract, Myanmar. Gems and Gemology 42, 147.

Harlow, G.E. and Bender, W., 2013. A study of ruby (corundum) compositions from the Mogok Belt, Myanmar: Searching for chemical fingerprints. American Mineralogist, 98(7), 1120-1132.

119

Hey, M.H., 1954. A new review of the chlorites. Mineralogical Magazine, 36(224), 377-374. Hut, G., 1987. Consultants' group meeting on stable isotope reference samples for geochemical and hydrological investigations. International Atomic Energy Agency, Vienna (Austria), 42.

Ivanov, Z., Dimov, D., Dobrev, S., Kolkovski, B. and Sarov, S., 2000, May. Structure, Alpine evolution and mineralizations of the Central Rhodopes area (South Bulgaria). Guide to excursion B, ABCD-GEODE Workshop, Borovets, Bulgaria,“St. Kliment Ohridski” University Press, Sofia. Jahn-Awe, S., Froitzheim, N., Nagel, T.J., Frei, D., Georgiev, N. and Pleuger, J., 2010. Structural and geochronological evidence for Paleogene thrusting in the western Rhodopes, SW Bulgaria: elements for a new tectonic model of the Rhodope Metamorphic Province. Tectonics, 37(3), 1-36. Jolivet, L. and Brun, J.P., 2010. Cenozoic geodynamic evolution of the Aegean. International Journal of Earth Sciences, 99(1), 109-138. Jolivet, L., Faccenna, C., Huet, B., Labrousse, L., Le Pourhiet, L., Lacombe, O., Lecomte, E., Burov, E., Denèle, Y., Brun, J.P. and Philippon, M., 2013. Aegean tectonics: Strain localisation, slab tearing and trench retreat. Tectonophysics, 597, 1-37. Jöns, N. and Schenk, V., 2008. Relics of the Mozambique Ocean in the central East African Orogen: evidence from the Vohibory Block of southern Madagascar. Journal of Metamorphic Geology, 26(1), 17-28.

Kane, R.E., Kammerling, R.C., 1992. Status of the ruby and sapphire mining in the Mogok stone tract. Gems and Gemology, 28, 152–174.

Karageorgiou, M.M.D., Karymbalis, E. and Karageorgiou, D.E., 2010. The use of the geographical information systems (GIS) in the geological-mineralogical mapping of the Paranesti area. Bulletin of the Geological Society of Greece, 43(3), 1601-1606. Kendall, C. and Caldwell, E.A., 1998. Chapter 2. Fundamentals of isotope geochemistry. In Isotope tracers in catchment hydrology, Elsevier Science B.V., Amsterdam, 51-86.

Keulen, N. & Kalvig, P., 2013. Fingerprinting of corundum (ruby) from Fiskenaesset, West Greenland. GEUS Geological Survey of Denmark and Greenland Bulletin, 28, 53–56.

Key, R.M. & Ochieng, J. 1991 The growth of rubies in south-east Kenya. Journal of Gemmology, 22(8), 484–496.

Kievlenko, E.Y. 2003. Geology of Gems. Ocean Pictures Ltd, Littleton, CO, USA.

120

Kirchenbaur, M., Pleuger, J., Jahn-Awe, S., Nagel, T.J., Froitzheim, N., Fonseca, R.O.C. and Münker, C., 2012. Timing of high-pressure metamorphic events in the Bulgarian Rhodopes from Lu–Hf garnet geochronology. Contributions to Mineralogy and Petrology, 163(5), 897- 921. Kissin, A.J. 1994. Ruby and sapphire from the Southern Ural Mountains, Russia. Gems and Gemology 30, 243–252.

Kohn, M.J., 1999. Why most “dry” rocks should cool “wet”. American Mineralogist, 84(4), 570- 580.

Kolcheva, K., Zeljazkova-Panajotova, M., Dobrecov, N.L. and Stojanova, V., 1986. Eclogites in Central Rhodope metamorphic group and their retrograde metamorphism. Geochemistry, Mineralogy and Petrology, 20, 136-21. Krebs, M.Y., Pearson, D.G., Fagan, A.J., Bussweiler, Y. and Sarkar, C., 2019. The application of trace elements and Sr–Pb isotopes to dating and tracing ruby formation: The Aappaluttoq deposit, SW Greenland. Chemical Geology, 523, 42-58.

Krenn, K., Bauer, C., Proyer, A., Mposkos, E. and Hoinkes, G. 2008: Fluid entrapment and reequilibration during subduction and exhumation: A case study from the high-grade Nestos shear zone, Central Rhodope, Greece. Lithos 104(1), 37-53. Krenn, K., Bauer, C., Proyer, A., Klötzli, U. and Hoinkes, G., 2010. Tectonometamorphic evolution of the Rhodope orogeny. Tectonics, 37(4), 1-25. Krohe A, Mposkos E. 2002. Multiple generations of extensional detachments in the Rhodope Mountains (northern Greece): evidence of episodic exhumation of high-pressure rocks. In: Blundell DJ, Neubauer F, von Quadt A, eds. The Timing and Location of Major Ore Deposits in an Evolving Orogen. Geological Society of London Special Publication 204: 151–178. Kroner, A., 1984. Late Precambrian plate tectonics and orogeny: a need to redefine the term Pan-African. Géologie Africaine, 23-28.

Krylov, D.P. and Glebovitsky, V.A., 2007, March. Oxygen isotopic composition and nature of fluid during the formation of High-Al corundum-bearing rocks of Mt. Dyadina, northern Karelia. Doklady Earth Sciences, 413(1), 210. Krylov, D.P., 2008. Anomalous 18O/16O ratios in the corundum-bearing rocks of Khitostrov, northern Karelia. Doklady Earth Sciences, 429A (3), 453–456.

Kucera, M., 1984, Industrial Minerals and Rocks, 1st ed, Amsterdam, Elsevier Science, 157.

Kydonakis, K., Brun, J.P. and Sokoutis, D., 2015. North Aegean core complexes, the gravity spreading of a thrust wedge. Journal of Geophysical Research: Solid Earth, 120(1), 595-616. 121

Lasnier, B.,1977. Persistance d'une série granulitique au coeur du Massif Central Français (Haut Allier)— Les termes basiques, ultrabasiques et carbonatés. Unpublished Ph.D. thesis, University of Nantes, France

Lawson, A.C., 1903. Plumasite, an oligoclase corundum rock, near Spanish peak, California. University of California Publications in Geological Sciences, 219–229.

Le Goff, E., Deschamps, Y. and Guerrot, C., 2010. Tectonic implications of new single zircon Pb- Pb evaporation data in the Lossogonoi and Longido ruby-districts, Mozambican metamorphic Belt of north-eastern Tanzania. Comptes Rendus Geoscience, 342(1), 36-45.

Liati, A. and Mposkos, E., 1990. Evolution of the eclogites in the Rhodope Zone of northern Greece. Lithos, 25(1-3), 89-99.

Liati, A. and Seidel, E., 1994. Sapphirine and hogbomite in overprinted kyanite-eclogites of central Rhodope, N. Greece: first evidence of granulite-facies metamorphism. European Journal of Mineralogy, 6(5), 733-738.

Liati, A. and Seidel, E., 1996. Metamorphic evolution and geochemistry of kyanite eclogites in central Rhodope, northern Greece. Mineralogy and Petrology, 123(3), 293-307.

Liati, A. and Gebauer, D., 1999. Constraining the prograde and retrograde PT path of Eocene HP rocks by SHRIMP dating of different zircon domains: inferred rates of heating, burial, cooling and exhumation for central Rhodope, northern Greece. Contributions to Mineralogy and Petrology, 137(4), 360-374. Liati, A., Gebauer, D., & Wysoczanski, R. 2002, U-Pb SHRIMP-dating of zircon domains from UHP garnet-rich mafic rocks and late pegmatoids in the Rhodope zone (N Greece); evidence for Early Cretaceous crystallization and Late Cretaceous metamorphism. Chemical Geology, 184(3–4), 361–379. Liati, A., 2005. Identification of repeated Alpine (ultra) high-pressure metamorphic events by U- Pb SHRIMP geochronology and REE geochemistry of zircon: the Rhodope zone of Northern Greece. Contributions to Mineralogy and Petrology, 150, 608–636. Liati, A., Gebauer, D., Fanning, C.M., Dobrzhinetskaya, L. (eds.), Faryad, S.W. (eds.), Wallis, S. (eds.) and Cuthbert, S. (eds.), 2011. Geochronology of the Alpine UHP Rhodope Zone: a review of isotopic ages and constraints on the geodynamic evolution. Ultrahigh-pressure metamorphism, 375-364. Elsevier. Liati, A., Theye, T., Fanning, C.M., Gebauer, D. and Rayner, N., 2016. Multiple subduction cycles in the Alpine orogeny, as recorded in single zircon crystals (Rhodope zone, Greece). Gondwana Research, 37(1), 199-207. 122

Marchev, P., Georgiev, S., Raicheva, R., Peytcheva, I., von Quadt, A., Ovtcharova, M. and Bonev, N., 2013. Adakitic magmatism in post-collisional setting: an example from the Early–Middle Eocene Magmatic Belt in Southern Bulgaria and Northern Greece. Lithos, 180, 159-180. Mbih, K.P., Meffre, S., Yongue, R.F., Kanouo, N.S. and Jay, T., 2016. Chemistry and origin of the Mayo Kila sapphires, NW region Cameroon (Central Africa): Their possible relationship with the Cameroon volcanic line. Journal of African Earth Sciences, 118, 263-273.

McDonough, W.F. and Sun, S.S., 1995. The composition of the Earth. Chemical Geology, 120(3- 4), 223-253.

Menant, A., Jolivet, L. and Vrielynck, B., 2016. Kinematic reconstructions and magmatic evolution illuminating crustal and mantle dynamics of the eastern Mediterranean region since the late Cretaceous. Tectonophysics, 675, 103-140.

Mendybaev, R.A., Richter, F.M., Spicuzza, M.J., Valley, J.W. and Davis, A.M., 2010, March. Oxygen Isotope Fractionation During Evaporation of Mg-and Si-rich CMAS-Liquids in Vacuum. Lunar and Planetary Science Conference, 41, 2725.

Menon, R.D. and Santosh, M., 1995. The Pan-African gemstone province of East Gondwana. Memoirs-Geological Society of India, 357-372.

Meert, J.G., Van Der Voo, R. & Ayub, S. 1995. Paleomagnetic investigation of the Neoproterozoic Gagwe lavas and Mbozi complex, Tanzania and the assembly of Gondwana. Precambrian Research, 74, 225–244.

Meert, J.G. 2003. A synopsis of events related to the assembly of eastern Gondwana. Tectonophysics, 362, 1–40.

Mercier, A., Debat, P. and Saul, J.M., 1999a. Exotic origin of the ruby deposits of the Mangari area in SE Kenya. Ore Geology Reviews, 14(2), 83-104.

Mercier, A., Rakotondrazafy, M. and Ravolomiandrinarivo, B., 1999b. Ruby mineralization in southwest Madagascar. Gondwana Research, 2(3), 433-438.

Merle, O., Michon, L., Camus, G. and de Goer, A., 1998. L'extension oligocène sur la transversale septentrionale du rift du Massif Central. Bulletin de la Société Géologique de France, 169(5), 615-626.

Moulas, E., Kostopoulos, D., Connolly, J. A., and Burg, J. P., 2013, PT estimates and timing of the sapphirine-bearing metamorphic overprint in kyanite eclogites from Central Rhodope, northern Greece. Petrology, 21(5), 507-521.

123

Mposkos, E. and Liati, A. 1993. Metamorphic evolution of metapelites in the high-pressure terrane of the Rhodope zone, Northern Greece. Canadian Mineralogist, 31(2), 401-424.

Mposkos, E. and Krohe, A., 2000. Petrological and structural evolution of continental high pressure (HP) metamorphic rocks in the Alpine Rhodope Domain (N. Greece). In Proceedings of the 3rd International Conference on the Geology of the Eastern Mediterranean (Vol. 221236). Mposkos, E. and Krohe, A., 2006. Pressure-temperature deformation paths of closely associated ultra-high-pressure (diamond-bearing) crustal and mantle rocks of the Kimi complex: implications for the tectonic history of the Rhodope Mountains, northern Greece. Canadian Journal of Earth Science, 43, 1755–1776. Mposkos E., Krohe A. and Baziotis I., 2010. Alpine polyphase metamorphism in metapelites from Sidironero complex (Rhodope domain, NE Greece). In: Proceedings of the 19th Congress of the Carpathian-Balkan Geological Association, Thessaloniki, Greece, Special Volume of the Scientific Annals of the School of Geology of the Aristotle University of Thessaloniki, 100, 173– 181. Muhongo, S., Errera, M., 1993. Gemmological Characteristics of Rubies in Eastern Uluguru Mountains, Tanzania, a Reconnaissance Study. Musée Royal de l'Afrique Centrale à Tervuren, Département de Géologie et Minéralogie, Rapport Annuel, 1991–1992.

Muhlmeister, S., Fritsch, E., Shigley, J.E., Devouard, B., and Laurs, B.M. 1998. Separating natural and synthetic rubies on the basis of trace element chemistry. Gems and Gemology, 34, 80– 101.

Nagel, T.J., Schmidt, S., Janak, M., Froitzheim, N., Jahn-Awe, S. and Georgiev, N., 2011. The exposed base of a collapsing wegde: the Nestos Shear Zone (Rhodope Metamorphic Province, Greece). Tectonics, 36(4), 1-17. Okrusch, M., Bunch, T., Bank, H., 1976. Paragenesis and petrogenesis of a corundum bearing marble at Hunza, Kashmir. Mineralium Deposita, 11, 278–297.

Palke, A.C., Renfro, N.D., and Berg, R.B. 2016. Origin of sapphires from a lamprophyre dike at Yogo Gulch, Montana, USA: Clues from their melt inclusions. Lithos, 260, 339–344.

Papanikolaou, D. and Panagopoulos A., 1981. On the structural style of Southern Rhodope, Greece. Geologica Balcanica, 11(3), 13–22. Pardieu, V., Sangsawong, S., Muyal, J., Chauviré, B., Massi, L. and Sturman, N., 2013. Montepeuz Area (Mozambique). Gemological Institute of America, Research and News, October, 1, 24.

124

Pardieu, V., Sangsawong, S., Muyal, J. and Sturman, N., 2014. Blue sapphires from the Mambilla Plateau, Taraba State, Nigeria. A preliminary exploration. GIA News from Research, GIA Laboratory Bangkok, 1-47.

Pêcher, A., Giuliani, G., Garnier, V., Maluski, H., Kausar, A.B., Malik, R.H. and Muntaz, H.R., 2002. Geology, geochemistry and Ar–Ar geochronology of the Nangimali ruby deposit, Nanga Parbat Himalaya (Azad Kashmir, Pakistan). Journal of Asian Earth Sciences, 21(3), 265-282.

Peretti, A., Mullis, J. & Kundig, R. 1990. Die Kaschmir-Saphire und ihr geologisches Erinnerungsvermögen. Forsch Techn. Neue Zürch Zeit. 187, 59.

Peretti, A., Mullis, J., Mouawad, F., 1996. The role of fluoride in the formation of color zoning in rubies from Mong Hsu (Myanmar, Burma). Journal of Gemmology, 25, 3–19.

Peucat, J.E., Raffault, P., Fritsch, E., Bouhnik-Le Coz, M., Simonet, C., and Lasnier, B. 2007. Ga/Mg ratio as a new geochemical tool to differentiate magmatic from metamorphic blue sapphires. Lithos, 98, 261–274.

Rakotondrazafy, A.F.M., Giuliani, G., Ohnenstetter, D., Fallick, A.E., Rakotosamizanany, S., Andriamamonjy, A., Ralantoarison, T., Razanatseheno, M., Offant, Y., Garnier, V. and Maluski, H., 2008. Gem corundum deposits of Madagascar: a review. Ore Geology Reviews, 34(1-2), 134-154.

Rakotosamizanany, S., Giuliani, G., Ohnenstetter , D., Rakotondrazafy, A.F.M. & Fallick, A.E. 2009. Les gisements de saphirs et de rubis associés aux basaltes alcalins de Madagascar: caractéristiques géologiques et minéralogiques. 2ème partie: Caractéristiques minéralogiques. Revue de l’Association Française de Gemmologie, AFG 170, 9–18.

Ricou, L.E., Burg, J.P., Godfriaux, I. and Ivanov, Z., 1998. Rhodope and Vardar: the metamorphic and the olistostromic paired belts related to the Cretaceous subduction under Europe. Geodinamica Acta, 11(6), 365-369. Ring, U., Glodny, J., Will, T. and Thomson, S., 2010. The Hellenic subduction system: high- pressure metamorphism, exhumation, normal faulting, and large-scale extension. Annual Review of Earth and Planetary Sciences, 38, 45-76. Robb, L.J., Robb, V.M., 1986. Archaean pegmatite deposits in the north-eastern Transvaal. In: Anhaeusser, C.R., Maske, S. (Eds.), Mineral Deposits of Southern Africa. Geological Society of South Africa, I–II, 437–449.

125

Robertson, A., Karamata, S. and Šarić, K., 2009. Overview of ophiolites and related units in the Late Palaeozoic–Early Cenozoic magmatic and tectonic development of Tethys in the northern part of the Balkan region. Lithos, 108(1-4), 1-36. Robertson S, 2012. Ruby Sources and Pricing Update, The Gem Guide Jan-Feb 4.

Rose, R.L., 1957. Andalusite-and corundum-bearing pegmatites in Yosemite National Park, California. American Mineralogist, 42(9-10), 635-647.

Rosing, M.T., Bird, D.K. and Dymek, R.F., 1987. Hydration of corundum-bearing xenoliths in the Qorqut granite complex, Godthaabsfjord, West Greenland. American Mineralogist, 72(1-2), 29-38.

Rozanski, K., Araguas-Araguas, L. and Gonfiantini, R., 1993. Isotopic patterns in modern global precipitation. Climate change in continental isotopic records, 78, 1-36.

Saminpanya, S., Manning, D., Droop, G. and Henderson, C., 2003. Trace elements in Thai gem corundums. Journal of Gemmology, 28(7), 399-416.

Santosh, M., and Collins, A.S., 2003. Gemstone Mineralization in the Palghat-Cauvery Shear Zone System (KarurKangayam Belt), Southern India. Gondwana Research, 64, 911–918

Schmidt, S., Nagel, T.J. and Froitzheim, N., 2010. A new occurrence of microdiamond-bearing metamorphic rocks, SW Rhodopes, Greece. European Journal of Mineralogy, 22(2), 189-198. Schwarz, D., Pardieu, V., Saul, J.M., Schmetzer, K., Laurs, B.M., Giuliani, G., Klemm, L., Malsy, A.K., Erel, E., Hauzenberger, C. and Du Toit, G., 2008. Rubies and sapphires from Winza, central Tanzania. Gems and Gemology, 44(4).

Seifert, A.V. & Hyrsl, J. 1999. Sapphire and garnet from Kalalani, Tanga Province, Tanzania. Gems and Gemology, 35, 108–120.

Seifert, W., Rhede, D., Förster, H.J., Naumann, R., Thomas, R. and Ulrych, J., 2014. Macrocrystic corundum and Fe–Ti oxide minerals entrained in alkali basalts from the Eger (Ohře) Rift: Mg− Fe 3+-rich ilmenite as tracer of an oxidized upper mantle. Mineralogy and Petrology, 108(5), 645-662.

Sharp, Z.D., 1992. In situ laser microprobe techniques for stable isotope analysis. Chemical Geology: Isotope Geoscience Section, 101(1-2), 3-19.

Sharp, Z., 2017. Principles of stable isotope geochemistry. 2nd Edition, University of New Mexico, Geochemistry Commons, Digital Commons Network™.

126

Shor, R., Weldon, R., 2009. Ruby and Sapphire production and distribution: a quarter Century of change. Gems and Gemology, 45, 236-259.

Simonet, C., Okundi, S., Masai, P., 2000. General setting of colored gemstone deposits in the Mozambique Belt of Kenya — preliminary considerations. Proceedings of the 9th Conference of the Geological Society of Kenya, Nairobi, November, 123–138.

Simonet, C., Paquette, J.L., Pin, C., Lasnier, B., Fritsch, E., 2004. The Dusi (Garba Tula) sapphire deposit, central Kenya—a unique Pan-African corundum-bearing monzonite. Journal of African Earth Sciences, 38, 401–410.

Simonet, C., Fritsch, E. & Lasnier, B. 2008. A classification of gem corundum deposits aimed towards gem exploration. Ore Geology Reviews, 34(1), 127-133.

Skinner, B.J., 1979. Earth resources. Proceedings of the National Academy of Sciences, 76(9), 4212-4217.

Smith, C.P., Clark, B., and Fagan, A.J., 2016. Ruby and Pink Sapphire from Aappaluttoq, Greenland. Journal of Gemmology, 35(4), 294.

Solesbury, F.W., 1967. Gem corundum pegmatites in NE Tanganyika. Economic Geology, 62(7), 983-991.

Sorokina, E.S., Rösel, D., Häger, T., Mertz-Kraus, R. and Saul, J.M., 2017. LA-ICP-MS U–Pb dating of rutile inclusions within corundum (ruby and sapphire): new constraints on the formation of corundum deposits along the Mozambique belt. Mineralium Deposita, 52(5), 641-649.

Spiridonov, E.M., 1998. Gemstone deposits of the former Soviet Union. Journal of Gemmology, 112.

Stern, R.J., Tsujimori, T., Harlow, G. and Groat, L.A., 2013. Plate tectonic gemstones. Geology, 41(7), 723-726.

Sun, S.S. and McDonough, W.F., 1989. Chemical and isotopic systematics of oceanic basalts: implications for mantle composition and processes. Geological Society, London, Special Publications, 42(1), 373-365. Sutherland, F.L. and Coenraads, R.R., 1996. An unusual ruby-sapphire-sapphirine-spinel assemblage from the Tertiary Barrington volcanic province, New South Wales, Australia. Mineralogical Magazine, 60(401), 623-638.

127

Sutherland, F.L., Schwarz, D., Jobbins, E.A., Coenraads, R.R. and Webb, G., 1998. Distinctive gem corundum suites from discrete basalt fields: a comparative study of Barrington, Australia, and West Pailin, Cambodia, gemfields. Journal of Gemmology, 26(2), 66-85.

Sutherland, F.L. and Schwarz, D., 2001. Origin of gem corundums from basaltic fields. Australian Gemmologist, 21(1), 30-33.

Sutherland, F.L., Duroc-Danner, J.M. and Meffre, S., 2008a. Age and origin of gem corundum and zircon megacrysts from the Mercaderes–Rio Mayo area, South-west Colombia, South America. Ore Geology Reviews, 34(1-2), 155-168.

Sutherland, L.F., Guiliani, G., Fallick, E.A., Garland, M. and Webb, G., 2008b. Sapphire–ruby characteristics west Pailin, Cambodia. The Australian Gemologist, 23, 329-368.

Sutherland, F.L., Giuliani, G., Fallick, A.E., Garland, M. & Webb, G.B., 2009a. Sapphire ruby characteristics, West Pailin, Cambodia: clues to their origin based on trace element and O isotope analysis. Australian Gemmologist. 23, 373–432.

Sutherland, F.L., Zaw, K., Meffre, S., Giuliani, G., Fallick, A.E., Graham, I.T. and Webb, G.B., 2009b. Gem-corundum megacrysts from east Australian basalt fields: trace elements, oxygen isotopes and origins. Australian Journal of Earth Sciences, 56(7),100 3-1022.

Sutherland, F.L. and Abduriyim, A. 2009c. Geographic typing of gem corundum: a text case from Australian Gemmologist. 31, 203–210.

Sutherland, F.L., Zaw, K., Meffre, S., Yui, T.F. and Thu, K., 2015. Advances in trace element “fingerprinting” of gem corundum, ruby and sapphire, Mogok area, Myanmar. Minerals, 5(1), 61-79.

Sutherland, F.L., Graham, I.T., Harris, S.J., Coldham, T., Powell, W., Belousova, E.A. and Martin, L., 2017. Unusual ruby–sapphire transition in alluvial megacrysts, Cenozoic basaltic gem field, New England, New South Wales, Australia. Lithos, 278, 347-360.

Sutthirat, C., Saminpanya, S., Droop, G.T.R., Henderson, C.M.B. and Manning, D.A.C., 2001. Clinopyroxene-corundum assemblages from alkali basalt and alluvium, eastern Thailand: constraints on the origin of Thai rubies. Mineralogical Magazine. 65, 277–295.

Tenthorey, E., Ryan, J.G., Snow, E.A., 1996. Petrogenesis of sapphirine-bearing metatroctolites from the Buck Creek ultramafic body, southern Appalachians. Journal of Metamorphic Geology, 14, 103–114.

128

Treloar, P., 1987. The Cr-minerals of Outokumpu. Their chemistry and significance. Journal of Petrology, 28, 867–886.

Turpaud, P., 2006. Characterization of igneous terranes by zircon dating: implications for the UHP relicts occurrences and suture identification in the Central Rhodope, Northern Greece. PhD dissertation, Univ. Johannes Gutenberg, Mainz, Germany. Turpaud, P. and Reischmann, T., 2010, Characterisation of igneous terranes by zircon dating: implications for UHP occurrences and suture identification in the Central Rhodope, northern Greece. International Journal of Earth Sciences, 99(3), 567-591. Uher, P., Giuliani, G., Szakáll, S., Fallick, A., Strunga, V., Vaculovič, T., Ozdín, D. and Gregáňová, M., 2012. Sapphires related to alkali basalts from the Cerová Highlands, Western Carpathians (southern Slovakia): composition and origin. Geologica Carpathica, 63(1), 71-82.

Valley, J.W., Kitchen, N., Kohn, M.J., Niendorf, C.R. and Spicuzza, M.J., 1995. UWG-2, a garnet standard for oxygen isotope ratios: strategies for high precision and accuracy with laser heating. Geochimica et Cosmochimica Acta, 59(24), 5223-5231.

Valley, J.W., 2001. Stable isotope thermometry at high temperatures. Reviews in Mineralogy and Geochemistry, 43(1), 365-413. van Hinsbergen, D.J.J., Hafkenscheid, E., Spakman, W., Meulenkamp, J.E. and Wortel, R., 2005. Nappe stacking resulting from subduction of oceanic and continental lithosphere below Greece. Geology, 37(4), 371-370. Voudouris, P., Graham, I., Melfos, V., Zaw, K., Sutherland, L., Giuliani, G. and Ionescu, M., 2010, April. Gem corundum deposits of Greece: Diversity, chemistry and origins. In Proceedings of the 13th Quadrennial IAGOD Symposium, Adelaide, Australia, 69, 429430.

Voudouris, P., Mavrogonatos, C., Graham, I., Giuliani, G., Melfos, V., Karampelas, S., Karantoni, V., Wang, K., Tarantola, A., Zaw, K. and Meffre, S., 2019. Gem Corundum Deposits of Greece: Geology, Mineralogy and Genesis. Minerals, 9(1), 49. Vysotskii, S.V., Ignat'ev, A.V., Yakovenko, V.V., Karabtsov, A.A., 2008. Anomalous light oxygen isotope composition inminerals of corundum-bearing rocks in northern Karelia. Doklady Earth Sciences, 423 (8), 1216–1219.

Vysotskiy, S.V., Ignat'ev, A.V., Levitskii, V.I., Nechaev, V.P., Velivetskaya, T.A., Yakovenko, V.V., 2014. Geochemistry of stable oxygen and hydrogen isotopes in minerals and corundum- bearing rocks in Northern Karelia as an indicator of their unusual genesis. Geochemistry International, 52 (9), 773–782.

129

Vysotskiy, S.V., Nechaev, V.P., Kissin, A.Y., Yakovenko, V.V., Ignat'ev, A.V., Velivetskaya, T.A., Sutherland, F.L. and Agoshkov, A.I., 2015. Oxygen isotopic composition as an indicator of ruby and sapphire origin: A review of Russian occurrences. Ore Geology Reviews, 68, 164-170.

Wakabayashi, J. 2004. Tectonic mechanisms associated with P–T paths of regional metamorphism: alternatives to single-cycle thrusting and heating. Tectonophysics, 392(1-4), 193-218.

Wang, K.K., Graham, I.T., Martin, L., Voudouris, P., Giuliani, G., Lay, A., Harris, S.J. and Fallick, A., 2019. Fingerprinting Paranesti rubies through oxygen isotopes. Minerals, 9(2), 91.

Wawrzenitz, N., Krohe, A., Baziotis, I., Mposkos, E., Kylander-Clark, A.R. and Romer, R.L., 2015. LASS U–Th–Pb monazite and rutile geochronology of felsic high-pressure granulites (Rhodope, N Greece): Effects of fluid, deformation and metamorphic reactions in local subsystems. Lithos, 236, 266-365. Whalen, J.B., Currie, K.L. and Chappell, B.W., 1987. A-type granites: geochemical characteristics, discrimination and petrogenesis. Contributions to Mineralogy and Petrology, 95(4), 407-419.

Wilson, A.F. and Baksi, A.K., 1983. Widespread 18O depletion in some Precambrian granulites of Australia. Precambrian Research, 23(1), 37-56. Wong, J., Verdel, C., and Allen, C.M. 2017. Trace-element compositions of sapphire and ruby from the eastern Australian gemstone belt, Mineralogical Magazine, 81, 1551–1576.

Wright L.E., 2017. Oxygen Isotopes. In: Gilbert A.S. (eds) Encyclopedia of Geoarchaeology. Encyclopedia of Earth Sciences Series. Springer, Dordrecht.

Wright, J.B., Hastings, D.A., Jones, W.B. and Williams, H.R., 1985. Geology and mineral resources of West Africa, 187. London: Allen & Unwin.

Yakymchuk, C. and Szilas, K., 2018. Corundum formation by metasomatic reactions in Archean metapelite, SW Greenland: Exploration vectors for ruby deposits within high-grade greenstone belts. Geoscience Frontiers, 9(3), 727-749. 130

Yui, T.F., Zaw, K. and Limtrakun, P., 2003. Oxygen isotope composition of the Denchai sapphire, Thailand: a clue to its enigmatic origin. Lithos, 67(1-2), 153-161. Yui, T.F., Wu, C.M., Limtrakun, P., Sricharn, W. and Boonsoong, A., 2006. Oxygen isotope studies on placer sapphire and ruby in the Chanthaburi-Trat alkali basaltic gemfield, Thailand. Lithos, 86(3-4), 197-211.

Zaw, K., Sutherland, L., Yui, T.F., Meffre, S. and Thu, K., 2015. Vanadium-rich ruby and sapphire within Mogok Gemfield, Myanmar: implications for gem color and genesis. Mineralium Deposita, 50(1), 25-39.

131

Appendix 1

The origin of a new pargasite-schist hosted ruby deposit from Paranesti, Northern Greece.

Published in The Canadian Mineralogist Vol. 55, pp. 535-560 (2017) DOI: 10.3749/canmin.1700014

132 535

The Canadian Mineralogist Vol. 55, pp. 535-560 (2017) DOI: 10.3749/canmin.1700014

THE ORIGIN OF A NEW PARGASITE-SCHIST HOSTED RUBY DEPOSIT FROM PARANESTI, NORTHERN GREECE

§ KANDY K. WANG ,IAN T. GRAHAM, ANGELA LAY, STEPHEN J. HARRIS, AND DAVID RONALD COHEN PANGEA Research Centre, School of Biological, Earth and Environmental Sciences, University of New South Wales Sydney, 2052, Australia

PANAGIOTIS VOUDOURIS Faculty of Geology & Geoenvironment, National and Kapodistrian University of Athens, Greece

ELENA BELOUSOVA Australian Research Council Centre of Excellence for Core to Crust Fluid Systems (CCFS) GEMOC, Department of Earth and Planetary Sciences, Macquarie University, Sydney, Australia

GASTON GIULIANI Universite´ de Lorraine, IRD and CRPG UMR 7358 CNRS-UL, BP 20, 15 rue Notre-Dame-des-Pauvres, 54501 Vandœuvre-les-Nancy,` France

ANTHONY E. FALLICK Isotope Geosciences Unit, S.U.E.R.C., Rankine Avenue, East Kilbride, Glasgow G75 0QF, Scotland, United Kingdom

ALAN GREIG School of Earth Sciences, University of Melbourne, Melbourne, Victoria, Australia

ABSTRACT

Gem-quality (cabochon) ruby-bearing occurrences (here termed PAR-1 and PAR-5) located near Paranesti, north eastern Greece have been systematically studied for the first time in this paper. Tectonically, the occurrences are located within the Nestos Shear Zone (NSZ). The NSZ separates two distinct geological units. The Rhodope Terrane is a heterogeneous unit of gneisses, mafic, ultramafic, and meta-sedimentary rocks in the hanging wall. The footwall Pangaion-Pirin Complex consists of marbles and acid gneisses of a Mesozoic carbonate platform on pre-Mesozoic continental basement. In this paper, a range of petrographic and geochemical techniques were used to determine (1) any similarities and differences to other mafic-ultramafic hosted ruby deposits worldwide; (2) distinctive geochemical fingerprints for Paranesti; and (3) the likely P-T conditions of formation. Detailed petrographic and whole-rock analyses utilizing ICP-MS, XRF, and XRD have found the Paranesti corundum to be of a mafic/ultramafic protolith with approximately 40 wt.% SiO2, 16 wt.% Mg, 11000 ppm Cr, and 440 ppm Ni. EMPA major element analysis determined the mineral inclusions within the corundum grains to be picotite and hercynite spinels. Pargasite is the dominant amphibole within the corundum-bearing amphibole schist host. The surrounding non-corundum bearing chlorite schist mainly comprises clinochlore. Petrographic examination of the mineral assemblages within the corundum-bearing schists revealed strong fracturing and alignment (parallel to the main regional foliation) of the corundum grains and margarite reaction rims around the corundum. The surrounding non-corundum amphibolites also contain anorthite, along with relict sillimanite, kyanite, and chlorite/muscovite/epidote overprinting. Detailed LA-ICP-MS trace element analysis of the color range of corundum from the two occurrences showed the corundum to be mainly of metamorphic origin, though pale rubies from PAR-5 suggest some metasomatic influence.

§ Corresponding author e-mail address: [email protected]

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 536 THE CANADIAN MINERALOGIST

The corundum displays distinctive geochemical locality signatures, with a combination of high Cr (average 2300 ppm with 15% sample points on core positions .5000 ppm and maximum 8600 ppm); high Si (average 1400 ppm with 40% over 1500 ppm and maximum 2500 ppm), low Mg (average 30 ppm), and very low V, Ti, and Ga. Based on the literature for similar occurrences, and the mineral assemblages observed at Paranesti, the estimated P-T conditions of corundum formation are ,7 kbar and ,750 8C, similar to the mafic African amphibolite-hosted rubies. This study has found the Paranesti occurrences to be most similar to the Winza, Tanzania ruby deposit, whilst there are some similarities to other high-Cr ruby deposits, primarily the Fiskenæsset, Greenland and metamorphic amphibolitic schist hosted African deposits. The Paranesti corundum most likely formed during regional amphibolite facies metamorphism which created the Nestos Shear Zone, along with a lesser influence (primarily observed in the PAR-5 occurrence) of more localized metasomatism. Subsequent multiple greenschist facies retrogression of the occurrences resulted in the current-day host amphibole-chlorite schist assemblages.

Keywords: Paranesti Greece, rubies, pargasite schist, margarite, spinel inclusion, Rhodope, Nestos Shear Zone, maﬁc/ultramaﬁc, retrogression, metasomatic, P-T conditions.

INTRODUCTION whole-rock analyses can be used to determine the protolith. Humanity has long had a fascination with the red In this paper, we describe corundum and their host gem variety of corundum – ruby. The rarity of the mafic-ultramafic metamorphic assemblages from the stone has seen prospectors roaming all the corners of little-known Paranesti corundum/ruby occurrences of the earth in search of it, from the jungles around north-eastern Greece. We present detailed in situ LA- Mogok in Myanmar (home of the world’s finest rubies ICP-MS trace element analyses of the corundum, famed for their pigeon’s blood color), through to along with whole-rock geochemical and petrographic Madagascar and Tanzania in Africa, and along the analyses of their host rocks, compare these occurrenc- eastern boundary of Australia to the snow-capped es with similar mafic-ultramafic hosted ruby deposits continent of Greenland. Ruby and sapphire are both worldwide, and evaluate their likely origin (Voudouris gem varieties of the mineral corundum. Pure corun- et al.2010,Wanget al.2016).Thecorundum dum is colorless. The pink to red color of ruby is due occurrences are located within the Nestos Shear Zone to the chromophore (coloring agent) Cr3þ and the blue, (NSZ), which is delineated by ultra-high pressure yellow, and green suite of colors of sapphire are due to (UHP) relicts and eclogites, associated with amphib- small amounts of Fe2þ,Mg2þ,Fe3þ, and Ti4þ. It is now olites of oceanic affinities and ultramafic rocks widely acknowledged that the origin of corundum can (Turpaud 2006). To date, these occurrences lack any generally be subdivided into two main categories: in-depth study, unlike the well-documented Asian, African, and Australian ruby deposits. This study seeks magmatic and metamorphic/(metasomatic), as re- to address three main geological questions regarding viewed in detail by Simonet et al. (2008) and the Paranesti corundum: (1) Under what conditions did Giuliani et al. (2014a). the corundum form? (2) How do these corundum and Distinctive characteristics of corundum are used to host rock assemblages compare with other mafic- identify the geological processes responsible for their ultramafic hosted rubies? (3) Are there any distinct genesis. These characteristics include color, trace geochemical signatures within the host rock assem- element chemistry, and inclusions of solids, liquids, blages that may aid in exploration for similar mafic- and gases that were trapped at some stage of crystal ultramafic hosted deposits within this region? growth (e.g., Dill 2010, Sorokina et al. 2016). Corundum specific indicators include trace element GEOLOGICAL SETTING analysis using the Ga/Mg ratio that has been used to Regional tectonic framework differentiate magmatic from metamorphic blue sapphires (Peucat et al. 2007). Based on trace element The in situ corundum occurrences of Paranesti are analysis of gem corundum grains, the FeO–Cr2O3– located within the Nestos Shear Zone of the Rhodope MgO–V2O3 versus FeOþTiO2þGa2O3 diagram can be Mountain Complex (RMC) in north-eastern Greece further used to constrain the geological environment of (Fig. 1). The Paranesti area is defined by a rugged formation of gem corundum (Giuliani et al. 2010, mountainous terrain. The western part is covered by Giuliani et al. 2014b). Corundum host specific the Rhodope Mountains and is cut by the Nestos River indicators include mineral phase assemblages provid- valley and its tributaries (Karageorgiou et al. 2010). ing valuable constraints on the metamorphic condi- The tectonic and metamorphic record of this Northern tions and P-T paths of corundum formation. Host rock Aegean region (including the RMC) reflects the

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 PARANESTI PARGASITE SCHIST HOSTED RUBIES 537

FIG. 1. Geological map of the Rhodope Mountain Complex, with Paranesti located within the Nestos Shear Zone (red star) (adapted from Turpaud & Reischmann 2010).

Middle Jurassic to Neogene northeast-dipping subduc- intermediate ductile mylonitic zones based on major tion and convergence of the African-Eurasian plates thrust events. This melange zone, which contains both which resulted in the closure of the Tethys Ocean high pressure (HP) and ultra-high pressure (UHP) (Bonev et al. 2006, Krenn et al. 2008). The RMC lies relicts and amphibolites, represents the Nestos Shear within the Eastern Mediterranean sector of the Alpine- Zone (Turpaud 2006), in which the corundum-bearing Himalayan orogenic system. The tectonic units of the amphibole schists of this present study are located. RMC generally comprise rocks of continental or The Nestos Shear Zone (NSZ) is a major high- mixed continental-oceanic origin and form a large strain zone separating the Lower and Upper metamor- part of northern Greece and southern Bulgaria (Dinter phic terranes. It is traceable for about 100 km from 1998, Ricou et al. 1998, Barr et al. 1999). The Xanthi in the east to the Bulgarian border in the west complex north-dipping nappe stacking of the RMC (Fig. 1). The Lower unit is commonly interpreted as a developed during the Cretaceous with at least 900 km Mesozoic platform on top of Variscan continental of continental lower crust and lithosphere subducted crust and displays an Eocene to Oligocene metamor- underneath the Aegean Sea (van Hinsbergen et al. phic overprint at upper greenschist to lower amphib- 2005). The poly-metamorphic history of the RMC has olite facies conditions. The Upper unit displays been the subject of much research. Four distinct high- regional upper amphibolite facies metamorphism, pressure metamorphic (HPM) to ultra-high pressure and several locations contain well-preserved eclogites metamorphic (UHPM) events were reported by Liati (Kolcevaˇ et al. 1986, Mposkos & Krohe 2000, 2006, (2005): (1) ca. 149 Ma (U)HPM; (2) ca.73Ma Krohe & Mposkos 2002). Microdiamond-bearing (U)HPM; (3) ca. 51 Ma HPM; and (4) ca.42Ma ultrahigh-pressure (UHP) rocks are found in the NSZ HPM. and interpreted to be a suture zone with subduction and Burg et al. (1996) proposed the subdivision of the exhumation of the UHP rocks and terrain accretion Rhodope into Lower and Upper Terranes separated by during the Mesozoic (Nagel et al. 2011). The NSZ was

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 538 THE CANADIAN MINERALOGIST

FIG. 2. Google Earth image showing the location of the Paranesti ruby deposits (PAR-1, red star; PAR-5, purple star) (Google Earth). GPS (PAR-1: 418 170 03.5600 N, 248 260 03.4800 E; PAR-5: 418 160 58.2200 N, 248 260 08.9700 E).

proposed by Nagel et al. (2011) to represent a mid- formation of biotite at the expense of white mica and crustal mylonite horizon into which brittle and brittle- garnet occurred at conditions of about 700 8C/8–10 ductile extensional faults in the hanging wall root. This kbar (Schmidt et al. 2010). extension at the surface corresponds to an event of In addition, meta-ultramaﬁc bodies have been crustal accretion and associated decoupling of the found in the amphibolite-marble series of the Xanthi downward slab from the thickened continental crust area (~50 km east of Paranesti) and characterized as above. This detachment triggered slab retreat and the metamorphic peridotite (tectonite) and possibly repre- rise of asthenospheric mantle into the gap above the sent dismembered ophiolite fragments (Dimadis et al. slab, thus explaining magmatic events in the Early 1990). Oligocene (e.g., Brun & Faccenna 2008, Jolivet & Brun 2010). LOCAL GEOLOGY Metamorphic zircons from metabasite and meta- Physical locations and dimensions pelites of the Kimi Complex (e.g., rocks similar to those of the Nestos Shear Zone) have indicated that To date, corundum in the Paranesti area has been both the eclogite and metapelite underwent (U)HP found to outcrop on the surface at one hillside (PAR-1) metamorphism together in the Jurassic, followed by and one roadside (PAR-5) location to the west of decompression and high-temperature (e.g., granulite) Perivlepto village (Fig. 2). This area is known to metamorphism at around 170–160 Ma (Bauer et al. individual prospectors and collectors with no com- 2007). Further metamorphism is recorded in the mercial mining of the corundum to date. A sample of eclogite possibly at ca. 115 Ma and at ca. 79 Ma for coarse-grained amphibole schist with subhedral co- the amphibolite facies event (Bauer et al. 2007). Ion rundum to 20 mm is shown in Figure 3E; they vary in probe (SHRIMP) zircon dating of sapphirine-bearing color from a pale pink to medium red to deep red and kyanite eclogites from the Thermes area by Moulas et are of cabochon ruby quality. The corundum grains al. (2013) indicates an Eocene age (42 6 2 Ma) for the range in size from ,1 mm up to 50 mm, with an end of amphibolite facies overprint. The microdia- average size of 5–10 mm. The corundum form mond-bearing garnet-kyanite-mica schist found at relatively ﬂat tabular grains, with their basal planes Sidironero, 50 km west of Paranesti, provides further paralleling the orientation of the main regional evidence for a former UHP metamorphic event in the foliation (Fig. 4E). Most grains are opaque to Rhodope Terrane (Schmidt et al. 2010). This high translucent with only rare transparent zones. Solid pressure eclogite assemblage [garnet þ clinopyroxene mineral inclusions in the corundum are rare (see þ biotite þ quartz ( þ phengite)], with phengite being below). The corundum grain size from PAR-1 completely retrogressed to biotite in the present rock, (average 10 mm and up to 20 mm) is generally greater stabilized at 750 8C and 22 kbar. Retrogressive than PAR-5 (average 5 mm and up to 10 mm). In

FIG. 3. Local geology of PAR-1 and PAR-5. (a) Original ruby discovery in road-cut along the main road – PAR-5. (b) Intrusive contact between amphibole schist and granite pegmatite at the main ruby workings marked by red outlines – PAR-1. (c) Ruby bearing amphibolite schist boudins along the road from PAR-5 towards Perivlepto village. The white layers above the boudin are thin marbles. (d) Kyanite schist at the road-side deposit PAR-5. (e) Coarse-grained amphibole schist with subhedral rubies to 20 mm. (f) Cluster of gem quality rubies ~1 cm in diameter, showing the range of colors found.

