Investigating the Potential Recovery of REY from Metalliferous Sediments in a Seafloor Analogue; the Troodos Ophiolite, Cyprus

UNIVERSITY OF SOUTHAMPTON

Faculty of Natural and Environmental Sciences

Ocean and Earth Science

Investigating the potential recovery of REY from metalliferous sediments in a seafloor analogue; The Troodos ophiolite, Cyprus

Pierre Josso

Thesis for the degree of Doctor of Philosophy

Submitted 16th of January 2017

UNIVERSITY OF SOUTHAMPTON

ABSTRACT

FACULTY OF NATURAL AND ENVIRONMENTAL SCIENCES Ocean and Earth Science

Geochemistry

Thesis for the degree of Doctor of Philosophy

INVESTIGATING THE POTENTIAL RECOVERY OF REY FROM METALLIFEROUS SEDIMENTS IN A SEAFLOOR ANALOGUE; THE TROODOS OPHIOLITE, CYPRUS

Pierre Josso

The perceived supply risk for essential materials used in the development of green energy and other state-of-the art technologies creates the need for investigation of new sources for these raw materials. Many of these raw materials are characterized as “critical” given supply risks posed by geographic location, the economic and political stability of producing countries, potential substitution and opportunities for recycling [European Commission, 2014]. At present, 20 raw materials are listed by the EU as critical and this inventory is likely to grow in the coming years as the world population increases, driven by the development of India, China, Africa, Brazil and others. Among these critical elements, the rare earth elements and yttrium (REY) form a group of 15 metals essential for the development of wind turbines, cell phones and batteries among other applications and their production has been under Chinese domination for the last three decades. More than 95 % of the consumed REY worldwide originated in China during the last thirty years, a monopole that reflects economical constrains rather than the unequal distribution of REY resources across the world. Indeed, important proven reserves are known outside China though their extraction is expensive and energy consuming. In addition, most REY-rich deposits possess important concentrations of actinides (U and Th) problematic for waste disposal. This study therefore investigates the potential recovery of REY from umbers, metalliferous sediments of the Troodos massif in Cyprus, as an alternative to the dominant magmatic-related REY deposits.

Field evidence and geochemical characterisation of umbers show strong similarities with high- temperature plume fall-out deposits observed in most mid-oceanic ridge settings. Umbers constitute fine-grained brown Fe-Mn-rich mudstones with an amorphous oxyhydroxides dominated mineralogy and total rare earth oxide contents of ≈0.05 wt. %. REY fractionation

trends show excellent comparison with signatures of hydrothermal particles settling around active vents. The umbers display a negative Ce anomaly in a convex upward REE trend when normalized to chondrite, characteristic of a hydrothermal signal overprinted by seawater. From an economic perspective, although the REE content is low, the absence of mineralogical control on the distribution of these elements in umbers and the extremely low radioactive content (Th + U < 5 ppm) makes their potential extraction attractive.

A protocol for the leaching of umbers is presented testing a variety of lixiviants used in the REY extractive industry. Results show a strong mobilisation of the lanthanides in the solution in comparison with non-targeted elements. Most importantly, the results presented highlight that 80 to 90 % of the initial REY content of umbers is leached out using weak acid concentration in a matter of hours at low temperature. Fractionation along the REY series during leaching usually favours the release of the middle and light REE with a decreasing trend towards the heavy REE, except for Yttrium. Ce recovery is minimal as a result of its tetravalent oxidation state allowing formation of acid-resistant Ce oxides.

Furthermore, a process of selective precipitation is presented for the purification of the leach solution and extraction of a solid REY phase using ammonium oxalate as a complexing and chelating agent. Precipitation experiments show the precipitation efficiency is a function of pH, between pH values ranging from 0.7 to 3.2, with more than 96 % of REY precipitated at pH > 1.1. Purity of the precipitate is adjusted using precise pH buffering to avoid Ca-oxalate formation as the major impurity. Indeed, mass balance calculations and direct EDS measurement of the oxalate precipitate by SEM show maximal purity at pH 1.1 (66 – 94 % REY) while increasing Ca precipitation decrease purity below 10 % at pH > 1.5. The fractionation observed along the lanthanide series during the precipitation experiments was successfully reproduced via numerical modelling using PHREEQC software. REE distribution within the precipitate therefore reflects the interplay of aqueous and solid REY-oxalate complexes stability constants as well as incorporation of REY within the structure of co-precipitating Ca- and Na-oxalates.