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 540 THE CANADIAN MINERALOGIST

FIG. 4. Ruby inclusions, reaction rims, and preferred alignment. (a) Euhedral color-zoned ruby grain (PAR-1). (b) Euhedral spinel inclusions in ruby (PAR-1). (c) Concentration of spinels along more intensely colored zone (PAR-1). (d) Margarite rim around ruby (PAR-5). (e) Ruby aligned parallel to main foliation. (f) Subhedral kyanite close to ruby.

hand-specimen, more PAR-1 rubies are observed with The PAR-1 locality is also conspicuous by the a deeper red color than for PAR-5. The corundum occurrence of a prominent white-colored quartz- grains in polished thin-section appear to be more feldspar-mica-garnet pegmatite (Fig. 3B) which in- fractured, ﬁner-grained, anhedral and twinned in PAR- truded into the corundum-bearing amphibole schists 1 than in PAR-5. During the present study, inclusions (subsequent to corundum genesis). This intrusive in corundum were only found to occur in PAR-1. pegmatite has previously been documented in the

central RMC as 65–63 Ma pegmatite veins cutting Samples were selected across the spectrum of rock amphibolitized eclogites in eastern Rhodope (Baziotis types found from the footwall and hanging wall of the et al. 2007). Furthermore, an Rb-Sr isochron age of corundum-bearing amphibole schists, along with 65.4 6 0.7 Ma is given for another crosscutting samples of similar-looking boudins of amphibole pegmatite (Liati et al. 2002). This intrusion event post- schists (but with no visible corundum) from outcrops dates the amphibolite facies during which the a few hundred meters and a few kilometers to the corundum is suggested to have formed. Hence the north. pegmatite intrusion bears no relationship to the genesis of the Paranesti corundum. Petrographic analysis

Geological characteristics of PAR-1 Petrographic analysis was performed using the Leica DM2500-P research polarizing microscope, PAR-1 comprises the main corundum workings which involved both transmitted light and reflected approximately 300 meters down a moderately steep light for thin-sections and polished thin-sections. hillside (Fig. 3B). The corundum is found within Photomicrographs were obtained through the Leica amphibole schist that occurs as ovoid-shaped boudin- DFC290 digital camera system. like lenses up to 5 m in length but generally only 1–1.5 m across. These boudins are elongated parallel to the Whole-rock XRF, ICP-MS, and XRD analysis main regional foliation (Fig. 3C). These are immediately surrounded by a narrow zone (generally ,0.3 m Whole rock X-ray fluorescence (XRF) analysis of across) of highly schistose clinochlore schist (Fig. 3C) the corundum-bearing amphibole schist (sample PAR- and rare small boudins of corundum-kyanite-amphi- 1) was undertaken within the Mark Wainright bole schists also occur. The corundum-bearing zone at Analytical Centre at UNSW, Australia. The instrument the main workings covers an area of approximately 50 detection limits for each of the major oxides and trace (length) 3 10 (width) m. elements are listed in Supplementary Data Table 1 (available from the Depository of Unpublished Data Geological characteristics of PAR-5 on the MAC website, document pargasite-schist ruby CM55-4_10.3749/canmin.1700014). PAR-5 is the original corundum discovery along For detailed trace element analysis, the crushed and the roadside on top of the ridge (Fig. 3A) close to milled sample was then analyzed at the School of Perivlepto village. PAR-5 is approximately 300 m east Earth Sciences, University of Melbourne using the of PAR-1; the footwall consists of thinly-banded Agilent 7700x inductively coupled plasma - mass (0.05–0.002 m) amphibolites and a thin shell of spectrometer (ICP-MS) following the procedure out- clinochlore (,0.3 m across) surrounding boudins of lined in Eggins et al. (1997). The detection limit of the corundum-bearing amphibole schist. The corundum- trace elements analyzed is listed in Supplementary bearing boudins are approximately 1 3 0.3 m in size. Data Table 2. The hanging-wall consists of kyanite-bearing quartz- X-ray diffraction (XRD) analysis on (3 PAR-1, 4 plagioclase-amphibolite gneisses and rarer kyanite- PAR-5, and 3 host lithologies) samples was undertak- amphibole-chlorite schists (Fig. 3D). Sheared contacts en at the Solid State & Elemental Analysis Unit occur between all rock types. Additionally, several 2 UNSW Analytical Centre using the PANalytical boudins occur within 10 km (including some further Empyrean II X-Ray Diffraction System with Co anode north along the same road), though none of these were X-ray tube. The samples were prepared using mortar found to contain any visible corundum. The corun- and pestle as well as a UNSW Rock Labs standard dum-bearing zone at the roadside workings covers an tungsten carbide ring mill into fine powders (~50 lm). area approximately 30 (length) 3 3 (width) m. Samples were packed into 20 mm aluminum holders (300 mg) and gently pressed in the front of the holder SAMPLES AND ANALYTICAL METHODS with a glass slide to reduce the effects of preferred Samples orientation. The operating conditions were 45 kV and 40 mA, with a 2h scan angle range of 5–708 and a step- Samples were collected in situ from both localities size of 51 s. These XRD patterns were then interpreted and comprised corundum enclosed within their host using XPLOT 32 and X’pert Highscore. amphibole schists, looser corundum from the surface which had weathered out of their host amphibole SEM and EMPA analysis schists, large pure masses of chlorite from the chlorite schist zones, kyanite-bearing amphibole schists, and Backscattered electron (BSE) images of polished loose kyanite grains which had weathered out of it. and carbon-coated corundum and amphibole samples

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 542 THE CANADIAN MINERALOGIST

were taken using Hitachi S3400 SEM at the UNSW 5). The only noticeable primary inclusions within the Electron Microscopy Unit (EMU), UNSW, Australia corundum grains were rare spinels (in PAR-1 samples, using secondary electron (SE) detectors. The compo- Fig. 4A–C). Color zoning is rare (Fig. 4A and C) and sition of selected grains was then determined using a only found within the one corundum grain from PAR- JEOL 8500F Electron Probe Microanalyzer, with 5 1. Most of the corundum grains exhibited a distinctive wavelength dispersive spectrometers and 2 energy margarite rim with continuous optical properties (Fig. dispersive spectrometers. Operational conditions were 4D). This material was later examined under SEM and 15 kV accelerating voltage, 20 nA beam current, and 1 EMPA and confirmed to be margarite. Euhedral lm beam size for corundum samples and 5 lm beam kyanite grains occur close, but not adjacent, to the size for amphiboles. corundum grains and are shown in Figure 4F and are found at both PAR-1 and PAR-5. No zircon, monazite, LA-ICP-MS or the more common rutile inclusions were identified in these samples, and hence in situ dating of the age of Six crystals from PAR-1 and 5 crystals from PAR-5 formation for the Paranesti corundum samples could covering the spectrum of colors from the occurrences not be undertaken. were embedded in polished mounts and used for the Samples from above the footwall in the layered LA-ICP-MS trace element analysis. The LA-ICP-MS kyanite-bearing amphibole gneiss with no corundum analysis was performed using an Excimer 193 nm Ar- association have sillimanite fringes around the amphi- F gas laser coupled with an Agilent Technologies 7700 boles (Fig. 5A), muscovite replacing amphibole (Fig. Series quadrupole inductively coupled plasma mass 5D), and muscovite intergrown with Mg-rich chlorite spectrometer (ICP-MS). Ablation was performed in a (Fig. 5E). Metamorphic episodes are evident through He atmosphere with a Mass Flow Control (MFC) 1 of clinozoisite overprinting plagioclase (Fig. 5B), amphi- 0.465 L/min and MFC 2 of 0.36 L/min. Ar was used as bole overprinting plagioclase (Fig. 5C), and micro- the carrier gas with MFC of 1.02 L/min. NIST610 and shear zones containing abundant epidote (Fig. 5F) NIST612 were used as primary standards and BCR2G from the footwall amphibolites. How these mineral as the secondary standard. The laser was configured to assemblages and relationships provide evidence for the perform two shots of pre-ablation on a 110-lm spot- metamorphic evolution of the corundum occurrences size at 5 Hz followed by a 30 s pause and 59 s of will be discussed in later sections. background gas, and then pulsed at 5 Hz on an 85-lm spot-size for 2 min with a burst count of 595 shots and Whole rock geochemistry fluence of 9.28 J/cm2. Cell voltage and pressure were 15 kV and 6490 mbar, respectively, with an overall Whole rock XRF geochemistry (Table 1) suggests energy reading of 5.5 mJ. The internal Al standard an ultramafic rather than mafic igneous precursor for the value was set at 98 wt.% (519,665 ppm), to allow for corundum-bearing amphibole schists. Chromium is maximum trace element contents. Spot analyses were prominent at 10745 ppm, while Ni (390 ppm), Co (60 performed at core and rim positions on each of the ppm), and Zn (250 ppm) are relatively high and all grains, with particular emphasis on color-zoned other trace elements are noticeably low (Table 1). The regions. Refer to Supplementary Data Table 4 for whole rock ICP-MS analyses (Table 2) support the XRF further information on the samples and methods. trace element analyses with the same high Cr (11800 ppm), high Ni (450 ppm), and Zn (320ppm) contents. RESULTS XRD analysis identified the corundum host assemblages of PAR-1 and PAR-5 as consisting of the same Petrography mineral phases (pargasite, corundum, clinochlore, and nimite), with pargasite being the dominant amphibole The corundum crystals exhibited a wide range of and clinochlore being the dominant chlorite. In habits. Most were heavily fractured or brecciated, and addition, tremolite was found in PAR-1 but not in were elongate and aligned parallel to the main PAR-5. The host rock non-corundum bearing assem- foliation within the host rock (Fig. 4E). Some rare blages (i.e., footwall and hanging wall lithologies) also corundum grains were euhedral. The corundum- contain pargasite as the main amphibole along with co- bearing rocks mostly comprise pargasite, corundum, existing anorthite and clinozoisite. chlorite (mostly clinochlore, rarely nimite), margarite, and in some cases, tremolite or monazite. In contrast, EMPA analyses the host rocks (which have sheared contacts with the corundum-bearing schists) comprise mostly pargasite The chemical composition of the Paranesti corun- with varying amounts of anorthite, clinozoisite, dum using the EMPA method is shown in Table 3. A chlorite, and monazite (Supplementary Data Table total of 38 points from PAR-1 (12 crystals) and 20

FIG. 5. Petrographic images of surrounding host rock mineral assemblage with no ruby association (PPL ¼ plane polarized light; UXP ¼ under crossed polarized light). (a) Sillimanite fringes around amphibole (Am-amphibole, Si-sillimanite). Two color amphiboles. (b) Clinozoisite overprinting plagioclase (Pl-plagioclase, Cl-clinozoisite). (c) Amphibole overprinting plagioclase (Am-amphibole, Pl-plagioclase). (d) Muscovite replacing amphibole (Mu-muscovite, Am-amphibole). (e) Muscovite intergrowth with Mg rich chlorite (Mu-muscovite, Mg-Cl-Mg rich chlorite). (f) Micro shear zone with ﬁne- grained epidote (Ep-epidote).

points from PAR-5 (6 crystals) were analyzed and enriched in Cr compared to those from the road-cut show that the Paranesti corundum are enriched in (in sample (PAR-5). Variations are shown in the concen- decreasing order of abundance) Cr, Fe, and Si and that trations of Cr (Cr2O3: PAR-1, 0.11–1.68 wt.%; PAR-5, corundum from the main site (PAR-1) are more 0.13–0.29 wt.%), Fe (FeO: PAR-1, 0.19–0.73 wt.%;

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 544 THE CANADIAN MINERALOGIST

TABLE 1. MAJOR AND TRACE ELEMENT XRF TABLE 2. WHOLE ROCK ICP-MS TRACE ELEMENT ANALYSES OF THE RUBY-BEARING HOST ROCK ANALYSIS OF THE RUBY-BEARING HOST ROCK (SAMPLE PAR-1) (SAMPLE PAR-1)

Major Oxides wt.% Trace Elements ppb

SiO2 40.12 Ca 80,231,522 TiO2 0.09 Sc 2,376 Al2O3 20.96 Ti 326,720 Fe2O3 5.66 V5 45,381 MnO 0.15 Cr 11,784,253 MgO 15.92 Co 59,725 CaO 9.95 Ni 447,696

Na2O 1.25 Cu 8,345 K2O 0.46 Zn 318,113 P2O5 0.01 Ga 9,264 SO3 0.05 As 606 LOI 4.55 Rb 19,672 Total 99.17 Sr 46,021 Y 476 Trace Elements ppm Zr 839 Nb 276 Pb 4 Mo 1,212 Sr 46 Cd 32 Rb 16 Sn 465 Y3 Sb 302 Zr 12 Cs 575 Zn 252 Ba 44,409 Ni 387 La 90 Ga 8 Ce 235 Co 63 Pr 36 As 3 Nd 176 Ba 55 Sm 55 Cr 10745 Eu 62 V55 Gd 68 Tb 12 Dy 76 PAR-5, 0.18–0.36 wt.%), and Si (SiO2: PAR-1, 0.01– Ho 16 0.37 wt.%; PAR-5, 0.01–0.22 wt.%). Er 48 SEM imaging (Fig. 6) and EMPA analysis (Table Tm 8 4) of spinel inclusions within the corundum grains Yb 55 show that they are dominantly composed of Al, Cr, Fe, Lu 9 and Mg (with Sp5 also being enriched in Zn, 7.1 wt.%) Hf 20 and that their composition varies widely, even though Ta 16 the grains are in close proximity. On spinel discrim- W 1,190 ination diagrams (Fig. 7), the spinel inclusions plot in Tl 98 the ﬁelds for Al-chromite (1), picotite (4), and with Pb 1,985 one result bordering hercynite. Considering the Mg-Fe Th 9 exchange, Figure 7B shows that the majority of the U24 results plot close to pleonaste with one plotting close Trace Element Ratios to hercynite. Nb/Ta 17.3 Some six points on the phyllosilicate reaction rim Zr/Hf 41.5 around the corundum grains from PAR-5 were Y/Ho 29.2 analyzed and found to be margarite (Table 5). One Rb/Sr 0.43 margarite analysis (Mar4) appears to have lower concentrations of SiO2 (23.4 wt.% versus other 5 points of 29 wt.%), lower CaO (8.9 wt.% versus 12 wt.%), and higher Al2O3 (60.6 wt.% versus 48 wt.%).

TABLE 3. REPRESENTATIVE EMPA ANALYSES OF PARANESTI RUBY GRAINS

PAR-1 PAR-5 wt.% Min Max l (mean) r2 (std dev) Min Max l (mean) r2 (std dev)

SiO2 0.01 0.37 0.08 0.07 - 0.22 0.07 0.05 Na2O - 0.03 0.00 0.01 - 0.01 0.00 0.00 K2O - 0.01 0.00 0.00 - 0.01 0.00 0.00 TiO2 - 0.01 0.02 0.03 - 0.06 0.01 0.02 MgO - 0.05 0.01 0.01 - 0.01 0.00 0.00

P2O5 - 0.04 0.01 0.01 - 0.03 0.01 0.01 CaO - 0.05 0.01 0.01 - 0.05 0.01 0.01

Cr2O3 - 1.68 0.55 0.45 - 0.29 0.14 0.07 Al2O3 96.83 100.05 98.58 0.90 97.74 99.09 98.32 0.38 MnO - 0.05 0.01 0.01 - 0.03 0.01 0.01 FeO 0.19 0.73 0.40 0.15 0.18 0.36 0.28 0.04

Ga2O3 - 0.04 0.01 0.01 - 0.04 0.01 0.01 Nb2O5 - 0.07 0.01 0.01 - 0.03 0.01 0.01 Ta2O5 - 0.10 0.02 0.03 - 0.10 0.02 0.03 ZnO - 0.07 0.02 0.02 - 0.07 0.02 0.03 Total 98.21 100.60 99.71 0.55 98.43 99.65 98.90 0.35 Calculated cations assuming stoichiometry Si 0.00 0.00 Ti 0.00 0.00 Al 3.97 3.96 Cr 0.01 0.00 Fe2þ 0.01 0.01 Mn 0.00 0.00 Mg 0.00 0.00 Ca 0.00 0.00 Na 0.00 0.00 K 0.00 0.00

Total of 38 points analyzed for PAR-1 rubies. Total of 20 points analyzed for PAR-5 rubies.

FIG. 6. SEM and EMPA image of inclusions within ruby grains in sample PAR-1. (a) Close-up view of spinel inclusions in the ruby grain. (b) Spinel inclusions with a range of chemical compositions between the grains.

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 546 THE CANADIAN MINERALOGIST

TABLE 4. REPRESENTATIVE ELECTRON- This contrast in results could be due to the proximity MICROPROBE ANALYSES OF SPINEL INCLUSIONS of the analysis point to the corundum grain, resulting WITHIN RUBY, TOTAL OF 5 POINTS in apparently enhanced Al2O3. The consistency of all other 5 points shows the representative data of the Oxide (wt.%) Sp1 Sp2 Sp3 Sp4 Sp5 margarite from the Paranesti occurrences.

SiO2 0.01 0.02 0.04 0.01 0.13 Some thirty-one points on the amphiboles from Na2O 0.01 - - - - PAR-1 were probed. They are clearly Mg-rich (Table K2O -----6) with only minor variation in their composition, TiO2 0.01 0.11 - 0.04 - primarily in Fe, Al, Si, Mg, and Ca. The amphiboles MgO 13.60 8.54 9.89 7.14 1.91 are low in Ti, Mn, K, and F. On the amphibole P2O5 - 0.02 0.04 - 0.02 classification diagram for the compositional range of CaO - - - - 0.05 pargasite–ferropargasite–hastingsite–magnesiohasting- Cr2O3 20.33 16.22 22.26 18.77 30.94 site (Fig. 8), they all plot close to the pargasite Al O 42.64 58.31 42.92 53.85 31.77 2 3 endmember. MnO 0.28 0.19 0.46 0.29 1.26 The chemical composition of the surrounding FeO 19.01 14.93 21.80 18.40 25.26 chlorite at PAR-1 using EMPA is shown in Table 7. Ga2O3 - - 0.01 0.07 0.12 These values agree with clinochlore (peninite) as Nb2O5 - 0.02 0.01 - 0.04 identified in XRD. These results are plotted in a Ta2O5 - - - - 0.02 ZnO 1.14 0.79 1.37 0.61 7.09 chlorite classification diagram (Fig. 9). Total 97.03 99.15 98.80 99.18 98.60 LA-ICP-MS analyses Calculated cations assuming stoichiometry Si 0.00 0.00 0.00 0.00 0.00 The chromophore and trace element LA-ICP-MS Ti 0.00 0.00 0.00 0.00 0.00 results from PAR-1 and PAR-5 (6 crystals and 23 Al 1.46 1.92 1.48 1.81 1.22 points and 5 crystals and 17 points, respectively; Table Cr 0.47 0.36 0.51 0.42 0.80 3þ 8A) show that the Paranesti corundum samples show a Fe 0.08 0.00 0.00 0.00 0.00 range of values: Ca (650–2100 ppm), Cr (4–8600 Fe2þ 0.38 0.35 0.53 0.44 0.69 ppm), Fe (1600–3800 ppm), and Si (780–2500 ppm). Mn 0.01 0.00 0.01 0.01 0.03 They have very low concentrations of V (max. 5 ppm), Mg 0.59 0.36 0.43 0.30 0.09 Zn 0.02 0.02 0.03 0.01 0.17 Ga (29 ppm), and Mg (mostly below detection limit but with a few exceptions up to 380 ppm). These results are in line with the EMPA analyses. The maximum Cr value (8600 ppm) for a deep red ruby from PAR-5 is unusually high. Results for the other trace elements analyzed by LA-ICP-MS are shown in

3þ 3þ 3þ FIG. 7. (a) Triangular spinel classiﬁcation diagram (Cr –Fe –Al ) (adapted from Gawlick et al. 2015). (b) Binary classiﬁcation diagram considering the Mg2þ–Fe2þ exchange (adapted from Gawlick et al. 2015).

TABLE 5. REPRESENTATIVE ELECTRON- TABLE 6. REPRESENTATIVE EMPA ANALYSES OF MICROPROBE ANALYSES OF MARGARITE – AMPHIBOLES SURROUNDING RUBY GRAINS REACTION RIM AROUND RUBY GRAINS, TOTAL OF 6 WITHIN THE RUBY-BEARING AMPHIBOLE SCHISTS POINTS PAR-1 (31 POINTS)

Oxide Pargasite l r2 (wt.%) Mar1 Mar2 Mar3 Mar4 Mar5 Mar6 Oxide (wt.%) Min Max (mean) (std dev) SiO 29.09 29.60 29.61 23.42 28.66 29.31 2 SiO 44.81 47.75 46.03 0.78 Na O 1.13 0.65 0.50 0.58 0.67 0.71 2 2 TiO - 0.26 0.06 0.06 K O 0.12 0.84 0.20 0.44 0.25 0.44 2 2 Al O 13.00 15.82 14.64 0.83 TiO 0.05 - - 0.02 0.04 0.05 2 3 2 FeO 5.24 8.89 5.94 0.99 MgO 0.12 0.10 0.11 0.05 0.09 0.10 MnO 0.06 0.24 0.17 0.04 P O 0.01 0.02 0.01 - 0.02 0.01 2 5 MgO 15.69 17.62 16.87 0.53 CaO 11.70 11.59 12.08 8.90 12.26 11.72 CaO 11.07 12.23 11.94 0.33 Cr O 0.05 - 0.02 0.06 - - 2 3 BaO - 0.07 0.01 0.02 Al O 48.74 48.40 49.45 60.58 49.82 48.46 2 3 Na O 1.52 1.87 1.71 0.10 MnO - 0.01 - 0.01 - - 2 K O 0.23 0.37 0.29 0.03 FeO 0.11 0.11 0.18 0.18 0.16 0.14 2 F - 0.17 0.03 0.05 Ga O ------2 3 H O - 3.59 2.31 1.05 Nb O - - - 0.01 - - 2 2 5 Total 99.93 100.00 99.99 0.02 Ta2O5 0.09 - 0.07 - 0.02 0.07 ZnO 0.01 - - 0.07 - 0.06 Calculated cations assuming stoichiometry Total 91.22 91.31 92.22 94.33 91.98 91.06 Si 8.00 8.00 8.00 - Ti 0.00 0.03 0.01 0.01 Calculated cations assuming stoichiometry Al 1.08 1.37 1.24 0.08 Si 2.02 2.06 2.06 1.63 1.99 2.04 Fe2þ 0.75 1.29 0.86 0.14 Ti 0.00 - - 0.00 0.00 0.00 Mn 0.01 0.04 0.02 0.01 Al 3.99 3.96 4.05 4.96 4.08 3.97 Ca 2.92 3.25 3.14 0.10 Fe2þ 0.01 0.01 0.01 0.01 0.01 0.01 Ba 0.75 0.85 0.81 0.03 Mn - 0.00 - 0.00 - - Na 0.51 0.64 0.58 0.04 Mg 0.01 0.01 - 0.00 - - K 0.05 0.08 0.06 0.01 Ca 0.87 0.86 0.90 0.66 0.91 0.87 Na 0.15 0.09 0.07 0.08 0.09 0.09 K 0.01 0.07 0.02 0.04 0.02 0.04 colored varieties. No noticeable variance based on core/rim positions was observed. Table 8B. The results show that, other than being Corundum crystals from PAR-1 show a higher enriched in Cr, Fe, Si, and Ca, Paranesti corundums Cr2O3 content than PAR-5 under EMPA (average 0.55 are low in Mg, Ti, and Ga, and most of the other trace versus 0.14 wt.% and maximum 1.68 versus 0.29 elements are below the detection limit. wt.%, respectively). However, the difference under For the elements that were analyzed by both EMPA LA-ICP-MS for Cr shows a higher maximum in PAR- and LA-ICP-MS techniques, the results show a general 5 (8600 ppm) versus PAR-1 (2900 ppm). The EMPA consistency. Core/rim analysis on average for all and LA-ICP-MS results were obtained from different samples combined did not show any significant samples (i.e., thin sections for EMPA versus pucks for variation. When color variation is taken into account, LA-ICP-MS). Therefore, the color variation may have on average the Fe contents at the rim for the paler been easier to notice in the thicker puck samples than corundums are ~200 ppm lower than at the core. The in the thinner thin section samples. The difference is paler PAR-1 samples show similar maximum and also likely due to the individual grains analyzed and average Fe contents to medium colored PAR-1 with the color variance between the samples. The highest less variance in the range. Titanium values are Cr sample analyzed (8600 ppm) showed a deep red marginally higher in medium colored PAR-1 samples color, as seen in Figure 10B. PAR-5 shows a higher Fe compared to the paler samples (70 versus 20 ppm) range (1800–3800 ppm) than PAR-1 (1600–2700 with no noticeable difference between core and rim. ppm). PAR-1 shows a slightly higher Ca content No noticeable trace element contents were observed (770–2100 ppm) than PAR-5 (650–1900 ppm). between the core/rim results for PAR-5. As expected, Both Paranesti corundum sets have low Ti, V, Ga, however, Cr is significantly higher in the darker and Mg values, mostly below the detection limit of the colored corundums compared to the medium and pale LA-ICP-MS. The low values result in magnified ratios

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 548 THE CANADIAN MINERALOGIST

FIG. 8. Diagram showing the range in composition of host amphiboles from the ruby-bearing amphibole schists. Note that all plot close to the pargasite endmember. Assumes 20% Fe3þ and 80% Fe2þ (adapted from Colville et al. 1966).

due to the close to zero denominations when compared Mg(3100)–Fe–Ti(310) to help categorize Fe-rich against known ratios in the elemental graph (Fig. 11). magmatic and Mg-rich metamorphic fields. The Para- The sample plots are therefore more skewed towards nesti corundums are shown on these various meta- one axis. morphic versus magmatic diagrams via a range of Elemental graphs based on LA-ICP-MS results trace element discrimination methods (Fig. 11). Ga/ show a general overlapping trend between PAR-1 and Mg was originally used to differentiate between blue PAR-5 (Fig. 11, C, and E). However, two clusters for sapphires of magmatic versus metamorphic origin PAR-5 appear on the FeO–Cr2O3–MgO–V2O3 dia- (Peucat et al. 2007). It has been proposed that the high gram (Fig. 11E). The paler and medium colored PAR- Mg content in corundum is due to fluid circulation 5b samples fall close to a value of zero on the y axis, during a metamorphic stage where metasomatic together with all PAR-1 corundum in a region of the exchange occurs between mafic and adjacent Si-rich diagram where the metasomatic domain overlaps that acidic rocks. Ga contents .100 ppm and Cr2O3/Ga2O3 of the corundum in the mafic-ultramafic domain, while ratios ,1 thus typify corundum of magmatic origin, the dark red colored PAR-5a rubies form a distinctive and Ga contents ,100 ppm and Cr2O3/Ga2O3 ratios cluster to the left of zero with distinct negative values .1 are more characteristic of corundums of metamor- on the FeO-Cr2O3-MgO-V2O3 axis, clearly in the phic origin (Saminpanya et al. 2003). Transitional mafic/ultramafic domain. values across this magmatic/metamorphic relationship may represent metasomatic processes (Sutherland et DISCUSSION al. 2002). Geochemical signatures Both PAR-1 and PAR-5 corundums demonstrate a linear trend in the chromophore trace elements. The A range of studies have used chromophore elemental diagrams Cr310–Fe–Ga3100 (Fig. 11A), Fe– elements (Fe, Cr, Ti, V) and genetic indicators (Ga Mg3100–Ti310 (Fig. 11B), Ga/Mg versus Fe/Mg (Fig. and Mg) to distinguish corundum from different 11C), and Fe/Ti versus Cr/Ga (Fig. 11D) based on LA- primary sources using elemental diagrams (e.g., ICP-MS results, show that the Paranesti corundums fall Peucat et al. 2007, Sutherland et al. 2009, 2014, within a general metamorphic origin. Further differen- Giuliani et al. 2014b). These include triangular plots of tiation of the metamorphic field shows that the dark-red

TABLE 7. REPRESENTATIVE EMPA ANALYSES OF CHLORITE FROM THE ZONE SURROUNDING THE RUBY-BEARING AMPHIBOLE SCHIST BOUDINS PAR-1 (20 POINTS)

Clinochlore Oxide (wt.%) Min Max M (mean) r2 (std dev)

SiO2 28.53 38.44 30.22 2.00 TiO2 0.01 0.39 0.05 0.08 Al2O3 22.03 22.94 22.49 0.29 FeO 6.24 10.69 7.14 1.27 MnO - 0.15 0.04 0.04 MgO 19.63 31.44 30.06 2.79 CaO - 0.27 0.02 0.07 BaO - 0.04 0.01 0.01

Na2O - 0.06 0.01 0.01 K2O - 1.97 0.12 0.44 F - 0.59 0.04 0.13

H2O 8.26 11.10 9.79 0.65 Total 99.75 100.00 99.98 0.06 Calculated cations assuming stoichiometry Si 2.71 3.48 2.79 0.17 Ti - 0.03 - 0.01 Al 2.38 2.49 2.45 0.03 Fe3þ - 0.12 0.04 0.03 Fe2þ 0.43 0.84 0.52 0.10 Mn - 0.01 - - Mg 2.65 4.32 4.14 0.39 FIG. 9. Classification diagram for chlorite from the Paranesti Ca - 0.03 - 0.01 deposit (adapted from Hey 1954). Na - 0.01 - - K - 0.23 0.01 0.05 F - 0.18 0.01 0.04 Comparison with rubies from deposits around the OH 7.82 8 7.99 0.04 world The Paranesti corundums show relatively high Si rubies from PAR-5a clearly fall in the field for ruby and Ca and very low Ga and V contents. Silicon values with a mafic-ultramafic origin, while the paler-colored of up to about 500 ppm were considered within the PAR-5b and most of the PAR-1 corundums fall in the solubility limits for the corundum structure, with overlapping region between ruby in mafic-ultramafic higher values signaling the presence of fine-scale rocks and metasomatic corundum (Fig. 11E). Most of inclusions (Harlow & Bender 2013). Nevertheless, the the corundums can clearly be attributed to a mafic- Paranesti Si range from 780 to 2450 ppm is well above the accepted solubility limits for the corundum ultramafic protolith, and this suggests that for the structure. No Si-bearing primary inclusions were Paranesti occurrences, the corundums formed in situ. found within the corundums during this study. The The low SiO (40.1 wt.%) and high Mg (15.9 wt.%) 2 possible source of the Si could potentially be attributed content shown by the whole-rock XRF analysis (Table to extremely fine-grained and overprinting margarite 1) also point to an ultramafic protolith. (reaction rim post corundum formation). This over- Overall, the geochemistry of PAR-1 and PAR-5 printing of margarite would explain both the high Si corundum crystals shows rather distinct differences and Ca content detected and is in line with the Si/Ca despite the sampling site proximity. The PAR-1 ratio from the LA-ICP-MS results. Alternatively, the occurrence has overall larger ruby grains which are high Ca content could be due to the leaching from the generally darker red in color than those from PAR-5 relict host amphibole (of pargasite-magnesiohasting- (albeit the deepest-colored rubies found within this site solid solutions) of Ca incorporated into the study came from PAR-5). This suggests that the two corundum during formation; this would suggest ruby occurrences formed under the same conditions but with formation after equilibration with the surrounding variable access to Fe and Cr during their formation. amphiboles. All the evidence shown above suggests

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 550 THE CANADIAN MINERALOGIST

TABLE 8A. CHROMOPHORE AND KEY TRACE ELEMENTS (Mg AND Fe) LA-ICP-MS ANALYSES (ppm), OF THE PARANESTI CORUNDUM AND THEIR RATIOS

PAR-1 Mg Ti V Cr Fe Ga Fe/Mg Ga/Mg Cr/Ga Fe/Ti 1M-R1 42 81 2 1,668 1,820 18 43 - 92 23 1M-C1 17 184 2 1,101 1,810 19 104 1 57 10 1M-C2 21 46 2 903 1,841 20 87 1 44 40 1M-R2 12 19 1 1,869 1,689 17 144 1 113 90 2P-R1 12 11 1 384 2,095 18 182 2 21 189 2P-C1 12 6 1 451 2,150 18 184 2 25 379 2P-C2 26 13 1 460 2,485 22 97 1 21 189 2P-R2 13 20 1 460 2,312 17 180 1 27 118 3P-R1 11 19 1 403 2,094 18 185 2 22 108 3P-C1 19 8 1 360 2,065 17 106 1 21 272 3P-C2 8 29 1 448 2,442 20 317 3 23 86 3P-C3 12 33 1 494 2,664 20 224 2 24 80 4M-R1 9 125 1 2,101 1,572 15 167 2 142 13 4M-C1 10 143 1 1,373 1,591 14 152 1 95 11 4M-C2 16 154 1 1,239 1,678 16 104 1 77 11 4M-R2 13 119 2 2,856 2,361 21 179 2 134 20 5M-C1 15 48 3 1,868 2,664 23 172 1 82 55 5M-C2 13 36 3 1,946 2,511 21 190 2 91 69 5M-C3 14 43 2 1,988 2,421 20 178 2 98 57 6M-R1 7 15 2 2,773 1,650 16 225 2 171 108 6M-C1 14 33 3 2,304 1,999 21 146 2 110 61 6M-C2 16 11 2 2,296 1,704 18 107 1 126 159 6M-C3 25 19 3 2,329 1,961 20 79 1 18 102

PAR-5 Mg Ti V Cr Fe Ga Fe/Mg Ga/Mg Cr/Ga Fe/Ti 1D-R1 82 69 5 8,627 3,822 29 46 - 299 56 1D-C1 78 73 5 8,157 3,800 28 49 - 296 52 1D-C2 76 62 5 8,272 3,762 27 50 - 310 60 1D-C3 61 56 4 6,614 3,008 22 50 - 301 54 2P-C1 42 45 3 646 2,431 18 58 - 35 54 2P-R1 23 30 2 524 2,479 18 108 1 30 83 2P-C1 376 46 3 417 2,643 19 7 - 22 58 3D-C1 11 16 2 4,173 1,969 19 175 2 219 124 3D-C2 8 22 2 5,483 1,980 19 246 2 291 88 3D-R1 10 17 3 7,548 1,987 15 195 2 489 116 3D-C3 26 13 3 3,961 2,075 17 78 1 230 164 4M-C1 12 10 2 1,522 2,170 17 188 1 89 227 4M-R1 12 16 2 1,369 2,206 17 179 1 79 142 4M-R2 9 14 2 1,652 2,294 18 246 2 91 165 5M-C1 15 152 2 172 1,833 13 119 1 13 12 5M-C2 44 153 2 4 2,163 14 49 - - 14 5M-R1 13 190 2 49 1,842 13 147 1 4 10

Polished mounts (position: C - core, R - rim; color intensity: P - pale, M - medium, D - dark).

that the corundums formed in situ and are either pre- which shows a prominent internal growth structure, the or just syn- to the major metamorphic event that Longido deposit of Tanzania (veins of anyolite zoisite- affected the region. amphibole), and the Mangari deposit of Kenya The Paranesti occurrences share similar character- (Mercier et al. 1999). Table 9 provides a comparison istics with a number of amphibolite-hosted maﬁc/ of the geological occurrence, estimated P-T condi- ultramaﬁc ruby deposits from Africa, including the tions, main paragenesis, and ruby chemical character- Ejeda-Fotadrevo deposit of southwest Madagascar, istics for selected African deposits and the Paranesti

TABLE 8B. OTHER TRACE ELEMENT LA-ICP-MS ANALYSES (ppm) OF THE PARANESTI CORUNDUM

PAR-1 Al B Na Si P K Ca Ni Cu Zn 1M-R1 518,665 10 2 2,456 47 6 817 1 ,0.087 BDL 1M-C1 518,665 11 3 1,805 21 11 812 10 ,0.101 3 1M-C2 518,665 8 ,1.02 1,823 26 24 982 1 BDL BDL 1M-R2 518,665 6 2 1,515 19 ,1.18 940 BDL ,0.051 BDL 2P-R1 518,665 7 ,1.35 1,743 19 14 926 1 ,0.107 BDL 2P-C1 518,665 9 ,1.31 1,892 15 6 1,069 2 1 ,0.118 2P-C2 518,665 7 2 1,663 19 6 1,484 BDL 1 ,0.181 2P-R2 518,665 10 1 2,332 22 ,1.58 852 BDL BDL BDL 3P-R1 518,665 7 10 2,045 16 5 1,033 BDL BDL 1 3P-C1 518,665 10 4 1,606 24 4 932 BDL BDL 1 3P-C2 518,665 7 2 1,676 17 ,2.70 972 BDL BDL BDL 3P-C3 518,665 7 ,1.27 1,532 24 ,2.39 1,319 BDL BDL BDL 4M-R1 518,665 6 5 1,689 21 ,2.27 769 BDL ,0.102 BDL 4M-C1 518,665 8 ,1.22 1,587 23 ,2.17 965 1 BDL ,0.167 4M-C2 518,665 6 ,1.09 1,523 24 ,1.96 1,170 1 1 3 4M-R2 518,665 4 1 1,113 31 ,1.40 1,900 BDL 1 ,0.107 5M-C1 518,665 3 4 1,286 20 2 2,119 BDL 2 ,0.112 5M-C2 518,665 3 2 956 21 4 1,855 BDL 3 BDL 5M-C3 518,665 3 1 999 19 2 1,815 BDL 2 BDL 6M-R1 518,665 3 ,1.18 1,276 18 ,1.66 1,017 BDL BDL BDL 6M-C1 518,665 3 1 781 13 ,1.18 1,820 BDL BDL BDL 6M-C2 518,665 5 12 1,279 20 ,1.42 1,360 BDL BDL BDL 6M-C3 518,665 6 13 1,008 12 1 1,590 1 1 ,0.098

PAR-5 Al B Na Si P K Ca Ni Cu Zn 1D-R1 518,665 3 ,0.70 837 10 ,0.97 1,903 1 BDL 1 1D-C1 518,665 3 ,0.73 959 13 ,1.02 1,902 1 1 BDL 1D-C2 518,665 4 ,0.71 875 10 ,0.99 1,756 1 BDL BDL 1D-C3 518,665 3 ,0.71 948 12 ,0.98 1,147 1 BDL BDL 2P-C1 518,665 6 3 1,130 21 5 1,434 1 ,0.31 3 2P-R1 518,665 5 ,0.81 993 11 ,1.09 1,654 1 1 1 2P-C1 518,665 5 30 2,123 20 177 1,663 2 BDL BDL 3D-C1 518,665 5 ,1.54 1,244 34 4 653 1 BDL 1 3D-C2 518,665 5 ,1.59 1,258 14 ,2.02 909 BDL BDL BDL 3D-R1 518,665 7 ,1.65 1,061 15 ,2.11 954 1 BDL ,0.163 3D-C3 518,665 6 ,1.56 1,371 24 ,2.02 809 1 ,0.139 ,0.206 4M-C1 518,665 4 ,1.96 1,313 22 ,2.55 734 1 ,0.119 ,0.253 4M-R1 518,665 5 3 1,396 20 ,2.22 1,048 1 ,0.155 ,0.179 4M-R2 518,665 4 3 1,269 25 ,2.09 1,394 1 BDL BDL 5M-C1 518,665 3 ,1.61 1,342 20 ,2.06 737 1 ,0.143 ,0.152 5M-C2 518,665 5 ,1.64 1,193 32 ,2.10 765 1 1 2 5M-R1 518,665 4 ,1.57 1,290 16 ,1.97 1,031 BDL BDL BDL

Note: Polished mounts (Position: C - core, R - rim; Color intensity: P - pale, M - medium, D - dark). BDL - below detection limit.

occurrences. The similarities in geochemistry of the well-foliated gneisses, indicative of upper amphibolite rubies provide further support for an ultramaﬁc origin to granulite facies conditions (Schwarz et al. 2008). for the Paranesti rubies, as well as guidelines on the The corundum crystals are embedded within dark- possible P-T conditions. colored amphibolite with accessory Cr-spinel, mica, The closest ruby deposit type in the worldwide kyanite, and allanite. The possible lithologies for the literature database to date is the Tanzanian Winza ruby protolith include high-alumina layered gabbro or deposit. The main rock types here are migmatitic and leucogabbro. Spinel that overgrew, or is included in,

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 552 THE CANADIAN MINERALOGIST

FIG. 10. Numbered ruby grains in grain mounts analyzed using LA-ICP-MS. (a) PAR-1 (6 grains). (b) PAR-5 (5 grains).

the corundum is Fe- and Mg-rich. Similar spinel signature of high Cr, Si, and Fe but low Ti and Ga. In inclusions have been found within the Paranesti rubies. addition, Paranesti corundum is also low in Mg. The average Cr content of Paranesti corundum from PAR-1 and PAR-5b are comparable to other world Possible P-T model for the formation of the Paranesti corundum occurrences. However, PAR-5a (Fig. 11A, rubies 4300–8600 ppm), is high compared to other known The pressure-temperature (P-T) evolution for ruby occurrences. Winza rubies recorded ~0.1–0.6 wt.% genesis is of geodynamic interest because it elucidates Fe, which is comparable to that found for the Paranesti the processes of burial and uplift in convergent rubies (~0.2–0.8 wt.% Fe). Also, the low V content tectonic settings. The secondary formation of marga- found in Paranesti (2 6 1 ppm) is also observed in rite around the ruby rims as well as the existence of Winza (2.5 6 2 ppm). Rubies from the Montepuez sillimanite indicates the retrogressive metamorphic Area, Mozambique, are also hosted in metamorphic history at Paranesti. green amphibolitic (actinolite) lenses with margarite Previous studies have shown that the poly-meta- associations; however, the Cr content is lower at morphic terrains of the Rhodope Mountain Complex ,2600 ppm (Pardieu et al. 2013). show a complex multi-stage development. Overprint- The high chromium (Cr) content based on the LA- ing structural relationships were originally interpreted ICP-MS and EMPA studies found in the Paranesti ruby to represent three separate metamorphic events in the (peak 0.9 wt.% LA-ICP-MS and .1 wt.% EMPA) is central Greek Rhodope. First, an early high pressure rare compared to in situ rubies (~0.3 wt.%) world- metamorphic record of pre-tectonic eclogite for which wide. Most rubies of metamorphic/metasomatic origin metamorphic conditions of 19 kbar and 700 8C have have signiﬁcantly lower values. One other ruby deposit been estimated (Liati & Mposkos 1990, Liati & Seidel with high Cr (up to 14,000 ppm) is the Aappaluttoq 1996). Based on studies of relict mineral assemblages deposit, Fiskenæsset Greenland (Keulen & Kalvig from the high grade gneisses, the pressure is estimated 2013). The Fiskenæsset rubies are found within an at .15 kbar, compatible with .45 km burial depth amphibolite-anorthosite contact zone in an assemblage (Mposkos & Liati 1993). This eclogite-facies event of corundum, sapphirine, pargasite, spinel, hornblende, was followed by amphibolite facies metamorphism gedrite, biotite, plagioclase, phlogopite 6 cordierite which overprinted the high pressure metamorphism (P and anthophyllite (Appel & Ghisler 2014, Keulen & ¼ 0.8–1.1 GPa, T ¼ 580–690 8C, Liati & Seidel 1996). Kalvig 2013, Smith et al. 2016). The Fiskenæsset The traditional view of kyanite in eclogites being rubies are relatively rich in Fe and Si, but relatively replaced by sapphirine-bearing symplectites is attri- poor in Ti and Ga (Keulen & Kalvig 2013). The host buted to granulite facies metamorphic events (Liati & rock anorthite is distinctly different from the Paranesti Seidel 1994, 1996). This view was challenged by pargasite schist, and the mineral assemblage shows a Moulas et al. (2013), where the metamorphic grade higher metamorphic grade and the occurrence of both was found to be at the much lower amphibolite facies plagioclase and sapphirine. The red corundum from (0.4 GPa , P , 0.7 GPa, 580 8C , T , 800 8C). Paranesti and Aappaluttoq share a similar chemical Through the various analytical methods deployed in

FIG. 11. Trace element discrimination diagrams showing the ﬁelds for magmatic, metamorphic, and metasomatic corundums, along with the plots for the Paranesti rubies. (A) Adapted from Sutherland et al. (2009); (B) adapted from Peucat et al. (2007); (C) adapted from Sutherland et al. (2014); (D) adapted from Sutherland et al. (2009); and (E) adapted from Giuliani et al. (2010, 2014b).

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 on 23 May 2019 by University of New South Wales user Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf

TABLE 9. COMPARISON BETWEEN THE PARANESTI AMPHIBOLE SCHIST HOSTED RUBY DEPOSIT AND 554 AFRICAN AMPHIBOLE-HOSTED RUBY DEPOSITS

Estimated Ruby Chemical Deposits Geological Occurrences Main Paragenesis P-T Conditions Characteristics

Malagasy - Amphibolites forming diffuse zones within -hornblende (tschermakitic) þ ruby þ plagioclase T: 730–870 8C Cr2O3: 0.2–1.2% Madagascar basic/ultrabasic complexes (An88–91 ) þ Cr-spinel þ phlogopite P: 9–11 kbar Fe2O3: 0.3–0.7% Ejeda-Fotadrevo In anorthosite layers and veins within basic/ -plagioclase (An94–96 ) þ rubyþgarnet þ gedrite þ TiO2: 0.0–0.03% Madagascar ultrabasic complexes sapphirine þ zoisite þ clinopyroxene þ horn- V2O3: 0.01–0.07% blende (pargasitic) þ Cr-zoisite þ ruby þ Cr- spinel þ margarite Tanzania In veins of zoisite-amphibolites (anyolites) and -hornblende þ Cr-zoisite þ ruby Cr-spinel þ T: 750–870 8C Cr2O3: 0.5–2.0% Longido anorthosites cutting through ultrabasic garnet(Prp-Alm) P: 8.5–11 kbar Fe2O3: 0.2–0.5% Lossogonoi bodies -plagioclase( An65–93 ) þ ruby hornblende þ Cr- TiO2: 0.0–0.03% spinel V2O3: 0.01–0.05% Tanzania Winza Mafic migmatitic and well-foliated gneisses -amphibole þ garnet þ plagioclase þ kyanite þ T: 800 6 50 8C Cr2O3: 0.07 6 0.02%

apatite þ spinel inclusions P: 8–10 kbar Fe2O3: 0.36 6 0.06% MINERALOGIST CANADIAN THE TiO2: 0.02 6 0.02% V2O3: 0.005–0.015% Mangari - Kyanite-plagioclasites and/or sapphirinites in -ruby þ plagioclase þ sapphirine þ spinel þ T: 700–750 8C Cr2O3: 0.23–1.4% SE Kenya the contact zones between ultrabasic bodies clinochlore þ phlogopite þ graphite P: 8–10.5 kbar Fe2O3: 0–0.7% and country rocks TiO2: 0.01–0.85% or veins of kyanite/sapphirine-bearing V2O3: 0–0.05% desilicated pegmatites cutting through ultrabasic bodies Montepuez Green actinolite with small lenses rich in -ruby þ actinolite þ margarite þ Chalcopyrite ? Cr2O3: 0.06–0.15% Mozambique anorthite Fe2O3: 0.003–0.02% TiO2: 0.002–0.005% Greenland Reaction zone formed from metasomatic -ruby þ sapphirine þ pargasite þ spinel þ ?Cr2O3: 0.15–1.94% Aappaluttoq interactions between ultramafic rock hornblende þ gedrite þ biotite þ plagioclase þ TiO2: 0.02–0.11% (peridotite) and mafic rock (leucogabbro) phlogopite 6 plagioclase 6 cordierite þ V2O3: 0.01–0.23% anthophyllite

French Massif Amphibolites from metatroctolites of Variscan -ruby þ pargasite þ spinel þ gedrite þ T: 800–820 6 36 8C Cr2O3: 1.96% Central ophiolitic complex tschermakite þ anorthite þ olivine þ kyanite þ P: 9.1–10.5 6 1.0 kbar ? Surrounded by Cr-amphibole or within sapphirine þ margarite plagioclase patches

Greece Pargasite schist boudin surrounded by -Pargasite þ ruby þ plagioclase þ spinel þ T: 580–800 8C Cr2O3: 0.01–1.3% Paranesti clinochlore schist kyanite P: 4–7 kbar Fe2O3: 0.2–0.5% -margarite þ muscovite þ clinochlore TiO2: 0–0.03% V2O3: 0–0% Adapted From Mercier et al. 1999; Schwarz et al. 2008, Pardieu et al. 2013, Thirangoon 2009, Berger et al. 2010, and Smith et al. 2016. PARANESTI PARGASITE SCHIST HOSTED RUBIES 555

Simonet et al. (2008) discussed the existence of a ‘‘gem corundum domain’’ in his corundum classiﬁca- tion model, with P-T conditions ranging from P . 3 kbar and T . 450 8CtoP . 7 kbar and T . 550 8C. Figure 12 illustrates the theoretical P-T path for the Paranesti corundum compared to some other metamorphic deposits from around the world. Based on the P-T constraints discussed above, the Paranesti corundum initially formed under amphibolite facies conditions with 4 kbar , P , 7 kbar and 580 8C , T , 750 8C (Fig. 12), similar to the African amphibolite ruby deposits in Table 9.

Origin of the Paranesti rubies

The euhedral to subhedral shape of the spinel inclusions indicates their syngenetic nature: the spinels were entrapped during the metamorphic formation of the rubies. The outer rim color-zoning of the ruby grains did not demonstrate sufficient trace element variation to indicate its cause. It is possible that during the last stage of the crystallization, depletion of Fe and Ti and consequent relative enrichment in Cr resulted in the enhanced color and zoning. However, further investigations beyond the scope of this present study are required to determine the exact cause. The FIG. 12. Estimated P-T path (marked in red) for Paranesti ruby alignment of the ruby grains indicates there to be a formation compared to other known metamorphic and pre- or syngenetic relationship with main foliations metasomatic ruby deposits (adapted from Giuliani et al. within the Nestos Shear Zone. 2014a). P–T conditions for the formation of corundum in Margarite in quartz/calcite-free rocks (higher metamorphic deposits (modified from Simonet et al. metamorphic grade margarite assemblage) within the 2008). P-T fields of North Carolina (Tenthorey et al. Central Alps were found to decompose at about 620 8C 1996), Mangare (Mercier et al. 1999), Morogoro (Altherr and 7 kbar (Bucher-Nurminen et al. 1983). The et al. 1982), southern Kenya (Key & Ochieng 1991, Simonet 2000), Hunza (Okrusch et al. 1976), Sri Lanka prograde metamorphism is supported by the changes (De Maesschalck & Oen 1989), Greenland (Garde & in the margarite-bearing assemblages where: Marker 1988), Kashmir (Peretti et al. 1990) with three P- 1 margarite ! 1 anorthite þ 1 corundum þ 1HO T boxes corresponding to the evolution of the fluids in the 2 sapphire crystals from the center (c) to intermediate (i) ð1Þ and outer (o) zones, Urals (Kissin 1994), and Mong Hsu Margarite is found to no longer be stable after this P-T (Peretti et al. 1996). constraint in prograde conditions. Petrographic observations of the Paranesti corundums suggest the margarite formed reaction rims around ruby grains, this study, no granulite facies mineral (or relict of suggesting that the corundum did not form due to the these minerals) was identified in the samples. Kyanite breakdown of margarite. Rather, the corundum formed is found within the corundum-bearing rock assem- first at a higher temperature and the margarite formed blages as well as in the wall rock that is devoid of subsequently at its lower stability temperature due to rubies. As these kyanite grains are more euhedral in retrogression. This indicates that the corundums shape and only weakly fractured compared to the formed at a temperature in excess of 620 8C. In a anhedral to subhedral highly fractured corundum corundum-bearing xenolith from the Qoˆrqut Granite grains, this suggests earlier formation of the corun- Complex, Godtha˚bsfjord, Greenland, Rosing et al. dum. Remnant sillimanite was found in the hanging (1987) found plagioclase and corundum to react with wall, providing evidence for retrogression from water to form margarite at ~5 kbar and ~580 8C. amphibolite facies to the current greenschist facies Margarite from the current study consists of reaction metamorphism in line with the findings of Moulas et rims around ruby grains. Based on the P-T constraints al. (2013). discussed above, margarite developed during the

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 556 THE CANADIAN MINERALOGIST

retrogressive phase of metamorphism, subsequently at Africa, Madagascar, south India, Sri Lanka, and east the expense of the corundum. Antarctica. The formation of this belt was due to Corundum-kyanite-sapphirine amphibolites (com- continental collision between the eastern and western prising an assemblage of pargasite/tschermakite, Gondwana blocks with associated high-grade meta- anorthite, corundum, kyanite, sapphirine, and spinel) morphism and fluid circulation (metasomatism) at from the French Massif Central (FMC) bear similar- amphibolite to granulite facies conditions. Rubies and ities to the Paranesti occurrences (Berger et al. 2010). sapphires have been reported to occur in various rock The FMC was found to have formed from subduction types of the metamorphic units of Greece (Voudouris of a plagioclase-rich troctolite, part of the Limousin et al. 2010). This study on Paranesti rubies contributes ophiolite. The difference between Paranesti and the to a potentially wider gem corundum province FMC is the complete lack of anorthite (or any other covering the Greek Hellenides and eastwards towards plagioclase) and sapphirine and lack of any direct the border with Turkey. evidence of ophiolitic indicators in the Paranesti Rubies of metamorphic origin are associated with corundum assemblages. continental collision zones and have been proposed as Based on the results presented above and the regional plate tectonic indicators by Stern et al. (2013). The geological and tectonic framework studies of others Paranesti occurrence is located within the NSZ (Alpine authors, the Paranesti corundums are inferred to have suture zone), in which remnants of a partially subducted formed within an ultramafic precursor. Metamorphic magmatic arc foundered on the attenuated margin of events of either ca. 149 Ma or ca. 73 Ma, prior to the Europe is preserved. Subduction and collision of the arc pegmatite intrusion at ca. 65 Ma, resulted in the with the incoming Lower Terrane continent on one side formation of the Paranesti corundums during amphibo- of the NSZ produced decoupling within the arc and lite facies metamorphism. No feldspar or quartz has been subduction of its deeper parts along with the frontal observed in the ruby-bearing assemblages, in contrast to parts of the Lower Terrane, one of the intra-Tethys the surrounding amphibolites. Based on the elevated Ca continental blocks derived from Gondwana (Burg content (up to 10 wt.%), the igneous protolith most 2012). Brun et al. (2016) now consider the Nestos likely had large proportions of clinopyroxene and may Suture Zone to correspond to the Vardar Suture Zone have been an aluminous clinopyroxenite. (e.g., subduction of the Vardar ocean beneath Eurasia). Due to the lack of mineral inclusions available for The Paranesti corundum occurrence thus adds support dating, it is difficult to ascertain whether PAR-1 and to the suggestion that metamorphic rubies hosted in PAR-5 formed during the same metamorphic event, mafic-ultramafic rocks can be used as plate tectonic though both contain similar mineral assemblages. The suture zone indicators. This broad tectonic framework trace element diagrams display a metasomatic influ- can thus be used to assist in future exploration for this ence on the paler colored PAR-5 corundum. This is valuable gemstone, both in northern Greece and other supported by the location of the occurrences within the collisional zones of the world. melange suture zone between two distinct tectonic units. The Paranesti corundum-bearing assemblages CONCLUSIONS went through retrogression from amphibolite facies to greenschist facies based on the observed mineral This is the first systematic research on the origin assemblage of remnant sillimanite to pargasite/kya- and genesis of the Paranesti corundum occurrences. nite/margarite/plagioclase and lastly replacement by The range of analytical results provides a framework clinochlore/muscovite and clinozoisite within the for the P-T conditions of corundum formation during surrounding corundum-free amphibolites. amphibolite facies conditions estimated at 4 kbar , P , 8 , , 8 There are no convincing field or petrographic 7 kbar and 580 C T 750 C and with relationships to support a primary character for the subsequent retrogression. protolith of the corundum-bearing pargasitic schists Geochemical signatures of the Paranesti corundums from Paranesti. These metamorphic rocks have been display a distinctive high Cr, low V, Ti, Mg, and Ga trace element fingerprint as a geographical locality affected by intense tectonic processes and no primary marker (Fig. 13). This distinctive geolocation signa- phases or textures have been preserved. ture depleted of all major trace elements other than Broader tectonic implications iron for corundum profiling could benefit future commercial exploration and marketing of the rubies Dissanayake & Chandrajith (1999) discussed the from the area. East African metamorphic belt, termed the Metamor- Further studies in quantitative thermobarometry, phic Mozambique Belt in the context of the existence oxygen isotope signatures, and age-dating would of a wide gem corundum province covering East complement the research data on these occurrences.

FIG. 13. Box and whisker plot for Paranesti corundum chromophore (ppm based on LA-ICP-MS results).

ACKNOWLEDGMENTS for this project came from UNSW Faculty of Science Research Grant to Ian Graham (and co-investigators) and The authors wish to thank Joanne Wilde from UNSW Australian Institute of Nuclear Science and Engineering for the assistance with petrographic sample preparation (AINSE) grants AINGRA07061 and AINGRA08025. and Dr. Karen Privat from the Electron Microscopy Unit Funding for some of the analytical equipment used in this (EMU) of the Mark Wainwright Analytical Centre project was provided through UNSW MREII Grants and (UNSW) for the assistance in EMPA analyses. Funding Australian Research Council (ARC) Large Infrastructure

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 558 THE CANADIAN MINERALOGIST

and Equipment Funds (LIEF) grant LE0989067. The BRUN, J.P., FACCENNA,C.,GUEYDAN,F.,SOKOUTIS,D., analytical data were obtained using instrumentation PHILIPPON, M., KYDONAKIS, K., & GORINI, C. (2016) The funded by DEST Systemic Infrastructure Grants, ARC two-stage Aegean extension, from localized to distributed, a result of slab rollback acceleration. Canadian Journal of LIEF, NCRIS/AuScope, industry partners, and Macquar- Earth Sciences 53(11), 1142–1157. ie University. This is contribution 982 from the ARC Centre of Excellence for Core to Crust Fluid Systems BUCHER-NURMINEN, K., FRANK, E., & FREY, M. (1983) A model (http://www.ccfs.mq.edu.au) and 1161 in the GEMOC for the progressive regional metamorphism of margarite- Key Centre (http://gemoc.mq.edu.au). The authors would bearing rocks in the Central Alps. American Journal of Science 283(A), 370–395. also like to acknowledge the XRF Facility within the Mark Wainwright Analytical Centre at the University of BURG, J.-P. (2012) Rhodope: From Mesozoic convergence to New South Wales, and the Q-ICP-MS unit in the School Cenozoic extension. Review of petro-structural data in the of Earth Sciences, University of Melbourne, Australia for geochronological frame. Journal of the Virtual Explorer their analytical support. We would also like to thank 42(1), 1–44.

David Turner and Julien Berger for their reviews, and the BURG, J.-P., RICOU, L.-E., IVANOV, Z., GODFRIAUX, I., DIMOV, Editor Lee Groat for his comments, all of which greatly D., & KLAIN, L. (1996) Syn-metamorphic nappe complex improved the final manuscript. in the Rhodope Massif: Structure and Kinematics. Terra Nova 8, 6–15. REFERENCES COLVILLE,P.A.,ERNST, W.G., & GILBERT, M.C. (1966) Relationships between cell parameters and chemical ALTHERR, R., OKRUSCH, M., & BANK, H. (1982) Corundum-and compositions of monoclinic amphiboles. American Min- kyanite-bearing anatexites from the Precambrian of eralogist 51, 1727. Tanzania. Lithos 15(3), 191–197. DE MAESSCHALCK, A.A. & OEN, I.S. (1989) Fluid and mineral APPEL, P.W. & GHISLER, M. (2014) Ruby- and sapphirine- inclusions in corundum from gem gravels in Sri Lanka. bearing mineral occurrences in the Fiskenaesset, Nuuk Mineralogical Magazine 53, 539–545. and Maniitsoq Regions, West Greenland. Danmarks og Grønlands Geologiske Undersøgelse 2014/72, 71. DILL,H.G.(2010)The‘‘chessboard’’ classification scheme of mineral deposits: mineralogy and geology from BARR, S.R., TEMPERLEY, S., & TARNEY, J. (1999) Lateral aluminum to zirconium. Earth-Science Reviews 100(1), growth of the continental crust through deep level 1–420. subduction–accretion: a re-evaluation of central Greek Rhodope. Lithos 46(1), 69–94. DIMADIS,E.,KASSOLI-FOURNARAKI,A.,KOLCHEVA,K., FILIPPIDIS, A., & TSIRAMBIDIS, A. (1990) Ultramafic body BAUER, C., RUBATTO, D., KRENN, K., PROYER, A., & HOINKES, in Gorgona area, North of Xanthi (Central Rhodope G. (2007) A zircon study from the Rhodope metamorphic Massif, Greece). Geologica Rhodopica 2, 117–126. complex, N-Greece: time record of a multistage evolution. Lithos 99(3), 207–228. DINTER, D.A. (1998) Late Cenozoic extension of the Alpine collisional orogen, northeastern Greece: origin of the BAZIOTIS,I.,MPOSKOS,E.,&PERDIKATSIS,V.(2007) north Aegean basin. Geological Society of America Geochemistry of amphibolitized eclogites and cross- Bulletin 110(9), 1208–1230. cutting tonalitic–trondhjemitic dykes in the Metamor- phic Kimi Complex in East Rhodope (N.E. Greece): DISSANAYAKE, C.B. & CHANDRAJITH, R. (1999) Sri Lanka– implications for partial melting at the base of a thickened Madagascar Gondwana linkage: evidence for a Pan- crust. International Journal of Earth Science 97(3) 459– African mineral belt. Journal of Geology 107(2), 223– 477. 235.

BERGER, J., FEMENIAS, O., OHNENSTETTER, D., PLISSART, G., & EGGINS, S.M., WOODHEAD, J.D., KINSLEY, L.P.J., MORTIMER, MERCIER, J.C. (2010) Origin and tectonic signiﬁcance of G.E., SYLVESTER, P., MCCULLOCH, M.T., HERGT, J.M., & corundum–kyanite–sapphirine amphibolites from the HANDLER, M.R. (1997) A simple method for the precise Variscan French Massif Central. Journal of Metamorphic determination of 40 trace elements in geological Geology 28(3), 341–360. samples by ICPMS using enriched isotope internal standardisation. Chemical Geology 134(4), 311–326. BONEV, N., BURG, J.P., & IVANOV, Z. (2006) Mesozoic– Tertiary structural evolution of an extensional gneiss GARDE,A.&MARKER, M. (1988) Corundum crystals with dome—the Kesebir–Kardamos dome, eastern Rhodope blue–red color zoning near Kangerdluarssuk, Sukkertop- (Bulgaria–Greece). International Journal of Earth Sci- pen district, West Greenland. Rapport Grønlands Geo- ences 95(2), 318–340. logiske Undersøkelse 140, 46–49.

BRUN, J.P. & FACCENNA, C. (2008) Exhumation of high- GAWLICK, H.J., AUBRECHT, R., SCHLAGINTWEIT, F., MISSONI, S., pressurerocksdrivenbyslabrollback.Earth and &PLASˇIENKA, D. (2015) Ophiolitic detritus in Kimmer- Planetary Science Letters 272, 1–7. idgian resedimented limestones and its provenance from

an eroded obducted ophiolitic nappe stack south of the LIATI, A. (2005) Identification of repeated Alpine (ultra) high- Northern Calcareous Alps (Austria). Geologica Carpath- pressure metamorphic events by U-Pb SHRIMP geochro- ica 66(6), 473–487. nology and REE geochemistry of zircon: the Rhodope zone of Northern Greece. Contributions to Mineralogy GIULIANI, G., LASNIER, B., OHNENSTETTER, D., FALLICK, A.E., & and Petrology 150(6), 608–630. PE´ GE` RE, G. (2010) Les gisements de corindon de France. le Regne` Mineral´ 93, 5–22. LIATI,A.&MPOSKOS, E. (1990) Evolution of the eclogites in the Rhodope Zone of northern Greece. Lithos 25(1–3), GIULIANI, G., OHNENSTETTER, D., FALLICK, A., GROAT, L., & 89–99. FAGAN, A. (2014a) The geology and genesis of gem nd corundum deposits. In Geology of Gem Deposits, 2 LIATI,A.&SEIDEL, E. (1994) Sapphirine and hogbomite¨ in Edition (L. Groat, ed.). Mineralogical Association of over-printed kyanite-eclogites of central Rhodope, N. Canada Short Course Series 44, 29–112. Greece: first evidence of granulite-facies metamorphism. European Journal of Mineralogy 6, 733–738. GIULIANI,G.,CAUMON,G.,RAKOTOSAMIZANANY,S., OHNENSTETTER,D.,&RAKOTONDRAZAFY M. (2014b) LIATI,A.&SEIDEL, E. (1996) Metamorphic evolution and Classification chimique des corindons par analyse factor- geochemistry of kyanite eclogites in central Rhodope, ielle discriminante: application a` la typologie des gise- northern Greece. Contributions to Mineralogy and ments de rubis et saphirs. Revue de l’Association Petrology 123, 293–307. Fran¸caisede Gemmologie A.F.G. 188, 14–22. LIATI A., GEBAUER, D., & WYSOCZANSKI, R. (2002) U-Pb HARLOW, G.E. & BENDER, W. (2013) A study of ruby SHRIMP-dating of zircon domains from UHP garnet-rich (corundum) compositions from the Mogok Belt, Myan- mafic rocks and late pegmatoids in the Rhodope zone (N mar: Searching for chemical fingerprints. American Greece); evidence for Early Cretaceous crystallization Mineralogist 98(7), 1120–1132. and Late Cretaceous metamorphism. Chemical Geology 184(3–4), 281–299. HEY, M.H. (1954) A review of the chlorites. Mineralogical Magazine 30, 277. MERCIER, A., RAKOTONDRAZAFY, M., & RAVOLOMIANDRINARIVO, B. (1999) Ruby mineralization in southwest Madagascar. JOLIVET,L.&BRUN, J.P. (2010) Cenozoic geodynamic Gondwana Research 2(3), 433–438. evolution of the Aegean. International Journal of Earth Science 99, 109–138. MOULAS,E.,KOSTOPOULOS,D.,CONNOLLY,J.A.,&BURG, J.P. (2013) PT estimates and timing of the sapphirine- KEULEN,N.&KALVIG, P. (2013) Fingerprinting of corundum bearing metamorphic overprint in kyanite eclogites from (ruby) from Fiskenaesset, West Greenland. GEUS Geo- Central Rhodope, northern Greece. Petrology 21(5), logical Survey of Denmark and Greenland Bulletin 28, 507–521. 53–56. MPOSKOS,E.&KROHE, A. (2000) Petrological and structural KEY, R.M. & OCHIENG, J. (1991) The growth of rubies in evolution of continental high pressure (HP) metamorphic south-east Kenya. Journal of Gemmology 22(8), 484–496. rocks in the Alpine Rhodope Domain (N Greece). In Proceedings of the 3rd International Conference on the KISSIN, A.J. (1994) Ruby and sapphire from the southern Ural Geology of the Eastern Mediterranean (I. Panayides, C. mountains, Russia. Gems and Gemology 30(4), 243–252. Xenophontos, & J. Malpas, eds.). (221–232).

ˇ KOLCEVA, K., ZELJASKOVA-PANAJOTOVA, M., BOBRECOV, N.L., MPOSKOS,E.&KROHE, A. (2006) Pressure-temperature- &STOJANOVA, V. (1986) Eclogites in Central Rhodope deformation paths of closely associated ultra-high-pres- metamorphic group and their retrograde metamorphism. sure (diamond-bearing) crustal and mantle rocks of the Geochemistry, Mineralogy, and Petrology 20–21, 130– Kimi complex: implications for the tectonic history of the 144. Rhodope Mountains, northern Greece. Canadian Journal of Earth Science 43, 1755–1776. KRENN, K., BAUER, C., PROYER, A., MPOSKOS, E., & HOINKES, G. (2008) Fluid entrapment and reequilibration during MPOSKOS,E.&LIATI, A. (1993) Metamorphic evolution of subduction and exhumation: A case study from the high- metapelites in the high pressure terrain of the Rhodope grade Nestos shear zone, Central Rhodope, Greece. Lithos Zone, Northern Greece. Canadian Mineralogist 31, 401– 104(1), 33–53. 424.

KROHE,A.&MPOSKOS, E. (2002) Multiple generations of NAGEL,T.J.,SCHMIDT,S.,JANA´ K, M., FROITZHEIM, N., JAHN-AWE, extensional detachments in the Rhodope Mountains S., & GEORGIEV, N. (2011) The exposed base of a (northern Greece): evidence of episodic exhumation of collapsing wedge: the Nestos shear zone (Rhodope high-pressure rocks. In The Timing and Location of Metamorphic Province, Greece). Tectonics 30(4),TC4009. Major Ore Deposits in an Evolving Orogen, (D.J. Blundell, F. Neubauer, & A. von Quadt, eds.). Geolog- OKRUSCH, M., BUNCH, T.E., & BANK, H. (1976) Paragenesis ical Society of London Special Publication 204,151– and petrogenesis of a corundum-bearing marble at Hunza 178. (Kashmir). Mineralium Deposita 11(3), 278–297.

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 560 THE CANADIAN MINERALOGIST

PARDIEU, V., SANGSAWONG, S., MUYAL, J., CHAUVIRE´ , B., MASSI, STERN, R.J., TSUJIMORI, T., HARLOW, G., & GROAT, L.A. (2013) L., & STURMAN, N. (2013) Montepeuz Area (Mozambi- Plate tectonic gemstones. Geology 41(7), 723–726. que). Gemological Institute of America, Research and News, October, 1–24. SUTHERLAND, F.L., GRAHAM, I.T., POGSON, R.E., SCHWARZ, D., WEBB, G.B., COENRAADS, R.R., FANNING, C.M., HOLLIS, PERETTI, A., MULLIS, J., & KUNDIG, R. (1990) Die Kashmir- J.D., & ALLEN, T.C. (2002) The Tumbarumba basaltic saphire und ihr geologisches erinnerungsvermogen.¨ Neue gem field, New South Wales: In relation to sapphire-ruby Zurcher¨ Zeitung 187, 58–59. deposits of Eastern Australia. Records of the Australian Museum 54, 215–248. PERETTI, A., MULLIS, J., & MOUAWAD, F. (1996) The role of fluorine in the formation of colour zoning in rubies from SUTHERLAND, F.L., ZAW, K., MEFFRE, S., GIULIANI, G., FALLICK, Mong Hsu, Myanmar (Burma). Journal of Gemmology A.E., GRAHAM, I.T., & WEBB, G.B. (2009) Gem corundum London 25, 3–19. megacrysts from East Australia basalt fields: Trace elements, O isotopes and origins. Australian Journal of PEUCAT, J.J., RUFFAULT, P., FRITSCH, E., BOUHNIK-LE, E., Earth Sciences 56, 1003–1020. SIMONET, C., & LASNIER, B. (2007) Ga/Mg ratio as a new geochemical tool to differentiate magmatic from meta- SUTHERLAND, F.L., ZAW, K., MEFFRE, S., YUI, T.F., & THU,K. morphic blue sapphires. Lithos 98, 261–274. (2014) Advances in trace element ‘‘fingerprinting’’ of gem corundum, ruby and sapphire, Mogok area, Myanmar. RICOU, L.E., BURG, J.P., GODFRIAUX, I., & IVANOV, Z. (1998) Minerals 5(1), 61–79. Rhodope and Vardar: the metamorphic and the olistostromic paired belts related to the Cretaceous subduction TENTHOREY, E.A., RYAN,J.G.,&SNOW, E.A. (1996) under Europe. Geodinamica Acta 11(6), 285–309. Petrogenesis of sapphirine-bearing metatroctolites from the Buck Creek ultramafic body, southern Appalachians. ROSING, M.T., BIRD, D.K., & DYMEK, R.F. (1987) Hydration of Journal of Metamorphic Geology 14(2), 103–114. corundum-bearing xenoliths in the Q0rqut Granite Complex, Godthibsfjord, West Greenland. American THIRANGOON,K.(2009)Ruby and Pink Sapphire from Mineralogist 72, 29–38. Aappaluttoq, Greenland Status of on-going research. GIA Laboratory, Bangkok, Thailand. SAMINPANYA,S.,MANNING, D.A.C., DROOP, G.T.R., & HENDERSON, C.M.B. (2003) Trace elements in Thai gem TURPAUD, P. (2006) Characterisation of igneous terranes by corundums. Journal of Gemmology 28(7), 399–416. zircon dating: implications for the UHP relicts occurrences and suture identification in the Central Rhodope, SCHMIDT, S., NAGEL, T.J., & FROITZHEIM, N. (2010) A new Northern Greece. Dissertation, Johannes Gutenberg-Uni- occurrence of microdiamond-bearing metamorphic rocks, versitat, Mainz, Germany. SW Rhodopes, Greece. European Journal of Mineralogy 22(2), 189–198. TURPAUD,P.&REISCHMANN, T. (2010) Characterisation of igneous terranes by zircon dating: implications for UHP SCHWARZ, D., PARDIEU, V., SAUL, J.M., SCHMETZER, K., LAURS, occurrences and suture identification in the Central B.M., GIULIANI, G., KLEMM, L., MALSY, A.K., EREL, E., Rhodope, northern Greece. International Journal of Earth HAUZENBERGER, C., & DU TOIT, G. (2008) Rubies and Sciences 99(3), 567–591. sapphires from Winza, central Tanzania. Gems and Gemology 44, 322–347. VAN HINSBERGEN, D.J.J., HAFKENSCHEID, E., SPAKMAN, W., MEULENKAMP, J.E., & WORTEL, R. (2005) Nappe stacking SIMONET, C. (2000) Geologie´ des gisements de saphir et de resulting from subduction of oceanic and continental rubis. L’exemple de la John Saul mine, Mangari, Kenya. lithosphere below Greece. Geology 33(4), 325–328. Doctoral dissertation, These` de Doctorat, Universitede´ Nantes, France, 349 pp. VOUDOURIS, P., GRAHAM, I., MELFOS, V., ZAW, K., SUTHERLAND, L, GIULIANI, G., & IONESCU, M. (2010) Gem corundum SIMONET, C., FRITSCH, E., & LASNIER, B. (2008) A classifica- deposits of Greece: Diversity, chemistry and origins. tion of gem corundum deposits aimed towards gem Proceedings of the 13th Quadrennial IAGOD Symposium exploration. Ore Geology Reviews 34(1), 127–133. 69, Adelaide, Australia, 429–430.

SMITH, C.P., CLARK, B., & FAGAN, A.J. (2016) Ruby and Pink WANG, K.K., GRAHAM, I., VOUDOURIS, P., GIULIANI,G., Sapphire from Aappaluttoq, Greenland. Journal of FALLICK, A.E., & COHEN, D. (2016) Origin of rubies from Gemmology 35(4), 294. the Paranesti region, North-Eastern Greece. 35th Interna- tional Geological Congress, Cape Town, South Africa, ¨ SOROKINA, E.S., HOFMEISTER, W., HAGER, T., MERTZ-KRAUS, R., 1766. BUHRE, S., & SAUL, J.M. (2016) Morphological and chemical evolution of corundum (ruby and sapphire): Crystal ontogeny reconstructed by EMPA, LA-ICP-MS, and Cr3þ Raman mapping. American Mineralogist Received January 30, 2017. Revised manuscript accepted 101(12), 2716–2722. June 13, 2017.

Downloaded from https://pubs.geoscienceworld.org/canmin/article-pdf/55/4/535/3951791/i1499-1276-55-4-535.pdf by University of New South Wales user on 23 May 2019 Appendix 2

Fingerprinting Paranesti rubies through oxygen isotopes.

Published in Minerals (MDPI) Vol. 9, pp. 1-14 (2019) DOI: 10.3390/min9020091

Wang, K.K., Graham, I.T., Martin, L., Voudouris, P., Giuliani, G., Lay, A., Harris, S.J. and Fallick, A., 2019. Fingerprinting Paranesti rubies through oxygen isotopes. Minerals, 9(2), 91.

159 minerals

Article Fingerprinting Paranesti Rubies through Oxygen Isotopes

Kandy K. Wang 1,*, Ian T. Graham 1, Laure Martin 2, Panagiotis Voudouris 3 , Gaston Giuliani 4, Angela Lay 1, Stephen J. Harris 1 and Anthony Fallick 5

1 PANGEA Research Centre, School of Biological, Earth and Environmental Sciences, University of NSW, 2052 Sydney, Australia; [email protected] (I.T.G.); [email protected] (A.L.); [email protected] (S.J.H.) 2 Centre for Microscopy Characterisation and Analysis, The University of Western Australia, 6009 Perth, Australia; [email protected] 3 Faculty of Geology and Geoenvironment, National and Kapodistrian University of Athens, 157 84 Athens, Greece; [email protected] 4 Université de Lorraine, IRD and CRPG UMR 7358 CNRS-UL, BP 20, 15 rue Notre-Dame-des-Pauvres, 54501 Vandœuvre-lès-Nancy, France; [email protected] 5 Isotope Geosciences Unit, S.U.E.R.C., Rankine Avenue, East Kilbride, Glasgow G75 0QF, UK; [email protected] * Correspondence: [email protected]; Tel.: +61-411418800

Received: 1 December 2018; Accepted: 30 January 2019; Published: 3 February 2019

Abstract: In this study, the oxygen isotope (δ18O) composition of pink to red gem-quality rubies from Paranesti, Greece was investigated using in-situ secondary ionization mass spectrometry (SIMS) and laser-fluorination techniques. Paranesti rubies have a narrow range of δ18O values between ~0 and +1 and represent one of only a few cases worldwide where δ18O signatures can be used to distinguishh them from other localities. SIMS analyses from this study and previous work by the authors suggests that the rubies formed under metamorphic/metasomatic conditions involving deeply penetrating meteoric waters along major crustal structures associated with the Nestos Shear Zone. SIMS analyses also revealed slight variations in δ18O composition for two outcrops located just ~500 m apart: PAR-1 with a mean value of 1.0 ± 0.42 and PAR-5 with a mean value of 0.14 ± 0.24 . This work adds to the growing useh of in-situh methods to determine the origin of gem-qualityh corundumh and re-confirms its usefulness in geographic “fingerprinting”.

Keywords: rubies; corundum; in-situ oxygen isotopes; Paranesti Greece; Nestos Shear Zone; Secondary ion mass spectrometry (SIMS)

1. Introduction

1.1. Oxygen Isotopic Studies in Corundums Oxygen is an abundant element in the Earth’s crust, mantle and ﬂuids. Oxygen consists of three naturally-occurring stable isotopes: 16O (99.76%), 17O (0.04%) and 18O (0.2%). δ18O expressed as Vienna standard mean ocean water (VSMOW) in per mil is the standard for the oxygen isotopic composition which is a measure of the ratio of the stable isotopes oxygen-18 (18O) and oxygen-16 (16O). There are numerous applications of oxygen isotope geochemistry including paleoclimatology, urban forensics, geological genesis and many more [1–3]. Oxygen isotope fractionation is a function of the initial Rayleigh evaporation-precipitation cycle, temperature of the system and degree of water-rock interaction and therefore great care must be taken when interpreting oxygen isotope values [4–7].

Minerals 2019, 9, 91; doi:10.3390/min9020091 www.mdpi.com/journal/minerals Minerals 2019, 9, 91 2 of 14 Minerals 2019, 9, x FOR PEER REVIEW 2 of 14

AlthoughAlthough worldwide worldwide corundumcorundum oxygen isotope va valueslues have have been been found found in in a awide wide range range from from - 27‰ (Khitostrov, Russia) to +23‰ (Mong Hsu, Myanmar), most are in the range of +3‰ to +21‰ [8– -27 (Khitostrov, Russia) to +23 (Mong Hsu, Myanmar), most are in the range of +3 to +21 [8–10]. 10]. This criterion has often been used to determine the geological origin of coloured corundum and Thish criterion has often been usedh to determine the geological origin of colouredh corundumh and especially the gem corundums, rubies and sapphires. δ18O has been particularly useful in determining especially the gem corundums, rubies and sapphires. δ18O has been particularly useful in determining the likely primary geological origin of placer corundums where the primary origin is uncertain [11]. the likely primary geological origin of placer corundums where the primary origin is uncertain [11]. As isotopic fractionation is a function of both temperature and geological processes, oxygen isotope As isotopic fractionation is a function of both temperature and geological processes, oxygen isotope data need to be treated with some degree of caution Thus, there are very few examples where oxygen data need to be treated with some degree of caution Thus, there are very few examples where oxygen isotopes have been used to “fingerprint” the geographic location [12]. isotopes have been used to “ﬁngerprint” the geographic location [12].

1.2.1.2. Geological Geological Setting Setting and and SampleSample BackgroundBackground The Paranesti rubies are found within the Nestos Shear Zone (NSZ) of the Rhodope Mountain The Paranesti rubies are found within the Nestos Shear Zone (NSZ) of the Rhodope Mountain Complex (RMC) in north-eastern Greece (Figure 1). The tectonic and polymetamorphic record of this Complex (RMC) in north-eastern Greece (Figure 1). The tectonic and polymetamorphic record of Northern Aegean region (including the RMC) reflects the Middle Jurassic to Neogene northeast this Northern Aegean region (including the RMC) reﬂects the Middle Jurassic to Neogene northeast dipping subduction and convergence of the African-Eurasian plates which resulted in the closure of dipping subduction and convergence of the African-Eurasian plates which resulted in the closure of the the Tethys Ocean [13,14]. The NSZ is thought to be a one of the syn-metamorphic thrusts in the RMC Tethys Ocean [13,14]. The NSZ is thought to be a one of the syn-metamorphic thrusts in the RMC that that are responsible for regional metamorphic inversion, placing higher amphibolite-facies are responsible for regional metamorphic inversion, placing higher amphibolite-facies intermediate intermediate terranes onto upper-greenschist to lower amphibolite-facies rocks of the lower terrane terranes onto upper-greenschist to lower amphibolite-facies rocks of the lower terrane [15,16]. [15,16].

FigureFigure 1. 1.Geological Geological map map of theof the Rhodope Rhodope Mountain Mountain Complex, Complex, with with Paranesti Paranesti located located within within the Nestos the ShearNestos Zone Shear (red Zone star) (red (Adapted star) (Adapt fromed Moulas from etMoulas al, 2017 et al, [17 2017]). [17]).

BasedBased onon an earlier earlier systematic systematic study study on onParanesti Paranesti [18], [the18], ruby-bearing the ruby-bearing occurrences occurrences were found were foundto be hosted to be in hosted pargasite in pargasite schist with schist a mafic/ul withtramafic a maﬁc/ultramaﬁc protolith. The protolith.surrounding The non-corundum- surrounding non-corundum-bearingbearing chlorite schist was chlorite found schistto mainly was be foundcomprised to mainly of clinochlore. be comprised The ruby-bearing of clinochlore. occurrence The ruby-bearingfound on the occurrencehillside is referred found onto as the PAR-1 hillside (Figure is referred 2a) and to the as road-side PAR-1 (Figure occurrence2a) and is termed the road-side PAR- 5 (Figure 2b). Not all of the pargasite boudins nor the pegmatite intrusion found within the vicinity of the two sites contained corundum (Figure 2c,d).

Minerals 2019, 9, 91 3 of 14 occurrence is termed PAR-5 (Figure 2b). Not all of the pargasite boudins nor the pegmatite intrusion found within the vicinity of the two sites contained corundum (Figure 2c,d). Minerals 2019, 9, x FOR PEER REVIEW 3 of 14

(a) (b)

FigureFigure 2.2. LocalityLocality diagramdiagram ofof thethe rubyruby occurrences.occurrences. ((aa)) PAR-1PAR-1 locationlocation onon toptop ofof thethe hill.hill. ((bb)) PAR-5PAR-5 locationlocation on on the the roadside.roadside. ((cc)) PargasitePargasite schistschist boudinboudin foundfound approx.approx. 500500 mm northnorth ofof PAR-5PAR-5 withoutwithout anyany corundum.corundum. ((dd)) PegmatitePegmatite onon toptop ofof thethe ruby-bearingruby-bearing pargasite pargasite schist schist at at PAR-1. PAR-1.

AA summarysummary of of the the main main ﬁndings findings from from this this previous previous study study is listedis listed in Tablein Table1. Detailed 1. Detailed LA-ICP-MS LA-ICP- traceMS trace element element analyses analyses showed showed that the that rubies the rubies are of are metamorphic of metamorphic origin (Figureorigin (Figure3a) with 3a) minor with partial minor metasomaticpartial metasomatic inﬂuences influences (Figure (Figure3b). The 3b). high The Rhigh2 value R2 value based based on theon the Fe/Mg Fe/Mg vs vs Ga/Mg Ga/Mg elemental elemental discriminationdiscrimination diagram diagram shows shows both both PAR-1 PAR-1 and and PAR-5 PAR- rubies5 rubies to contain to contain highly consistenthighly consistent trace element trace compositionselement compositions (Figure 3 a).(Figure 3a).

Minerals 2019, 9, x FOR PEER REVIEW 4 of 14

Table 1. Summary of prior Paranesti ruby results (Wang et al. 2017 [18]). Minerals 2019, 9, 91 4 of 14 Attributes PAR-1 PAR-5 Physical Characteristics Site TableHillside 1. Summary surface of prior outcrop Paranesti ruby Roadside results (Wang surface et al. outcrop—500 2017 [18]). m east of PAR1 Grain-sizeAttributes 10 mm–20 PAR-1 mm 5 mm–10 PAR-5 mm Colour Deeper red than PAR-5 (generally) Medium red Physical Characteristics InclusionsSite HillsideSpinels surface outcrop Roadside surface outcrop—500None m east of PAR1 MicroscopeGrain-size view More fractured, 10 mm–20 finer-grained mm 5 mm–10- mm Host rockColour DeeperPargasite red than schist PAR-5 (generally) PargasiteMedium schist red Inclusions SpinelsEMPA Analyses (wt. %) None Microscope view More fractured, ﬁner-grained - Cr2OHost3 rock 0.11–1.68Pargasite schist Pargasite0.13–0.29 schist FeO 0.19–0.73 EMPA Analyses (wt. %) 0.18–0.36 TiO2 Cr2O3 0–0.010.11–1.68 0.13–0.290–0.06 FeO 0.19–0.73 0.18–0.36 Ga2O3 0–0.04 0–0.04 TiO 0–0.01 0–0.06 2 LA-ICP-MS: Trace Element Analysis (ppm) Ga2O3 0–0.04 0–0.04 Cr 360–2856LA-ICP-MS: Trace Element Analysis (ppm) 4–8627 Fe Cr 1572–2664360–2856 1833–38224–8627 V Fe 1–31572–2664 1833–38222–5 Mg V 7–42 1–3 8–3762–5 Mg 7–42 8–376 Ti Ti 6–1846–184 10–190 Ga Ga 14-2314-23 13–29 Si Si 781–2456781–2456 837–2123 Ca Ca 769–2119769–2119 653–1903

(a)

Figure 3. Cont.