This study demonstrates the feasibility of extracting efficiently REY from Fe-Mn oxide-rich metalliferous sediments. These deposits constitute interesting alternatives to high-grade deposits and their processing for REY production could be valuated as a by-product of pigment production. Alternatively, the process presented here could be applied to other oxide-based formations including marine ferromanganese deposits, or industrial wastes containing comparable high-tech metals concentration and enrichment process.

Table of Contents

Table of Contents ...... i List of Tables ...... vii List of Figures ...... ix Declaration of authorship ...... xvii Acknowledgements ...... xix Definitions and Abbreviations ...... xxi Chapter 1: Introduction and context ...... 1

1.1 Rare earth elements ...... 1

1.1.1 Fundamental chemical properties of rare earth elements ...... 1 1.1.2 REE in the economy ...... 4

1.1.2.1 Applications, market and deposits classification ...... 4 1.1.2.2 Resources and production...... 8 1.1.2.3 Extraction and Processing ...... 8

1.2 REE in the ocean: sources and behaviour ...... 10

1.2.1 Rivers input and estuarine mixing ...... 11 1.2.2 REE in seawater ...... 12

1.2.2.1 REE fractionation and particle associations ...... 12 1.2.2.2 REE distribution in the water column ...... 12

1.3 REE in hydrothermal systems ...... 15

1.3.1 REE in hydrothermal fluids in the oceanic crust ...... 15 1.3.2 REE patterns in hydrothermal fluids ...... 15 1.3.3 Fluid-rock interactions and REE enrichment in hydrothermal fluids ...... 17 1.3.4 Factors of control for REE speciation in hydrothermal fluids ...... 19 1.3.5 Conclusion on hydrothermal REE signature ...... 20

1.4 REE behaviour during mixing of hydrothermal solution with seawater ...... 21

1.4.1 The hydrothermal REE budget to open ocean ...... 21 1.4.2 Particle formation and reaction in the buoyant plume ...... 22 1.4.3 REE fractionation by Fe particles ...... 23

i 1.5 Hydrothermal metalliferous sediments in the ocean: diversity and mode of formation ...... 24 1.6 The Troodos Ophiolite, Cyprus ...... 25

1.6.1 Ophiolites: history and terminology ...... 25 1.6.2 Location and regional geology of Cyprus ...... 26

1.6.2.1 The Mamonia Complex ...... 26 1.6.2.2 The Kyrenia range ...... 27 1.6.2.3 The Southern Transform Fault Zone ...... 27 1.6.2.4 The Troodos massif ...... 28 1.6.2.5 Rotation and uplift of the Troodos massif ...... 31

1.6.3 Fe-Mn metalliferous sediments of the Troodos ophiolite ...... 33

1.6.3.1 Ochre ...... 33 1.6.3.2 Umbers ...... 33 1.6.3.3 Umbers alteration facies ...... 37

1.6.3.3.1 Mn-depleted umbers ...... 37 1.6.3.3.2 Silicified umbers ...... 37

1.7 Thesis rational and objectives ...... 38 1.8 Thesis structure ...... 40

Chapter 2: Geology and geochemistry of umbers ...... 43

2.1 Introduction ...... 43 2.2 Fieldwork ...... 43

2.2.1 Objectives ...... 43 2.2.2 Sampling method ...... 44 2.2.3 Field observations and lithologies ...... 44 2.2.4 Mapping...... 49

2.2.4.1 Kampia area ...... 49 2.2.4.2 Margi area ...... 49

2.3 Geochemical characterization of umber deposits ...... 54

2.3.1 Methods ...... 54

2.3.1.1 Mineralogy ...... 54 2.3.1.2 Major elements analysis ...... 54

2.3.1.3 Trace elements ...... 55 2.3.1.4 Sr isotopes ...... 56 2.3.1.5 Principal component analysis ...... 57 2.3.1.6 Isocon diagrams ...... 58