Minerals 2019, 9, 91 5 of 14 Minerals 2019, 9, x FOR PEER REVIEW 5 of 14

Minerals 2019, 9, x FOR PEER REVIEW 5 of 14

(b)

FigureFigure 3. Trace 3. Trace element element discrimination discrimination diagrams diagrams( showingshowingb) the the fields fields fo forr magmatic, magmatic, metamorphic metamorphic and and metasomaticmetasomatic corundums, corundums, along along with with the the plots plots forfor thethe Paranesti rubies. rubies. (a) ( aFeMg) FeMg vs. vs.GaMg GaMg elemental elemental Figurediagram 3. Traceshowing element the metamorphicdiscrimination vs di agramsmagmatic showing fields theand fields SD linesfor magmatic, with Paranesti metamorphic ruby plots. and diagram showing the metamorphic vs magmatic fields and SD lines with Paranesti ruby plots. Adapted metasomaticAdapted with corundums, permission alongfrom Sutherland with the plots et al. for 2014 the [19]. Paranesti (b) FeO rubies. + TiO (2a +) GaFeMg2O3 vs.vs. FeO-CrGaMg elemental2O3-MgO- with permission from Sutherland et al. 2014 [19]. (b) FeO + TiO2 + Ga2O3 vs. FeO-Cr2O3-MgO-V2O3 diagramV2O3 elemental showing diagram the metamorphic showing a metasomatic vs magmatic origin fiel dsas welland asSD a linesmafic-ultramafic with Paranesti influence ruby onplots. the elementalAdaptedParanesti diagram with rubies. permission showingAdapted from with a metasomatic Sutherland permission et from al. origin 2014 Giuliani [19]. as well (etb) al. FeO as 2014 + a TiO mafic-ultramafic[20].2 + Ga2O3 vs. FeO-Cr influence2O3-MgO- on the ParanestiV2O3 elemental rubies. Adapted diagram withshowing permission a metasomatic from Giulianiorigin as well et al. as 2014 a mafic-ultramafic [20]. influence on the 2. MaterialsParanesti and rubies. Methods Adapted with permission from Giuliani et al. 2014 [20]. 2. Materials and Methods 2. MaterialsTwo different and Methods oxygen isotopic analytical methods have been used in this study in order to verify Twothe oxygen different isotope oxygen values isotopic of the analyticalParanesti rubies. methods The haverubies been were used mechanically in this study extracted in order from tothe verify the oxygenpargasiteTwo isotope differenthost matrix values oxygen and of isotopic thecarefully Paranesti analytical cleaned rubies. priormethods Theto being have rubies beensent were usedfor analysis. mechanically in this study In many inextracted order samples, to verify fromthe the pargasitetheruby oxygen hostcrystals matrix isotope occur and valuesin clusters carefully of the of platy cleanedParanesti crystals prior rubies. where to The being the rubies grain sent were forsizes analysis. mechanically generally In range many extracted between samples, from 0.5–1.5 thethe ruby crystalspargasitecm occur(Figure host in 4a–c). clusters matrix Importantly, and of platy carefully in crystals the cleaned previous where prior study the to grain[18]being the sizessent ruby for generally grains analysis. were range In generally many between samples, found 0.5–1.5 to the be cm ruby crystals occur in clusters of platy crystals where the grain sizes generally range between 0.5–1.5 (Figurefree4 a–c).of inclusions Importantly, and thus in amenable the previous to in-situ study analysis. [18] the ruby grains were generally found to be free cm (Figure 4a–c). Importantly, in the previous study [18] the ruby grains were generally found to be of inclusions and thus amenable to in-situ analysis. free of inclusions and thus amenable to in-situ analysis.

(a) (b)

Figure 4. Cont. Minerals 2019, 9, 91 6 of 14 Minerals 2019, 9, x FOR PEER REVIEW 6 of 14

(c)

Figure 4. Images of ruby samples from Paranesti. (a) Dark red ruby samples from PAR-1 in pargasite Figure 4. Images of ruby samples from Paranesti. (a) Dark red ruby samples from PAR-1 in pargasite schist host rock 0.5–1.0 cm; (b) Cluster of pale ruby samples from PAR-5 in pargasite schist host rock schist host rock 0.5–1.0 cm; (b) Cluster of pale ruby samples from PAR-5 in pargasite schist host rock 0.5–1.5 cm; (c) Clean PAR-1 ruby sample free from inclusions used for the 2009 Oxygen Isotope 0.5–1.5 cm; (c) Clean PAR-1 ruby sample free from inclusions used for the 2009 Oxygen Isotope analysis. analysis. 2.1. Laser-Fluorination Method (2009) 2.1. Laser-Fluorination Method (2009) In 2009,In 2009, a reconnaissancea reconnaissance studystudy was was conducted, conducted, whereby whereby five individual five individual grains, one grains, each from one each 18 fromdifferent different corundum corundum localities/geological localities/geological environments environments in Greece, in Greece, were werestudied studied for their for theirδ18O δ O composition.composition. These These included included two two colourless colourless to to blue blue sapphires sapphires in in desilicified desilicified pegmatitepegmatite from Naxos, Naxos, one pink marble-hostedone pink marble-hosted ruby from ruby Kimi from and Kimi one purpleand one marble-hosted purple marble-hosted ruby from ruby Xanthi. from OneXanthi. medium One red intensitymedium ruby red in pargasiteintensity ruby schist in frompargasite Paranesti schist (PAR-1)from Paranesti was also (PAR-1) included. was also Oxygen included. isotope Oxygen analyses wereisotope performed analyses using were a modification performed us ofing the a modification laser-fluorination of the laser- techniquefluorination described technique by Sharp described [21] that was similarby Sharp to [21] that that applied was similar by Giuliani to that applied et al. in by 2005 Giuliani [11]. et al. in 2005 [11]. The method involves the complete reaction of ~1 mg of ground corundum. This powder is then The method involves the complete reaction of ~1 mg of ground corundum. This powder is then heated by a CO2 laser, with ClF3 as the fluorine reagent. The released oxygen is passed through an in- heated by a CO2 laser, with ClF3 as the fluorine reagent. The released oxygen is passed through line Hg-diffusion pump before conversion to CO2 on platinized graphite. The yield is then measured an in-lineby a capacitance Hg-diffusion manometer. pump before The ga conversions-handling vacuum to CO2 lineon platinizedis connected graphite. to the inlet The system yield of isa then measureddedicated by a VG capacitance PRISM 3 manometer.dual inlet isotope-ratio The gas-handling mass spectrometer. vacuum line All is oxyg connecteden isotope to the ratios inlet are system of a dedicatedreported in VGδ18O PRISM (‰) relative 3 dual to inlet Vienna isotope-ratio standard mean mass ocean spectrometer. water (VSMOW). All oxygen The secondary isotope ratios are reportedstandard inusedδ18 forO( the) laser-fluorination relative to Vienna method standard was an mean internal ocean quartz water standard, (VSMOW). NBS28 quartz The secondary that standardgave usedan average for the δ18h laser-fluorinationO value of 9.6‰. Oxygen method yields was differing an internal significantly quartz standard,from the theoretical NBS28 quartz value that gaveof an 14.07 average µmol.δ 18perO mg value were of taken 9.6 as. Oxygen likely evidence yields differingof analytical significantly artefact. Precision from theand theoretical accuracy on value of 14.07the µinternalmol. per quartz mg standard were taken areh as± 0.1‰ likely (1 evidenceσ). Duplicate of analyticaland triplicate artefact. analyses Precision of sapphire and and accuracy ruby on suggested that this is appropriate for such materials. the internal quartz standard are ± 0.1 (1σ). Duplicate and triplicate analyses of sapphire and ruby suggested2.2. Secondary that this Ion is Mass appropriate Spectrometry for such(SIMS)h materials. Method (2017) 2.2. SecondaryThe 2017 Ion Massanalyses Spectrometry were performed (SIMS) exclusively Method (2017) on a range of coloured ruby grains from the two distinct Paranesti locations described in the previous study [13]. Unlike the 2009 analyses, these Theanalyses 2017 were analyses made using were secondary performed ionisation exclusively mass spectrometry on a range of(SIMS) coloured to analyse ruby different grains areas from the two distinctof selected Paranesti ruby grains locations in-situ to described measure oxygen in the isotope previous ratios study with less [13 ].than Unlike one per the mil 2009 (‰) level analyses, theseprecision. analyses Oxygen were made isotope using ratios secondary (18O/16O) in ionisation ruby were mass determined spectrometry using a Cameca (SIMS) IMS to analyse 1280 multi- different areascollector of selected ion microprobe ruby grains within in-situ the Centre to measure for Microscopy, oxygen isotope Characterisation ratios with and lessAnalysis than (CMCA), one per mil ( ) levelUniversity precision. of Western Oxygen Australia isotope (UWA). ratios The (18 materialsO/16O) inexamined ruby were in 2017 determined included 3 usingruby grains a Cameca from IMS 1280h multi-collector3 different samples ion (57 microprobe analyses) from within PAR-1 the and Centre 5 ruby for grains Microscopy, from 5 different Characterisation samples (44 analyses) and Analysis from PAR-5. Each analysis point is shown in Figure 5a,b. (CMCA), University of Western Australia (UWA). The materials examined in 2017 included 3 ruby grains from 3 different samples (57 analyses) from PAR-1 and 5 ruby grains from 5 different samples (44 analyses) from PAR-5. Each analysis point is shown in Figure 5a,b. Minerals 2019, 9, 91x FOR PEER REVIEW 77 of of 14

(a) (b)

Figure 5. (a) SIMS in-situin-situ analysisanalysis spotspot location location individually individually marked marked on on the the ruby ruby grain–PAR-5 grain–PAR-5 Grain Grain A; (A;b) ( SIMSb) SIMS in-situ in-situ analysis analysis spot spot location location individually individually marked marked on theon the ruby ruby grain–PAR-5 grain–PAR-5 Grain Grain C. C.

The sample mounts were carefullycarefully cleaned with detergent, distilled water and ethanol in an ultrasonic bath bath and and then then coated coated with with gold gold (30 (30 nm nm in inthickness) thickness) prior prior to SIMS to SIMS O isotope O isotope analyses. analyses. For Foroxygen oxygen isotopic isotopic analyses, analyses, secondary secondary ions were ions spu werettered sputtered from the from sample the sample by bombarding by bombarding its surface its surfacewith a Gaussian with a Gaussian Cs+ beam Cs and+ beam a total and impact a total energy impact of energy 20 keV. of The 20 keV.surface The of surface the sample of the was sample rastered was rasteredwith a 2.5 with nA aprimary 2.5 nA primarybeam over beam a 15 over × 15 a 15µm× area.15 µ mAn area. electron An electrongun was gun used was to usedensure to ensurecharge chargecompensation compensation during duringthe analyses. the analyses. Secondary Secondary ions were ions admitted were admitted in the indouble the double focusing focusing mass massspectrometer spectrometer within within a 100 aµm 100 entranceµm entrance slit and slit focus anded focused in the incentre the centre of a 4000 of a µm 4000 fieldµm aperture field aperture (×100 (magnification).×100 magnification). They were They energy were filtered energy filteredusing a 30 using eV band a 30 eVpass band with pass a 5 eV with gap a 5toward eV gap the toward high- theenergy high-energy side. 16O side.and 1618O andwere18 Ocollected were collected simultaneously simultaneously in multicollection in multicollection mode modein Faraday in Faraday Cup Cupdetectors detectors fitted fittedwith 10 with10 Ω10 and10 Ω1011and Ω, 10respectively.11 Ω, respectively. Each analysis Each analysisincludes includesa pre-sputtering a pre-sputtering over a 20 over× 20 µm a 20 area× 20 duringµm area 30 durings and the 30 sautomatic and the automaticcentring of centring the secondary of the secondaryions in the ions field in aperture, the field aperture,contrast aperture contrast and aperture entrance and slit entrance and consisted slit and consistedof 20 four-second of 20 four-second cycles which cycles give which an average give an averageinternal precision internal precision of ~0.16‰ of (2 ~0.16 SE). (2 SE). External reproducibility during h the analytical sessions was evaluated by repeating analyses in one single fragment of PAR-1. External reproducibility in this fragment was 0.3 and 0.4 per mil (2SD) (2SD) during the two analytical sessions. In In total, total, three three large fragments of PAR-1 were analysed for their oxygen isotope composition, altogether yielding an average value of 0.9 ± 0.60.6 per per mil mil (2SD, n = 57, Table 22).). Raw oxygen isotope ratios were corrected forfor instrumental massmass fractionationfractionation usingusing thethe δδ1818O composition of PAR-1, which oxygen isotope compositioncomposition was obtained by laser fluorination fluorination method (from thethe 20092009 study).study). Uncertainty on each δ1818OO spot has been calculated by propagating the errors on instrumental mass fractionation determination, which include the standard deviation of the mean oxygen isotope ratio measured on thethe primary standard during the session and internal error on each sample data point. Corrected δ18OO (quoted (quoted with with respect respect to to Vienna Vienna Standard Standard Mean Mean Ocean Ocean Water or VSMOW) are presented inin SupplementarySupplementary MaterialMaterial TableTable S1.S1.

3. Results

3.1. Laser-Fluorination Results Using the laser-ﬂuorination method, the oxygen isotope ratio for the pargasite schist hosted PAR-1 ruby was found to be δ18O 1.0‰. This analytical run also included a number of rubies and sapphires from different geological environments. Sapphires from desilicified pegmatites were found

Minerals 2019, 9, 91 8 of 14

Table 2. Oxygen isotope results from 2017 using the SIMS method.

Grain δ18O Min δ18O Max δ18O Mean Number of Analyses PAR-1a 0.64 1.62 1.00 ± 0.42 31 PAR-1b 0.44 1.17 0.67 ± 0.37 13 PAR-1c 0.77 1.68 1.27 ± 0.47 13 PAR-1 Total 0.44 1.68 1.00 ± 0.42 57 PAR-5central −0.04 0.51 0.27 12 PAR-5a −0.14 0.85 0.25 10 PAR-5b −0.31 0.42 0.03 8 PAR-5c −0.22 0.16 −0.06 9 PAR-5d 0.08 0.27 0.17 5 PAR-5 Total −0.31 0.85 0.14 ± 0.24 44 Combined PAR-1 and PAR-5 −0.31 1.31 0.60 93

3. Results

3.1. Laser-Fluorination Results Using the laser-ﬂuorination method, the oxygen isotope ratio for the pargasite schist hosted PAR-1 ruby was found to be δ18O 1.0 . This analytical run also included a number of rubies and sapphires from different geological environments.h Sapphires from desiliciﬁed pegmatites were found to range from 4.8 to 5.0 and rubies from marble-hosted deposits were found to range from 20 to 22 (Table3).h h h h

Table 3. Oxygen isotope results from the 2009 reconnaissance using the laser-fluorination method, n = 1.

Sample Location Sample Type Deposit Type δ18O NAX2 Naxos, Greece Colourless sapphire Desiliciﬁed pegmatite 4.80 NAX3 Naxos, Greece Colourless to blue sapphire Desiliciﬁed pegmatite 5.05 PAR-1 Paranesti, Greece Red ruby Pargasite schist 1.00 KIM2 Kimi, Greece Pink ruby Marble-hosted 20.50 Xanthi Xanthi, Greece Purple-pink ruby Marble-hosted 22.09

3.2. Secondary Ionisation Mass Spectrometry (SIMS) Results The oxygen isotope ratios δ18O (VSMOW) are presented in Table 2. PAR-5 results show values of −0.31 to 0.85 (0.14 ± 0.24), on average slightly lower compared to PAR-1 results 0.44 to 1.68 (1.00 ±h0.42) evenh though the two occurrences are only 500 m apart. h h

4. Discussion

4.1. Corundum Oxygen Isotopes as An Identiﬁer for Geological Origin A framework on the interpretation of the geological origin of gem corundums using the δ18O ratio proposed by Giuliani et al is now widely adopted [11]. Based on this framework, rubies and pink sapphires can be classiﬁed into 5 categories based on their δ18O value range.

1. Mafic gneiss hosted from 2.9 to 3.8 ; 2. Mafic-ultramafic rocks (amphibolite,h h serpentinite) from 3.2 to 6.8 ; 3. Desilicated pegmatites from 4.2 to 7.5 ; h h 4. Shear zones cross-cutting ultramafich h lenses and pegmatites within sillimanite gneisses 11.9 –13.1 ; 5. Marble-hostedh h rubies 16.3 –23 . This framework has been further validated by numerous subsequent corundum oxygenh isotopeh studies [22–25]. Minerals 2019, 9, 91 9 of 14

The reconnaissance 2009 laser fluorination results on the sapphires and rubies from different geological environments very closely fits the oxygen isotope value ranges from the model of Giuliani et al. where over 200 corundum samples were analysed under the same method [1]. That is the marble-hosted value from 20 to 22 is within the range of 16.3 to 23 and the sapphires from the desilicified pegmatites withh a valueh from 4.8 to 5 fits withinh the frameworkh range from 4.2 to 7.5 . Therefore, the δ18O results obtained usingh theh laser-fluorination method in 2009 are furtherh validatedh as accurate measurements.

4.2. PAR-1 vs. PAR-5 Variations The oxygen isotope values obtained using SIMS indicates that the Paranesti rubies have a narrow defined band of oxygen isotope signatures with a mean on +1 (ranging from −0.31 to 1.31 ). This is lower than any ratios based on the existing framework forh rubies. There are furtherh distinctiveh constrained values between PAR-1 (+0.65 to 1.31 ) and PAR-5 (−0.31 to 0.85 ). There is a slight overlap of the individual highest value inh PAR-5 toh the lowest value in PAR-1.h Theh average for PAR-1 is +1 whilst the average for PAR-5 is +0.14 . hThere may be some differences betweenh core-rim oxygen isotope values observed in the PAR-5 SIMS results where the core average (−0.02) is lower than the rim average (0.29). However, this is within the range of uncertainty when the errors are taken into account. This narrow range within individual localities and between the two localities that are 500 m apart is in stark contrast to the findings of Bindeman et al (2010) [7] who found variation within single 10 cm samples of up to 3 and variation within single ruby grains of up to 1.5 . It also does not rule-out variances dueh to partitioning in individual crystals during growth. Therefore,h a detailed cathodoluminescence analysis to determine the homogeneity or heterogeneity of the sample grains is suggested for future studies. As the traditional laser fluorination method consumes the entire grain, such subtle zoning would not be seen using this technique. Thus, the greater spatial resolution of the SIMS technique enables us to analyse discrete isotopic domains (i.e., rims, cores, sectors) within single corundum crystals.

4.3. Global Low to Ultra-Low Oxygen Isotope Corundum Comparison δ18O (SMOW) values for gem corundums below 1 are very rare and not shown on the original systematic framework by Giuliani et al 2005 [11]. Otherh than the Paranesti rubies shown above, the only negative value for corundums are from Karelia in north-western Russia and sapphire from a secondary deposit in Madagascar. Table 4 lists the global low to ultra-low oxygen isotope analyses for corundums. The Madagascar sapphire deposit of Ilakaka with δ18O of −0.3 to 16.5 is a consolidated placer formed in a sandstone environment. The geological origin of theh differenth ranges of isotopic values found for the sapphires corresponds to at least ﬁve different geological environments [26]. The low δ18O delta values for some sapphires correspond up to an unknown geological sapphire type. The PAR-1 result of +1.0 is lower than most known primary corundum oxygen isotope values other than the unique ultra-lowh values of corundums from Karelia [28,30] and one instance of ruby from the Soamiakatra area of Madagascar [26]. The Karelia corundum formed under unique circumstances (see discussion below) and can be easily distinguished from the Paranesti rubies. The Madagascar rubies show much higher average δ18O values and the minimum value obtained corresponds to the maximum value from Paranesti. Therefore, oxygen isotope analysis is a valuable tool that can be used to ﬁngerprint the Paranesti rubies Figure 6 from other worldwide occurrences. Minerals 2019, 9, 91 10 of 14

Table 4. Global comparison of corundums with low oxygen isotope values.

δ18O δ18O Primary vs. Corundum Country District Host Rock (Min)h (Max)h Secondary Type Greece 1 Paranesti-1 * 0.65 1.31 Pargasite schist Primary Ruby Greece 1 Paranesti-5 * −0.31 0.85 Pargasite schist Primary Ruby MineralsMadagascar 2019, 92, x FORSoamiakatra PEER REVIEW * 1.25 4.70 Pyroxenitic enclaves in basalt Primary Ruby10 of 14 Madagascar 2 Ilakaka * −0.30 16.5 Placer in sandstone Secondary Sapphire Madagascar 2 Andilamena * 0.50 3.9 Placer in basalt Secondary Ruby 6 Thailand Russia 3,4,7 Bo RaiKhitostrov * ˆ 1.30 −26.3 4.20− 17.7 Placerplagiogneiss in basalt SecondaryPrimary CorundumRuby Crn-St-Gt-Bi-Prg-Pl rocks with coarse Russia* Individual4,7 grainKhitostrov analysis * ^ Whol−26e-rock - analysis # Only one analysis result, no range.Primary Bi—biotite, Corundum grained Crn Cam—Ca-amphibole,Russia 7 Khitostrov Crn—corundum, * −18.6 -Ged—gedriteKy-Crn-Pl, amphibole, leucocratic Gt—garnet, Ky—kyanite,Primary CorundumPl— plagioclase,Russia 3 Prg—pargasiticVarastskoye ˆ amphibole,−19.2 − St—staurolite.11.3 1 Wangplagiogneiss et al. (2017) [18]; 2 GiulianiPrimary et al. (2007) Corundum 4 − [26];Russia 3 VysotskiyVarastskoye et al. (2015) # [27];17.3 4 Bindeman - and SerebryakovCrn-Cam rock,coarse (2011) grained [28]; 5 Vysotskiy Primary et al. (2014) Corundum Varastskoye # −19.2 - Crn and Crn-St-Pl substituting Ky Primary Corundum [12];Russia 6 4Yui et al. (2006)Dyadina [29]; # 7 Bindeman 0.49 et -al. (2010) Inclusion[7]. of Cam-Crn in giant Gt Inclusion Corundum Russia 4 Dyadina # 0.10 - Crn-Cam rock, coarse grained Primary Corundum Russia 5 Dyadina * 0.4 0.8 Corundum amphibolite Primary Corundum 18 TheRussia Madagascar4 Kulezhma sapphire # deposit 0.31 of - Ilakaka with δCam-CrnO of rock−0.3‰ to 16.5‰Primary is a consolidated Corundum placerRussia formed4 in a Pulongasandstone # environment. 0.67 - The geologicalCrn-Gt-Ged origin rock of the differentPrimary ranges of Corundum isotopic Russia 4 Perusel’ka * 0.26 3.45 Crn-Cam rock, coarse grained Primary Corundum valuesRussia found5 for thePerusel’ka sapphires # corresponds 0.6 - to at leasCorundum-kyanitet five different amphibolite geological environments Primary Corundum [26]. The low δRussia18O delta5 valuesPerusel’ka for some # sapphires 1.5 correspond - upCorundum to an amphiboliteunknown geologicalPrimary sapphire Corundum type. Russia 5 Notozero * −1.7 −1.5 Ged-Gt rocks with Crn and St Primary Corundum TheRussia PAR-14 resultMironova of Guba +1.0‰ ˆ is (2.34) lower than - most knownCam-Crn primary rock corundum oxygenPrimary isotope Corundum values otherThailand than 6the uniqueBo Raiultra-low * values 1.30 of 4.20 corundums fromPlacer Karelia in basalt [28,30] and oneSecondary instance of Ruby ruby from *the Individual Soamiakatra grain analysis area of ˆ Whole-rockMadagascar analysis [26]. # The Only Karelia one analysis corundum result, noformed range. under Bi—biotite, unique Cam—Ca-amphibole, Crn—corundum, Ged—gedrite amphibole, Gt—garnet, Ky—kyanite, Pl—plagioclase, circumstancesPrg—pargasitic (see amphibole,discussion St—staurolite. below) and1 canWang be et ea al.sily (2017) distinguished [18]; 2 Giuliani from et al. the (2007) Paranesti [26]; 3 Vysotskiy rubies. The Madagascaret al. (2015) rubies [27]; 4 showBindeman much and Serebryakovhigher average (2011) [28δ18];O5 Vysotskiy values etand al. (2014)the minimum [12]; 6 Yui et al.value (2006) obtained [29]; 7 correspondsBindeman to etthe al. maximum (2010) [7]. value from Paranesti. Therefore, oxygen isotope analysis is a valuable tool that can be used to fingerprint the Paranesti rubies Figure 6 from other worldwide occurrences.

FigureFigure 6. 6. OxygenOxygen isotopic isotopic comparison comparison of of rubies rubies from from Paranesti Paranesti occurrences occurrences compared compared with with low low δδ1818OO corundumscorundums from from Karelia Karelia in in northwestern northwestern Russia Russia and and Soamiakatra Soamiakatra in in Madagascar. Madagascar.

4.4.4.4. Possible Possible Causes Causes for for Low Low Oxygen Oxygen Isotope Isotope Corundum Corundum Formation Formation 18 ThereThere are are several several current current hypotheses hypotheses on on how how corundums corundums can can form form with with low low δδ18OO isotope isotope ratios. ratios. TheseThese range range from from hydrothermal hydrothermal alteration alteration of of deeply deeply penetrating penetrating surface surface meteoric meteoric waters waters to to isotope isotope separationseparation by by thermal thermal diffusion diffusion during during endogenous endogenous fluid fluid flow flow [31,32]. [31,32]. 4.4.1. Kinetic Isotope Fractionation 4.4.1. Kinetic Isotope Fractionation Kinetic isotope fractionation occurs when rapid thermal decomposition of hydrous phases Kinetic isotope fractionation occurs when rapid thermal decomposition of hydrous phases results in isotope disproportionation into a high-δ18O residue and a low-δ18O fluid [33,34]. However, results in isotope disproportionation into a high-δ18O residue and a low-δ18O fluid [33,34]. However, the high-δ18O residue material should also be found within proximity of the studied samples for this hypothesis to apply. The water-rock interaction is kinetically restricted in supracrustal rocks and isotope fractionation factors are large at low temperatures, favouring higher-δ18O solids [35]. In contrast, isotopic exchange is more rapid within a hydrothermal system. As the Paranesti rubies formed under amphibolite-facies conditions above 600 °C [18], significant kinetic isotope fractionation is highly unlikely and therefore rules out this hypothesis.

4.4.2. Thermal Diffusion

Minerals 2019, 9, 91 11 of 14 the high-δ18O residue material should also be found within proximity of the studied samples for this hypothesis to apply. The water-rock interaction is kinetically restricted in supracrustal rocks and isotope fractionation factors are large at low temperatures, favouring higher-δ18O solids [35]. In contrast, isotopic exchange is more rapid within a hydrothermal system. As the Paranesti rubies formed under amphibolite-facies conditions above 600 ◦C[18], signiﬁcant kinetic isotope fractionation is highly unlikely and therefore rules out this hypothesis.

4.4.2. Thermal Diffusion For thermal diffusion, the oxygen in a temperature gradient is redistributed with low δ18O at the hotter end and high δ18O at the colder end of melt or hydrous solution [31]. Akimova (2015) [32] has proposed a model of cascading thermo-diffusion within shear zones to explain the Karelian ultra-low δ18O corundums. This scenario would require several individual thermal cells to align in the correct position simultaneously with a similar convection rate and timing. Given that only two locations have shown ruby-bearing pargasite schist with other very similar pargasite boudins nearby being ruby absent, this model appears to be less likely than the hydrothermal scenario.

4.4.3. Other Ultra-Low δ18O Protoliths Ultra-low δ18O protoliths could potentially provide the low δ18O during corundum syn-metamorphic formation [35]. However, a source for the ultra-low δ18O protolith would be needed under this scenario such as a low δ18O mantle reservoir or previously surface-exposed and then rapidly buried metamorphic rocks. Neither were observed at Paranesti. As oxygen isotope analyses were not performed on the whole-rock and associated mineral phases for the Paranesti occurrences, this hypothesis cannot be ruled out.

4.4.4. Hydrothermal Alteration Model This hypothesis involves the conservation of the initial isotopic ratios of the protolith in the corundum-bearing rocks and then isotopic exchange between these rocks and meteoric waters before metamorphism [10]. Wang et al. [18] demonstrated that the rubies from Paranesti were syn-metamorphic and were largely free of inclusions. However, as shown by Bindeman et al (2010) [7], this does not preclude preservation of the initial ratios within the protolith for the Paranesti occurrence. Therefore, it is unlikely that the low isotopic ratios observed within the Paranesti rubies were due to preservation of the initial ratios within the protolith. For granulite facies metamorphism, Wilson and Banksi (1983) [36] proposed three processes that could produce a low oxygen isotope value. These are (1) pre-granulite reaction between heated seawater and hot basic intrusives or an initial protolith such as a palaeosol for the sapphirine–spinel–(cordierite) assemblages; (2) syn-granulite depletion in 18O related to dehydration during granulite metamorphism and removal of the resultant products of partial melting with a depletion in 18O by up to 2 or 3 for the restite; and (3) post-granulite facies metamorphism with recrystallization under theh effect ofh biotite and/or amphibole-metasomatism with depletion in δ18O up to 4 . Based on the previous study [18], the Paranesti rubies were found to have formed under amphiboliteh facies conditions and there is no evidence that they ever reached granulite facies within the specified zone. However, there are other locations within the Rhodope Mountain Complex (RMC) where regional metamorphism reached granulite facies conditions, though these are some distance away from Paranesti and no rubies are known from these locations. The glacial meltwater influence during formation of corundums was proposed to explain the ultra-low δ18O isotopic ratios observed for corundums from Karelia in north-western Russia [10,12,27]. However, there is no evidence suggesting the existence of glaciers in the Mediterranean region based on the reconstruction of the tectonic evolution of the East Mediterranean region since the late Cretaceous [37]. Could the RMC be a higher mountain with glaciers that have melted during ruby genesis? There is no evidence in the literature to suggest such and this would only be a remote Minerals 2019, 9, 91 12 of 14 possibility. Therefore, it is unlikely that a glacial meltwater source played a role during ruby formation at Paranesti. However, it is likely that meteoric water (but not glacial melt) interaction caused by downward flow of surface waters along deep crustal fractures/structures during the formation of the corundum would contribute in producing low δ18O values for the Paranesti rubies.

5. Conclusions The in-situ SIMS oxygen isotope analyses on the Paranesti rubies is the ﬁrst time that a primary (and exclusively) ruby occurrence was found to have ~+1 for its δ18O-isotope composition. Based on the low δ18O value and the local geology, it is mosth likely that the Paranesti rubies formed under metamorphic/metasomatic conditions involving deeply penetrating meteoric waters along major crustal structures related to the Nestos Shear Zone. PAR-5 is potentially closer to the source of the hydrothermal inﬂuence during ruby formation compared to PAR-1 and thus has a lower δ18O. Importantly, this study shows that in-situ gem corundum oxygen isotope analysis using the SIMS method may be used to determine the likely geographic origin for corundums lacking any provenance details. Importantly, with the SIMS method being only minimally destructive, with the analysis spot (15 × 15 µm) amenable to repolishing, a wider adoption of this technique has important applications/implications for the international gem and jewellery industry. A future area of research would be to apply this methodology for more gem mineral varieties other than corundum and emeralds. The aim of such future work would be to determine if in-situ SIMS oxygen isotope analysis can be used to both better understand gem formation and to see if it can be used to clearly separate the same gem mineral from different geographic locations.

Supplementary Materials: The following are available online at http://www.mdpi.com/2075-163X/9/2/91/s1, Table S1: Secondary Ionisation Mass Spectrometry (SIMS) Results. Author Contributions: K.K.W. wrote the manuscript and interpreted the results of the analyses. I.T.G. collected the samples, provided technical input, funding and supervised the project. L.M. provided technical input and ran the SIMS analyses. P.V. collected the samples and provided geological expertise on the Paranesti region. G.G. and A.F. coordinated the laser fluorination analyses and provided technical input. A.L. and S.J.H. provided technical input. Funding: This research received no external funding. Acknowledgments: The authors would like to thank Joanne Wilde, formerly of the School of Biological, Earth and Environmental Sciences, UNSW Sydney, for making the polished mounts required for SIMS analysis. The authors acknowledge the facilities and the scientific and technical assistance of the Australian Microscopy & Microanalysis Research Facility at the Centre for Microscopy, Characterisation & Analysis, The University of Western Australia, a facility funded by the University, State and Commonwealth Governments. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Hoefs, J. Stable Isotope Geochemistry; Springer: Berlin/Heidelberg, Germany, 1997; ISBN 9783662033791. 2. Bowen, G.J.; Wassenaar, L.I.; Hobson, K.A. Global application of stable hydrogen and oxygen isotopes to wildlife forensics. Oecologia 2005, 143, 337–348. [CrossRef][PubMed] 3. Miller, K.G.; Fairbanks, R.G.; Mountain, G.S. Tertiary oxygen isotope synthesis, sea level history and continental margin erosion. Paleoceanography 1987, 2, 1–19. [CrossRef] 4. Criss, R.E.; Taylor, H.P., Jr. Meteoric-hydrothermal systems. Rev. Mineral. 1986, 16, 373–424. 5. Valley, J.W. Stable Isotope Thermometry at High Temperatures. Rev. Mineral. Geochem. 2001, 43, 365–413. [CrossRef] 6. Bindeman, I. Oxygen Isotopes in Mantle and Crustal Magmas as Revealed by Single Crystal Analysis. Rev. Mineral. Geochem. 2008, 69, 445–478. [CrossRef] 7. Bindeman, I.N.; Schmitt, A.K.; Evans, D.A.D. Limits of hydrosphere-lithosphere interaction: Origin of the lowest-known δ18O silicate rock on Earth in the Paleoproterozoic Karelian rift. Geology 2010, 38, 631–634. [CrossRef] Minerals 2019, 9, 91 13 of 14

8. Giuliani, G.; Ohnenstetter, D.; Garnier, V.; Fallick, A.E.; Rakotondrazafy, M.; Schwarz, D. The geology and genesis of gem corundum deposits. In Geology of Gem Deposits; Groat, L.A., Ed.; Mineralogical Association of Canada: Québec, QC, Canada, 2007; Volume 37, pp. 23–78. ISBN 9780921294375. 9. Yui, T.-F.; Khin, Z.; Limtrakun, P. Oxygen isotope composition of the Denchai sapphire, Thailand: A clue to its enigmatic origin. Lithos 2003, 67, 153–161. [CrossRef] 10. Krylov, D.P.; Glebovitsky, V.A. Oxygen isotopic composition and nature of fluid during the formation of high-Al corundum-bearing rocks of Mt. Dyadina, northern Karelia. Doklady Earth Sci. 2007, 413, 210–212. [CrossRef] 11. Giuliani, G.; Fallick, A.E.; Garnier, V.; France-Lanord, C.; Ohnenstetter, D.; Schwarz, D. Oxygen isotope composition as a tracer for the origins of rubies and sapphires. Geology 2005, 33, 249. [CrossRef] 12. Vysotskiy, S.V.; Ignat’ev, A.V.; Levitskii, V.I.; Nechaev, V.P.; Velivetskaya, T.A.; Yakovenko, V.V. Geochemistry of stable oxygen and hydrogen isotopes in minerals and corundum-bearing rocks in northern Karelia as an indicator of their unusual genesis. Geochem. Int. 2014, 52, 773–782. [CrossRef] 13. Bonev, N.; Burg, J.-P.; Ivanov, Z. Mesozoic–Tertiary structural evolution of an extensional gneiss dome—The Kesebir–Kardamos dome, eastern Rhodope (Bulgaria–Greece). Int. J. Earth Sci. 2006, 95, 318–340. [CrossRef] 14. Krenn, K.; Bauer, C.; Proyer, A.; Mposkos, E.; Hoinkes, G. Fluid entrapment and reequilibration during subduction and exhumation: A case study from the high-grade Nestos shear zone, Central Rhodope, Greece. Lithos 2008, 104, 33–53. [CrossRef] 15. Barr, S.R.; Temperley, S.; Tarney, J. Lateral growth of the continental crust through deep level subduction–accretion: A re-evaluation of central Greek Rhodope. Lithos 1999, 46, 69–94. [CrossRef] 16. Nagel, T.J.; Schmidt, S.; Janák, M.; Froitzheim, N.; Jahn-Awe, S.; Georgiev, N. The exposed base of a collapsing wedge: The Nestos Shear Zone (Rhodope Metamorphic Province, Greece): The Nestos Shear Zone. Tectonics 2011, 30.[CrossRef] 17. Moulas, E.; Schenker, F.L.; Burg, J.-P.; Kostopoulos, D. Metamorphic conditions and structural evolution of the Kesebir-Kardamos dome: Rhodope metamorphic complex (Greece-Bulgaria). Int. J. Earth Sci. 2017, 106, 2667–2685. [CrossRef] 18. Wang, K.K.; Graham, I.T.; Lay, A.; Harris, S.J.; Cohen, D.R.; Voudouris, P.; Belousova, E.; Giuliani, G.; Fallick, A.E.; Greig, A. The Origin of a New Pargasite-Schist Hosted Ruby Deposit from Paranesti, Northern Greece. Can. Mineral. 2017, 55, 535–560. [CrossRef] 19. Sutherland, F.; Zaw, K.; Meffre, S.; Yui, T.-F.; Thu, K. Advances in Trace Element “Fingerprinting” of Gem Corundum, Ruby and Sapphire, Mogok Area, Myanmar. Minerals 2014, 5, 61–79. [CrossRef] 20. Giuliani, G.; Ohnenstetter, D.; Fallick, A.E. The geology and genesis of gem corundum deposits. In Geology of Gem Deposits. Series: Mineralogical Association of Canada Short Course Series, 44; Groat, L.A., Ed.; Mineralogical Association of Canada: Québec, QC, Canada, 2014; pp. 35–112. ISBN 9780921294542. 21. Sharp, Z.D. In situ laser microprobe techniques for stable isotope analysis. Chem. Geol. 1992, 101, 3–19. [CrossRef] 22. Garnier, V.; Giuliani, G.; Ohnenstetter, D.; Fallick, A.E.; Dubessy, J.; Banks, D.; Vinh, H.Q.; Lhomme, T.; Maluski, H.; Pêcher, A.; et al. Marble-hosted ruby deposits from Central and Southeast Asia: Towards a new genetic model. Ore Geol. Rev. 2008, 34, 169–191. [CrossRef] 23. Zaw, K.; Sutherland, L.; Yui, T.-F.; Meffre, S.; Thu, K. Vanadium-rich ruby and sapphire within Mogok Gemfield, Myanmar: Implications for gem color and genesis. Miner. Deposita 2015, 50, 25–39. [CrossRef] 24. Graham, I.; Sutherland, L.; Zaw, K.; Nechaev, V.; Khanchuk, A. Advances in our understanding of the gem corundum deposits of the West Pacific continental margins intraplate basaltic fields. Ore Geol. Rev. 2008, 34, 200–215. [CrossRef] 25. Sutherland, F.L.; Duroc-Danner, J.M.; Meffre, S. Age and origin of gem corundum and zircon megacrysts from the Mercaderes–Rio Mayo area, South-west Colombia, South America. Ore Geol. Rev. 2008, 34, 155–168. [CrossRef] 26. Giuliani, G.; Fallick, A.; Rakotondrazafy, M.; Ohnenstetter, D.; Andriamamonjy, A.; Ralantoarison, T.; Rakotosamizanany, S.; Razanatseheno, M.; Offant, Y.; Garnier, V.; et al. Oxygen isotope systematics of gem corundum deposits in Madagascar: Relevance for their geological origin. Miner. Deposita 2007, 42, 251–270. [CrossRef] Minerals 2019, 9, 91 14 of 14

27. Vysotskiy, S.V.; Nechaev, V.P.; Kissin, A.Y.; Yakovenko, V.V.; Ignat’ev, A.V.; Velivetskaya, T.A.; Sutherland, F.L.; Agoshkov, A.I. Oxygen isotopic composition as an indicator of ruby and sapphire origin: A review of Russian occurrences. Ore Geol. Rev. 2015, 68, 164–170. [CrossRef] 28. Bindeman, I.N.; Serebryakov, N.S. Geology, Petrology and O and H isotope geochemistry of remarkably 18O depleted Paleoproterozoic rocks of the Belomorian Belt, Karelia, Russia, attributed to global glaciation 2.4 Ga. Earth Planet. Sci. Lett. 2011, 306, 163–174. [CrossRef] 29. Yui, T.-F.; Wu, C.-M.; Limtrakun, P.; Sricharn, W.; Boonsoong, A. Oxygen isotope studies on placer sapphire and ruby in the Chanthaburi-Trat alkali basaltic gemﬁeld, Thailand. Lithos 2006, 86, 197–211. [CrossRef] 30. Krylov, D.P. Anomalous 18O/16O ratios in the corundum-bearing rocks of Khitostrov, northern Karelia. Doklady Earth Sci. 2008, 419, 453–456. [CrossRef] 31. Bindeman, I.N.; Lundstrom, C.C.; Bopp, C.; Huang, F. Stable isotope fractionation by thermal diffusion through partially molten wet and dry silicate rocks. Earth Planet. Sci. Lett. 2013, 365, 51–62. [CrossRef] 32. Akimova, E.Y.; Lokhov, K.I. Ultralight Oxygen in Corundum-Bearing Rocks of North Karelia, Russia, as a Result of Isotope Separation by Thermal Diffusion (Soret Effect) in Endogenous Fluid Flow. J. Mater. Sci. Chem. Eng. 2015, 3, 42–47. [CrossRef] 33. Clayton, R.N.; Mayeda, T.K. Kinetic Isotope Effects in Oxygen in the Laboratory Dehydration of Magnesian Minerals. J. Phys. Chem. 2009, 113, 2212–2217. [CrossRef] 34. Mendybayev, R.A.; Richter, F.M.; Spicuzza, M.J.; Davis, A.M. Oxygen isotope fractionation during evaporation of Mg- and Si- rich CMAS-liquid in vacuum. Lunar Planet. Inst. Sci. Conf. Abstr. 2010, 41, 2725. 35. Bindeman, I.N.; Serebryakov, N.S.; Schmitt, A.K.; Vazquez, J.A.; Guan, Y.; Azimov, P.Y.; Astaﬁev, B.Y.; Palandri, J.; Dobrzhinetskaya, L. Field and microanalytical isotopic investigation of ultradepleted in 18O Paleoproterozoic “Slushball Earth” rocks from Karelia, Russia. Geosphere 2014, 10, 308–339. [CrossRef] 36. Wilson, A.F.; Baksi, A.K. Widespread 18O depletion in some precambrian granulites of Australia. Precambrian Res. 1983, 23, 33–56. [CrossRef] 37. Menant, A.; Jolivet, L.; Vrielynck, B. Kinematic reconstructions and magmatic evolution illuminating crustal and mantle dynamics of the eastern Mediterranean region since the late Cretaceous. Tectonophysics 2016, 675, 103–140. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). Appendix 3

Gem Corundum Deposits of Greece: Geology, Mineralogy and Genesis.

Published in Minerals (MDPI) Vol. 9, pp. 1-41 (2019) DOI: 10.3390/min9010049

174 minerals

Article Gem Corundum Deposits of Greece: Geology, Mineralogy and Genesis

Panagiotis Voudouris 1,* , Constantinos Mavrogonatos 1 , Ian Graham 2 , Gaston Giuliani 3, Vasilios Melfos 4 , Stefanos Karampelas 5, Vilelmini Karantoni 4, Kandy Wang 2, Alexandre Tarantola 6 , Khin Zaw 7, Sebastien Meffre 7 , Stephan Klemme 8, Jasper Berndt 8, Stefanie Heidrich 9, Federica Zaccarini 10, Anthony Fallick 11, Maria Tsortanidis 1 and Andreas Lampridis 1

1 Faculty of Geology and Geoenvironment, National and Kapodistrian University of Athens, 15784 Athens, Greece; [email protected] (C.M.); [email protected] (M.T.); [email protected] (A.L.) 2 School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia; [email protected] (I.G.); [email protected] (K.W.) 3 Université Paul Sabatier, GET/IRD et Université de Lorraine, C.R.P.G./C.N.R.S., 54501 Vandoeuvre, CEDEX, France; [email protected] 4 Faculty of Geology, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece; [email protected] (V.M.); [email protected] (V.K.) 5 Bahrain Institute for Pearls & Gemstones (DANAT), WTC East Tower, P.O. Box 17236 Manama, Bahrain; [email protected] 6 UMR GeoResources, Faculté des Sciences et Technologies, Université de Lorraine, 54506 Vandoeuvre-lès-Nancy, France; [email protected] 7 CODES Centre of Ore Deposit and Earth Sciences, University of Tasmania, Tas 7001, Australia; [email protected] (K.Z.); [email protected] (S.M.) 8 Institut für Mineralogie, Westfälische-Wilhelms Universität Münster, 48149 Münster, Germany; [email protected] (S.K.); [email protected] (J.B.) 9 Mineralogisch-Petrographisches Institut, Universität Hamburg, 20146 Hamburg, Germany; [email protected] 10 Department of Applied Geosciences and Geophysics, University of Leoben, 8700 Leoben, Austria; [email protected] 11 Isotope Geosciences Unit, S.U.E.R.C., Glasgow G75 0QF, UK; [email protected] * Correspondence: [email protected]; Tel.: +30-210-727-4129

Received: 13 December 2018; Accepted: 14 January 2019; Published: 15 January 2019

Abstract: Greece contains several gem corundum deposits set within diverse geological settings, mostly within the Rhodope (Xanthi and Drama areas) and Attico-Cycladic (Naxos and Ikaria islands) tectono-metamorphic units. In the Xanthi area, the sapphire (pink, blue to purple) deposits are stratiform, occurring within marble layers alternating with amphibolites. Deep red rubies in the Paranesti-Drama area are restricted to boudinaged lenses of Al-rich metapyroxenites alternating with amphibolites and gneisses. Both occurrences are oriented parallel to the ultra-high pressure/high pressure (UHP/HP) Nestos suture zone. On central Naxos Island, colored sapphires are associated with desilicated granite pegmatites intruding ultramaﬁc lithologies (plumasites), occurring either within the pegmatites themselves or associated metasomatic reaction zones. In contrast, on southern Naxos and Ikaria Islands, blue sapphires occur in extensional ﬁssures within Mesozoic metabauxites hosted in marbles. Mineral inclusions in corundums are in equilibrium and/or postdate corundum crystallization and comprise: spinel and pargasite (Paranesti), spinel, zircon (Xanthi), margarite, zircon, apatite, diaspore, phlogopite and chlorite (Naxos) and chloritoid, ilmenite, hematite, ulvospinel, rutile and zircon (Ikaria). The main chromophore elements within the Greek corundums show a wide range in concentration: the Fe contents vary from (average values) 1099 ppm in the blue sapphires of Xanthi, 424 ppm in the pink sapphires of Xanthi, 2654 ppm for Paranesti rubies, 4326 ppm

Minerals 2019, 9, 49; doi:10.3390/min9010049 www.mdpi.com/journal/minerals Minerals 2019, 9, 49 2 of 41

for the Ikaria sapphires, 3706 for southern Naxos blue sapphires, 4777 for purple and 3301 for pink sapphire from Naxos plumasite, and finally 4677 to 1532 for blue to colorless sapphires from Naxos plumasites, respectively. The Ti concentrations (average values) are very low in rubies from Paranesti (41 ppm), with values of 2871 ppm and 509 in the blue and pink sapphires of Xanthi, respectively, of 1263 ppm for the Ikaria blue sapphires, and 520 ppm, 181 ppm in Naxos purple, pink sapphires, respectively. The blue to colorless sapphires from Naxos plumasites contain 1944 to 264 ppm Ti, respectively. The very high Ti contents of the Xanthi blue sapphires may reflect submicroscopic rutile inclusions. The Cr (average values) ranges from 4 to 691 ppm in the blue, purple and pink colored corundums from Naxos plumasite, is quite fixed (222 ppm) for Ikaria sapphires, ranges from 90 to 297 ppm in the blue and pink sapphires from Xanthi, reaches 9142 ppm in the corundums of Paranesti, with highest values of 15,347 ppm in deep red colored varieties. Each occurrence has both unique mineral assemblage and trace element chemistry (with variable Fe/Mg, Ga/Mg, Ga/Cr and Fe/Ti ratios). Additionally, oxygen isotope compositions confirm their geological typology, i.e., with, respectively δ18O of 4.9 ± 0.2‰ for sapphire in plumasite, 20.5‰ for sapphire in marble and 1‰ for ruby in mafics. The fluid inclusions study evidenced water free CO2 dominant fluids with traces 3 of CH4 or N2, and low CO2 densities (0.46 and 0.67 g/cm ), which were probably trapped after the metamorphic peak. The Paranesti, Xanthi and central Naxos corundum deposits can be classified as metamorphic sensu stricto (s.s.) and metasomatic, respectively, those from southern Naxos and Ikaria display atypical magmatic signature indicating a hydrothermal origin. Greek corundums are characterized by wide color variation, homogeneity of the color hues, and transparency, and can be considered as potential gemstones.

Keywords: corundum megacrysts; ruby; sapphire; plumasite; metamorphic-metasomatic origin; Greece

1. Introduction Rubies and sapphires, the two different varieties of the mineral corundum, are among the most popular gemstones used in jewelry. The color in corundum varies from brown, pink to pigeon-blood-red, orange, yellow, green, blue, violet etc. The main chromophore trace elements in corundum are Cr, Fe, Ti and V. Cr3+ produces pink and red, while Fe2+–Ti4+ pairs produce blue. Distinction of primary gem quality-rubies and sapphires in magmatic and metamorphic is mainly based on the classification schemes presented by Garnier et al. [1], Giuliani et al. [2] and Simonet et al. [3]. Corundum deposits were mainly formed in three periods, the Pan-African orogeny (750–450 Ma), the Himalayan orogeny (45–5 Ma) and the Cenozoic continental rifting with related alkali basalt volcanism (65–1 Ma) [2,4–10]. Magmatic and metamorphic ruby and sapphire deposits related to the Pan-African orogeny are present in southern Madagascar, East Africa, South India and Sri Lanka; those related to the Himalayan orogeny mostly occur in marbles from Central and Southeast Asia. Alkali basalt-related sapphires occur in Australia, northern and central Madagascar, Nigeria, Cameroon, French Massif Central and southeast Asia. The use of trace element content of gem corundums, in association with their oxygen isotopic signature, is an effective tool in characterizing and interpreting their origin, and can also be used for the exploration, classification and comparison of gem corundum deposits especially those of disputed origin (e.g., placers, xenoliths, etc.) [2,7,11–17]. The presence of corundum in Greece (Xanthi, Naxos and Ikaria islands) has been known of since several years ago, but mostly from unpublished company reports [18–20]. It was only during the 80s, when detailed mineralogical, petrological studies investigated the geological environment of formation for the emery-hosted corundum on Naxos, Samos and Ikaria island [21], and the marble-hosted corundum in Xanthi area [22,23]. Iliopoulos and Katagas [24] described metamorphic Minerals 2019, 9, 49 3 of 41 conditions of formation for corundum-bearing metabauxites from Ikaria island. Fieldwork during 1998–2010 by the first two authors resulted in additional findings of corundum megacrysts in Greece, mostly within the Rhodope (Xanthi and Drama areas) and Attico-Cycladic (Naxos and Ikaria islands) tectono-metamorphic units [25–29]. According to these authors, both occurrences in the Rhodope are classified as metamorphic deposits related to meta-limestones (Xanthi deposit) and mafic granulites (Paranesti deposit), based on the scheme of Simonet et al. [3], and supported the theory of their formation during the high temperature-medium pressure retrograde metamorphic episode of carbonates and eclogitic amphibolites during the Cenozoic collision along the Nestos Suture Zone. The plumasite-hosted corundum deposits at Naxos island and those hosted in metabauxites from Ikaria island are classified as metasomatic and metamorphic, respectively [27,29]. Recently, Wang et al. [30] compiled multi-analytical geochemical, mineralogical and petrological studies on rubies from Paranesti Drama area, discussing their origin and comparing the deposit to those in other mafic-ultramafic complexes especially that of Winza in Tanzania. In addition to the above localities, corundum has been also reported from the Koryfes Hill prospect, which is a telescoped porphyry-epithermal system hosted within Tertiary granitoids in the Kassiteres-Sapes area, on the south of the Rhodope massif (Figure 1). Hydrothermal corundum occurs within a transitional sericitic-sodic/potassic alteration of a quartz-feldspar porphyry, which forms the root zone of an advanced argillic lithocap comprising diaspore, topaz, pyrophyllite and alunite supergroup minerals [31]. Hydrothermal corundum forms up to 3-mm-large aggregates of deep blue-colored crystals that are separated from quartz (and K-feldspar/albite) by a sericite rim. Electron microprobe analyses indicate minor amounts of Fe (up to 0.34 wt. % FeO), Mg (up to 0.02 wt. % MgO) Ti (up to 0.13 wt. % TiO2) and Cr (up to 0.12 wt. % Cr2O3) substituting for Al in the structure. The Koryfes Hill corundum occurrence belongs to those present in porphyry deposits elsewhere and is attributed to the magmatic class of corundum deposits, according to the classification scheme of Simonet et al. [3]. It was suggested by Voudouris [31] that the assemblage corundum-sericite at Koryfes Hill formed as a result of rapid cooling of ascending magmatic-hydrothermal solutions under pressures between 0.6 and 0.3 kb. However, due to its small grain size, which probably rather prevent its use as a gemstone this corundum occurrence will not be further discussed in the present paper. The aim of this work is to present detailed information on the geology, mineralogy, geochemistry and fluid characteristics involved in the formation of rubies and sapphires in Greece, by expanding on the previous work of the Wang et al. [30], which was only focused on the Paranesti area. New laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) and electron probe microanalysis (EPMA) data on corundum will help define their origin and compare them with other famous occurrences elsewhere.

2. Materials and Methods For this study, the following 18 samples, collected by the first author, were investigated: Naxos samples (Nx1a, Nx1b, Nx2a, Nx2b, Nx3a, Nx3b, Nx4, Nx5a, Nx5b), Ikaria samples (Ik1a, Ik1b), Xanthi samples (Go1a, Go1b, Go5a, Go5b) and Drama samples (Dr1a, Dr1b, Dr2), ranging in color from colorless, to pink, purple, red and blue (Table 1). All occurrences are primary, which is rare for blue sapphire and relatively rare for ruby (S. Karampelas, pers. communication). Thirty-five thin and ten thin-and-polished sections of corundum-bearing samples and host rocks were studied by optical and a JEOL JSM 5600 scanning electron microscope (SEM) equipped with back-scattered imaging capabilities, respectively, at the Department of Mineralogy and Petrology at the University of Athens (Athens, Greece). Quantitative analyses were carried out at the “Eugen F. Stumpfl” Laboratory installed at the University of Leoben, Austria, using a Superprobe Jeol JXA 8200 wavelength-dispersive electron microprobe (WDS), and at the Department of Mineralogy and Petrology, University of Hamburg (Hamburg, Germany) using a Cameca-SX 100 WDS. Analytical conditions were as follows: at Leoben, 20 kV accelerating voltage, 10 nA beam current, 1 µm beam diameter. Peak and backgrounds counting times were 20 and 10 s for major and 40 and 20 s for trace Minerals 2019, 9, 49 4 of 41 elements, respectively. The X-ray lines used were AlKα, SiKα, TiKα, GaKα, FeKα, CeLα, VKα, MgKα, CrKα, and CaLα. The standards used were: corundum for Al, wollastonite for Si and Ca, rutile for Ti, GaAs for Ga, ilmenite for Fe, synthetic V for V, chromite for Mg and Cr. At Hamburg, accelerating voltage of 20 kV, a beam current of 20 nA and counting time of 20 s. The X-ray lines used were: AlKα, SiKα, TiKα, GaKα, FeKα, CeLα, VKα, MgKα, CrKb, and CaLα. The standards used were: andradite and vanadinite (for Si, Ca and V), and synthetic Al2O3 (for Al), TiO2 (for Ti), Fe2O3 (for Fe), Cr2O3 (for Cr), Ga2O3 (for Ga) and MgO (for Mg). Corrections were applied using the PAP online program [32].

Table 1. Sample description of the studied Greek corundum crystals.

Sample Variety Color Host Rock Locality Dr1a, Dr1b, Dr2 Ruby Red Pargasite-schist Paranesti, Drama Go1a, Go1b Sapphire Blue-colorless Marble Gorgona, Xanthi Go5a Sapphire Pink-purple Marble Gorgona, Xanthi Go5b Sapphire Pink Marble Gorgona, Xanthi Nx1a, Nx2a, Nx2b, Sapphire Blue-colorless Plumasite Naxos (Kinidaros) Nx3a, Nx3b, Nx4 Nx1b Sapphire Purple Plumasite Naxos (Kinidaros) Nx5a Sapphire Blue Metabauxite Naxos (Kavalaris) Nx5b Sapphire Pink Plumasite Naxos (Kinidaros) Ik1a, Ik1b Sapphire Blue Metabauxite Ikaria

LA-ICP-MS analyses were conducted at the CODES ARC Centre of Excellence in Ore Deposits of the University of Tasmania, Australia, and the Institute of Mineralogy, University of Münster, Germany. Analytical conditions were as follows: The LA-ICP-MS analysis at the CODES was performed under standard procedures using a New Wave UP-213 Nd: YAG Q-switched Laser Ablation System coupled with an Agilent HP 4500 Quadrupole ICP-MS. The international standard NIST 612 was used as the primary standard to calculate concentrations and correct for ablation depth, and the basaltic glass BCR-2 was used as the secondary standard. Analyses were standardized on the theoretical Al content for corundum at Al 529,227 ppm, with an error in analytical precision of 2–3%. Corundum sample ablation at the University of Münster was done with a pulsed 193 nm ArF excimer laser (Analyte G2, Photon Machines). A repetition rate of 10 Hz and an energy of ~4 J/cm2 were used throughout the entire session. The beam spot diameter was set to 35 µm. Trace element analysis has been carried out with an Element 2 mass spectrometer (ThermoFisher). Forward power was 1250 W and reflected power <1 W, gas flow rates were 1.2 L/m for He carrier gas, 0.9 L/m and 1.2 L/m for the Ar-auxiliary and sample gas, respectively. Argon cooling gas flow rate was set to 16 L/min. Before starting analysis, the system was tuned on a NIST 612 reference glass measuring 139La, 232Th and 232Th16O to get stable signals and high sensitivity, as well as low oxide rates (232Th16O/232Th < 0.1%) during ablation. A total of 30 elements were quantitatively analyzed using the NIST 612 glass as an external standard and 27Al as internal standard, which had previously been determined by electron microprobe. Overall time of a single analysis was 60 s (20 s for background, 40 s for peak after switching laser on). Concentrations of measured elements were calculated using the Glitter software [33,34]. Standard reference glass BHVO-2G and BIR1-G were analyzed in order to monitor for precision and accuracy in the silicate phases in the course of this study. The obtained results match the published range of concentrations given in the GeoReM database (version 23). Fluid inclusions studies were carried in a total of 14 double-polished sections prepared at the Department of Mineralogy, Petrology and Economic Geology of the Aristotle University of Thessaloniki (Thessaloniki, Greece). Routine heating and freezing runs were performed at a LINKAM THM-600/TMS 90 heating–freezing stage coupled to a Leitz SM-LUX-POL microscope. Calibration of the stage was achieved using organic standards with known melting points, and the H2O phase transition from ice to liquid; the precision of the measurements, including reproducibility, was ±0.2 ◦C. The SoWat program [35] was used to process fluid inclusion data. Minerals 2019, 9, 49 5 of 41

Oxygen isotope analyses were performed at the Scottish Universities Environmental Research Centre, Glasgow, Scotland. The oxygen isotope composition of corundum was obtained using a modification of the CO2 laser fluorination system similar to that described by Sharp [36], which was similar applied by Giuliani et al. [14]. The method involves the complete reaction of ~1 mg of ground corundum. This powder is heated by a CO2 laser, with ClF3 as the fluorine reagent. The released oxygen is passed through an in-line Hg-diffusion pump before conversion to CO2 on platinized graphite. The yield is then measured by a capacitance manometer. The gas-handling vacuum line is connected to the inlet system of a dedicated VG PRISM 3 dual-inlet isotope-ratio mass spectrometer. Duplicate analyses of corundum samples suggest that precision and accuracy are ±0.1‰ (1σ). All oxygen isotope ratios are shown in δ18O(‰) relative to Vienna standard mean ocean water (VSMOW).

3. Geological Setting

3.1. Regional Geology The Hellenide orogen formed as a result of the collision between the African and Eurasian plates above the north-dipping Hellenic subduction zone from the Late Jurassic to the present [37–39]. From north to south, it consists of three continental blocks (Rhodopes, Pelagonia, and Adria-External Hellenides) and two oceanic domains (Vardar and Pindos Suture Zones). In the Aegean region, continuous subduction of both oceanic and continental lithosphere beneath the Eurasian plate since the Early Cretaceous resulted in a series of magmatic arcs from the north (Rhodope massif) to the south (Active South Aegean Volcanic Arc) [40]. The geodynamic evolution of Hellenides is characterized by a collisional phase taking place during the Mesozoic, followed by a continuous southward slab retreat in a back-arc setting, started at the Eocene but still ongoing, that triggered large-scale extension concomitant with thrusting at the southern part of the Hellenic domain [39,41]. The gem corundum deposits in Greece are spotted in two tectono-metamorphic units of the greater Hellenides Orogen: the Rhodope- and the Attico-Cycladic Massifs [29,30,42].

3.1.1. Rhodope Massif The Rhodope Massif is considered part of the European continental margin [41,43]. It is a heterogeneous crustal body composed in its eastern and central parts of two sub-domains (Figure 1): the Northern Rhodope Domain and the Southern Rhodope Core Complex (both separated by the Nestos thrust fault and the Nestos Suture Zone) [41]. The Northern Rhodope Domain consists of a Lower unit of high-grade basement including orthogneisses derived from Permo-Carboniferous protoliths; this Unit includes four metamorphic core complexes (the Arda, Biala Reka-Kechros, and Kesebir-Kardamos migmatitic domes) which are considered to be equivalent to orthogneisses in the Southern Rhodope Core Complex (SRCC) [41,44]. The latter also consists of Triassic marbles with intercalated amphibolitic and metapelitic rocks [44,45]. The upper tectonic unit of the Northern Rhodope Domain includes high-grade basement rocks that have both continental and oceanic afﬁnities, and with protoliths ranging in age from Neoproterozoic through Ordovician, and Permo-Carboniferous to Early Cretaceous [45]. The rocks of the Intermediate unit experienced high- to ultra-high-pressure metamorphism with subsequent high-grade amphibolite-facies overprint [46]. The Rhodope Massif has undergone a polycyclic alpine orogeny including UHP-eclogitic metamorphism, followed by HP granulite-facies, amphibolite-facies and ﬁnally by greenschist facies metamorphic events starting from Jurassic (~200–150 Ma) and lasting up to the Oligocene [47,48]. In the Rhodope region, syn-orogenic exhumation of the metamorphic pile initiated in the early Eocene (~55 Ma) and core complex extension followed the Cretaceous syn-metamorphic SW-directed thrusting [41,43,49,50]. The metamorphic core complexes were progressively exhumed along several ductile- to brittle shear zones, active from ~42–12 Ma [49,51]. Minerals 2019, 9, 49 6 of 41 Minerals 2018, 8, x FOR PEER REVIEW 6 of 42

FigureFigure 1.1.( (aa)) SimplifiedSimplified geologicalgeological mapmap ofof easterneastern RhodopeRhodope massif,massif, GreeceGreece showingshowing thethe mainmain tectonictectonic zoneszones andand thethe distributiondistribution ofof CenozoicCenozoic igneousigneous rocksrocks (after(after MelfosMelfos andand VoudourisVoudouris[ [52]52] andand referencesreferences therein),therein), showingshowing locationlocation of corundums corundums at at Xanthi Xanthi,, Paranesti Paranesti and and Kassiteres/Sapes Kassiteres/Sapes areas areas (stars); (stars); (b) (bGeological) Geological map map of Naxos of Naxos island island showing showing locations locations of sapphires of sapphires in plumasites in plumasites at Kinidaros at Kinidarosand meta- andbauxites meta-bauxites at Kavalaris at Hill Kavalaris (stars) (after Hill (stars) Duchêne (after et al. Duch [53];ê nemodified et al. [after53]; modifiedOttens and after Voudouris Ottens [54]); and Voudouris(c) Geological [54]); map (c) of Geological Ikaria showing map of the Ikaria location showing of blue the sapphire location in of Atsida blue sapphire area (after in AtsidaBeaudoin area et (afteral. [55]; Beaudoin modified et al. after [55]; Ottens modified and after Voudouris Ottens and [54]). Voudouris Yellow [stars54]). Yellowmark the stars studied mark the corundum studied corundumoccurrences. occurrences.

Minerals 2019, 9, 49 7 of 41

Separating the upper and lower tectonic units in the Rhodope complex, the Nestos Suture Zone a NW trending shear zone was extensively studied by Papanikolaou and Panagopoulos [56], Krenn et al. [57], Nagel et al. [58] and Turpaud and Reischmann [45] and is composed from the bottom to the top of the following lithological types: (a) a lower 1-km-thick highly sheared “mélange” zone, consisting of amphibolites, garnet-kyanite schists, migmatites, orthogneisses and marbles; (b) a 1-km-thick sequence of augen-gneisses; (c) two layers of marbles intercalated with amphibolites and mylonitic amphibolites; (d) a layer of ortho-gneisses, characterized by biotite-gneisses with highly migmatized base (Sidironero). Gautier et al. [48] suggest a continuous thrusting tectonism along the Nestos Suture Zone until 33 Ma ago, and Nagel et al. [58] indicate that the Nestos Suture Zone constitutes the base of an Eocene thrusting wedge that also includes UHP units which were probably merged with the Nestos Zone during the thrusting event.

3.1.2. Attico-Cycladic Massif The Cyclades consist of a lowermost Pre-Alpidic Basement Unit, the intermediate Cycladic Blueschist Unit, and the uppermost Pelagonian Unit [59]. The Blueschist Unit represents a polymetamorphic terrane which tectonically overlies the basement gneiss and consists of a metamorphosed volcano-sedimentary sequence of clastic metasedimentary rocks, marbles, calc-schists, and mafic and felsic meta-igneous rocks [38,39,60]. The Cyclades have been exhumed since the Eocene as metamorphic core complexes formed in low-, medium-, and/or high-temperature environments [50]. An Eocene (~52–53 Ma) high-pressure eclogite- to blueschist-facies metamorphism was followed by syn-orogenic exhumation of the blueschists in a cold retrograde path and then by early Oligocene amphibolite to greenschist facies, and finally, by high-temperature medium-pressure metamorphism and associated migmatites in the early Miocene [37,60–63]. The amphibolite to greenschist metamorphic event occurred during extension-related exhumation and was coeval with back-arc extension at the Rhodopes in northern Greece [64–66]. Exhumation of the Cycladic rocks as metamorphic core complexes was accommodated during the Oligocene–Miocene by several ductile to brittle detachment systems. The extensional event also allowed for various granitoids (granite, granodiorite, leucogranite) to be intruded throughout the Cyclades between 15 and 7 Ma [64]. On Naxos island, a migmatitic dome (pre-alpine basement), is overlain by the Blueschist unit containing alternating layers of marbles, schists and gneisses, and by the upper Pelagonian tectonic unit (Figure 1b). Blueschist relics at the SE part of the island indicate temperatures from 400 to 460 ◦C at minimum pressures of 7–9 kbar [21] for the high-pressure/low-temperature (HP/LT) Eocene metamorphic event. The metamorphic grade of the Oligocene–Miocene event increases from 400 ◦C at 6 kbar in marbles, schists, gneisses and amphibolites at the SE of the island, towards the core of the dome, where migmatites in the leucogneissic and amphibolite-facies sillimanitic schists rocks, suggest temperatures up to 700 ◦C at 6–8 kbar [21,67–70]. Jansen and Schuiling [67] distinguished six metamorphic zones (I–VI) at Naxos island based on mineral isograds. Maximum temperatures are 420 ◦C for zone I, inferred by the presence of diaspore; 420 to 500 ◦C (appearance of biotite) for zone II; 500 ◦C to 540–580 ◦C (disappearance of chloritoid) for zone III, 540–580 ◦C to 620–660 ◦C (kyanite-sillimanite transition) for zone IV, 620–660 ◦C to 660–690 ◦C (onset of melting) for zone V; and >690 ◦C for zone VI. The Naxos migmatitic core is surrounded by a discontinuous block of ultramafic (meta-peridotites) lenses, representing a thrust zone along which the metamorphic complex lies on top of the pre-alpine bedrock [71]. The peridotites underwent high-P metamorphism and then, after cooling amphibolite-facies conditions (T~600 ◦C), were finally re-heated and metamorphosed together with the country rocks. According to Jansen and Schuiling [67], pegmatites penetrating the ultramafic bodies in the sillimanite stability zone, are desilicated and composed of phlogopite, anorthite, corundum, chlorite, zoisite, tourmaline and beryl. Anorthite crystals are composed almost entirely of anorthite (98% An), while margarite can be found in places. According to Andriessen et al. [72] and Katzir et al. [71], aplites and pegmatites are of Early Miocene age (19–20 Ma), and were formed as a result of the crystallization of an in situ anatectic liquid, that was produced during the high-T Minerals 2019, 9, 49 8 of 41 metamorphic event. They represent a system that channeled fluids through the leucogneissic core to the lower metamorphic unit and resulted in the metasomatization of peridotites and in situ recrystallization of the peridotitic blackwalls. Corundum-bearing emery deposits occur in the metamorphic zones I–V following the increase in metamorphic grade from SE towards NW to the migmatite core [21,73]. Ikaria Island (Figure 1c), similarly to Naxos, represents a Miocene migmatite-cored metamorphic core complex, where peak-metamorphic temperatures range from 390 ◦C in the upper parts of the structure down to 625 ◦C in the core of the dome in the vicinity of migmatites and S-type granite [55,74]. Three main tectonic units are distinguished and are, from bottom to top, the Ikaria, Agios Kirykos and Fanari units, limited by two major shear zones [55,75,76]. The non-metamorphic Fanari unit consists of Miocene to Pliocene sandstones, conglomerates and ophiolitic molasses. The intermediate Agios Kirykos unit (previously called the Messaria unit) consists of alternating marble and metapelite layers, metamorphosed in greenschist-facies conditions. Finally, the Ikaria unit is composed of amphibolite-facies (ca. 6–8 kbar and 600–650 ◦C) metasediments including micaschist and marble layers and minor metabasites. Peak-metamorphic conditions were retrieved from the basal parts of the succession. Two synkinematic granitic bodies intrude the Ikaria unit: (a) an I-type Bt-granite (Raches granite) in the western part of the island, with K-Ar ages of 22.7 Ma [64]; and (b) an S-type Bt-Ms-granite (Xilosirtis granite) in the southern part of the island, with Rb-Sr ages of 18.1 Ma [64]. Liati and Skarpelis [77], Iliopoulos and Katagas [24], and Iliopoulos [75] studied the Ikaria metabauxitic rocks hosted in marbles of the Ikaria unit, and recorded a Jurassic age for the formation of bauxitic deposits, and upper greenschist to lower amphibolite-facies conditions for their metamorphism.

3.2. Local Geology

3.2.1. Xanthi The Gorgona-Stirigma corundum occurrence is stratiform and distributed in marble layers (Figure 2a), reaching up to 50 m width, alternating with eclogitic amphibolites and gneisses [22,27,42]. The chemical composition of the marbles varies widely, highly locally impure marbles, rich in alumino-silicate minerals, are characterized by a mineralogical assemblage composed of calcite, dolomite, corundum, spinel, pyrophyllite, muscovite, paragonite, amphibole, zoisite, margarite, chlorite, olivine, serpentine, phlogopite, rutile, titanite, anorthite, garnet and Ni-tourmaline [22,23,27,42]. Spinel occurs as a rim around corundum crystals. The color of spinel ranges from blue to green and brown. Liati [22] deﬁned the following contact assemblages: calcite-anorthite-zoisite, calcite-corundum-spinel, calcite-corundum-zoisite and margarite-zoisite-anorthite-chlorite. The corundum crystals occur within phyllosilicate-rich micro-shear zones along marble layers. Sapphire is of a pink, orange, purple to blue color, usually of tabular or barrel-shaped euhedral form and reaches sizes of up to 4 cm (Figure 3a–c). In some cases, blue corundum alters to green spinel. Red corundum has also been reported in this area, mainly from zoisite-bearing amphibolites [23], but could not be identiﬁed in the present study.

3.2.2. Paranesti The Paranesti corundum occurrence is stratiform, oriented parallel to the main regional foliation, and distributed mainly in boudin-like lenses of amphibole schist (Figure 2b) [27,30,42]. The outcrops on the surface are spotted west of Perivlepto village at one hillside and one roadside location [30]. In the ﬁrst site, the corundum-bearing amphibole schist lenses are surrounded by a narrow clinochlore schist zone and occur together with boudins of corundum-kyanite-amphibole schists. The schists are intruded by white-colored quartz-feldspar-mica-garnet pegmatitic veins, which do not present characteristics of desilication (Figure 2c). In the second site, the lenses are hosted by amphibolites, intercalated with kyanite-bearing gneisses and kyanite-amphibole-chlorite schists. Again, the transition from amphibolitic lithology towards corundum mineralization is made through a clinochlore-rich schist zone. Ruby crystals from Paranesti, range in size from <1 mm to 50 mm size (average size 5–10 mm, and are of pale pink to deep red color [30]. Their morphology is mainly ﬂat tabular with Minerals 2019, 9, 49 9 of 41

Minerals 2018, 8, x FOR PEER REVIEW 9 of 42 basal planes paralleling the orientation of the main regional foliation [30], and less commonly prismatic andcommonly barrel-shaped prismatic (Figure and barrel-shaped3d,e). They occur (Figure together 3d,e). with They pargasite, occur together which iswith the mainpargasite, amphibole which ofis thethe assemblage,main amphibole forming of the a similar assemblage, image withforming Tanzanian a similar rubies image which with are Tanzanian surrounded rubies by zoisite. which The are rubysurrounded crystals by coexist zoisite. locally The ruby with kyanitecrystals (Figurecoexist locally3f). with kyanite (Figure 3f).

FigureFigure 2.2.Field Field photographs photographs demonstrating demonstrating the geological the geolog settingical ofsetting the ruby of andthe sapphireruby and occurrences sapphire inoccurrences Greece. (a )in Marble Greece. layers (a) Marble alternating layers with alternating eclogitic amphibolites with eclogitic at amphibolites Gorgona/Xanthi; at Gorgona/Xanthi; (b) Boudinaged hornblende(b) Boudinaged schists, hornblende intercalated schists, within intercalated eclogitic amphibolites, within eclogitic marbles amphibolites, and metapegmatites marbles and at Paranesti/Drama;metapegmatites at (Paranesti/Drama;c) The main ruby (c working) The main at Paranesti.ruby working The at pegmatite Paranesti. body The pegmatite at the lower body part at ofthe the lower photograph part of the shows photograph no desiliﬁcation; shows no desilification; (d,e) Ultrabasic (d,e rocks) Ultrabasic between rocks the between migmatite the dome-likemigmatite coredome-like and the core overlying and the kyanite-sillimaniteoverlying kyanite-sillimanit grade schistse grade and schists marbles and in marbles the two in plumasite the two plumasite localities eastlocalities of Kinidaros, east of NaxosKinidaros, island; Naxos (f) Corundum-bearing island; (f) Corundum-bearing plumasite at the plumas contactite to at ultrabasites the contact atthe to localityultrabasites shown at the in locality Figure 2showne; (g) Panoramicin Figure 2e; view (g) Panoramic of the metabauxite view of the bearing metabauxite marbles bearing at Kavalaris marbles Hill,at Kavalaris southern Hill, Naxos southern island; Naxos (h) Panoramic island; (h) view Panoramic for the view metabauxite-bearing for the metabauxite-bearing marbles of Ikariamarbles unit; of (Ikariai) Corundum-bearing unit; (i) Corundum-bearing metabauxite metabauxite lenses within lenses the Ikaria within unit the marbles. Ikaria unit marbles.

3.2.3.3.2.3. Naxos TheThe corundumscorundums studiedstudied fromfrom NaxosNaxos islandisland occuroccur inin twotwo localities,localities, representingrepresenting differentdifferent geologicalgeological environments.environments. (a) In the central part of the island and about 2 km east and southeast of Kinidaros, two corundum-bearing(a) In the central plumasites part of (desilicatedthe island and pegmatites) about 2 werekm east studied and (Figuresoutheast2d–f) of [Kinidaros,27,29,42]. Thetwo meta-peridotitescorundum-bearing occurred plumasites initially (desilicated as lenses of pegmatites) few to several were tens studied of meters (Figure size and 2d–f) are surrounded[27,29,42]. The by phlogopitite,meta-peridotites which occurred formed initially at the contact as lenses between of few theto several metaperidotites tens of meters and the size intruding and are surrounded pegmatites. Today,by phlogopitite, most of the which plumasites formed and at metaperidotites the contact between in these the two metaperidotites localities occur asand loose the fragmentsintruding mainlypegmatites. due toToday, erosion most and of agricultural the plumasites activities and metaperidotites in the area. in these two localities occur as loose fragments mainly due to erosion and agricultural activities in the area.

Minerals 2019, 9, 49 10 of 41 Minerals 2018, 8, x FOR PEER REVIEW 10 of 42

FigureFigure 3.3.Field Field photographsphotographs andand handspecimenshandspecimens demonstratingdemonstrating gemgem corundumcorundum crystalscrystals ofof Greece:Greece: (a(a––cc)) Pink, Pink, blue, blue, to to purple purple sapphires sapphires within within Xanthi Xanthi marbles; marbles; (d(d,e,e)) DeepDeepred-colored red-colored rubyruby withwith both both tabulartabular and and barrel-shaped barrel-shaped crystals crystals within within pargasite pargasite schist schist from from Paranesti/Drama; Paranesti/Drama; (f(f)) Ruby Ruby associated associated (replacing)(replacing) kyanite, kyanite, Paranesti/Drama; Paranesti/Drama; ((gg)) Barrel-shapedBarrel-shaped blueblue andand colorlesscolorless sapphiressapphires hostedhosted within within anorthite-richanorthite-rich desilicateddesilicated pegmatitepegmatite (plumasite)(plumasite) andand associatedassociated withwith phlogopitephlogopite andand tourmalinetourmaline (schorlite)(schorlite) from from the the first first Kinidaros Kinidaros locality, locality, Naxos Naxos island; island; ( h(h)) Purple Purple to to pink pink sapphire sapphire within within plumasite plumasite thethe laterlater rimmedrimmed byby phlogopitephlogopite atat thethe leftleft part,part, secondsecond KinidarosKinidaros locality,locality, NaxosNaxos island;island; (i()i) BlueBlue sapphiresapphire in in fissures fissures within within metabauxite metabauxite at at Kavalaris Kavalaris Hill, Hill, southern southern Naxos Naxos island; island; ( j(,jk,k)) Blue Blue sapphires sapphires andand margarite margarite as freeas free growing growing crystals crystals of probably of prob hydrothermalably hydrothermal origin within origin fissure within of metabauxites, fissure of Ikariametabauxites, island. Ikaria island.

Minerals 2019, 9, 49 11 of 41

The plumasites are surrounded by an up to 20-cm-wide zone of phlogopite and chlorite (blackwall) (Figure 3g). Within the plumasites, colorless to blue, purple and pink corundum may occur either as isolated crystals within the plagioclase matrix, and/or associated with tourmaline, and phlogopite (Figure 3g,h). Purple to pink sapphires have been also found entirely enclosed in phlogopitite in the blackwalls. The sapphire crystals are barrel-shaped, display macroscopic color zoning from colorless cores to blue rims or pink cores to purple rims and reach sizes up to 3 cm [27,42]. (b) In the southern part of the island, close to Kavalaris Hill (Figure 2g), corundum represents a single mineral rock termed “corundite” by Feenstra and Wunder [73], which was formed by the dissociation of former diasporites, in meta-carstic bauxites during prograde regional metamorphism. This ﬁrst corundum occurrence in metabauxites also recorded as corundum-in isograd (T~420–450 ◦C and P~6–7 kbar, [73]), separates diasporites with the assemblage diaspore-chloritoid-muscovite-paragonite-calcite-hematite-rutile from emery characterized by corundum-chloritoid-muscovite-paragonite-margarite-hematite-rutile. Corundum and chloritoid intergrowths occur as a rim in metabauxites, while the corundum crystals occur mainly in the contact between marbles and metabauxites. The corundum from this locality does not occur in well-shaped crystals; it is, rather, massive, and of blue color (Figure 3i).

3.2.4. Ikaria The metabauxite occurrence is located on mountain Atheras (Figure 2h) [20,75]. Already Iliopoulos [75], reported the presence of corundum as mineralogical constituent of meta-bauxite (emery) lenses occurring within dolomitic and calcic marble, which lies on top of gneiss, both being part of the Ikaria unit. In this area, four main layers of metabauxites, of up to 1 m width, were detected (Figure 2i). However, the corundum megacrysts described by Voudouris et al. [27,42] represent a second, late, generation of corundum formation within the emeries, since they ﬁll together with margarite extensional ﬁssures and networks of veins discordant to the metabauxite foliation (Figure 3j,k) [27,42]. The corundums are deep-blue in color, tabular to euhedral, reaching sizes up to 4 cm and are accompanied by Fe-chlorite, hematite, rutile and diaspore.

4. Mineralogy The Paranesti rubies are dull to transparent, exhibiting clear parting and polysynthetic twinning. The rubies are associated with edenitic hornblende and tremolite, and rimmed by margarite, muscovite, chlorite and chromian spinel (Figure 4a,b and Figure 5a,b). The ruby crystals are surrounded by a margarite rim. Spinel appears as the main component of primary inclusions in the rubies. The association of ruby in this deposit is pargasite, chlorite (mainly clinochlore and rarely nimite), margarite, tremolite and/or monazite, while for the host rocks the assemblage is pargasite, anorthite, clinozoisite, chlorite and monazite. The Paranesti rubies are rimmed by Cr-bearing spinel and also include ulvospinel. Corundum is in equilibrium with amphibole, while in a later, retrograde stage margarite-muscovite, Cr-bearing spinel, orthoclase, chlorite, smectite and vermiculite, rim and/or replace corundum. Cr-bearing spinel is replaced by chlorite. The corundums from Xanthi are transparent, with very clear parting and ﬁne cracks. Polysynthetic twinning is common. Corundum may be isolated within the carbonate matrix (Figure 4c) or rimmed by amesite (an Al-rich serpentine member), margarite, titanite or by brown and green spinel. Blue sapphires are zoned, with alternating deep blue and colorless domains (Figure 4c). This zoning, or irregular color distribution in the sapphires, is attributed to different Fe and Ti contents in the crystals (see next paragraph). Ni-bearing brown tourmaline accompanying corundum in the paragenesis contains up to 4.4 wt. % NiO, much higher than the Ni content reported in tourmaline from Samos by Henry and Dutrow [78]. The sapphire crystals from Naxos present strong pleochroism with their color ranging from dark blue to colorless (Figure 4d–f). They are transparent, with smaller or bigger inner fractures. Hexagonal barrel-shaped corundum crystals exhibit blue color zoning. The blue color of corundum is Minerals 2019, 9, 49 12 of 41 observed either as a blue core surrounded by a white rim or as a blue-zoned outer rim surrounding Minerals 2018, 8, x FOR PEER REVIEW 12 of 42 a colorless core. Polysynthetic twinning is observed along the colorless parts of the crystals. Sapphiresare associated are associated with anorthite, with anorthite, phlogopite, phlogopite, zircon, margarite, zircon, margarite, muscovite, muscovite, tourmaline tourmaline and chlorite and chlorite(Figures (Figures4d–f and4d–f 5c–i). and 5c–i).The sapphires The sapphires are areenveloped enveloped by byan an assemblage assemblage consisting consisting ofof oligoclase/labradorite-orthoclase-muscoviteoligoclase/labradorite-orthoclase-muscovite andand thenthen byby phlogopite,phlogopite, chloritechlorite (Figure(Figure5 5c–h).c–h). InIn samplesample Nx1,Nx1, blueblue sapphiresapphire associatedassociated withwith phlogopitephlogopite isis rimmedrimmed byby anan intergrowthintergrowth ofof oligoclase,oligoclase, anorthite,anorthite, orthoclaseorthoclase andand muscovite.muscovite. Margarite,Margarite, zircon,zircon, diaspore,diaspore, phlogopitephlogopite andand chloritechlorite areare includedincluded inin blueblue sapphire;sapphire; however,however, itit isis consideredconsidered thatthat margarite,margarite, diasporediaspore andand chloritechlorite areare retrograderetrograde mineralsminerals postdatingpostdating corundumcorundum formationformation (this(this study).study). In samples Nx4 andand Nx4a,Nx4a, blueblue sapphiresapphire includesincludes phlogopite,phlogopite, zirconzircon andand apatiteapatite andand surroundedsurrounded byby anorthite,anorthite, muscovitemuscovite andand tourmalinetourmaline withwith minorminor amountsamounts ofof monazite monazite and and chlorite. chlorite. Phlogopite Phlogopite and and anorthite anorthite are are replaced replaced by muscovite,by muscovite, margarite margarite and chlorite.and chlorite. Tourmaline Tourmaline postdates postdates anorthite anorthite and muscovite.and muscovite.

FigureFigure 4.4. MicrophotographsMicrophotographs (plane(plane polarizedpolarized light)light) demonstratingdemonstrating mineralogical assemblages, color- andand chemicalchemical zoningzoning ofof corundum-bearingcorundum-bearing assemblagesassemblages fromfrom Greece.Greece. ValuesValues inin microphotographsmicrophotographs 2 2 3 ((cc,,ee,,ff,,hh,,ii)) correspondcorrespond toto TiOTiO2 andand FeFe2OO3 contentcontent inin wt.wt. %% (upper(upper andand lowerlower values,values, respectively)respectively) asas obtainedobtained byby electronelectron probeprobe microanalyses.microanalyses. ((aa,,bb)) RubyRuby associatedassociated withwith pargasitepargasite (Prg)(Prg) fromfrom Paranesti,Paranesti, DramaDrama area. Note Note replacement replacement of ofruby ruby by byspinel spinel (Spl) (Spl) and andmargarite margarite (Mrg) (Mrg) in Figure in Figure 4b (sample4b (sample Dr1a); Dr1a);(c) blue (c )to blue colorless to colorless sapphires sapphires included included in calcite, in calcite, Xanthi Xanthi area area (sample (sample Go1b); Go1b); (d,e (d) ,Zonede) Zoned blue blue to tocolorless colorless sapphires sapphires within within anorthite anorthite (Plg) (Plg) and and rimmed rimmed by by tourmaline tourmaline (Tur) (Tur) and muscovitemuscovite (Mus),(Mus), Kinidaros,Kinidaros, NaxosNaxos islandisland (sample(sample Nx4);Nx4); ((ff)) SapphireSapphire withwith blueblue corecore andand colorlesscolorless rimrim associatedassociated withwith margaritemargarite (Mrg)(Mrg) andand anorthiteanorthite (Plg),(Plg), Kinidaros,Kinidaros, NaxosNaxos islandisland (sample(sample Nx3);Nx3); ((gg)) IdiomorphicIdiomorphic deepdeep blueblue coloredcolored sapphiresapphire crystalscrystals surroundedsurrounded byby margarite,margarite, IkariaIkaria islandisland (sample(sample Ik1b);Ik1b); ((hh,,ii)) ZonedZoned blueblue toto colorlesscolorless idiomorphicidiomorphic sapphires,sapphires, surroundedsurrounded byby margarite,margarite, Ikaria Ikaria island island (sample (sample Ik2). Ik2) .