2.3.2 Results ...... 59

2.3.2.1 Mineralogy...... 59 2.3.2.2 Scanning electron microscopy...... 59 2.3.2.3 Geochemical characterization of umbers and associated lithologies... 66 2.3.2.4 Massive umber geochemistry ...... 66

2.3.2.4.1 Whole rock geochemistry ...... 66 2.3.2.4.2 Statistical analysis ...... 69 2.3.2.4.3 Rare earth elements ...... 71

2.3.2.5 Alteration of Umber ...... 72

2.3.2.5.1 Mn-depleted umbers...... 72 2.3.2.5.2 Silicified umbers ...... 76

2.3.2.6 Sr isotopes ...... 79 2.3.2.7 Stratigraphic profiles ...... 81

2.4 Discussion...... 87

2.4.1 Comparison of umbers with other metalliferous sediments ...... 87

2.4.1.1 Major elements ...... 87 2.4.1.2 Evidence from REE ...... 90

2.4.2 Model of emplacement ...... 92 2.4.3 Umbers in Cyprus, resources and limits ...... 94

2.5 Conclusion ...... 97

Chapter 3: REY extraction by leaching experiments ...... 99

3.1 Introduction ...... 99 3.2 Material and methods ...... 100 3.3 Results of the leaching experiments ...... 102

3.3.1 Ion exchange solution ...... 102 3.3.2 Acid concentration effect ...... 104

iii 3.3.3 Liquid-to-solid ratio effect ...... 106 3.3.4 Time of reaction ...... 106 3.3.5 Effect of Temperature ...... 108 3.3.6 REE fractionation in the recovery...... 110 3.3.7 Multiple stage leaching ...... 110 3.3.8 X-ray diffraction analysis on residues ...... 112

3.4 Discussion ...... 113

3.4.1 Ionic solutions ...... 113 3.4.2 REE leachability ...... 113 3.4.3 Optimal leaching conditions ...... 114

3.5 Conclusions ...... 115

Chapter 4: Selective precipitation of REY from a leach solution using oxalate: an experimental and modelling approach ...... 117

4.1 Introduction ...... 117 4.2 Oxalate Chemistry ...... 117 4.3 Material and methods ...... 119 4.4 Results ...... 123

4.4.1 Mass Balance ...... 123 4.4.2 Elemental partitioning between solution and precipitate in various pH123 4.4.3 Purity of the precipitate ...... 125 4.4.4 Scanning electron microscopy on precipitate ...... 126 4.4.5 Overall REY recovery and distribution ...... 129

4.5 Modelling REY oxalate precipitation from umber leach solution...... 130

4.5.1 Solution modelling ...... 130

4.5.1.1 Stability constant of aqueous REY-oxalate complexes...... 130 4.5.1.2 Stability constant of solid REY oxalate complexes...... 132 4.5.1.3 Estimation of missing constant via linear free-energy relationship ... 134 4.5.1.4 Parameters of the model ...... 136 4.5.1.5 Inherent limits of the model...... 136

4.5.2 Results ...... 137

4.5.2.1 Comparison between experimental and modelled results ...... 139

4.6 Conclusion ...... 142

Chapter 5: Conclusion...... 143

5.1 General conclusions of the study...... 143 5.2 New knowledge gained from this study ...... 145 5.3 From laboratory to industrial scale...... 146 5.4 Economic feasibility and future of REY extraction in Europe ...... 147 5.5 Prospective research ...... 149

Appendices ...... 151 Appendix A ...... 153 Appendix B ...... 161 Appendix C ...... 169 Appendix D ...... 175 Bibliography ...... 189

List of Tables

Table 1.1: Chemical and physical properties of rare earth, Yttrium and Scandium, modified after Henderson (1996), [Laveuf and Cornu, 2009]...... 2

Table 1.2: Major end uses of REY (modified after Weng et al. (2015))...... 5

Table 1.3: Classification of rare earth ore deposits (modified after Kanazawa and Kamitani (2006))...... 6