Minerals 2019, 9, 49 13 of 41

On Ikaria Island, blue sapphires in the metabauxites are accompanied by the assemblage margarite + chloritoid + hematite/ilmenite + rutile (Figures 4g–i and 5j–l). Chloritoid ﬁll fractures and also occurs as inclusions in margarite. The retrogression has continued to low temperatures as suggested by the presence of diaspore replacing corundum (Figure 4j–l) and chlorite replacing chloritoid. Corundum forms colorless cores and deep blue rims (Figure 4h,i). While transparent, their blue color is quite dark, with colorless spots appearing all over their surface. Corundum includes ilmenite, hematite, ulvospinel, rutile and zircon. In addition, a Zn-bearing green colored spinel, which forms solid solution between hercynite and gahnite, has been reported with chlorite and/or margarite by Iliopoulos [75] and Iliopoulos and Katagas [24]. In sample Ik1a, Ik1b, and Ik2 blue sapphire is surrounded by margarite, chlorite and green-spinel. Chlorite ﬁlls fractures and represents latest deposition. Chloritoid, ilmenite, ulvospinel, rutile, Ti-hematite and zircon occur as inclusions in sapphire. Sapphire is associated with hematite, zircon and chloritoid and is replaced by diaspore and margarite.

5. Mineral Chemistry

5.1. EPMA Electron probe microanalyses (EPMA) for the Greece corundums are presented in Table 2. EPMA data from Paranesti/Drama rubies showed an extreme enrichment in Cr2O3, which reaches up to 2.65 wt. %, with the higher concentrations being consistent with intense red-colored grains. Beyond Cr2O3, the Paranesti rubies are characterized by moderate Fe2O3 vaues (up to 0.44 wt. %) and by traces of MgO, Ga2O3, SiO2, TiO2 and CaO, which range from below detection up to 0.17, 0.15, 0.09, 0.05 and 0.02 wt. %, respectively. Finally, V2O3 content is mostly below detection, but in some cases values up to 0.07 wt. % were detected. Colored sapphires from Gorgona/Xanthi are characterized by elevated contents of TiO2, MgO and Fe2O3. The blue color varieties show the highest TiO2 content, which reaches up to 1.05 wt. %, while MgO and Fe2O3 reach up to 0.27 and 0.22 wt. %, respectively. In the pink-colored grains, TiO2 and Fe2O3 contents are significantly lower (up to 0.17 and 0.06 wt. %, respectively) in comparison to the blue ones, but MgO content increases up to 0.36 wt. %. SiO2 and CaO values are generally below 0.08 wt. %, but in one analysis SiO2 is 0.51 wt. %, probably indicating a fine-grained silicate inclusion. V2O3 content tends to be higher in the pink varieties rather than the colorless to blue, with values ranging from below detection to 0.06 and 0.01 wt.%, respectively. Pink-colored grains contain more than double Ga2O3 content (up to 0.17 wt. %) compared to the blue ones (up to 0.07 wt. %). Finally, colorless grains diasplay very low TiO2 content (from below detection up to 0.12 wt. %) and similar Fe2O3 contents to the blue domains. In colored sapphires from the Naxos plumasite, Fe2O3 is the dominant impurity. In the blue vareties, the Fe2O3 content reaches up to 0.93 wt.%, and decreases significantly in the colorless (up to 0.69 wt. %), pink- (up to 0.50 wt. %) and the purple-colored grains (up to 0.34 wt. %). A similar decreasing trend is remarked for the Ga2O3 content, which reaches up to 0.18 wt. % in the white to blue sapphires, and decreases slightly in the pink and purple varieties, reaching up to 0.15 and 0.10 wt. %, respectively. On the other hand, TiO2 and MgO seem both to follow a different trend: the highest values for both oxides characterize the purple-colored grains (up to 0.58 and 0.54 wt. %, respectively. Blue grains display intermediate values (up to 0.38 and 0.49 wt. %), followed by the pink grains (up to 0.27 and 0.35 wt. %), respectively. TiO2 content in the colorless areas ranges from below detection to 0.12 wt. % similarly to the blue domains. The Cr2O3 content is up to 0.1 wt. % in the colorless to blue and purple varieties, and almost doubles in the case of pink sapphires (up to 0.18 wt. %). Traces of SiO2,V2O3 and CaO were also detected in all samples (up to 0.13, 0.09 and 0.03 wt. %, respectively). Blue sapphires from Ikaria island appear enriched in TiO2 and Fe2O3 values, which reach up to 0.69 and 0.57 wt. %, respectively. SiO2 shows a wide range from below detection to 0.44 wt. %, while MgO content is constantly low (up to 0.05 wt. %). The Cr2O3 and Ga2O3 content are almost fixed, Minerals 2019, 9, 49 14 of 41

reaching up to 0.14 wt. %. Traces of V2O3 and CaO were also detected (up to 0.04 and 0.02 wt. %, respectively). Colorless domains are characterized by very low TiO2 values, which range from below detection up to 0.12 wt. %. Minerals 2018, 8, x FOR PEER REVIEW 14 of 42

FigureFigure 5.5. MicrophotographsMicrophotographs (SEM-BSE images) demonstrating mineralogicalmineralogical assemblagesassemblages ofof corundum-bearingcorundum-bearing assemblagesassemblages fromfrom Greece.Greece. ((aa,,b)) RubyRuby (Crn)(Crn) includingincluding andand rimmedrimmed byby chromianchromian spinelspinel (Spl),(Spl), margarite margarite (Mrg), (Mrg), muscovite muscovite (Mu), (Mu), chlorite chlorite (Chl) and(Chl) pargasitic and pargas hornblendeitic hornblende (Amp) (sample (Amp) Dr1,(sample Paranesti/Drama); Dr1, Paranesti/Drama); (c,d) Blue (c sapphire,d) Blue (Crn)sapphire in association (Crn) in association with anorthite with (An) anorthite and phlogopite (An) and (Phl)phlogopite replaced (Phl) by replaced muscovite by muscovite (Mus), margarite (Mus), marg (Mrg),arite and (Mrg), oligoclase and oligoclase (Olg) (Naxos (Olg) island, (Naxos sample island, Nx1);sample (e )Nx1); Zircon (e (Zrn)) Zircon is included (Zrn) is in included sapphire, in diaspore sapphire, (Dsp) diaspore and margarite (Dsp) and (Mrg) margarite ﬁll fractures (Mrg) and fill arefractures retrograde and are minerals. retrograde Naxos minerals. island, sampleNaxos Nx1);island, ( fsample,g) Blue Nx1); sapphire (f,g) including Blue sapphire phlogopite including and ﬂuoritephlogopite (Fl) isand associated fluorite (Fl) with is anorthiteassociated (An) with and anorthite replaced (An) by tourmalineand replaced (Tur) by andtourmaline muscovite (Tur) (Mus) and (Naxosmuscovite island, (Mus) sample (Naxos Nx4); island, (h) Blue sample sapphire Nx4); and ( anorthiteh) Blue sapphire are rimmed and by anorthite muscovite, are margarite rimmed and by chloritemuscovite, (Chl) margarite the late probablyand chlorite replacing (Chl) the phlogopite late probably (Naxos replacing island, phlogopite sample Nx4a); (Naxos (i) Blueisland, sapphire sample includingNx4a); (i) zirconBlue sapphire is replaced including by margarite zircon (Naxosis replaced island, by samplemargarite Nx4a); (Naxos (j–l island,) Blue sapphire sample Nx4a); associated (j–l) withBlue margaritesapphire associated includes and with replaces margarite chloritoid includes (Cld) and andreplaces are retrogressed chloritoid (Cld) to diaspore and are andretrogressed chlorite, respectively.to diaspore and Zircon chlorite, and Ilmenite respective (Ilm),ly. Zircon rutile (Rt) and are Ilmenite included (Ilm), in corundum rutile (Rt) (Ikariaare included island, in sample corundum Ik2). (Ikaria island, sample Ik2).

Minerals 2019, 9, 49 15 of 41

Table 2. Chromophores and key trace elements electron probe microanalyses (EPMA) (wt. %) of the Greek corundum crystals.

Sample Color ppm MgO Al2O3 SiO2 CaO TiO2 V2O3 Cr2O3 Fe2O3 Ga2O3 Total aver bd 96.96 0.03 0.01 0.01 0.01 1.82 0.39 0.02 99.23 Dr1a (n = 8) red min bd 96.52 0.03 bd bd bd 1.74 0.34 bd 98.80 max bd 97.39 0.04 0.02 0.01 0.01 1.99 0.43 0.09 99.66 aver bd 96.98 0.04 0.01 0.01 0.01 2.14 0.37 0.02 99.55 Dr1b (n = 13) red min bd 96.13 0.02 bd bd bd 1.48 0.33 bd 98.95 max bd 97.58 0.09 0.01 0.01 0.01 2.65 0.44 0.07 100.10 aver 0.02 99.79 0.01 0.01 0.01 0.01 0.32 0.19 0.05 100.41 Dr2 (n = 8) red min bd 99.18 bd bd bd bd 0.18 0.15 bd 99.78 max 0.17 100.65 0.03 0.02 0.05 0.074 0.43 0.21 0.15 101.30 aver 0.07 98.02 0.09 0.01 0.59 0.01 0.04 0.11 0.01 98.94 Go1a (n = 11) blue-colorless min 0.02 97.37 bd bd bd bd bd 0.07 bd 98.19 max 0.27 98.61 0.51 0.01 0.94 0.01 0.08 0.14 0.04 99.36 aver 0.04 99.21 0.03 0.02 0.56 0.01 0.01 0.11 0.02 100.00 Go1b (n = 15) blue-colorless min bd 98.46 0.01 bd bd bd bd 0.031 bd 99.70 max 0.10 99.73 0.07 0.08 1.05 0.01 0.05 0.21 0.07 100.38 aver 0.05 100.93 0.01 0.01 0.06 0.01 0.03 0.04 0.05 101.18 Go5b (n = 14) pink min bd 100.22 bd bd bd bd bd bd bd 100.40 max 0.36 101.58 0.03 0.02 0.17 0.06 0.06 0.06 0.17 101.99 aver bd 97.88 0.35 0.01 0.23 0.01 0.05 0.36 0.03 98.90 Ik1a (n = 8) blue min bd 97.08 0.24 bd bd bd bd 0.27 bd 98.16 max bd 98.45 0.44 0.016 0.53 0.04 0.13 0.50 0.12 99.36 aver 0.01 99.13 0.04 0.01 0.25 0.01 0.03 0.43 0.03 99.91 Ik1b (n = 16) blue min bd 98.44 0.02 bd bd bd bd 0.14 bd 99.61 max 0.01 99.68 0.06 0.01 0.68 0.04 0.10 0.57 0.14 100.21 aver 0.01 98.25 0.09 0.01 0.22 bd 0.04 0.31 0.02 98.94 Nx1a (n = 8) blue-colorless min bd 97.85 0.07 bd bd bd bd 0.23 bd 98.77 max 0.05 98.80 0.12 0.02 0.38 bd 0.12 0.45 0.08 99.42 aver 0.12 99.80 0.02 0.01 0.27 0.01 0.01 0.27 0.04 100.54 Nx1b (n = 14) purple min bd 98.08 bd bd 0.08 bd bd 0.20 bd 98.79 max 0.54 100.57 0.07 0.02 0.58 0.06 0.04 0.34 0.11 101.28 aver 0.10 100.26 0.02 0.01 0.07 0.01 0.01 0.45 0.05 100.98 Nx2b (n = 22) blue-colorless min bd 93.73 bd bd bd bd bd 0.136 bd 99.21 max 0.49 101.81 0.07 0.03 0.19 0.08 0.04 0.69 0.18 101.96 aver 0.01 98.98 0.05 0.01 0.08 0.01 0.01 0.66 0.01 99.80 Nx3b (n = 15) blue-colorless min bd 98.33 0.01 bd 0.01 bd bd 0.19 bd 99.30 max 0.02 99.47 0.13 0.01 0.16 0.01 0.04 0.93 0.05 100.23 aver 0.01 98.09 0.09 0.01 0.10 0.01 0.01 0.72 0.01 99.02 Nx4 (n = 16) blue-colorless min bd 97.69 0.05 bd bd bd d 0.52 bd 98.67 max 0.01 98.55 0.12 0.01 0.17 0.01 0.01 0.85 0.05 99.43 aver 0.10 100.64 0.01 0.01 0.06 0.01 0.05 0.35 0.04 101.27 Nx5b (n= 12) pink min bd 97.20 bd bd bd bd bd 0.16 db 98.91 max 0.35 100.7 0.06 0.02 0.27 0.09 0.18 0.45 0.15 102.70 bd = below detection; aver = average; min = minimum; max = maximum.

5.2. LA-ICP-MS The trace element results for averages, ranges and critical ratios are listed in Tables 3 and 4. Analyzed rubies from Paranesti/Drama are particularly enriched in Cr (up to 15,347 ppm), and less so in Fe (up to 4348 ppm), with the highest values for both elements coming from the most dark-colored grains. The Mg, Ga and V values are consistently low, (up to 31, 24 and 6 ppm, respectively). The low V contents result in consistent very low V/Cr ratios, typically <1. Calcium and Ti values display wide variations between 117–1444 ppm and 8–148 ppm, respectively, with higher values characterizing the most light-colored grains. Other trace elements are close or below detection limits. Minerals 2019, 9, 49 16 of 41

Table 3. Chromophores and key trace elements (Mg and Fe) LA-ICP-MS analyses (ppm) of the Greek corundum crystals and their different chemical ratios.

Sample. Color ppm Mg Ti V Cr Fe Ga V/Cr V + Cr/Ga Fe/Mg Ga/Mg Fe/Ti Cr/Ga aver 15 17 2.4 9814 2020 14 0.0003 718 162 1.11 133 718 Dr1a (n = 8) red min 8 8 2 8326 1799 12 0.0002 602 86 0.53 75 602 max 25 27 2.7 12,116 2181 16 0.0003 879 247 1.68 225 879 aver 17 39 3.79 14,322 3795 14 0.0003 998 295 1.13 109 998 Dr1b (n = 13) red min 9 22 3.01 11,625 3352 12 0.0002 750 122 0.46 68 750 max 31 58 5.71 15,347 4348 16 0.0004 1252 440 1.72 187 1252 aver 17 67 2.4 3289 2149 20 0.0008 162 157 1.48 58 161 Dr2 (n = 8) red min 13 27 2.1 2431 1918 19 0.0006 122 91 0.95 16 122 max 24 148 2.9 4056 2438 24 0.0011 190 209 2 83 189 aver 380 5007 90 97 1096 89 1.89 2.98 3.16 0.26 0.26 1.91 Go1a (n = 11) blue-colorless min 227 1915 40 35 554 29 0.23 1.15 1.32 .07 0.10 0.29 max 601 6462 159 251 1339 119 4.15 7.73 5.73 0.49 0.68 6.30 aver 264 3503 66 70 910 89 1.19 1.59 4.33 0.46 0.35 0.86 Go1b (n = 15) blue-colorless min 78 1007 23 35 363 64 0.26 0.84 1.66 0.14 0.12 0.30 max 536 6361 207 187 1309 121 2.96 3.71 9.25 0.94 1.20 2.92 aver 28 102 20 104 1291 47 0.37 2.88 60 2.46 24 2.46 Go5a (n = 8) pink-purple min 13 30 5 4 962 9 0.04 0.34 21 0.43 4.98 0.43 max 49 205 60 298 1782 100 1.36 8.13 130 7.45 56 7.45 aver 35 509 65 297 424 78 0.30 4.67 14 2.54 1.45 3.83 Go5b (n = 14) pink min 17 39 47 105 294 72 0.04 2.13 8.17 4.42 0.51 1.27 max 53 810 79 1082 494 87 0.68 14.68 24 6.57 9.19 14.08 aver 13 1135 120 222 3237 85 0.57 4.04 445 12 3.10 2.65 Ik1a (n = 8) blue min 5 653 70 168 2710 72 0.22 3.30 62 1.50 1.93 1.97 max 54 1942 154 313 3627 93 0.84 5.29 711 19 4.73 4.32 aver 16 1390 111 224 5415 94 0.49 3.54 348 6.07 8.49 2.38 Ik1b (n = 16) blue min 13 267 31 103 3903 68 0.20 2.13 207 3.54 0.93 1.52 max 24 4508 164 279 7324 114 0.67 5.24 491 9.76 20.3 3.21 aver 47 368 16 691 1532 46 0.02 15 35 1.06 7.32 15 Nx1a (n = 8) blue-colorless min 32 105 14 504 1377 42 0.02 11 24 0.70 1.78 10 max 71 943 19 851 1804 53 0.04 19 43 1.37 14 18 Minerals 2019, 9, 49 17 of 41

Table 3. Cont.

Sample. Color ppm Mg Ti V Cr Fe Ga V/Cr V + Cr/Ga Fe/Mg Ga/Mg Fe/Ti Cr/Ga aver 64 520 22 43 4677 63 0.52 1.03 82 1.12 12 0.48 Nx1b (n = 14) purple min 39 124 15 36 3324 55 0.34 0.82 27 0.37 4.16 0.67 max 199 848 31 48 6670 76 0.86 1.14 123 1.57 44 0.83 aver 45 489 24 3.6 3210 49 14 0.55 84 1.30 16 0.07 Nx2a (n = 8) blue-colorless min 24 40 19 0.9 2852 47 1.83 0.44 42 0.67 4.15 0.02 max 82 761 36 11 3474 53 42 0.68 127 1.99 86 0.22 aver 63 560 26 51 4241 60 0.50 2.23 94 1.23 2.14 1.65 Nx2b (n = 22) blue-colorless min 16 89 13 31 1758 48 0.09 0.91 24 0.31 0.97 0.51 max 208 966 42 73 6678 72 1.01 6.01 217 3.01 7.29 4.42 aver 50 264 23 115 3189 52 1.11 2.68 68 1.12 94 2.24 Nx3a (n = 8) blue-colorless min 35 10 17 5 2843 47 0.10 0.51 42 0.68 5.57 0.08 max 79 598 30 247 3922 59 5.10 4.99 88 1.37 316 4.46 aver 105 1944 28 71 2773 62 0.50 2.23 30 0.68 1.86 1.65 Nx3b (n = 15) blue-colorless min 98 454 25 23 2220 53 0.09 0.91 16 0.34 0.97 0.41 max 198 3222 34 246 3317 78 1.01 6.01 50 0.93 7.29 4.42 aver 51 377 21 45 3828 83 5.98 0.87 81 1.85 41 0.57 Nx4 (n = 16) blue-colorless min 16 18 9 1 1812 50 0.11 0.18 41 0.96 3.47 0.01 max 117 846 32 111 6361 184 30 1.83 126 4.61 189 1.38 aver 5 462 52 262 3706 87 0.20 3.64 1069 24 9.17 3.04 Nx5a (n = 8) blue min 2 238 43 227 3301 84 0.13 3.16 241 6.28 4.32 2.64 max 14 784 65 339 4268 90 0.24 4.48 2309 47 18 3.97 aver 52 181 20 428 3096 45 0.05 9.95 77 1.08 24 9.52 Nx5b (n = 12) pink min 17 60 14 274 2138 37 0.03 6 34 0.36 7.13 5.61 max 133 348 43 548 4716 52 0.10 13 208 2.07 59 12 Minerals 2019, 9, 49 18 of 41

Table 4. Other trace elements LA-ICP-MS analyses (ppm) of the Greek corundum crystals.

Sample ppm Be Na Si Ca Mn Ni Zn Rb Sr Zr Nb Mo Sn Ba Ta W Pb Th U aver bd 4.5 294 175 0.19 0.26 0.04 0.05 0.13 0.02 0.03 0.01 0.18 0.45 0.01 0.04 0.02 0.01 0.01 Dr1a (n = 8) min bd 1.0 185 117 0.14 0.10 0.03 0.00 0.00 0.02 0.02 0.01 0.07 0.01 0.01 0.02 0.01 0.01 0.01 max bd 20 520 282 0.20 0.47 0.08 0.37 0.59 0.03 0.04 0.01 0.47 3.15 0.01 0.12 0.02 0.01 0.01 aver bd 236 1807 780 bd 57 bd bd bd bd 0.07 0.36 bd 9.80 0.05 0.17 bd 0.02 0.01 Dr1b (n = 13) min bd 152 bd 352 bd 32 bd bd bd bd 0.03 0.09 bd 0.39 0.01 0.11 bd 0.01 0.01 max bd 323 2099 1202 bd 309 bd bd bd bd 0.11 1.27 bd 47 0.14 0.29 bd 0.02 0.01 aver bd bd bd 1414 bd bd bd bd bd bd 0.15 bd bd bd 0.02 0.48 bd 0.01 0.01 Dr2 (n = 8) min bd bd bd 1384 bd bd bd bd bd bd 0.04 bd bd bd 0.02 0.23 bd 0.01 0.01 max bd bd bd 1444 bd bd bd bd bd bd 0.21 bd bd bd 0.02 0.73 bd 0.01 0.01 aver bd bd bd 1748 bd bd bd bd bd bd 28 bd 38 0.62 1.60 68 0.73 6.98 0.32 Go1a (n = 11) min bd bd bd 1482 bd bd bd bd bd bd 0.46 bd 11 0.55 0.09 1.11 0.70 2.14 0.03 max bd bd bd 2014 bd bd bd bd bd bd 90 1.46 79 0.69 4.01 311 0.76 21 0.72 aver bd 492 bd 2265 bd bd bd bd bd bd 8.76 0.22 16 0.52 1.64 44 5.43 10 0.10 Go1b (n = 15) min bd 434 bd 1859 bd bd bd bd bd bd 0.29 0.11 2.14 0.23 0.11 0.48 2.23 1.07 0.01 max bd 524 bd 2491 bd bd bd bd bd bd 55 0.34 97 0.74 11 498 8.63 53 0.26 aver 0.2 1.3 224 136 0.20 0.75 0.22 0.01 0.04 0.02 0.11 0.01 27 0.02 0.02 0.16 0.06 0.40 0.01 Go5a (n = 8) min 0.1 bd 14 88 0.20 0.04 0.03 0.01 0.01 0.01 0.01 0.01 0.09 0.01 0.00 0.03 0.01 0.01 0.01 max 0.5 3.3 369 217 0.20 2.04 0.44 0.01 0.24 0.05 0.52 0.01 213 0.06 0.04 0.46 0.34 1.17 0.03 aver bd bd bd bd bd bd bd bd bd bd 0.16 0.45 bd 0.87 0.07 bd bd 0.03 0.02 Go5b (n = 14) min bd bd bd bd bd bd bd bd bd bd 0.16 0.33 bd 0.57 0.06 bd bd 0.01 0.01 max bd bd bd bd bd bd bd bd bd bd 0.16 0.61 bd 1.17 0.08 bd bd 0.06 0.05 aver 0.45 116 2479 1700 0.44 0.56 2.25 0.16 1.45 0.07 2.56 0.01 4.97 0.35 0.96 1.19 2.95 1.74 0.12 Ik1a (n = 8) min 0.10 24 353 312 0.23 0.37 0.94 0.02 0.30 0.04 1.59 0.01 3.23 0.11 0.34 0.79 1.40 0.85 0.04 max 0.70 313 6492 4158 1.05 0.81 4.30 0.48 3.39 0.11 4.03 0.01 7.66 0.71 1.73 1.89 4.85 2.45 0.23 aver bd bd bd bd bd bd bd bd bd bd 7.92 0.34 6.88 1.74 0.39 4.64 bd 4.34 0.04 Ik1b (n = 16) min bd bd bd bd bd bd bd bd bd bd 0.40 0.11 2.83 1.13 0.10 0.74 bd 0.04 0.01 max bd bd bd bd bd bd bd bd bd bd 37 0.84 23 2.34 1.06 12 bd 31 0.22 aver 2.08 5.30 543 14 0.20 1.10 0.03 0.03 0.66 0.05 1.16 0.01 32 0.39 15 2.74 0.02 0.40 0.01 Nx1a (n = 8) min 0.18 1.00 399 14 0.20 0.79 0.03 0.00 0.01 0.00 0.11 0.01 5.61 0.01 3.82 0.16 0.02 0.01 0.01 max 11.03 34 1110 14 0.20 1.65 0.03 0.18 4.75 0.15 3.89 0.01 108 3.01 46 8.74 0.02 2.45 0.02 aver 2.91 205 2303 1551 bd bd bd bd bd bd 7.77 0.24 1040 0.11 0.39 43 bd 0.72 0.01 Nx1b (n = 14) min 1.31 141 2069 1351 bd bd bd bd bd bd 0.41 0.12 166 0.10 0.16 4.04 bd 0.02 0.01 max 4.50 253 2537 1874 bd bd bd bd bd bd 19 0.49 1712 0.12 0.65 147 bd 2.57 0.02 aver 1.21 36 804 120 0.21 0.71 0.03 0.20 1.73 1.23 9.87 0.01 566 0.05 1.58 105 0.02 0.13 0.01 Nx2a (n = 8) min 0.10 1 466 14 0.11 0.35 0.03 0.01 0.01 0.76 0.23 0.01 197 0.01 0.18 0.63 0.02 0.01 0.01 max 3.62 271 2209 648 0.35 1.39 0.03 0.04 11 2.25 36 0.01 887 0.17 7.14 480 0.02 0.48 0.03 Minerals 2019, 9, 49 19 of 41

Table 4. Cont.

Sample ppm Be Na Si Ca Mn Ni Zn Rb Sr Zr Nb Mo Sn Ba Ta W Pb Th U aver 3.53 112 1998 2118 bd bd bd 0.92 0.88 bd 12 0.33 608 bd 3.12 147 0.53 1.48 0.08 Nx2b (n = 22) min 2.71 90 1848 1844 bd bd bd 0.66 0.68 bd 0.11 0.09 5 bd 0.18 1.08 0.51 0.01 0.01 max 5.64 141 2252 2418 bd bd bd 1.25 1.15 bd 117 0.68 1842 bd 18 877 0.56 13 0.49 aver 2.79 1.58 501 14 0.20 1.06 0.03 0.02 0.22 0.83 7.71 0.01 340 0.27 2.16 100 0.02 0.07 0.01 Nx3a (n = 8) min 0.25 1.00 412 14 0.20 0.59 0.03 0.01 0.01 0.01 0.98 0.01 4.95 0.01 0.53 11 0.02 0.01 0.00 max 17 5.65 693 14 0.20 1.67 0.03 0.11 1.56 1.64 36 0.01 574 2.05 6.02 424 0.02 0.21 0.01 aver 4.25 bd bd bd bd bd bd bd bd bd 9.87 0.23 560 0.17 4.85 49 bd 3.36 0.02 Nx3b (n = 15) min 3.28 bd bd bd bd bd bd bd bd bd 0.16 0.12 198 0.12 0.34 0.49 bd 0.04 0.01 max 5.82 bd bd bd bd bd bd bd bd bd 48 0.34 947 0.22 13 245 bd 20 0.08 aver 7.05 1.69 500 14 0.23 0.83 0.03 0.01 0.02 1.80 112 0.01 244 0.06 111 34 0.03 0.83 0.06 Nx4 (n = 16) min 0.10 1.00 267 14 0.15 0.28 0.03 0.00 0.00 0.01 1.16 0.01 9.19 0.01 0.31 1.75 0.02 0.01 0.01 max 38 12 920 14 0.61 1.92 0.03 0.17 0.12 7.56 769 0.01 684 0.38 549 157 0.10 4.69 0.55 aver 1.12 1.12 372 163 16 1.17 0.79 0.01 0.04 0.06 1.53 0.01 3.64 0.66 0.23 6.70 1.80 7.40 0.02 Nx5a (n = 8) min 0.10 1.00 328 98 0.20 0.70 0.22 0.01 0.01 0.02 1.14 0.01 2.31 0.07 0.11 3.80 0.18 4.20 0.01 max 1.86 1.99 415 239 110 1.89 2.25 0.01 0.08 0.10 1.91 0.01 5.45 2.75 0.35 9.45 7.72 10 0.03 aver bd bd bd bd bd bd bd bd bd bd 1.49 0.63 4.30 bd 12 6.28 0.67 0.02 0.01 Nx5b (n = 12) min bd bd bd bd bd bd bd bd bd bd 0.10 0.28 3.22 bd 0.19 bd 0.49 0.01 0.01 max bd bd bd bd bd bd bd bd bd bd 2.89 0.97 6.60 bd 96 13.1 1.03 0.02 0.01 Minerals 2019, 9, 49 20 of 41

Colored (colorless to blue, pink) sapphires from Gorgona/Xanthi display significant differences regarding the concentration of some chromophore elements. Colorless to blue sapphires are mostly characterized by significant variations of their Ti content which reflects their zoned coloration. Bluish areas display high Ti values, which are up to 6 times higher compared to colorless/white areas and reach up to 6462 ppm. Iron content is quite fixed in both colorless and blue areas, reaching up to 1339 ppm, but is slightly higher in the pink varieties (up to 1782 ppm). Pink sapphires are also characterized by much less Ti (up to 810 ppm) and significantly higher Cr (up to 1082 ppm) concentrations, compared to the colorless/blue areas where Cr values are generally less than 298 ppm. Vanadium and Ga content remains relatively fixed, regardless of the color, with maximum values of up to 227 and 121 ppm, respectively. Mg values are generally higher in the colorless to blue grains, with values up to 536 ppm, while in the pink grains is generally below 65 ppm. Si and Ca display values up to 369 and 217 ppm, respectively, while other elements, are generally close to or below the detection limit, with the exception of Sn and Ta, which reach up to 213 and 4 ppm, respectively, in the colorless to blue grains. Iron is the most abundant trace element in analyzed sapphires form Ikaria island, with values reaching up to 7324 ppm. Ti displays high values as well, up to 4508 ppm. Cr, Mg and V values are relatively constant with maximum values reaching up to 313, 54 and 164 ppm, respectively. Gallium shows elevated concentrations, up to 114 ppm, resulting in very high values of the Ga/Mg critical ratio. Si and Ca display wide variations between 353–6492 ppm and 312–4158 ppm, with the highest values correlating and thus indicating submicroscopic silicate inclusions (possibly margarite). Other trace elements display generally low concentrations except for Sn, whose values reach up to 23 ppm. Corundums from Naxos island are mostly characterized by high Fe concentrations, which reach up to 6678 ppm. Higher Fe values are related to areas with more intense blue coloration. Ti is considerably lower, reaching values up to 966 ppm, with the exception of one colorless to blue corundum, where higher values were detected, up to 3222 ppm. Chromium varies significantly, between 1 and 851 ppm. The higher Cr values characterize corundum grains with pink and purple hues. Mg and Ga values range between 2–208 ppm and 42-184 ppm, respectively. The higher Ga concentrations are remarked in the metabauxite-hosted sapphires, that accordingly display the highest Ga/Mg ratio values compared to any other Greek corundums (this study). Vanadium content is quite fixed in all samples and varies between 10 and 65 ppm, with the majority of the measured values being in the range of 20–40 ppm. Si and Ca reach values up to 2537 and 2418 ppm, respectively, probably reflecting the presence of submicroscopic inclusions. Sodium is generally low, but in some cases values up to 271 ppm were detected, probably due to submicroscopic inclusions of silicate minerals (plagioclase and/or paragonite). Almost all corundums contain traces of Be, which ranges between 0.10 and 38 ppm. Manganese is generally close to or below detection, but in the case of a corundum from metabauxites reaches up to 110 ppm. Most samples are characterized by high Sn values, which reach up to 1812 ppm, while Nb is also enriched, reaching up to 769 ppm. Finally, Ta reaches up to 549 ppm, while the rest trace elements are mostly below detection (Table 4).

6. Fluid Inclusions The ﬂuid inclusions (FI) in the studied corundums range in size between <1 µm and 55 µm. Only one type of two-phase vapor-liquid CO2 (LcarVcar) FI is observed within all samples [79]. (Figure 6a–d,g). They often have an elongated negative crystal shape of the host corundum (Figure 6a,k). They occur as isolated, and based on their similarity in shape and the mode of occurrence, they are considered as primary FI (Figure 6a,c,f,j,k). However, the most signiﬁcant criterion for identifying primary FI is when they occur along the growth zones of the corundum (Figure 6i), based on the criteria suggested by Bodnar [80]. Minerals 2018, 8, x FOR PEER REVIEW 22 of 42

6. Fluid Inclusions The fluid inclusions (FI) in the studied corundums range in size between <1 μm and 55 μm. Only one type of two-phase vapor-liquid CO2 (LcarVcar) FI is observed within all samples [79]. (Figure 6a– d,g). They often have an elongated negative crystal shape of the host corundum (Figure 6a,k). They occur as isolated, and based on their similarity in shape and the mode of occurrence, they are considered as primary FI (Figure 6a,c,f,j,k). However, the most significant criterion for identifying primaryMinerals 2019 FI, is9, 49when they occur along the growth zones of the corundum (Figure 6i), based on21 ofthe 41 criteria suggested by Bodnar [80].

FigureFigure 6. Photomicrographs of of FI FI in in corundums from the Rhodope massif of Greece, Paranesti and GorgonaGorgona ( (aa––ff)) and and from NaxosNaxos islandisland thethe Attic-CycladicAttic-Cycladic massif ((g–l). (a) IsolatedIsolated primaryprimary two-phasetwo-phase vapor-liquid inclusions composed of CO2 ± (CH4 and/or N2). The fluid inclusion at the upper part has a vapor-liquid inclusions composed of CO2 ± (CH4 and/or N2). The fluid inclusion at the upper part negativehas a negative crystal crystal shape, shape,and the and inclusion the inclusion undernea underneathth shows shows necking-down necking-down deformation deformation (Drama, (Drama, thin- sectionthin-section Dr2). Dr2).(b) Pseudosecondary (b) Pseudosecondary two-phase two-phase carbonic carbonic FI along FI a along fracture a fracture healed during healed crystal during growth crystal (Xanthi,growth thin-section (Xanthi, thin-section Go5a). (c) Primary Go5a). (two-phasec) Primary carbonic two-phase FI (Xanthi, carbonic thin-section FI (Xanthi, Go5a). thin-section (d) Two-phase Go5a). carbonic(d) Two-phase inclusion carbonic which inclusionshows evidence which of shows post-entrapment evidence of modification post-entrapment (brittle modification crack and halo (brittle of neoformedcrack and haloinclusions) of neoformed (Xanthi, inclusions)thin-section (Xanthi,Go5b). (e thin-section) Tiny FI along Go5b). fracture (e) Tiny plane FI trails, along but fracture due to plane their smalltrails, size but no due phases to their can smallbe identifi sizeed no (Xanthi, phases thin-section can be identified Go1b). (Xanthi, (f) Isolated thin-section two-phase Go1b). carbonic (f )inclusion Isolated (Xanthi,two-phase thin-section carbonic Go1b); inclusion (g) Oriented (Xanthi, thin-sectioncluster of pseudosecondary Go1b); (g) Oriented carbonic cluster FI along of pseudosecondary fractures healed duringcarbonic crystal FI along growth fractures (thin-section, healed duringNx1a). crystal(h) Isolated growth primary (thin-section, two-phase Nx1a). carbonic (h) Isolated inclusion primary with necking-downtwo-phase carbonic deformation inclusion (thin-section, with necking-down Nx1a). (i) FI along deformation growth zones (thin-section, of the corundum Nx1a). (thin-section, (i) FI along Nx1a).growth (j zones,k) Primary of the two-phase corundum carbonic (thin-section, FI (thin-section, Nx1a). (j, kNx2a).) Primary (l) Pseudosecondary two-phase carbonic two-phase FI (thin-section, carbonic FINx2a). along (al )fracture Pseudosecondary healed during two-phase crystal growth carbonic (thin-section, FI along Nx2b). a fracture healed during crystal growth (thin-section, Nx2b).

Minerals 2019, 9, 49 22 of 41

Oriented clusters of Lcar−Vcar FI along planes are very common in the studied samples. They are related to micro-cracks and subsequent fractures healed during crystal growth, and therefore are considered as pseudosecondary (Figure 6b,g,l). Numerous tiny FI (length <2 µm) along trails are usually observed, but due to their small size, the nature of the present phase(s) cannot be identified (Figure 6e). They are considered as secondary FI. The majority of the FI appears stretched and necked or empty due to leaking phenomena (Figure 6a,d,h). Microthermometric results were based on primary and pseudosecondary inclusions, without necking-down and post-entrapment modification evidence. However, many measurements were unsuccessful due to the black color of the carbonic fluid in the inclusions and the difficulty to observe phase changes [79]. Table 5 presents the microthermometric data of all studied FI from Drama, Xanthi, Ikaria and ◦ Naxos. The melting temperatures of CO2 (TmCO2) range between −57.3 and −56.6 C, at or slightly ◦ lower than the triple point of CO2 (−56.6 C). This indicates that the fluid is dominated by CO2 with very small quantities of CH4 and/or N2. Temperatures of Tm were not obtainable in the sample Dr1a from Paranesti (Drama). Clathrate nucleation was not observed in any measured fluid inclusion, demonstrating that liquid water (H2O) was not incorporated in the whole process of the corundums formation. All FI homogenized to the liquid carbonic phase (LCO2) at temperatures (ThCO2) varying ◦ from 27.3 to 31.0 C. This Th is close to the critical temperature of pure CO2 and corresponds to densities 3 of the source fluids from 0.46 to 0.67 g/cm . Figure 7 shows the histograms of the ThCO2 of the FI from the four different studied corundum occurrences in Greece (Paranesti, Gorgona, Naxos, Ikaria).

Table 5. Microthermometric data of studied primary and pseudosecondary ﬂuid inclusions from corundum crystals in Greece.

◦ ◦ 3 Sample Locality n Tm CO2 ( C) Th CO2 ( C) Densities (g/cm ) Dr1a Paranesti, 7 −57.2 to −56.7 28.7–31.1 0.46–0.64 Dr2Drama 7 −57.1 to −56.6 27.4–30.4 0.57–0.67 Go1a 4 −57.0 to −56.8 29.2–30.2 0.58–0.62 Go1bGorgona, 3 -56.7 to -56.6 27.8–30.5 0.57–0.66 Go5aXanthi 6 −57.2 to −56.8 28.2–30.9 0.53–0.65 Go5b 6 −57.2 to −56.7 27.4–29.2 0.62–0.67 Ik1a 3 −56.9 to −56.6 30.1–30.8 0.54–0.59 Ikaria Ik1b 6 −57.0 to −56.6 27.4–28.3 0.65–0.67 Nx1a 8 −57.2 to −56.6 27.3–31.0 0.51–0.67 Nx1b 10 −57.3 to −56.7 28.1–30.5 0.57–0.65 Nx2aNaxos 4 −57.2 to −56.6 27.9–29.3 0.62–0.66 Nx2b 6 −57.1 to −56.8 28.9–31.0 0.51–0.63 Nx4b 4 −57.2 to −56.9 27.5–30.4 0.57–0.67 Minerals 2019, 9, 49 23 of 41 Minerals 2018, 8, x FOR PEER REVIEW 24 of 42

FigureFigure 7.7. Histograms of homogenization temperatures temperatures of of CO CO22 (ThCO(ThCO2)2 to) to the the liquid liquid phase phase (LCO (LCO2) 2in) inthe the fluid ﬂuid inclusion inclusion from from the the Rhodope Rhodope massif massif (Par (Paranetsi,anetsi, Gorgona) and thethe Attic-CycladicAttic-Cycladic massifmassif (Naxos,(Naxos, Ikaria)Ikaria) fromfrom Greece. Greece.

7.7. OxygenOxygen IsotopeIsotope DataData OxygenOxygen isotopeisotope compositionscompositions (Table(Table 66)) confirmconfirm their geological typology, i.e., with, respectively, 18 δδ18O of 4.9 ± 0.2‰0.2‰ forfor sapphire sapphire in in plumasite, plumasite, 20.5‰ 20.5‰ forfor sapphire sapphire in marble andand 11‰‰ forfor rubyruby inin maficsmafics (Figure(Figure8 8).). TheThe desilicateddesilicated blueblue sapphire-bearingsapphire-bearing pegmatitepegmatite fromfrom NaxosNaxos havehave similarsimilar oxygenoxygen isotopeisotope valuesvalues toto those from desilicated pegmatites pegmatites in in ma maficfic host host rocks rocks from from elsewhere elsewhere [2,7,14,81,82]. [2,7,14,81,82 The]. The O- 18 O-isotopeisotope composition composition of ofsapphire sapphire is isbuffered buffered by by the the δδ18OO value value of of the the mafic mafic host rock. The rubyruby inin 18 pargasitepargasite schistsschists fromfrom ParanestiParanesti hashas aa veryvery lowlowδ δ18O-valueO-value ofof 11‰‰ thatthat cancan bebe interpretedinterpreted inin differentdifferent ways:ways: (i) (i) inherited inherited pre-metamophic pre-metamophic reactions reactions between between sea-water sea-water and andhot basic/ultrabasic hot basic/ultrabasic rocks before rocks 18 beforesubduction subduction and metamorphism; and metamorphism; (ii) syn-metamorphism (ii) syn-metamorphism depletion depletion in 18O related in O relatedto hydration to hydration during duringamphibolite amphibolite facies faciesmetamorphic metamorphic conditions; conditions; (iii) (iii) post-amphibolite post-amphibolite facies facies metamorphism withwith recrystallizationrecrystallization underunder thethe effecteffect ofof metasomatismmetasomatism ofof thethe metamorphosed metamorphosed mafic/ultramafic mafic/ultramafic rocksrocks withwith 18 aa highhigh depletiondepletion inin 18O;O; andand (iv)(iv) metamorphic/metasomaticmetamorphic/metasomatic conditions conditions involving deeply penetratingpenetrating meteoricmeteoric waterswaters along major crustal structures, structures, see see Wang Wang et et al. al. [83]. [83]. At At the the moment, moment, the the absence absence of ofmore more O-isotope O-isotope data data on onthis this type type of ruby of ruby precludes precludes any anyof these of these possible possible hypotheses. hypotheses. The two The δ two18O- 18 δvaluesO-values of ruby of rubyin marble, in marble, between between 20.5‰ 20.5 and‰ 22.1‰,and 22.1 are‰ in, areagreement in agreement with the with range the rangeof values of values found foundfor this for type this of type ruby of worldwide ruby worldwide [2,14]. [2,14].

Table 6. Oxygen isotope values of corundum from Greece. δ18O corundum (‰, V-SMOW) (after Table 6. Oxygen isotope values of corundum from Greece. δ18Ο corundum (‰, V-SMOW) (after Wang et al. [83]). Wang et al. [83]). δ18 Sample Sample δ18Ο O DescriptionDescription Nx2aNx2a 4.8 4.8 Blue to colorless Blue sapphire to colorless in sapphireplumasite in plumasite Nx3aNx3a 5.1 5.1 Blue to colorless Blue sapphire to colorless in sapphireplumasite in plumasite Dr1b 1.0 Red ruby in pargasite-schist Go5aDr1b 1.0 22.1 Red ruby in Pink-purple pargasite-schist sapphire in marble Go5bGo5a 22.1 20.5 Pink-purple sapphire Pink sapphire in marble in marble Ik1aGo5b 20.5 22.4 Pink sapphire Blue in sapphire marble in metabauxite Ik1a 22.4 Blue sapphire in metabauxite

Minerals 2019, 9, 49 24 of 41 Minerals 2018, 8, x FOR PEER REVIEW 25 of 42

Figure 8. OxygenOxygen isotopic isotopic composition composition of Greek Greek corundum corundum (after Wang et al. [83]) [83]) compared with the oxygen isotopicisotopic ranges ranges from from corundum corundum deposits deposits worldwide worldwide from the from data ofthe Giuliani data of et al.Giuliani [2,7,14 ,81et ,82al.]. [2,7,14,81,82].Color in diamonds Color represents in diamonds the colorrepresents of the the studied color corundums.of the studied Colorless corundums. sapphires Colorless are represented sapphires areby whiterepresented diamonds. by white diamonds.

8. Discussion

8.1. Trace Elements Fingerpring: Metamorphic versus Magmatic Origin of Greek Corundum 8.1. Trace Elements Fingerpring: Metamorphic versus Magmatic Origin of Greek Corundum Chromophore and genetic indicato indicatorr elements elements (Fe, (Fe, Cr, Cr, Ti, Ti, V, Ga and Mg) are commonly used to distinguish corundum from different primary sources using elemental diagrams [[16,84–88].16,84–88]. In the (Cr ++ V)/Ga versus versus Fe/Ti Fe/Ti diagram diagram (Figure (Figure 9a,9a, [84,85]), [84,85]), displaying displaying the the fields fields for metamorphic and magmatic corundums, the the majority majority of of the the sample sampless plot plot in in the field field of metamorphic corundum, exhibiting a large variationvariation ofof Fe/TiFe/Ti ratios. Some Some sapphires sapphires from from Naxos Naxos island island that that plot plot in the magmatic field field have have very very low low Cr/Ga Cr/Ga ratios. ratios. Rubies Rubies from from Paranesti/Drama Paranesti/Drama show show high highCr/Ga Cr/Ga and Fe/Ti and ratios,Fe/Ti ratios,followed followed by the pink by the and pink purple and sapphires purple sapphires from Naxos from island. Naxos Both island. colorless Both colorlessto blue and to pink blue sapphiresand pink sapphiresfrom Gorgona/Xanthi from Gorgona/Xanthi and blue sapphires and blue sapphiresfrom Ikaria from island Ikaria are islandcharacterized are characterized by relatively by fixedrelatively Cr/Ga, fixed but Cr/Ga, display but variations display variationswith respect with to their respect Fe/Ti to theirratios. Fe/Ti ratios. × × In the Fe – Cr × 1010 – – Ga Ga × 100100 discrimination discrimination diagram diagram (Figure (Figure 9b;9b; after after Sutherland Sutherland et et al. [84]), [84]), the rubies from Paranesti are extremely rich in Cr2O3 and display a linear trend inside the metamorphic rubies from Paranesti are extremely rich in Cr2O3 and display a linear trend inside the metamorphic corundum field. field. Sapphires from Gorgona/Xanthi,Gorgona/Xanthi, Naxos Naxos and and Ikaria Ikaria is islandslands are are scattered scattered in both the metamorphic and magmatic corundum fields.fields. ColoredColored sapphires from Gorgona/XanthiGorgona/Xanthi are are scattered in bothboth thethe metamorphicmetamorphic and and magmatic magmatic fields fields and and display display two two trends, trends, along along the Ga–Fethe Ga–Fe and Ga–Crand Ga–Cr lines, with the pink varieties plotting closer to the Cr O edge, indicating a direct relationship between color lines, with the pink varieties plotting closer to the2 3 Cr2O3 edge, indicating a direct relationship between and Cr O content. Colorless to blue sapphires from Naxos island mostly plot along the Fe–Ga line, in color and2 3Cr2O3 content. Colorless to blue sapphires from Naxos island mostly plot along the Fe–Ga line,the magmatic in the magmatic corundum corundum field, while field, the while pink andthe purplepink and varieties purple display varieties a linear display trend a linear towards trend the Cr O edge, inside the metamorphic field, which is subparallel to the trend of pink sapphires from towards2 3 the Cr2O3 edge, inside the metamorphic field, which is subparallel to the trend of pink sapphiresGorgona/Xanthi. from Gorgona/Xanthi. Blue sapphires fromBlue sapphires Ikaria island from clearly Ikaria plot island in the clearly metamorphic plot in the field. metamorphic field.

Minerals 2019, 9, 49 25 of 41 Minerals 2018, 8, x FOR PEER REVIEW 26 of 42

Figure 9. 9. GreekGreek corundum corundum LA-ICP-MS LA-ICP-MS analyses analyses plotted plottedon a (a) on(Cr a+ V)/Ga (a) (Cr versus + V)/Ga Fe/Ti discrimination versus Fe/Ti diagramdiscrimination separating diagram the fields separating for magmatic the fields and formeta magmaticmorphic corundums and metamorphic (after Sutherland corundums et al. (after [84] andSutherland Harris et et al. al.[85]); [84 (]b) andtrace Harris element et Fe al. – Cr [85 × ]);10 – ( bGa) × trace 100 discrimination element Fe – diagram Cr × 10 after – Sutherland Ga × 100 etdiscrimination al. [84]; (c) diagramFe/Mg versus after SutherlandGa/Mg diagram et al. [ 84discrimination]; (c) Fe/Mg versus separating Ga/Mg the diagram fields discriminationfor magmatic, transitionalseparating the and fields metamorphic for magmatic, corundums transitional (after and Pe metamorphicucat et al. [16], corundums Sutherland (after et al. Peucat [84]; et(d al.) V/Cr [16], versusSutherland Ga/Mg et al. discrimination [84];(d) V/Cr diagram versus Ga/Mg separating discrimination the fields diagramfor magmatic, separating transitional the fields and for metamorphicmagmatic, transitional corundums and (after metamorphic Sutherland corundums et al. [86]). (after Sutherland et al. [86]).

Minerals 2019, 9, 49 26 of 41

In the Fe/Mg versus Ga/Mg plot (Figure 9c), most samples plot in the metamorphic corundum field, except for sapphires from metabauxites of Naxos and Ikaria islands and two analyses of pink sapphires fom Gorgona/Xanthi. A few blue sapphires from Naxos island and pink sapphires from Gorgona/Xanthi plot in the in-between area of transitional corundum. Two distinct trends can be remarked in the metamorphic corundum field: Sapphires from Gorgona/Xanthi form a linear trend, with the pink varieties exhibiting higher both Fe and Ga concentrations. The rest of the analyzed corundums plot in a relatively small area, parallel to the previous trend, but are also characterized by higher Fe values, the highest of which characterize the rubies from Paranesti/Drama. The V/Cr versus Ga/Mg diagram is useful for deciphering the genetic environments of corundum (Figure 9d). The majority of the analyzed samples plot in the metamorphic corundum field, with the exception of a few analyses that fit in the transitional field and the metabauxite-hosted sapphires from Ikaria and Naxos islands that have a magmatic signature (Figure 9d). In this diagram, rubies from Paranesti/Drama plot close to the Ga/Mg axis, as they are characterized by very low V content. On the other hand, colored sapphires from the other localities plot in two vertical trends. The first one, is characterized by fixed Ga/Mg content and refers to pink, purple and the majority of the colorless to blue sapphires from Naxos island, and expresses a decrease in the Cr content from the purple to the pink and finally the blue varieties and subsequent increase of their V content. The second trend is characterized by stable V/Cr ratio, but shows a significant variation regarding its Ga/Mg content. Gallium concentration increases by three orders from the blue sapphires of Gorgona/Xanthi towards the pink varieties of the same locality and finally to the metabauxite-hosted sapphires of Ikaria and Naxos islands which show the highest Ga values. In the (FeO–Cr2O3–MgO–V2O3) versus (FeO + TiO2 + Ga2O3) diagram after Giuliani et al. [87,88], which further differentiates the metamorphic environments (Figure 10), all rubies from Paranesti/ Drama clearly fall in the field of rubies with mafic/ultramafic origin. Pink sapphires from Gorgona/Xanthi, plot in an overlapping area between the marble-hosted rubies and the metasomatic corundum fields, while the rest of sapphires from this locality, which are colorless to blue in color, plot in a linear trend along the y axis of the diagram, indicating variable contents of Fe, Ti and Ga. The rest of the studied samples (Naxos and Ikaria sapphires) plot mostly in the field of metasomatic corundum, with some analyses plotting in the borders with the field of syenite-related sapphires. In general, the trends observed in Figures 9 and 10 reflect variations in chromophore element concentrations, even in samples from the same locality. This could be attributed to changes in chemistry and physicochemical conditions of the corundum-forming environment, perhaps due to metasomatic processes as described by Harris et al. [85].

8.2. Comparison with Corundum Deposits around the World Various plots, based on trace element fingerprints, have been proposed as a useful tool for distinguishing the origin and genesis of gem corundum deposits (Figures 11 and 12;[4,16,89–93]). In the Mg × 100 – Fe – Ti × 10 diagram (Figure 11a), rubies from Paranesti/Drama plot both in the magmatic and metamorphic fields, and mostly along the Fe–Mg line. Analyses that plot in the magmatic field are also of metamorphic origin and indicate the variation of the Fe/Mg ratio, thus suggesting a limitation of this chemical diagram. Sapphires from Gorgona/Xanthi and the majority of the Naxos island’s sapphires plot in the metamorphic field. The blue and pink sapphires from Gorgona/Xanthi, as well as some blue sapphires from Naxos island, plot along the Mg–Ti line of low-Fe metamorphic sapphires (e.g., Ilakaka/Madagascar and Ratnapura Balangoda/Sri Lanka), subparallel to the so-called Kashmir trend. Other colored sapphires from Naxos plot subparallel to the Kashmir trend of metasomatic blue sapphires, but they are characterized by elevated contents of Fe in the purple- and Mg in the pink-colored varieties, and are scattered through the Bo Phloi/Thailand atypical blue sapphires. Metabauxite-hosted sapphires from Naxos and Ikaria islands display a Ti-rich magmatic trend and plot preferably along the Fe–Ti line and follow the Fe–Ti Pailin (Cambodia) of magmatic blue sapphires. Minerals 2019, 9, 49 27 of 41 Minerals 2018, 8, x FOR PEER REVIEW 28 of 42

Figure 10. 10. GreekGreek corundum corundum LA-ICP-MS LA-ICP-MS analyses analyses plotted plotted on a on FeO–Cr a FeO–Cr2O3–MgO–V2O3–MgO–V2O3 versus2O3 versusFeO + TiOFeO2 ++ TiOGa2O2 +3 discrimination Ga2O3 discrimination diagram diagram after Giuliani after Giuliani et al. [87,88]. et al. [Symbols87,88]. Symbols as in Figure as in 9. Figure 9.

In the FeFe ×× 0.10.1 − (Cr(Cr + + V) V) ×× 1010 − −TiTi plot, plot, after after Peucat Peucat et et al. al. [16], [16], rubies rubies from from Paranesti/Drama, as well asas pinkpink andand purple purple sapphires sapphires plot plot close close to to the the Cr Cr + V+ edge,V edge, as theyas they are enrichedare enriched in Cr in (Figure Cr (Figure11b). 11b).Moreover, Moreover, pink sapphires pink sapphires from Gorgona/Xanthi, from Gorgona/Xa alongnthi, with along metabauxite-hosted with metabauxite-hosted sapphires fromsapphires Ikaria fromand Naxos Ikaria islands,and Naxos plot islands, close to plot the close same to edge, the same but due edge, to theirbut due relative to their enrichment relative enrichment in V, rather in than V, ratherCr. The than blue Cr. sapphires The blue from sapphires Gorgona/Xanthi, from Gorgon alonga/Xanthi, with some along blue sapphireswith some from blue Naxos, sapphires plote from close Naxos,to the Ti plote edge, close in a to linear the Ti array edge, parallel in a linear to the array (Cr + parallel V)–Ti line. to the Sapphires (Cr + V)–Ti from line. Naxos Sapphires are scarce from in Naxosthe diagram, are scarce mainly in the reflecting diagram, their mainly variable reflecting content their of Ti,variable with the content weak-colored of Ti, with areas the weak-colored plotting close areasto or alongplotting the close Fe–(Cr to or + along V) line. the TheFe–(Cr majority + V) line. of theThe Naxos majority sapphires of the Naxos form sapphires a linear array, form parallela linear array,to the parallel Mogok marble-hostedto the Mogok marble-hosted trend, but they trend, differ bu int respectthey differ to their in respect lower Feto their and higherlower Fe Cr and + V highercontents. Cr Some+ V contents. Naxos sapphires Some Naxos plot closesapphires to the plot fields close for Colombiato the fields and for Umba Colombia blue sapphires and Umba hosted blue sapphiresin desilicated hosted pegmatites. in desilicated pegmatites. In thethe FeFe versus versus Ga/Mg Ga/Mg diagram diagram (Figure (Figure12a; 12a; after after Peucat Peucat et al. et [16 al.], Zwaan [16], Zwaan et al. [89 et]), al. the [89]), majority the majorityof the analyzed of the corundumsanalyzed corundums are scattered are in scatte the fieldred ofin metamorphicthe field of metamorphic sapphires, and sapphires, specifically and in specificallythe plumasitic in the sub-field. plumasitic Some sub-field. sapphires Some from sapphires Naxos from island, Naxos with island, blue and with pink blue color and plotpink in color the plotfield in of alluvialthe field sapphires of alluvial from sapphires Yogo Culch from Montana, Yogo Culch while Montan blue sapphiresa, while fromblue Gorgona/Xanthisapphires from Gorgona/Xanthiare characterized are by characterized very low Ga/Mg by very ratios low and Ga thus/Mg plot ratios slightly and outsidethus plot the slightly metamorphic outside field. the metamorphicMetabauxite-hosted field. Metabauxite-host sapphires from Naxosed sapphires and Ikaria from islands Naxos plot and alongIkaria the islands Main plot Asian along Field the (MAF) Main Asianof sapphires Field (MAF) related of to sapphires alkali basalt. related They to alkali form basalt. a linear They array form indicating a linear variationsarray indicating in their variations Ga/Mg inratio their and Ga/Mg their Fe ratio content and is their slightly Fe highercontent compared is slightly to sapphireshigher compared from Pailin to sapphires (Cambodia) from and Pailin Ilmen (Russia).(Cambodia) Finally, and aIlmen few analyses(Russia). of Finally, pink sapphires a few analyses from Gorgona/Xanthi of pink sapphires are from plotted Gorgona/Xanthi in the MAF field, are plottedand are characterizedin the MAF field, by slightly and are higher characterized both Fe and by Ga/Mg slightly values, higher compared both Fe and to the Ga/Mg rest pink values, and comparedcolorless to to blue the sapphiresrest pink and of this colorles location.s to blue sapphires of this location. The Cr 2OO33 versusversus Fe Fe22OO3 3diagramdiagram (Figure (Figure 12b),12b), adapted adapted from from Schwarz Schwarz et et al. al. [90], [90], shows shows different different types of African African deposits: deposits: marble-type marble-type from from Mong Mong Hsu Hsu and and Mogok Mogok (Myanmar), (Myanmar), desilicated desilicated pegmatite pegmatite from MangariMangari (Kenya), (Kenya), amphibolitic-type amphibolitic-type from fr Chimwadzuluom Chimwadzulu (Malawi), (Malawi), Songea andSongea Winza and (Tanzania) Winza (Tanzania)and basaltic-type and basaltic-type from Thai (Thailand-Cambodia from Thai (Thailand-Cambodia border region). border The majority region). of theThe Paranesti/Drama majority of the Paranesti/Dramarubies plot outside rubies the fields plot of outside rubies fromthe elsewhere,fields of rubies because from they elsewhere, contain significantly because they more contain Cr2O3. significantlySome samples, more though, Cr2O poorer3. Some in samples, Cr2O3, fit though, well into poorer the Winza in Cr (Tanzania)2O3, fit well ruby into field. the Winza Sapphire (Tanzania) samples plotruby mostly field. Sapphire along the sample Fe2O3 axis,s plot with mostly the majorityalong the of Fe the2O3 compositions axis, with the being majority comparable of the compositions with colored beingsapphires comparable from Winza. with colored sapphires from Winza.

Minerals 2019, 9, 49 28 of 41 Minerals 2018, 8, x FOR PEER REVIEW 29 of 42

Figure 11. 11. GreekGreek corundum corundum LA-ICP-MS LA-ICP-MS analyses analyses plotted plotted on a (a on) Mg a × ( a100) Mg − Fe× – Ti100 × 10− discriminationFe – Ti × 10 diagram.discrimination Also shown diagram. are the Also metamorphic shown are the Fe–Mg metamorphic and Fe–Ti Fe–Mg trends andof Mogok Fe–Ti and trends low-Fe of Mogok sapphires and oflow-Fe Pailin, sapphires Ilakaka and of Pailin, Ratnapura Ilakaka Balangoda, and Ratnapura the Umba Balangoda, and Kashmir the Umba trends and of Kashmirmetasomatic trends blue of sapphires,metasomatic and blue the Bo sapphires, Phloi scattering and the of Bo atypical Phloi scattering blue sapphires of atypical (modified blue after sapphires Peucat (modiﬁed et al. [16]); after (b) FePeucat × 0.1 et− (Cr al.[ 16+ V)]); × (b 10) Fe − Ti× 0.1diagram− (Cr for +V) Colombia,× 10 − Ti Mogok diagram and for Umba Colombia, blue sapphires Mogok and along Umba with blue the plotssapphires from alongthe Greek with corundums, the plots from modified the Greek after corundums, Peucat et modiﬁedal. [16]. Symbols after Peucat as in et Figure al. [16 9.]. Symbols as in Figure 9.

Minerals 2019, 9, 49 29 of 41 Minerals 2018, 8, x FOR PEER REVIEW 30 of 42

FigureFigure 12.12. GreekGreek corundumcorundum LA-ICP-MSLA-ICP-MS analysesanalyses plottedplotted onon aa (a(a)) Fe Fe versus versus Ga/Mg Ga/Mg discriminationdiscrimination

diagramdiagram (after Peucat et al. [16], [16], Zwaan Zwaan et et al. al. [89]); [89]); (b (b) )Cr Cr2O2O3 versus3 versus Fe Fe2O23O diagram3 diagram demonstrating demonstrating the thefields ﬁelds of different of different African African deposits deposits (modified (modiﬁed after after Schwarz Schwarz et al. et [90]); al. [ 90(c]);) Cr (c2O)3 Cr/Ga2O2O3/Ga3 versus2O3

versusFe2O3/TiO Fe22O 3plot/TiO 2demonstratingplot demonstrating composition composition of diverse of diverse corundum corundum deposits deposits (modified (modiﬁed afterafter Rakotondrazafy et al. [91], Sutherland et al. [4], Pham Van et al. [92], Simonet et al. [93]. Symbols as in Figure 9.

Minerals 2019, 9, 49 30 of 41

The Cr2O3/Ga2O3 versus Fe2O3/TiO2 diagram (Figure 12c; modified from Rakotondrazafy et al. [91]), displays the geochemical fingerprint of corundums from different types. The majority of analyzed corundums plot in the field of metamorphic origin. It is clear that rubies from Paranesti/Drama show higher Cr2O3 values, and thus cannot plot in fields of rubies from elsewhere. Some rubies from Paranesti also fall close to the Soamiakatra rubies, which are hosted in clinopyroxenite enclaves of alkali basalts. Pink and purple sapphires from Naxos island display variation of their Fe2O3/TiO2 ratio. The pink sapphires plot in a common field of the Soamiakatra and Sahambano sapphires (Madagascar), while the purple varieties are scattered in a sector overlapping both the marble- and the desilicated pegmatite in marble-types of Vietnam. Colorless to blue sapphires from Naxos plumasites are scattered along the lower limit of the metamorphic field, indicating significant variation of their Fe2O3/TiO2 values, while the Cr2O3/Ga2O3 values stay relatively fixed. Pink-colored sapphires from Gorgona/Xanthi plot partly inside the marble-hosted corundum from Vietnam, while the blue varieties plot in the extension of this trend, slightly outside the field, indicating minor Fe2O3 content. Metabauxite-hosted corundum from Naxos island plot slightly outside the magmatic corundum field, and their ratios are not comparable with any other corundum of the plot, except a few analyses, which plot close to the Vietnamese field of gneiss-hosted corundumite. Ikaria metabauxite-hosted samples plot well into the metamorphic corundum field and show a wide range of Fe2O3/TiO2 values, covering the fields between marble-hosted and desilicated pegmatite in marbles (Vietnam), Sahambano and Zafafotsy (Magadascar) corundum deposits.

8.3. Fluid Characterization The fluid inclusions study in the corundum (rubies and sapphires) from the four Greek occurrences revealed the presence of CO2-dominated fluids with very small quantities of CH4 and/or N2, and relatively low densities, varying between 0.46 and 0.67 g/cm3 [79]. Primary and pseudosecondary water-free carbonic fluid inclusions represent the main fluid, which was incorporated during the crystallization of rubies and sapphires. The absence of any significant change in the fluid composition and density of the primary and pseudosecondary inclusions possibly suggests that (i) the fluid was homogeneous and related to the same source; and (ii) the host rocks were not affected by the circulation of any external fluids. The CO2-rich fluids are likely of metamorphic origin and probably derived from devolatilization of marble in most cases (Naxos and Ikaria). Most of the corundum in different geological environments worldwide generally contains pure or nearly pure CO2-bearing fluid inclusions. Previous studies have shown that in metamorphic complexes, CO2-bearing fluids were incorporated in the corundum formation from granulite facies rocks in Sri Lanka [94], in the marble hosted ruby deposits from Luc Yen in North Vietnam [95] and in pegmatoids from the Nestos Shear Zone in Greece [57]. Corundum occurrences with pure or almost pure CO2-bearing fluids, without any water included, were documented in sapphires from pegmatites in the Kerala district of India [96] and in a corundum bearing skarn from granulites in Southeast Madagascar [97]. In other corundum occurrences, such as in the Kashmir blue sapphires and in the Thailand sapphires, CO2 is an important component of the source fluids [98,99]. The occurrence of high-density CO2-rich fluid inclusions in granulite facies rocks shows that large amounts of CO2 infiltrate the lower crust during the peak of metamorphism [100–102]. However, Hollister et al. [103] have shown that low-density CO2 fluid inclusions must have been trapped after the peak of metamorphism. Two main sources for input of CO2 fluids in the lower crust have been suggested [101,104,105]: from the mantle, or from the metamorphism of previously dehydrated crust. High-Al and low-Si protoliths in a high regional metamorphic grade can produce a pure supercritical CO2 fluid and lead to the formation of corundum. A mantle-derived CO2 has been suggested for Naxos by Schuiling and Kreulen [106]. In the present study, similar fluids containing almost pure CO2 have been documented. Water was not identified in the fluid inclusions, neither by optical microscopy nor by any phase transition during Minerals 2019, 9, 49 31 of 41 microthermometry. This excludes the possibility of fluid immiscibility for the corundum formation and implies the presence of a primary water-free (or very poor) CO2 dominated fluid. However, the presence of minor amounts of water in the paleofluid in all studied corundums cannot be excluded, due to the presence of hydrous mineral inclusions in the corundum. Already, Buick and Holland [107] have argued that the “primary” fluid inclusions in the metamorphic complex of Naxos have been compositionally modified by selective leakage of H2O during uplift, and Krenn et al. [57] reported on recrystallization-induced leakage resulting in minor admixture of H2O from former hydrous inclusions in corundum from pegmatites along the Nestos suture zone, Xanthi area. It is very likely that the studied primary low-density fluid inclusions (d = 0.46–0.67 g/cm3) were entrapped after the peak of metamorphism, suggesting that corundum formation took place during retrogression. The pseudosecondary, also low-density, carbonic fluid inclusions were entrapped in trails during the corundums formation process and are related to the evolution of the metamorphic events at a retrograde metamorphic regime of cooling and uplift and the subsequent exhumation, also after the peak of metamorphism. These variations in the density of the fluid inclusions can be interpreted as the result of pressure variation associated with successive localized microfracturing, a process which was suggested in the case of Luc Yen rubies of North Vietnam [95]. However, a continuous entrapment of the carbonic fluids during growth of corundum with pressure decreasing is not excluded.

8.4. Conditions of Greek Corundum Formation The Xanthi and Paranesti corundums belong to metamorphic s.s. deposits, according to the classification of Giuliani et al. [2,7] and Simonet et al. [3], and more specifically, to those related to meta-limestones and mafic granulites, respectively. Both occurrences formed during the retrograde metamorphic path of high-temperature/medium-pressure metamorphism of platform carbonates and amphibolites during the Cenozoic collision that resulted in the Nestos Suture Zone. Wang et al. [30] concluded that Paranesti rubies were formed within an ultramafic precursor, most probably an aluminous clinopyroxenite, during amphibolite facies metamorphism. The estimated P–T conditions for their formation are 4 kbar < P < 7 kbar and 580 ◦C < T < 750 ◦C, and with subsequent retrogression. The Paranesti (and Gorgona/Xanthi) corundum-bearing assemblages followed a nearly isothermal decompression within the amphibolite facies and then a further evolution towards the greenschist facies, along the path that was proposed by Krenn et al. [57] for the Nestos suture zone (Figure 13). This path records a transition from the kyanite to the sillimanite stability field during retrogression, as observed by former kyanite surrounded by fibrolitic sillimanite ([57,108] and this study). In the studied samples from Paranesti, corundum surrounds kyanite, and was probably formed after kyanite in the stability field of sillimanite during the retrogression. Mineralogical observations of the corundums from Paranesti ([30] and this study) indicate that the margarite occurs as reaction rims around ruby grains, suggesting that it was formed subsequently due to retrogression and according to the following reaction: Margarite = Anorthite + Corundum + H2O (1) In contrast to Paranesti, margarite in the Gorgona/Xanthi corundum-bearing marbles seems to be a prograde mineral ([22,109] and this study], indicating that corundum along with anorthite could have been formed by the breakdown of margarite (see Reaction (1)). In the studied samples, margarite is either in contact or is separated from corundum by a kaolinite rim, which probably represents earlier anorthite. As described by Storre and Nitsch [110] and Chatterjee [111,112], margarite breakdown to anorthite and corundum takes place at temperatures of 610, 625 and 650 ◦C for pressures of 6, 7 and 8 kbar, respectively. On the other hand, and under the assumption that margarite was formed together with corundum, a more complex reaction:

8 Diaspore + Pyrophyllite + 2 Calcite = 2 Margarite + Corundum + 2 CO2 + 3 H2O (2) Minerals 2019, 9, 49 32 of 41 as described by Okrusch et al. [113] and Haas and Holdaway [114] may explain this assemblage. This reaction requires the metastable persistence of diaspore + pyrophyllite, which should have otherwise reacted to form Al-silicate about 40–60 ◦C below the lower stability limit of corundum [113]. The assemblage corundum and chlorite within the Gorgona marbles can be formed according to the following reaction, as experimentally demonstrated by Seifert [115]:

Mg-chlorite + 6 Corundum + 2 Spinel = sapphirine + 4 H2O (3) but the absence of sapphirine in the studied samples indicates that P–T did not exceed 6 kbar for temperatures between 620 and 720 ◦C[115]. The formation of the Xanthi corundum under dissociation of Mg-spinel + calcite into corundum + dolomite during the retrograde metamorphism of spinel-bearing dolomites according to the reaction:

Spinel + Calcite = Corundum + Dolomite + CO2 (4) was not observed, but cannot be ruled out. This reaction path has been described by Buick and Holland [116] in spinel-bearing dolomites from the leucocratic core of Naxos, where spinel crystals are separated from the calcite of the host rock by corundum + dolomite mantles, and they are often surrounded by Mg-chlorite + pargasite. The above authors suggested synchronous formation of Mg-chlorite + pargasite in place of spinel + calcite. Decomposition of spinel into corundum during the retrograde metamorphism in marbles as described by Reaction (4) was described as the major corundum-forming mechanism for the rubies at Jegdalek (Afghanistan), Hunza (Pakistan) and Luc Yen (Vietnam) [117]. In summary, conditions of corundum formation in the Gorgona marbles as suggested by Liati [22,109] and this study, are in accordance to those estimated by Wang et al. [30] for the Paranesti rubies. As an alternative hypothesis, corundum in the Gorgona marbles could have formed at lower pressures of about 3–4 kbar, during late stages of shear deformation from CO2-rich fluids as proposed by Krenn et al. [57] for corundum–anorthite assemblages postdating kyanite in desilicated pegmatoid veins within gneisses and amphibolites of the Nestos suture zone, about 4 km south of the Gorgona locality. In the absence of geochronological data for corundum-bearing assemblages within the Nestos suture zone, it remains difficult to state if they all represent the same metamorphic event, or can be attributed to different P–T conditions corresponding to different ages. The plumasites from Kinidaros, Naxos contain tourmaline and beryl, in addition to corundum, anorthite and phlogopite, and originate from desilication of leucogranitic pegmatites. For these pegmatites, Matthews et al. [118] reported temperature variations from >700 ◦C to ~400 ◦C, based on oxygen isotope fractionation among quartz, tourmaline and garnet, and attributed their formation from anatectic melts during regional high-temperature metamorphism. According to these authors, crystallization of the pegmatitic magmas should initiate under water-undersaturated conditions, but with crystallization of anhydrous minerals and ascent, the magma should evolve to water-saturated conditions at 630 to 640 ◦C. Thus, some of the higher temperatures (T = 650 ◦C) given by the isotope thermometry could represent the crystallization at reduced water activity. In addition, Siebenaller [119] used fluid inclusion measurements in tourmaline, garnet and beryl from leucogranite pegmatites from Naxos island, to estimate P–T formation conditions for the pegmatites of between 5 kbar/600 ◦C and <2 kbar/450 ◦C, along the exhumation path of leucogranite (Figure 13). The absence of andalusite from the studied corundum-bearing assemblages constrains the lower limit of corundum formation along the retrograde path at about 3 kbar and 550 ◦C. Finally, metabauxite occurrences on southern Naxos (Kavalaris Hill) correspond to the thermal dissociation of diaspore and the formation of corundum (α-AlOOH ↔ Al2O3 + H2O, Haas [120]), as described by Feenstra and Wunder [73]. The first corundum occurrence in Naxos metabauxites has been recorded as corundum-in isograd (T~420–450 ◦C and P~6 kbar, [73]). Similarly, the metabauxites from Minerals 2019, 9, 49 33 of 41

Minerals 2018, 8, x FOR PEER REVIEW 34 of 42 Ikaria were formed at T in the range 450–550 ◦C and P ~5–6kbar [75,77]. According to Iliopoulos [75], the[75], latter the correspond latter correspond to corundum-chloritoid-bearing to corundum-chloritoid-bearing metabauxites metabauxites from from Naxos Naxos island island (zones (zones II andII IIIand of Feenstra III of Feenstra [21]; Figure [21]; Figure13). 13).

FigureFigure 13. 13.(a ()a P–T) P–T diagram diagram showing showing mineralmineral equilibria related related to to the the formation formation of of corundum corundum in inGreece Greece (modified(modified from from Garnier Garnier et et al. al. [117 [117]).]). Upper Upper blueblue area:area: Paranesti Paranesti ruby ruby and and Gorgona Gorgona sapphires sapphires ([30] ([30 and] and thisthis study); study); Lower Lower blue blue area: area: Alternative Alternative conditionsconditions for the Gorgona Gorgona sapphires sapphires (for (for explanation explanation see see text);text); Orange Orange area: area: Ikaria Ikaria and and southern southern NaxosNaxos metabauxites.metabauxites. The The thick thick orange orange arrow arrow indicates indicates possible possible pathpath for for formation formation of of vein-type vein-type sapphires sapphires fromfrom both localities; Green Green area: area: Central Central Naxos Naxos plumasites. plumasites. BlueBlue arrow: arrow: P–T–t P–T–t path of of high-grade high-grade rocks rocks from from the Nestos the Nestos suture suture zone (from zone Krenn (from etKrenn al. [57]); et Green al. [57 ]); arrow: P–T–t path of the tourmaline-garnet-beryl-bearing leucogranite from Naxos island (from Green arrow: P–T–t path of the tourmaline-garnet-beryl-bearing leucogranite from Naxos island Siebenaller [119]); Orange arrow: P–T–t path of the lower limit of metamorphic zone III (which hosts (from Siebenaller [119]); Orange arrow: P–T–t path of the lower limit of metamorphic zone III (which corundum-chloritoid-bearing metabauxites) from Naxos island (from Duchêne et al. [64]). The hosts corundum-chloritoid-bearing metabauxites) from Naxos island (from Duchêne et al. [64]). The equilibrium curve “chloritoid + O2 = staurolite + magnetite + quartz”, which marks the disappearance equilibrium curve “chloritoid + O = staurolite + magnetite + quartz”, which marks the disappearance of chloritoid, defines an upper limi2 t for the studied chloritoid-bearing metabauxites at southern Naxos of chloritoid, defines an upper limit for the studied chloritoid-bearing metabauxites at southern Naxos (Kavalaris) and Ikaria islands is from Feenstra [21]. Abbreviations: An = anorthite; And = andalusite; (Kavalaris) and Ikaria islands is from Feenstra [21]. Abbreviations: An = anorthite; And = andalusite; Cc = calcite; Cld = chloritoid; Clin = clinochlore; Co = corundum; Do = dolomite; Dsp = diaspore; Ky = Cc = calcite; Cld = chloritoid; Clin = clinochlore; Co = corundum; Do = dolomite; Dsp = diaspore; kyanite; Ma = margarite; Mag = magnetite; Qz = quartz; Sill = sillimanite; Spr = sapphirine; Sp = spinel; KySt = kyanite= staurolite;; Ma =(b margarite;) Comparison Mag between = magnetite; the hypothetical Qz = quartz; P–T Sill conditions = sillimanite; for the Spr formation = sapphirine; of Spcorundum = spinel; St in = Greece staurolite; and (inb) metamorphic Comparison betweendeposits thearound hypothetical the world P–T (modified conditions from for Giuliani the formation et al. of corundum[7], Simonet in et Greece al. [3]). and Boxes in metamorphicindicate P–T fields deposits of known around deposits: the world North (modified Carolina, from Rubies Giuliani from etmafic al. [ 7], Simonetgranulites et al. [121]; [3]). BoxesMorogoro, indicate co P–Trundum-bearing fields of known anatexites deposits: [122] North; Mangari, Carolina, Southern Rubies from Kenya, mafic granulitesMetasomatic [121]; rubies Morogoro, [123–125]; corundum-bearing Hunza, rubies anatexitesin marbles [ 122[113];]; Mangari, Sri Lanka, Southern sapphires Kenya, from Metasomaticgranulites rubies[94]; [123Greenland,–125]; Hunza, metasomatic rubies inrubies marbles [126]; [113 Kashmi]; Sri Lanka,r metasomatic sapphires sapphires from granulites with three [94 ];P–T Greenland, boxes metasomaticcorresponding rubies to the [126 evolution]; Kashmir of the metasomatic fluids in the sapphires sapphire crystals with three from P–T the boxescenter corresponding(c), to intermediate to the evolution(i) and outer of the (o) fluids zones in[98]; the Urals, sapphire rubies crystals in marb fromles [127], the centerand Mong (c), Hsu, to intermediate rubies in marbles (i) and [128]. outer (o) zones [98]; Urals, rubies in marbles [127], and Mong Hsu, rubies in marbles [128]. However, the studied sapphires from both localities correspond to open space-filling material in However,extensional the fissures studied and sapphires indicate a from second, both late localities stage of correspond corundum toformation. open space-filling We interpret material this incorundum extensional formation fissures to and be indicateof metasoma a second,tic origin late from stage a low-temperature of corundum formation. CO2-bearing We metamorphic interpret this corundumfluid, as formationalready proposed to be of for metasomatic similar sapphire origin from+ margarite a low-temperature + tourmaline-filled CO2-bearing veins from metamorphic Naxos fluid,island as already[116]. This proposed is in agreement for similar with sapphire the findings + margarite of Tropper + tourmaline-filledand Manning [129], veins who from consider Naxos islandcorundum-filled [116]. This ishydrofractures in agreement in withthe Naxos the findings metabaux ofites Tropper to be products and Manning of retrograde [129], whocooling consider and corundum-filled hydrofractures in the Naxos metabauxites to be products of retrograde cooling and

Minerals 2019, 9, 49 34 of 41 decompression, consistent with kinematically late textures. This hypothesis is also supported by the fact that trace element fingerprints of Naxos and Ikaria metabauxite-hosted sapphire suggest magmatic affinities (this study). A hydrothermal origin of diaspore has been reported for the gem-quality diaspore crystals (var. zultanite) occurring in fissures of metabauxites in the Ilbir˙ Mountains, SW Turkey [130]. According to the above authors, the diaspore mineralization is caused by hydrothermal remobilization of primary bauxite components into crosscutting structures. Figure 13 displays an evolution (thick orange arrow) towards lower P–T conditions, along the P–T–t retrograde path of the metamorphic zone III (hosting corundum-chloritoid-bearing metabauxites) from Naxos island (according to Duchêne et al. [53]). Since the tectonometamorphic evolution of Ikaria is quite similar to that of Naxos [55,74], this path could probably reflect the formation of the vein-type sapphire assemblages in both Naxos and Ikaria islands.

8.5. Greece: A New Gem Corundum Province? Greek corundum occurrences display a wide color variation, ranging from deep red, pink, purple, and blue to colorless, with crystal sizes of up to 5 cm. Among the studied occurrences, some corundums from Gorgona/Xanthi and the plumasite-hosted sapphires from Naxos display signiﬁcant transparency and homogeneity of color and should be further examined for their suitability as potential cut gemstones. The rest of the studied corundums, especially the vivid-colored varieties, could be considered suitable for their use as cabochons. Although gem-quality corundums are considered to be absent in emery deposits [7], the Naxos and especially Ikaria blue sapphires are of gem (cabochon) quality, and are atypical for other metabauxite-hosted corundums. Future exploration is required in order to establish the potential for economic exploitation of the corundum-bearing areas in Greece, pointing towards a new gem corundum province in the world. We highlight here the enrichment in Be of blue-colorless sapphires from Naxos plumasites, a feature which has only been reported from Ilakaka, Madagaskar, Sar-e-Sang, Afganistan and Weldborough, Tasmania, and was attributed to nano-inclusons of unidentiﬁed Be-rich phases [131]. The Naxos plumasitic sapphires, as well as those from the Xanthi marbles and Ikaria metabauxites, display very high values in Nb, Ta, W and Sn, much higher than those reported from the above occurrences. A further study of the concentration of these elements and especially of Be in the Greek corundums could be a useful tool for determining their origin [131].

9. Conclusions Gem corundum in Greece covers in a variety of geological environments located within the Rhodope (Xanthi and Drama areas) and Attico-Cycladic (Naxos and Ikaria islands) tectono-metamorphic units. Pargasite-schist hosted a ruby deposit in the Paranesti/Drama area and marble hosted pink to blue sapphires in the Xanthi area, occurring along the UHP-HP Nestos suture zone. Plumasite-hosted sapphires from Naxos island display a wide color variation (blue to colorless and pink). Deep blue colored sapphires from Naxos and Ikaria islands are hosted within extensional fissures in metabauxite lenses within marbles. Various mineral inclusions in corundums are in equilibrium and/or postdate corundum crystallization, and reflect the surrounding mineralogical assemblages. Included in corundums are: spinel and pargasite (Paranesti), spinel, zircon (Xanthi), margarite, zircon, apatite, diaspore, phlogopite and chlorite (Naxos) and chloritoid, ilmenite, hematite, ulvospinel, rutile and zircon (Ikaria). The chromophores of the studied corundums show a wide range in concentration and a unique trace element chemistry with variable critical ratios (Fe/Mg, Ga/Mg, Ga/Cr and Fe/Ti). Be, Nb, Sn, Ta and W are anomalously enriched in the plumasite-related sapphires from Naxos. Based on the geological setting of formation and trace element fingerprints the Paranesti and Xanthi corundum occurrences can be classified as metamorphic s.s hosted mafics/ultramafics and marbles, respectively. Those from central Naxos are of metasomatic origin and are related to desilicated pegmatites crosscutting ultramafic rocks. Finally, blue sapphires from southern Naxos and Ikaria, Minerals 2019, 9, 49 35 of 41 hosted in metabauxites, display an atypical magmatic signature indicating a metasomatic origin by fluid-rock interaction. The study of fluid inclusions in corundum showed that they are CO2-dominant with low density with very small quantities of CH4 and/or N2, and water-free. CO2-rich fluids are probably of metamorphic origin and derived from devolatilization of carbonate formations. Greek corundums are characterized by a wide color variation, homogeneity of the color hues and transparency and could be considered as potential gemstones.

Author Contributions: P.V. and C.M. collected the studied samples. P.V. assisted by C.M., I.G., K.Z., S.M., S.K., K.W., S.H., St.K., J.B., F.Z., V.K., M.T. and G.L. obtained and evaluated the mineralogical data. G.G. and A.F. obtained and evaluated the oxygen isotopic data. V.M. and A.T. conducted the fluid inclusion measurements. P.V., C.M., G.G., V.M., A.T. and S.K. wrote the manuscript. Funding: This research was partly funded by UNSW MREII Grants and Australian Research Council (ARC) Large Infrastructure and Equipment Funds (LIEF) grant LE0989067 and by LABEX ANR-10-LABX-21—Ressources21, Nancy, France. Acknowledgments: The authors would like to thank Evangelos Michailidis for his kind help with the SEM in the University of Athens. Two anonymous reviewers and the Editor are especially thanked for their constructive comments that greatly improved the manuscript. Conflicts of Interest: The authors declare no conflict of interest.