Table 1.4: World annual production of rare earth oxides (including yttrium) in tonnes. Data for 2007 – 2011 from Golev et al. (2014), data for 2012 -2015 from USGS reports on mineral commodities [U.S.G.S, 2013, U.S.G.S, 2014, U.S.G.S, 2015, U.S.G.S, 2016]. .. 9

Table 1.5: Comparison of selected REE concentration in open seawater and filtered water from MAR hydrothermal plume [Mitra et al., 1994]...... 22

Table 2.1: Average ± 95 % CI of measured trace element concentration of rock standards run as triplicates and their recommended published values for accuracy and reproducibility checking...... 56

Table 2.2: Comparison of measured and published 87Sr/86Sr values for rock standards [Jochum et al., 2005]...... 57

Table 2.3: SEM quantitative analysis on veins contained in umbers from sample PJ-CY-2014-36 compared with palygorskite data from (a) Newman (1987) and (b) Boyle (1984). ... 62

Table 2.4: Geochemical composition of the main umberiferous lithologies for major (in wt.%) and trace elements data (in ppm) expressed as the median ± 95% confidence interval. Comparison with data on supra lava umbers from (a) Boyle (1990) (average n = 63), (b) Robertson and Fleet (1976) and (c) Ravizza et al. (1999)...... 68

Table 2.5: Pearson coefficient correlation matrix for unaltered massive umber samples (n = 59).70

Table 2.6: Sr isotope data for the different umber lithologies facies and sedimentary cover of the Perapedhi Formation of the Troodos Ophiolite (dpl = duplicate)...... 80

Table 3.1: Range of experimental parameters and type of lixiviant used in the leaching experiments...... 100

vii Table 3.2: Elements concentration as a mean of triplicate ± 1 standard deviation measured by ICP- MS for international rock standards BHVO2, BIR1 and JB3 compared to their published values from Jochum et al. (2005)...... 103

Table 3.3: Recapitulative table of rare earth and yttrium yields from acid leaching experiments at the threshold and maximum test parameter values...... 114

Table 4.1: Composition of the stock leach solution (SLS) presented as the mean of triplicate and absolute standard deviation...... 120

Table 4.2: Measurements comparison for BHVO2, BIR1 and JB3 international rock standards prepared with or without oxalates to study potential matrix effect of oxalate during ICP analysis...... 121

Table 4.3: Elements concentration measured by ICP-MS for international rock standards BHVO2, BIR1 and JB3 compared to their published values from Jochum et al. (2005)...... 122

Table 4.4: Energy-dispersive X-ray spectroscopy analysis of the oxalate precipitates at pH 1.1. Note that EDS analysis were made on a free C and O basis. Area and crystals analysed are displayed in Figure 4.5 ...... 128

Table 4.5: Energy-dispersive X-ray spectroscopy analysis of the oxalate precipitates at different pH. Note that EDS analysis were made on a free C and O basis. Area and crystals analysed are displayed in Figure 4.5...... 129

Table 4.6: Data of log β (RE2Ox3.nH2O) at 25°C and infinite dilution...... 133

Table 5.1: Calculation of produced volume of reactant in experimental settings using prices and conditions of industrial products...... 148

Table 5.2: Example of high- and low-cost estimates of consumables for the production of 1Kg of an oxalate precipitate containing 1 Kg of REY using data from Table 5.1 and an initial REY concentration of 500 ppm in umbers...... 148

viii

List of Figures

Figure 1.1: Diagram showing the evolution of REY ionic radius as a function of their atomic number illustrating the lanthanide contraction and changes brought by the divalent and tetravalent ionic state of Ce and Eu respectively. This diagram also depicts why Y is classified as a heavy rare earth based on its ionic radius and its geochemical twining with Ho [Bau and Dulski, 1999]...... 3

Figure 1.2: Elements abundances in Earth’s upper continental crust as a function of atomic number. Many of the elements are classified into (partially overlapping) categories: (1) rock-forming elements (major elements in green field and minor elements in light green field); (2) rare earth elements (lanthanides, La–Lu, Y and Sc; labelled in blue); (3) major industrial metals (global production >~3x107 kg/year; labelled in red); (4) precious metals (purple); and (5) the nine rarest “metals”—the six platinum group elements plus Au, Re, and Te (a metalloid) in the yellow field [Haxel et al., 2002]. ... 6

Figure 1.3: Schematic diagram illustrating the main environment of formation of alkaline igneous rocks and carbonatites [Goodenough et al., 2016]...... 7

Figure 1.4: Vertical profiles for REE (pmol/kg), Nd/Er, La/Yb ratio and for main nutrients (PO4 and

SiO2 in µmol/kg) in filtered seawater in total acid digestion; triangle (East Caroline Basin), grey square (Coral Sea Basin) and black diamonds (Tasman Sea Basin) from Zhang and Nozaki (1996), black asterisk (western North Pacific ocean near Japan) from Alibo and Nozaki (1999)...... 14

Figure 1.5: PAAS-normalized REE trends for unfiltered seawater (total acid digestion) samples from the East Caroline Basin according to their depth [Zhang and Nozaki, 1996]. Trends exhibit a progressive enrichment in total REE content and more developed Ce negative anomaly as depth increases...... 15

Figure 1.6: Chondrite normalized REE patterns for (A) end-member hydrothermal fluids from MOR and back-arc basin settings displaying LREE enrichment with positive Eu anomaly and smooth decrease until HREE [Klinkhammer et al., 1994, Mitra et al., 1994, Douville et al., 1999, Douville et al., 2002, Craddock et al., 2010] and seawater data from Mitra et al. (1994). (B) end-member hydrothermal fluids from the back-arc Manus Basin [Craddock et al., 2010]; flat REE patterns from DESMOS and SuSu Knolls highlight influence of sulphate complex on REE fractionation. (C) end-member filtered hydrothermal fluids from Sisters Peak vent. Differences in shape and enrichment of

ix REE trends suggest fluid composition variations over time at Sisters Peak vent: The increasing REE content over time is negatively correlated to Ca evolution directly depending on anhydrite precipitation [Schmidt et al., 2010]. (D) end-member filtered hydrothermal fluids from Two Boats vent; Variable amounts of particles (mainly anhydrite) in collected samples are thought to explain variation in REE enrichment and the negative Eu anomaly in the most REE enriched sample [Schmidt et al., 2010]. (E) end-member hydrothermal fluids from MAR and the Manus Basin [Douville et al., 1999, Douville et al., 2002]: strong LREE enrichment and Eu anomaly influenced by ultramafic substratum in Snake pit and Rainbow vents fluid, signatures from PacManus show LREE depletion with relatively flat MREE-HREE trends and pattern from Desmos vent show LREE depletion and HREE enrichment without Eu anomaly highlighting influence on REE fractionation of magmatic volatiles input in hydrothermal fluids. (F) hydrothermal anhydrite sampled from the active chimney in Two Boats [Schmidt et al., 2010]. Trends display flat LREE to MREE patterns, variable Eu anomaly and HREE depletion. The size of the Eu anomaly appear to be inversely related to the total REE enrichment ...... 16

Figure 1.7: Schematic geochemical model of metalliferous formations distributed around an active high temperature vent in an axial valley. The diagram presents the relationship between a high temperature hydrothermal system and associated metallic deposits. The mound is made of accumulated particulate sulphides and collapse debris of chimneys being oxidized on the seafloor (ochres) sitting on top of a VMS deposit. Later oxidation of venting products forms Fe and Mn oxyhydroxides that deposit by fall-out as metalliferous sediments with preferential accumulation in horst and graben structure. Box on the upper right highlights Fe-Mn oxyhydroxides scavenging process by electrostatic interaction of seawater trace elements complexed either as carbonates, hydroxides or as dissolved species [Koschinsky and Halbach, 1995]. ... 25

Figure 1.8: Location map and tectonostratigraphic terranes of Cyprus [Robertson and Xenophontos, 1993]...... 27

Figure 1.9: Schematic presentation of the genesis of the Troodos Ophiolite (A) and the evolution of the Island of Cyprus (B-D) [Cyprus Geological Survey, 2017] ...... 29