References

1. Garnier, V.; Ohnenstetter, D.; Giuliani, G.; Schwarz, D. Saphirs et rubis. Classification des gisements de corindon. Règne Minéral 2004, 55, 4–47. 2. Giuliani, G.; Ohnenstetter, D.; Garnier, V.; Fallick, A.E.; Rakotondrazafy, M.; Schwarz, D. The geology and genesis of gem corundum deposits. In Geology of Gem Deposits; Mineralogical Association of Canada Short Course Series; Groat, L., Ed.; Mineralogical Association of Canada: Quebec City, QC, Canada, 2007; Volume 37, pp. 23–78. 3. Simonet, C.; Fritsch, E.; Lasnier, B. A classification of gem corundum deposits aimed towards gem exploration. Ore Geol. Rev. 2008, 34, 127–133. [CrossRef] 4. Sutherland, F.L.; Schwarz, D.; Jobbins, E.A.; Coenraads, R.R.; Webb, G. Distinctive gem corundum from discrete basalt fields: A comparative study of Barrington, Australia, and west Pailin, Cambodia gem fields. J. Gemmol. 1998, 26, 65–85. [CrossRef] 5. Garnier, V.; Giuliani, G.; Ohnenstetter, D.; Schwarz, D.; Kausar, A.B. Les gisements de rubis associés aux marbres de l’Asie centrale et du Sud-est. Règne Minéral 2006, 67, 17–48. 6. Garnier, V.; Maluski, H.; Giuliani, G.; Ohnenstetter, D.; Schwarz, D. Ar–Ar and U–Pb ages of marble-hosted ruby deposits from central and southeast Asia. Canad. J. Earth Sci. 2006, 43, 509–532. [CrossRef] 7. Giuliani, G.; Ohnenstetter, D.; Fallick, A.E.; Groat, L.; Fagan, A.J. The geology and genesis of gem corundum deposits. In Geology of Gem Deposits, 2nd ed.; Mineralogical Association of Canada Short Course Series; Groat, L.A., Ed.; Mineralogical Association of Canada: Quebec City, QC, Canada, 2014; Volume 44, pp. 113–134. ISBN 9780921294375. 8. Graham, I.T.; Khin, Z.; Cook, N.J. The genesis of gem deposits. Ore Geol. Rev. 2008, 34, 1–2. [CrossRef] 9. Graham, I.; Sutherland, L.; Zaw, K.; Nechaev, V.; Khanchuk, A. Advances in our understanding of the gem corundum deposits of the West Pacific continental margins intraplate basaltic fields. Ore Geol. Rev. 2008, 34, 200–215. [CrossRef] 10. Sorokina, E.S.; Karampelas, S.; Nishanbaev, T.R.; Nikandrov, S.N.; Semiannikov, B.S. Sapphire megacrysts in syenite pegmatites from the Ilmen Mountains, South Urals, Russia: New mineralogical data. Can. Miner. 2017, 55, 823–843. [CrossRef] 11. Muhlmeister, S.; Fritsch, E.; Shigley, E.J.; Devouard, B.; Laurs, M.B. Separating natural and synthetic rubies on the basis of trace-element chemistry. Gems Gemol. 1998, 34, 80–101. [CrossRef] 12. Sutherland, F.L.; Schwarz, D. Origin of gem corundums from basaltic fields. Aust. Gemmol. 2001, 21, 30–33. 13. Saminpanya, S. Mineralogy and Origin of Gem Corundum Associated with Basalt in Thailand. Ph.D. Thesis, University of Manchester, Manchester, UK, 2000. 14. Giuliani, G.; Fallick, A.E.; Garnier, V.; France-Lanord, C.; Ohnenstetter, D.; Schwarz, D. Oxygen isotope composition as a tracer for the origins of rubies and sapphires. Geology 2005, 33, 249. [CrossRef] Minerals 2019, 9, 49 36 of 41

15. Abduriyim, A.B.; Kitwaki, H. Determination of the origin of blue sapphire using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). J. Gemmol. 2006, 30, 23–26. [CrossRef] 16. Peucat, J.J.; Ruffault, P.; Fritsch, E.; Bouhnik-Le Coz, M.; Simonet, C.; Lasnier, B. Ga/Mg ratio as a new geochemical tool to differentiate magmatic from metamorphic blue sapphires. Lithos 2007, 98, 261–274. [CrossRef] 17. Sutherland, F.L.; Duroc-Danner, J.M.; Meffre, S. Age and origin of gem corundum and zircon megacrysts from the Mercaderes–Rio Mayo area, South-west Colombia, South America. Ore Geol. Rev. 2008, 34, 155–168. [CrossRef] 18. Papastamatiou, I. The Emery of Naxos. Geol. Geoph. Res. 1951, 1, 37–68. 19. Andronopoulos, V.; Tsoutrelis, C. Investigation of a Red Corundum Occurence in Xanthi, Greece. Unpublished Report; Greek Geol. Survey, Athens, Greece. 1964; 8p. 20. Ktenas, C. La géologie de l’ île de Nikaria (rédigée des restes de l’auteur par G. Marinos). Geol. Geoph. Res. 1969, 13, 57–85. 21. Feenstra, A. Metamorphism of Bauxites on Naxos, Greece. Ph.D. Thesis, Instituut voor Aardwetenschappen, Rijksuniversiteit te Utrecht, Utrecht, The Netherlands, 1985. 22. Liati, A. Corundum-and zoisite-bearing marbles in the Rhodope Zone, Xanthi area (N. Greece): Estimation of the fluid phase composition. Mineral. Petrol. 1988, 38, 53–60. [CrossRef] 23. Zannas, I. Corundum occurrence at Stirigma-Gorgona area, northern Xanthi. Petrogenesis and color variations. Miner. Wealth 1995, 97, 9–18. 24. Iliopoulos, I.; Katagas, C. Corundum bearing metabauxites from Ikaria Island (Greece): Mineralogy and geochemistry. In Proceedings of the 10th International Congress of the Geological Society of Greece, Thessaloniki, Greece, 15–17 April 2004; pp. 423–424. 25. Voudouris, P. The minerals of Eastern Macedonia and Western Thrace: Geological framework and environment of formation. Bull. Geol. Soc. Greece 2005, 37, 62–77. 26. Voudouris, P.; Melfos, V.; Katerinopoulos, A. Precious stones in Greece: Mineralogy and geological environment of formation. Understanding the genesis of ore deposits to meet the demand of the 21st century. In Proceedings of the 12th Quadrennial IAGOD Symposium, Moscow, Russia, 21–24 August 2006. 27. Voudouris, P.; Graham, I.; Melfos, V.; Zaw, K.; Lin, S.; Giuliani, G.; Fallick, A.; Ionescu, M. Gem corundum deposits of Greece: Diversity, chemistry and origins. In Proceedings of the 13th Quadrennial IAGOD Symposium, Adelaide, Australia, 6–9 April 2010; Volume 69, pp. 429–430. 28. Voudouris, P.; Katerinopoulos, A.; Magganas, A. Mineralogical geotopes in Greece: Preservation and promotion of museum specimens of minerals and gemstones. Sofia Initiative “Mineral diversity preservation”. In Proceedings of the IX International Symposium on Mineral Diversity Research and Reservation, Sofia, Bulgaria, 16–18 October 2017; pp. 149–159. 29. Graham, I.; Voudouris, P.; Melfos, V.; Zaw, K.; Meffre, S.; Sutherland, F.; Giuliani, G.; Fallick, A. Gem corundum deposits of Greece: A spectrum of compositions and origins. In Proceedings of the 34th IGC Conference, Brisbane, Australia, 5–10 August 2012. 30. Wang, K.K.; Graham, I.T.; Lay, A.; Harris, S.J.; Cohen, D.R.; Voudouris, P.; Belousova, E.; Giuliani, G.; Fallick, A.E.; Greig, A. The Origin of A New Pargasite-Schist Hosted Ruby Deposit From Paranesti, Northern Greece. Can. Mineral. 2017, 55, 535–560. [CrossRef] 31. Voudouris, P. Hydrothermal corundum, topaz, diaspore and alunite supergroup minerals in the advanced argillic alteration lithocap of the Kassiteres-Sapes porphyry-epithermal system, western Thrace, Greece. Neues Jahrbuch für Mineralogie Abhandlungen 2014, 191, 117–136. [CrossRef] 32. Pouchou, J.L.; Pichoir, F. Quantitative analysis of homogeneous or stratified microvolumes applying the model “PAP”. In Electron Probe Quantitation; Heinrich, K.F.J., Newbury, D.E., Eds.; Plenum Press: New York, NY, USA, 1991; pp. 31–75. 33. Griffin, W.L.; Powell, W.J.; Pearson, N.J.; O’Reilly, S.Y. GLITTER: Data reduction software for laser ablation ICP-MS. In Laser Ablation ICP-MS in the Earth Sciences: Current Practices and Outstanding Issues; Mineralogical Association of Canada Short Course Series; Sylvester, P., Ed.; Mineralogical Association of Canada: Quebec City, QC, Canada, 2008; Volume 40, pp. 308–311. 34. Van Achterbergh, E.; Ryan, C.G.; Jackson, S.E.; Griffin, W.L. Data reduction software for LA-ICP-MS: Appendix. In Laser Ablation-ICPMass Spectrometry in the Earth Sciences: Principles and Applications; Minerals 2019, 9, 49 37 of 41

Mineralogical Association of Canada Short Course Series; Sylvester, P.J., Ed.; Mineralogical Association of Canada: St John’s, NL, Canada, 2001; Volume 29, pp. 239–243.

35. Driesner, T.; Heinrich, C.A. The system H2O-NaCl. I. Correlations for molar volume, enthalpy, and isobaric heat capacity from 0 to 1000 degrees C, 1 to 5000 bar, and 0 to 1 X-NaCl. Geochim. Cosmochim. Acta 2007, 71, 4880–4901. [CrossRef] 36. Sharp, Z.D. In situ laser microprobe techniques for stable isotope analysis. Chem. Geol. 1992, 101, 3–19. [CrossRef] 37. Jolivet, L.; Brun, J.P. Cenozoic geodynamic evolution of the Aegean region. Int. J. Earth. Sci. 2010, 99, 109–138. [CrossRef] 38. Ring, U.; Glodny, J.; Thomson, S. The Hellenic subduction system: High-pressure metamorphism, exhumation, normal faulting and large-scale extension. Earth Plan. Sci. 2010, 38, 45–76. [CrossRef] 39. Jolivet, L.; Faccenna, C.; Huet, B.; Labrousse, L.; Le Pourhiet, L.; Lacombe, O.; Lecomte, E.; Burov, E.; Denèle, Y.; Brun, J.-P.; et al. Aegean tectonics: Strain localization, slab tearing and trench retreat. Tectonophysics 2013, 597, 1–33. [CrossRef] 40. Fytikas, M.; Innocenti, F.; Manneti, P.; Mazzuoli, R.; Peccerillo, A.; Villari, L. Tertiary to Quaternary evolution of volcanism in the Aegean region. In The Geological Evolution of the Eastern Mediterranean; Dixon, J.E., Robertson, A.H.F., Eds.; Geological Society Special Publications: Oxford, UK, 1984; Volume 17, pp. 687–699. 41. Kydonakis, K.; Brun, J.-P.; Sokoutis, D. North Aegean core complexes, the gravity spreading of a thrust wedge. J. Geophys. Res. Solid Earth 2015, 120, 595–616. [CrossRef] 42. Voudouris, P. Gemstones and Mineral-Megacrysts in Greece; University New South Wales: Sydney, Australia, 2010. 43. Burg, J.-P. Rhodope: From Mesozoic convergence to Cenozoic extension. Review of petro-structural data in the geochronological frame. J. Virtual Explor. 2012, 42, 1–44. [CrossRef] 44. Brun, J.-P.; Sokoutis, D. Kinematics of the Southern Rhodope Core Complex (North Greece). Int. J. Earth Sci. 2007, 96, 1079–1099. [CrossRef] 45. Turpaud, P.; Reischmann, T. Characterization of igneous terranes by zircon dating: Implications for UHP occurrences and suture identiﬁcation in the Central Rhodope, Northern Greece. Int. J. Earth Sci. 2010, 99, 567–591. [CrossRef] 46. Mposkos, E.; Kostopoulos, D. Diamond, former coesite and supersilicic garnet in metasedimentary rocks from the Greek Rhodope: A new ultrahigh-pressure metamorphic province established. Earth Plan. Sci. Lett. 2001, 192, 497–506. [CrossRef] 47. Liati, A. Identiﬁcation of repeated Alpine (ultra) highpressure metamorphic events by U-Pb SHRIMP geochronology and REE geochemistry of zircon: The Rhodope zone of Northern Greece. Contrib. Mineral. Petrol. 2005, 150, 608–630. [CrossRef] 48. Gautier, P.; Bosse, V.; Cherneva, Z.; Didier, A.; Gerdjikov, I. Polycyclic alpine orogeny in the Rhodope metamorphic complex: The record in migmatites from the Nestos shear zone (N. Greece). Bull. Geol. Soc. France 2017, 188, 36. [CrossRef] 49. Bonev, N.; Burg, J.-P.; Ivanov, Z. Mesozoic–Tertiary structural evolution of an extensional gneiss dome—The Kesebir–Kardamos dome, eastern Rhodope (Bulgaria–Greece). Int. J. Earth Sci. 2006, 95, 318–340. [CrossRef] 50. Menant, A.; Jolivet, L.; Vrielynck, B. Kinematic reconstructions and magmatic evolution illuminating crustal and mantle dynamics of the eastern Mediterranean region since the late Cretaceous. Tectonophysics 2016, 675, 103–140. [CrossRef] 51. Wüthrich, E. Low Temperature Thermochronology of the Northern Aegean Rhodope Massif. Unpublished Ph.D. Thesis, ETH Zürich, Zürich, Switzerland, 2009. 52. Melfos, V.; Voudouris, P. Cenozoic metallogeny of Greece and potential for precious, critical and rare metals exploration. Ore Geol. Rev. 2017, 59, 1030–1057. [CrossRef] 53. Duchêne, S.; Aïssa, R.; Vanderhaeghe, O. Pressure-temperature-time evolution of metamorphic rocks from Naxos (Cyclades, Greece): Constraints from thermobarometry and Rb/Sr dating. Geodin. Acta 2006, 19, 301–321. [CrossRef] 54. Ottens, B.; Voudouris, P. Griechenland: Mineralien-Fundorte-Lagerstätten; Christian Weise Verlag: Munchen, Germany, 2018; 480p. Minerals 2019, 9, 49 38 of 41

55. Beaudoin, A.; Augier, R.; Laurent, V.; Jolivet, L.; Lahfid, A.; Bosse, V.; Arbaret, L.; Rabillard, A.; Menant, A. The Ikaria hightemperature Metamorphic Core Complex (Cyclades, Greece): Geometry, kinematics and thermal structure. J. Geodesy 2015, 92, 18–41. 56. Papanikolaou, D.; Panagopoulos, G. On the structural style of Southern Rhodope (Greece). Geol. Balc. 1981, 11, 13–22. 57. Krenn, K.; Bauer, C.; Proyer, A.; Mposkos, E.; Hoinkes, G. Fluid entrapment and reequilibration during subduction and exhumation: A case study from the high-grade Nestos shear zone, Central Rhodope, Greece. Lithos 2008, 104, 33–53. [CrossRef] 58. Nagel, T.J.; Schmidt, S.; Janák, M.; Froitzheim, N.; Jahn-Awe, S.; Georgiev, N. The exposed base of a collapsing wedge: The Nestos Shear Zone (Rhodope Metamorphic Province, Greece): The Nestos shear zone. Tectonics 2011, 30.[CrossRef] 59. Bonneau, M. Correlation of the Hellenic nappes in the south-east Aegean and their tectonic reconstruction. Geol. Soc. Lond. Spec. Publ. 1984, 17, 517–527. [CrossRef] 60. Scheffer, C.; Vanderhaeghe, O.; Lanari, P.; Tarantola, A.; Ponthus, L.; Photiades, A.; France, L. Syn- to post-orogenic exhumation of metamorphic nappes: Structure and thermobarometry of the western Attic-Cycladic metamorphic complex (Lavrion, Greece). J. Geodyn. 2016, 96, 174–193. [CrossRef] 61. Grasemann, B.; Schneider, D.A.; Stöckli, D.F.; Iglseder, C. Miocene bivergent crustal extension in the Aegean: Evidence from the western Cyclades (Greece). Lithosphere 2012, 4, 23–39. [CrossRef] 62. Kruckenberg, S.C.; Vanderhaeghe, O.; Ferré, E.C.; Teyssier, C.; Whitney, D.L. Flow of partially molten crust and the internal dynamics of a migmatite dome, Naxos, Greece. Tectonics 2011, 30.[CrossRef] 63. Laurent, V.; Huet, B.; Labrousse, L.; Jolivet, L.; Monie, P.; Augier, R. Extraneous argon in high-pressure metamorphic rocks: Distribution, origin and transport in the Cycladic Blueschist Unit (Greece). Lithos 2017, 272, 315–335. [CrossRef] 64. Altherr, R.; Kreuzer, H.; Wendt, I.; Lenz, H.; Wagner, G.A.; Keller, J.; Harre, W.; Höhndorf, A. A late Oligocene/early Miocene high temperature belt in the Attic-Cycladic crystalline complex (SE Pelagonian, Greece). Geol. Jahrb. 1982, 23, 97–164. 65. Lister, G.S.; Banga, G.; Feenstra, A. Metamorphic core complexes of cordilleran type in the Cyclades, Aegean Sea, Greece. Geology 1984, 12, 221–225. [CrossRef] 66. Gautier, P.; Brun, J.P. Crustal-scale geometry and kinematics of late-orogenic extension in the central Aegean (Cyclades and Ewia Island). Tectonophysics 1994, 38, 399–424. [CrossRef] 67. Jansen, J.B.H.; Schuiling, R.D. Metamorphism on Naxos; petrology and geothermal gradients. Am. J. Sci. 1976, 276, 1225–1253. [CrossRef] 68. Wybrans, J.R.; McDougall, I. Metamorphic evolution of the Attic Cycladic Metamorphic Belt on Naxos (Cyclades, Greece) utilizing 40Ar/39Ar age spectrum measurements. J. Metam. Geol. 1988, 6, 571–594. [CrossRef] 69. Buick, I.S.; Holland, T.J.B. The P-T path associated with crustal extension, Naxos, Cyclades, Greece. In Evolution on Metamorphic Belts; Daly, J.S., Cliff, R.A., Yardley, B.W.D., Eds.; Geological Society Special Publications: Lomdon, UK, 1989; Volume 43, pp. 365–369. 70. Baker, J.; Matthews, A. The stable isotopic evolution of a metamorphic complex, Naxos, Greece. Contrib. Mineral. Petrol. 1995, 120, 391–403. [CrossRef] 71. Katzir, Y.; Valley, J.W.; Matthews, A.; Spicuzza, M.J. Tracking fluid flow during deep crustal anatexis: Metasomatism of peridotites (Naxos, Greece). Contrib. Mineral. Petrol. 2002, 142, 700–713. [CrossRef] 72. Andriessen, P.A.M.; Hebedaa, E.H.; Simonb, O.J.; Verschurea, R.H. Tourmaline K/Ar ages compared to other radiometric dating systems in Alpine anatectic leucosomes and metamorphic rocks (Cyclades and southern Spain). Chem. Geol. 1991, 91, 33–48. [CrossRef] 73. Feenstra, A.; Wunder, B. Dehydration of diasporite to corundite in nature and experiment. Geology 2002, 30, 119–122. [CrossRef] 74. Laurent, V.; Beaudoin, A.; Jolivet, L.; Arbaret, L.; Augier, R.; Rabillard, A.; Menant, A. Interrelations between extensional shear zones and synkinematic intrusions: The example of Ikaria Island (NE Cyclades, Greece). Tectonophysics 2015, 651–652, 152–171. [CrossRef] 75. Iliopoulos, I. Petrogenesis of Metamorphic Rocks from Ikaria Island. Ph.D. Thesis, University of Patras, Patra, Greece, 2005. Minerals 2019, 9, 49 39 of 41

76. Ring, U. The Geology of Ikaria Island: The Messaria extensional shear zone, granites and the exotic Ikaria nappe. J. Virtual Explor. 2007, 27, 3. 77. Liati, A.; Skarpelis, N. The metabauxites of Ikaria island, Eastern Aegean, Greece. In Proceedings of the 5th International Symposium on eastern Mediterranean Geology, Thessaloniki, Greece, 14–20 April 2004; pp. 1427–1430. 78. Henry, D.J.; Dutrow, B.L. Compositional zoning and element partitioning in nickeloan tourmaline from metamorphosed karstbauxite from Samos, Greece. Am. Mineral. 2001, 86, 1130–1142. [CrossRef] 79. Karantoni, V.D. Gemological Study of Corundum, Variety of Ruby and Sapphire from Greece. Unpublished Master’s Thesis, School of Geology, Aristotle University of Thessaloniki, Thessaloniki, Greece, 2018. 80. Bodnar, R.J. Introduction to fluid inclusions. In Fluid Inclusions: Analysis and Interpretation; Mineralogical Association of Canada Short Course Series; Samson, I.M., Anderson, A.J., Marshall, D.D., Eds.; Mineralogical Association of Canada: Quebec City, QC, Canada, 2003; Volume 32, pp. 1–8. 81. Giuliani, G.; Fallick, A.; Rakotondrazafy, M.; Ohnenstetter, D.; Andriamamonjy, A.; Ralantoarison, T.; Rakotosamizanany, S.; Razanatseheno, M.; Offant, Y.; Garnier, V.; et al. Oxygen isotope systematics of gem corundum deposits in Madagascar: Relevance for their geological origin. Mineral. Depos. 2007, 42, 251–270. [CrossRef] 82. Giuliani, G.; Fallick, A.E.; Ohnenstetter, D.; Pegre, G. Oxygen isotope composition of sapphire from the French Massif Central: Implications for the origin of gem corundum in basalt fields. Mineral. Depos. 2009, 44, 221–231. [CrossRef] 83. Wang, K.K.; Graham, I.T.; Martin, L.; Voudouris, P.; Giuliani, G.; Lay, A.; Harris, S.; Fallick, A.E. Fingerprinting Paranesti rubies through in-situ oxygen isotopes. Minerals. (under review). 84. Sutherland, F.L.; Zaw, K.; Meffre, S.; Giuliani, G.; Fallick, A.E.; Graham, I.T.; Webb, G.B. Gem corundum megacrysts from East Australia basalt fields: Trace elements, O isotopes and origins. Aust. J. Earth Sci. 2009, 56, 1003–1020. 85. Harris, S.J.; Graham, I.T.; Lay, A.; Powell, W.; Belousova, E.; Zappetini, E. Trace element geochemistry and metasomatic origin of alluvial sapphires from the Orosmayo region, Jujuy Province, Northwest Argentina. Canad. Mineral. 2017, 55, 595–617. [CrossRef] 86. Sutherland, F.; Zaw, K.; Meffre, S.; Yui, T.-F.; Thu, K. Advances in Trace Element “Fingerprinting” of Gem Corundum, Ruby and Sapphire, Mogok Area, Myanmar. Minerals 2015, 5, 61–79. [CrossRef] 87. Giuliani, G.; Lasnier, B.; Ohnenstetter, D.; Fallick, A.E.; Pégère, G. Les gisements de corindon de France. Le Règne Minéral 2010, 93, 5–22. 88. Giuliani, G.; Caumon, G.; Rakotozamisanny, S.; Ohnenstetter, D.; Rakotondrazafy, M. Classification chimique des corindons par analyse factorielle discriminante: Application à la typologie des gisements de rubis et saphirs. Rev. Gemmol. 2014, 188, 14–22. 89. Zwaan, J.H.; Buter, E.; Mertz-Kraus, R.; Kane, R.E. Alluvial sapphires from Montana: Inclusions, geochemistry and indications of a metasomatic origin. Gems Gemmol. 2015, 51, 370–391. 90. Schwarz, D.; Pardieu, V.; Saul, J.M.; Schmetzer, K.; Laus, B.M.; Giuliani, G.; Klemm, L.; Malsy, A.; Erel, E.; Hauzenberger, C.; et al. Rubies and Sapphires from Winza, Central Tanzania. Gems Gemmol. 2008, 44, 322–347. [CrossRef] 91. Rakotondrazafy, A.F.M.; Giuliani, G.; Ohnenstetter, D.; Fallick, A.E.; Rakotosamizanany, S.; Andriamamonjy, A.; Ralantoarison, T.; Razanatseheno, M.; Offant, Y.; Garnier, V.; et al. Gem corundum deposits of Madagascar: A review. Ore Geol. Rev. 2008, 34, 134–154. [CrossRef] 92. Pham Van, L.; Hoáng Quang, V.; Garnier, V.; Giuliani, G.; Ohnenstetter, D.; Lhomme, T.; Schwarz, D.; Fallick, A.E.; Dubessy, J.; Phan Trong, T. Gem corundum deposits in Vietnam. J. Gemmol. 2004, 29, 129–147. 93. Simonet, C.; Paquette, J.L.; Pin, C.; Lasnier, B.; Fritsch, E. The Dusi (Garba Tula) sapphire deposit, Central Kenya—A unique Pan-African corundum-bearing monzonite. J. Afr. Earth Sci. 2004, 38, 401–410. [CrossRef] 94. De Maesschalck, A.A.; Oen, I.S. Fluid and mineral inclusions in corundum from gem gravels in Sri Lanka. Mineral. Mag. 1989, 53, 539–545. [CrossRef] 95. Giuliani, G.; Dubessy, J.; Banks, D.; Quang, V.H.; Lhomme, T.; Pironon, J.; Garnier, V.; Trong Trinhd, P.;

Van Longf, P.; Ohnenstetter, D.; et al. CO2-H2S-COS-S8-AlO(OH)-bearing ﬂuid inclusions in ruby from marble-hosted deposits in Luc Yen area, North Vietnam. Chem. Geol. 2003, 194, 167–185. [CrossRef] 96. Menon, R.D.; Santosh, M.; Yoshida, M. Gemstone mineralization in southern Kerala, India. J. Geol. Soc. India 1994, 44, 241–252. Minerals 2019, 9, 49 40 of 41

97. Rakotondrazafy, A.F.M.; Moine, B.; Cuney, M. Mode of formation of hibonite (CaAl12O19) within the U-Th skarns from the granulites of S-E Madagascar. Contrib. Mineral. Petrol. 1996, 123, 190–201. [CrossRef] 98. Peretti, A.; Mullis, J.; Kündig, R. Die Kaschmir-Saphire und ihr geologisches Erinnerungsvermögen. Neue Zürcher Zeitung 1990, 187, 59. 99. Srithai, B.; Rankin, A.H. Fluid inclusion characteristics of sapphires from Thailand. In Mineral Deposits: Processes to Processing. Proceedings of the 5th Biennial SGA Meeting and the 10th Quadrennial IAGOD Symposium, London, UK, 22–25 August 1999; Stanley, C.J., Ed.; Balkema: Rotterdam, The Netherlands, 1999; pp. 107–110. 100. Touret, J.L.R. Le facies granulite en Norvege meridionale. II. Les inclusions fluids. Lithos 1971, 4, 423–436. [CrossRef] 101. Newton, R.C.; Smith, J.V.; Windley, B.F. Carbonic metamorphism, granulites and crustal growth. Nature 1980, 288, 45–50. [CrossRef] 102. Morrison, J.; Valley, J.W. Post-granulite facies fluid infiltration in the Adirondack Mountains. Geology 1988, 16, 513–516. [CrossRef] 103. Hollister, L.S.; Burruss, R.C.; Henry, D.L.; Hendel, E.M. Physical conditions during uplift of metamorphic terrains, as recorded by fluid inclusions. Bull. Minéral. 1979, 102, 555–561.

104. Lamb, W.M.; Valley, J.W. Metamorphism of reduced granulites in low-CO2 vapor-free environment. Nature 1984, 312, 56–58. [CrossRef] 105. Newton, R.C. Fluids of granulite facies metamorphism. In Fluid—Rock Interactions during Metamorphism; Walther, J.V., Wood, B.J., Eds.; Springer: New York, NY, USA, 1986; Volume 5, pp. 36–59.

106. Schuiling, R.D.; Kreulen, R. Are thermal domes heated by CO2-rich ﬂuids from the mantle? Earth Plan. Sci. Lett. 1979, 43, 298–302. [CrossRef] 107. Buick, I.S.; Holland, T.J.B. The nature and distribution of ﬂuids during amphibolite facies metamorphism, Naxos (Greece). J. Metam. Geol. 1991, 9, 301–314. [CrossRef] 108. Mposkos, E.; Krohe, A.; Baziotis, I. Alpine polyphase metamorphism in metapelites from Sidironero complex (Rhodope domain, NE Greece). In Proceedings of the XIX CBGA Congress, Thessaloniki, Greece, 23–26 September 2010; Volume 100, pp. 173–181. 109. Liati, A. Regional Metamorphism and Overprinting Contact Metamorphism of the Rhodope Zone, Near Xanthi (N. Greece): Petrology, Geochemistry, Geochronology. Unpublished Ph.D. Thesis, Technical University of Braunschweig, Braunschweig, Germany, 1986.

110. Storre, B.; Nitsch, K.-H. Zur Stabilität von Margarit im System CaO-Al2O3-SiO2-H2O. Contrib. Mineral. Petrol. 1974, 43, 1–24. [CrossRef]

111. Chatterjee, N. Margarite stability and compatibility relations in the system CaO-Al2O3-SiO2-H2O as a pressure-temperature indicator. Am. Mineral. 1976, 61, 699–709.

112. Chatterjee, N. The system CaO-Al2O3-SiO2-H2O: New phase equilibria data, some calculated phase relations, and their petrological applications. Contrib. Mineral. Petrol. 1984, 88, 1–13. [CrossRef] 113. Okrusch, M.; Bunch, T.E.; Bank, H. Paragenesis and petrogenesis of a corundum bearing marble at Hunza (Kashmir). Mineral. Depos. 1976, 11, 278–297. [CrossRef]

114. Haas, H.; Holdaway, M.J. Equilibria in the system Al2O3-SiO2-H2O involving the stability limits of pyrophyllite. Am. J. Sci. 1973, 273, 449–464. [CrossRef]

115. Seifert, F. Stability of sapphirine: A study of the aluminous part of the system MgO-Al2O3-SiO2-H2O. J. Geol. 1974, 82, 173–204. [CrossRef] 116. Markl, G.; Scharlau, A. Grosse Sapphire, Margarit und Tremolit aus den Smirgel-gruben auf Naxos, Griechenland. Lapis Mineral. Mag. 2008, 33, 26–35. 117. Garnier, V.; Giuliani, G.; Ohnenstetter, D.; Fallick, A.E.; Dubessy, J.; Banks, D.; Hoang Quang, V.; Lhomme, T.; Maluski, H.; Pêcher, A.; et al. Marble-hosted ruby deposits from Central and Southeast Asia: Towards a new genetic model. Ore Geol. Rev. 2008, 34, 169–191. [CrossRef] 118. Matthews, A.; Putlitz, B.; Hamiel, Y.; Hervig, R.L. Volatile transport during the crystallization of anatectic melts: Oxygen, boron and hydrogen stable isotope study on the metamorphic complex of Naxos, Greece. Geochim. Cosmochim. Acta 2003, 67, 3145–3163. [CrossRef] 119. Siebenaller, L. Fluid Circulations during Collapse of an Accretionary Prism: Example of the Naxos Island Metamorphic Core Complex (Cyclades, Greece). Ph.D. Thesis, University of Lorraine, Lorraine, France, 2008. 120. Haas, H. Diaspore-corundum equilibria determined by epitaxis of diaspore on corundum. Am. Mineral. 1972, 57, 1375–1385. Minerals 2019, 9, 49 41 of 41

121. Tenthorey, E.A.; Ryan, J.G.; Snow, E.A. Petrogenesis of sapphirine-bearing metatroctolites from the Buck Creek ultramaﬁc body, southern Appalachians. J. Metam. Geol. 1996, 14, 103–114. 122. Altherr, R.; Okrusch, M.; Bank, H. Corundum- and kyanite-bearing anatexites from the Precambrian of Tanzania. Lithos 1982, 15, 191–197. [CrossRef] 123. Mercier, A.; Debat, P.; Saul, J.M. Exotic origin of the ruby deposits of the Mangari area in SE Kenya. Ore Geol. Rev. 1999, 14, 83–104. [CrossRef] 124. Key, R.M.; Ochieng, J.O. The growth of rubies in south-east Kenya. J. Gemmol. 1991, 22, 484–496. [CrossRef] 125. Simonet, C. Géologie des Gisements de Saphir et de Rubis. L’exemple de la John Saul Mine, Mangari, Kenya. Ph.D. Thesis, Université de Nantes, Nantes, France, 2000. 126. Garde, A.; Marker, M. Corundum crystals with blue-red color zoning near Kangerdluarssuk, Sukkertoppen district, West Greenland. Report Gronlands Geologiske Undersokelse 1988, 140, 46–49. 127. Kissin, A.J. Ruby and sapphire from the Southern Ural Mountains, Russia. Gems Gemmol. 1994, 30, 243–252. [CrossRef] 128. Peretti, A.; Mullis, J.; Mouawad, F. The role of ﬂuorine in the formation of color zoning in rubies from Mong Hsu, Myanmar (Burma). J. Gemmol. 1996, 25, 3–19. [CrossRef]

129. Tropper, P.; Manning, C.E. The solubility of corundum in H2O at high pressure and temperature and its implications for Al mobility in the deep crust and upper mantle. Chem. Geol. 2007, 240, 54–60. [CrossRef] 130. Hatipo˘glu,M.; Türk, N.; Chamberlain, S.; Akgün, M. Gem-quality transparent diaspore (zultanite) in bauxite deposits of the Ibir˙ Mountains, Menderes Massif, SW Turkey. Mineral. Depos. 2010, 45, 201–205. [CrossRef] 131. Pardieu, V. Blue sapphires and Beryllium: An unﬁnished world quest. InColor 2013, 23, 36–43.

35th International Geological Congress 27 August – 4 September 2016, Cape Town, South Africa

Origin of rubies from the Paranesti region, North-Eastern Greece Wang, K.K.1, Graham, I.1, Cohen, D. 1and Voudouris, P2

1School of Biological, Earth and Environmental Sciences, University of New South Wales, Australia, [email protected]. 2Department of Mineralogy-Petrology, University of Athens, Greece

Gem-quality rubies occur in the Paranesti region of north-eastern Greece and were found by individual prospectors after road construction. To date, no systematic detailed research has been conducted on this gem deposit. The rubies are hosted in boudins of metamorphic actinolite schists, along with uncommon muscovite, kyanite and rare feldspars. The deposit is structurally bound by the syn-metamorphic nappe-system Figure 1: Primary ruby location map [1] (Rhodope Mountain) of Alpine age that formed during the Cretaceous to Mid-Tertiary collision of the paleo Europe tectonic regime.

A review of published data on existing corundum deposits compared to the chemical and mineralogical studies undertaken on the samples allows for a better understanding of the conditions of formation for these Greek rubies. To date, the deep-red rubies collected have been analysed for oxygen isotopes (δ18O, by laser fluorination) and trace element signatures (LA-ICP-MS method). Oxygen isotope values (δ18O) for the Paranesti rubies have been found to be ~1%, the same as sea-water, and possibly indicating a potential subduction zone genetic environment [2]. Whole rock XRF analyses suggest a mafic precursor with SiO2 (40 wt%), Al2O3 (21 wt%), MgO (16wt%), Fe2O3 (6 wt%) along with trace Cr (10745 ppm), Ni (388 ppm) and Zn (282 ppm).

216 Recent studies have used trace elements to “fingerprint” corundum gem deposits. The deep-red ruby samples from Paranesti show have an average concentration of Cr (3284 ppm) for rubies, moderate Fe (678 ppm), very low Ti (6 ppm), Ga (5 ppm) and V (1 ppm) and no appreciable concentrations of any other trace elements (86ppm) [3]. Almost all Cr concentrated in the rubies with minimal detection in the host rock.

Mineral inclusions of kyanite and rutile textural evidence of Figure 2. Coarse-grained pressure shadows around associated muscovite grains, actinolite schist with combined with enclosing gneisses around the deposit and the subhedral rubies to regional metamorphic history indicate a potential 20mm. retrogression from ultrahigh-pressure (UHP) metamorphic facies such as eclogite to lower P-T amphibolite facies. These rubies thus provide important additional information on the metamorphic/tectonic evolution of the Rhodope Metamorphic Province which has been shown to consist of at least three difference stages in terms of P-T evolution [4].

References: [1] Google Earth v 7.1.5.1557 (14 Dec 2015) Paranesti, Greece (Basarsoft 2016 & Orion Me 2016) [2] Gaston, G et al. (2005) Geology 33.4: 249-252. [3] Voudouris, P.et al. (2010), 13th Quad IAGOD Symp. 69: 429430. Figure 3. Gem quality [4] Mposkos, Evripidis, et al. (2001) Conference Abstract of the ruby grain, Paranesti Workshop on Fluid/Slab/Mantle Interactions and Ultrahigh-P Minerals, Waseda University

217 Appendix 5 – Conference abstract

XXII Meeting of the International Mineralogical Association 13-17 August 2018, Melbourne, Australia

IMA2018 Abstract submission Recent advances in our understanding of gem minerals

IMA2018-1121 Geographic typing of gem corundum taken a step further via in-situ oxygen isotope and trace element analysis: the example of Paranesti, Greece Kandy Wang* 1, Ian Graham1, Laure Martin2, Panagiotis Voudouris3, Angela Lay1, Stephen Harris1, Elena Belousova4, Gaston Giuliani5, Anthony Fallick6 1UNSW Sydney, Sydney, 2Centre for Microscopy, University of Western Australia, Perth, Australia, 3Mineralogy and Petrology, National and Kapodistrian University of Athens, Athens, Greece, 4GEMOC, Macquarie University, Sydney, Australia, 5IRD and CRPG, Universit´e de Lorraine, Paris, France, 6Scottish Universities Environmental Research Centre), Glasgow, United Kingdom

What is your preferred presentation method? Oral presentation Gem quality rubies and sapphires occur across a wide range of stable oxygen isotope δ18O‰ V- SMOW ratios (Giuliani et al., 2014). This technique is increasingly used as a unique identifier in corundum location typing and genesis given that mantle and crustal rocks usually have distinctive O-isotope compositions. Furthermore, trace element compositions and ratios have been used widely in the literature in order to identify likely sources of facetted gem corundums of unknown location. Oxygen (δ18O / δ16O) isotope ratios for gem-quality (cabochon) corundum occurrences from Paranesti, north-eastern Greece (here termed PAR-1 and PAR-5) have been analysed using high precision in-situ ion microprobe (SIMS) for the first time.

218 The ruby occurrence is located within the Nestos Shear Zone (NSZ) of the Rhodope Mountain Complex (RMC) in north-eastern Greece. The NSZ is a major high-strain intermediate ductile mylonitic zone with a polymetamorphic history which separates the hanging wall (Rhodope Terrane) from pre-Mesozoic continental basement in the footwall (Pangaion-Pirin Complex). Notable trace element chemistry differences (Table 1) have been found for the PAR-1 (main site) and PAR-5 (road-cutting) locations (Wang et. al, 2017).

The Paranesti O-isotope compositions range from -0.3‰ to +1.3‰. This consistently narrow δ18O range straddling -1‰ to +1‰ has not previously been found to exist for metamorphic corundum occurrences worldwide (Fig 1). Furthermore, the range of O-isotope compositions of the two suites PAR-1 (+0.6‰ to +1.3‰; n=49; mean value 1.0 ± 0.4‰) and PAR-5 (-0.3‰ to +0.8‰; n=44; mean value 0.1 ± 0.4‰) shows distinct range limits albeit being only few hundred metres apart. Globally, this is not the lowest observed in-situ oxygen isotope composition for corundum (i.e. Karelia -26‰). There are also documented Russian corundums from the Karelia region (Notozero - 1‰ to -2‰ and Perusel’ka +0.4‰ to +5.0‰) with similarly low oxygen isotope composition but these are either in the negative domain or close to/higher than 1‰ (Vysotskiya et. al, 2014) but no deposits were found to be overlapping between -1‰ to +1‰. In addition, PAR-1 and PAR-5 do not overlap each other considering the standard deviations.

The unique signature in oxygen isotopes is further confirmed by the different chemical signatures as shown in the table above. This suggests the possibility of one localised instance of meteoric fluid induced mesomatic influence for PAR-5 whilst both occurrences could be attributed to an overall metamorphic origin. An alternative hypothesis would suggest the preservation of the initial oxygen isotope signature (seawater alteration) which is preserved for an oceanic protolith that was then subducted to higher pressure and temperature during the amphibolite to granulite metamorphism.

219 References: Giuliani G., Ohnenstetter D. Fallick A. E., Groat L., & Fagan A. J. (2014): Chapter 2: The Geology and Genesis of Gem Corundum Deposits. Geology of Gem Deposits. Mineralogical Association of Canada, 2 (2): 29–112.

Wang K. K., Graham I. T., Lay A., Harris S.J., Cohen D.R., Voudouris P., Belousova E., Giuliani G., Fallick A. E. & Greig A. (2017): The origin of a new pargasite-schist hosted ruby deposit from Paranesti, Northern Greece. Canadian Mineralogist, 55: 535 – 560.

Vysotskiy, S.V., Ignat'ev, A.V., Levitskii, V.I., Nechaev, V.P., Velivetskaya, T.A., Yakovenko, V.V., (2014): Geochemistry of stable oxygen and hydrogen isotopes in minerals and corundum-bearing rocks in Northern Karelia as an indicator of their unusual genesis. Geochem. Int. 52 (9), 773–782.

220 Appendix 6 - Tables

Table 1. Detection limits for major and trace elements of the PANalytical PW2400 Sequential WDXRF Spectrometer.

Major oxides Limits of Detection Trace element Limits of detection (ppm) (ppm)

SiO2 60 As 2.5

TiO2 57 Ba 8.0

Al2O3 59 Cd 15.4

Fe2O3 43 Co 2.9

MgO 53 Cr 2.9

CaO 45 Cu 2.0

Mn3O4 35 Ga 1.0

Na2O 60 Mo 1.8

K2O 39 Nb 1.0

P2O5 27 Ni 2.0

SO3 55 Pb 2.0

Cr2O3 31 Rb 1.0

NiO 15 Sb 2.9

Sn 3.9

Sr 0.9

Th 2.8

U 2.9

V 2.9

Y 1.0

Zn 1.8

Zr 1.0

221 Table 2. Detection limit of Agilent 7700x ICP-MS used for trace element analysis.

222 Table 3. Analytical setting for EMPA analyses

Corundum Channel Crystal Element/Line Standards(s) Peak Pos (mm) Background (-) Background (+) 1 TAP F Fluorite 199.287 -7 6 Na Albite 129.446 Mg Periclase 107.493 -9 6

Al Al2O3 synthetic/albite 90.646 Si Various 77.45 -3 3.5 2 PET K Sanidine 120.101 Ca Diopside 107.931

3 LIF Ti Rutile/TiO2 191.371 -6 8 Ba Benitoite/barite 193.177/193.150 -6 6 4 LIF Mn Rhodonite 146.343 Fe Hematite 134.834

Amphiboles Channel Crystal Element/Line Standards(s) Peak Pos (mm) Background (-) Background (+) 1 TAP F Fluorite 199.287 -7 6 Na Albite 129.446 Mg Periclase 107.493 -9 6

Al Al2O3 synthetic/albite 90.646 Si Various 77.45 -3 3.5 2 PET K Sanidine 120.101 Ca Diopside 107.931

3 LIF Ti Rutile/TiO2 191.371 -6 8 Ba Benitoite/barite 193.177/193.150 -6 6 4 LIF Mn Rhodonite 146.343 Fe Hematite 134.834

223 Table 4. Corrected δ18O results (V-SMOW) from SIMS analyses.

Delta Delta Sample points values Sample points values Position PAR-1 δ18O PAR-5 δ18O O_PAR-1@02 0.80 O_PAR-5central@02 0.19 O_PAR-1@03 1.19 O_PAR-5central@03 0.33 O_PAR-1@04 0.98 O_PAR-5central@04 0.45 O_PAR-1@06 1.17 O_PAR-5central@09 0.31 O_PAR-1@07 1.11 O_PAR-5central@1 0.25 O_PAR-1@08 0.89 O_PAR-5central@10 0.40 O_PAR-1@09 0.92 O_PAR-5central@11 0.12 O_PAR-1@1 1.12 O_PAR-5central@12 -0.04 O_PAR-1@10 0.92 O_PAR-5central@13 0.10 O_PAR-1@11 1.12 O_PAR-5central@14 0.25 O_PAR-1@12 0.91 O_PAR-5central@15 0.51 O_PAR-1@13 0.84 O_PAR-5central@16 0.42 O_PAR-1@14 1.02 O_PAR-5grainA@1 0.85 rim O_PAR-1@15 1.02 O_PAR-5grainA@10 0.13 core O_PAR-1@16 0.89 O_PAR-5grainA@2 0.51 rim O_PAR-1@17 0.99 O_PAR-5grainA@3 0.35 core O_PAR-1@18 1.20 O_PAR-5grainA@4 0.19 core O_PAR-1@19 1.14 O_PAR-5grainA@5 0.27 rim O_PAR-1@20 1.02 O_PAR-5grainA@6 0.10 rim O_PAR-1b@21 1.11 O_PAR-5grainA@7 0.24 core O_PAR-1b@22 1.31 O_PAR-5grainA@8 -0.14 core O_PAR-1b@23 1.15 O_PAR-5grainA@9 -0.02 core O_PAR-1b@24 1.26 O_PAR-5grainB@1 0.01 rim O_PAR-1b@25 1.17 O_PAR-5grainB@2 0.15 rim O_PAR-1b@26 1.12 O_PAR-5grainB@3 -0.14 core O_PAR-1b@27 0.86 O_PAR-5grainB@4 0.23 core

224 O_PAR-1b@28 0.73 O_PAR-5grainB@5 0.42 rim O_PAR-1b@29 0.71 O_PAR-5grainB@6 -0.31 core O_PAR-1b@30 0.65 O_PAR-5grainB@7 -0.27 core O_PAR-1b@31 1.08 O_PAR-5grainB@8 0.18 rim O_PAR-1b@32 0.77 O_PAR-5grainC@1 0.16 rim O_PAR-1b@33 0.88 O_PAR-5grainC@2 0.13 core O_PAR-1b@34 0.84 O_PAR-5grainC@3 0.03 core O_PAR-1b@35 0.71 O_PAR-5grainC@4 -0.12 core O_PAR-1c@36 0.75 O_PAR-5grainC@5 -0.18 core O_PAR-1c@37 1.01 O_PAR-5grainC@6 -0.22 core O_PAR-1c@38 1.13 O_PAR-5grainC@7 -0.10 core O_PAR-1c@39 1.06 O_PAR-5grainC@8 -0.18 core O_PAR-1c@40 0.91 O_PAR-5grainC@9 -0.03 core O_PAR-1c@41 1.06 O_PAR-5grainD@1 0.21 O_PAR-1c@42 1.08 O_PAR-5grainD@2 0.08 Delta Delta Sample points values Sample points values Position PAR-1 δ18O PAR-5 δ18O O_PAR-1c@44 1.16 O_PAR-5grainD@4 0.18 O_PAR-1c@45 0.98 O_PAR-5grainD@5 0.11 O_PAR-1c@46 0.84 O_PAR-1c@47 1.07 O_PAR-1c@48 1.14 O_PAR-1c@49 1.03 O_PAR-1c@50 1.20

225