Figure 1.10: Alternative tectonic models of the rotation of the Troodos microplate: a) Expulsion from the Isparta angle; b) Collision, subduction, erosion and under thrusting of the Mamonia microcontinent; c) The collision of a trench with the Arabian continental margin to the east; d) Similar to b but with the Mamonia microcontinent on the subducting plate (from Robertson and Xenophontos (1993))...... 32

Figure 1.11: Schematic lithostratigraphic units of the Troodos ophiolitic sequence and sedimentary cover. Seismic stratigraphy from Moores and Vine (1971); Houtz and Ewing (1976). Note that due to tectonic structuration of the paleoridge, some units might be lacking in some localities, like the UPL near Skourioutissa bringing umbers and VMS deposits in normal contact...... 36

Figure 1.12: Reconstruction of the field relationship of a typical small umber hollow related to seafloor faulting, Troodos Massif (redrawn after Robertson and Boyle (1983))...... 37

Figure 2.1: Geological map of the eastern part of the Troodos Massif and sampling location. (Modified after 1:250 000 Geological map of Cyprus in Naden et al. (2006))...... 46

Figure 2.2: Photograph of an umber outcrop near Asgata, with the umber infilling a depression in the lava topography. The umbesr also show a good example of layering with an alternation of massive beds and thinner more friable ones as schematised in the stratigraphic column. Lateral extent of the outcrop is 11 m...... 47

Figure 2.3: Photographs of the basal “Mn-depleted” umber layer in Margi (A) and Kampia (B). (C); internal layering within umbers in Asgata, (D); umber outcrop evolution from massive beds at the base of the stratigraphy to a more clay-rich and layered deposit at the base of the stratigraphy to a more clay-rich and layered deposit at the top in Kampia, (E); silicified nodules within umbers near Margi, (F) massive bulbous silicified layer on top of umber outcrop and internal quartz veins (G) in the Margi Area...... 48

Figure 2.4: Geological map showing location of detailed mapping near Kampia (1) and Margi (2). Data from Naden et al. (2006)...... 50

Figure 2.5: (A) Geological map of the Kampia area, Cyprus. (B) Close up on the North West of the geological map of Kampia displaying location of cross-section transects referred to as a, b and c. (C) Schematic assemblage of cross-section transects showing layout of umbers...... 51

Figure 2.6: Geological map of the Mathiati-Margi [redrawn from Boyle and Robertson, 1984] presenting a more detailed geological context of area 2 on Figure 2.4. The red rectangle shows location of detailed mapping...... 52

Figure 2.7: Geological map of the Margi area, Cyprus. A-B cross section of the northern part of the outcrop presenting lithologies layout. Locally, umbers thickness up to 7 meter was observed. Note the cross cutting relationship of Mn-depleted umbers with intial umber layering...... 53

xi Figure 2.8: Measured versus recommended values for rock standards measured by XRF...... 55

Figure 2.9: Graphic representation of 87Sr/86Sr measurement on standard NBS987, 2σ individual error bars below points, the solid and dotted lines (area in grey) represent the published value of NBS987 ±2σ...... 57

Figure 2.10: X-ray diffraction patterns of apatite-rich umber, pyrolusite concretion and radiolarian cherts from the Perapedhi Formation...... 60

Figure 2.11: SEM electron backscattered image of carbon-coated polished thin section of sample PJ-CY-2014-36. In this image the dotted line represents the area scanned for elemental mapping (Figure 2.12) whereas the solid squares represent zones analysed with X-ray

for the elaboration of a geochemical profile considering the following oxides: Fe2O3,

SiO2, MnO, Al2O3, MgO, CaO and CoO (Figure 2.11Error! Reference source not found.)...... 62

Figure 2.12: X-ray element map by SEM of sample PJ-CY-2014-36. In these maps, the lighter shades indicate higher concentrations of the considered element...... 63

Figure 2.13: Geochemical profile by X-ray quantitative analysis on sample PJ-CY-2014-36 (± 2σ). Each square on the thin section represents the area scanned by X-ray for 3 min. .. 64

Figure 2.14: localised X-ray spectrum for geochemical identification of the various types and morphologies of grain identified by SEM in the matrix of sample PJ-CY-2014-36 .... 65

Figure 2.15: Fe-Si-Mn ternary diagram displaying the three main umberiferous lithologies encountered in the Troodos Ophiolite...... 66

Figure 2.16: Ternary discriminative diagrams for the genetic classification of oceanic ferromanganese deposits [Bonatti et al., 1972, Josso et al., 2016]...... 67

Figure 2.17: Correlation circle from the principal component analysis realised on samples of massive umbers (n = 59). Representation depicts factorial axes 1 and 2, respectively 39.4 % and 19.2 % of the total variance of the data set...... 69

Figure 2.18: PAAS-normalized REE plot for unaltered umbers, Mn-depleted umbers and silicified umbers (PAAS values from Taylor and McLennan (1985))...... 71

Figure 2.19: Immobile elements 2D plots for unaltered umbers, silicified umbers and Mn-depleted umbers. Legend similar to Figure 2.16...... 72

Figure 2.20: Examples of isocon diagrams plotting concentration of Mn-depleted umbers sample 88 (upper) and 98 (lower) rescaled against unaltered umbers for the analysis of

xii

element mobility during alteration. The grey line, determined by immobile elements, forms the line of no mass transfer; elements falling above constitute elemental gains while the zone of elemental loss is under the isochemical line. The concentration data have been normalized so that the sum of squares equal 1 [Humphris et al., 1998] avoiding errors in graphic interpretation due to arbitrary scaling ...... 74

Figure 2.21: pH vs Eh diagram presenting stability field for Fe and Mn as oxides and oxyhydroxides. Mn species left of the bold line are in the ionic form Mn2+, the grey area represent the Eh-pH window for diagenetic remobilisation in Mn-depleted umbers. The two diagonals long dotted lines define the stability field of water (298.15 K, 105 Pa) (modified from Takeno (2005))...... 76

Figure 2.22: Silicified umbers recalculated on an average SiO2 concentration normalized to average unaltered umber composition...... 78

Figure 2.23: 87Sr/86Sr isotopic ratio for unaltered umbers, silicified umbers, Mn-depleted umbers, pyrolusite concretion and sediments of the Perapedhi formation compared with data for late Cretaceous and Modern Seawater [Mc Arthur et al., 2001]...... 81

Figure 2.24: Geochemical profile for selected major and trace elements from Margi outcrop. 83

Figure 2.25: Geochemical profile for selected major and trace elements from Kampia outcrop. On each plot, the horizontal dotted line indicates the separation between lithologies.84

Figure 2.26: Geochemical profile for selected major and trace elements from Asgata. On each plot, the horizontal dotted line indicates the separation between lithologies...... 85

Figure 2.27: Fe2O3 vs MnO plot for the Cyprus umbers recalculated on a carbonate-free basis (CFB) relative to (A) modern submarine ferromanganese precipitates, note that our samples

are relatively poor in carbonates (av. < 5% CaCO3) while most data from cores incorporate a large portion of carbonates. Data sources: Back-arc Fe-rich sediments [Mc Murtry et al., 1991, Murphy et al., 1991, Binns et al., 1993, Sun et al., 2011]; arc Fe-rich sediments [Smith and Cronan, 1983, Savelli et al., 1999, Dekov et al., 2011]; metalliferous sediments from the Atlantic ridge [Metz et al., 1988, German et al., 1993, Mills et al., 1993]; metalliferous sediments from the East Pacific ocean [Piper, 1973, Heath and Dymond, 1977, Marchig and Erzinger, 1986, Rhulin and Owen, 1986b, Hrischeva and Scott, 2007]; Hawaii metalliferous sediments [Edwards et al., 2011]; metalliferous sediments from Japan [Kato et al., 2005a]; polymetallic nodules [Calvert and Price, 1977, Calvert and Piper, 1984, Dymond et al., 1984, Ohta et al., 1999,

xiii Baturin, 2009, Wegorzewski and Kuhn, 2014]; Hydrogenetic crusts [Bau, 1996, Kuhn et al., 1998, Hein et al., 2005, Asavin et al., 2010, Muiños et al., 2013]. (B) Fe2O3 vs MnO plot relative to on-land Fe-Mn deposits [Robertson and Hudson, 1972, Robertson and Fleet, 1976, Boyle, 1984, Robertson and Fleet, 1986, Karpoff et al., 1988, Robertson and Degnan, 1998, Kato et al., 2005a] ...... 89

Figure 2.28: PAAS-normalized REE data for umbers of the Troodos Ophiolite compared with various hydrothermally-derived actual and past sediments (see text for explanations). Data for hydrothermal fluids from Douville et al. (1999)and seawater from Alibo and Nozaki (1999)...... 91

Figure 2.29: Al index (100*Al)/(Al+Fe+Mn) vs Fe/Ti ratio displaying evolution of the umber deposit of Kampia from a hydrothermal end-member (sample 88-89) at the base of the profile towards a pelagic and detrital end-member (samples 104 and 105) at the top of the stratigraphic sequence...... 93

Figure 2.30: Schematic diagram outlining the formation and evolution of umber deposits in the Troodos Massif, Cyprus ...... 96

Figure 2.31: Close up of a typical umber outcrop accumulating in a depression of the basaltic basement, the depression can be related either to a small hollow in the topography or to a structurally controlled half graben. This figure is modified from Robertson and Boyle (1983) to highlight detailed field relations between massive umbers and alteration facies. The illustration shows: the irregular basal Mn-depleted umbers cross- cutting umber layering with associated pyrolusite concretion; phosphate-rich horizons; the various morphologies of silicified umber as nodules, layers and bulbous masses conserving undeformed palygorskite veins and overprinting phosphate-rich umbers and Mn-depleted umbers. The top of the outcrop shows the variation from massive umber layers to more clay-rich umbers and the transition to radiolarian cherts of the Perapedhi Formation. Late stage silicification along fault zones as quartz veins are best preserved with the more competent silicified masses...... 96

Figure 3.1: Element yields in the leach solution using ammonium sulphate and sodium chloride solutions at 20°C. Elements not presented in this figure have less than 1.5 % recovery. Details on conditions for each experiment are given on figures...... 104

Figure 3.2: Effect of acid molarity on the yield in the leach solution at 20°C for 120 min and liquid to solid ratio of 25:1 by weight...... 105

xiv

Figure 3.3: Effect of liquid to solid ratios on the yield in the leach solution using 1M HCl, 1M HNO3, 0.5M H2SO4 at 20°C for 120 min...... 107

Figure 3.4: Effect of the time of reaction on the yield in the leach solution...... 108

Figure 3.5: Effect of leaching temperature on element recovery from umbers using 1M acid solution at 25:1 liquid to solid ratio for 2h without stirring at 40 and 70°C...... 109

Figure 3.6: Detailed recovery for REY in each leaching experiment...... 111

Figure 3.7: Two step leaching experiment on sample PJ-CY-91. The mass for each element (µg) is presented for the most represented leach fraction...... 112

Figure 3.8: Comparison of X-ray diffraction patterns of sample 91 and the residue collected after filtration of the leaching experiment using 1M HCl at a liquid to solid ratio of 25, 20°C for 2h...... 113

Figure 4.1: Speciation of oxalic acid and conjugate oxalates as a function of pH using acid dissociation constant K1 = 5.9*10-2 and K2 = 6.4*10-5 [Chi and Xu, 1998]...... 119

Figure 4.2: Distribution of measured masses by ICP-MS in the solution and precipitate at pH 0.9, 1.6 and 2.3. For each element, measurements are normalized to the initial mass introduced in the experiments. In an ideal case, values should therefore add up to 100%...... 124

Figure 4.3: Element fractionation in the oxalate precipitate (mass percentage) as a function of pH from the stock leach solution...... 125

Figure 4.4: Element masses within the oxalate precipitate as a function of pH. The right axis represents the REY fraction or purity of the precipitate (%)...... 126

Figure 4.5: SEM electron back-scatter image of oxalate precipitate at pH 1.1 (A and B) and at pH 2.5 (C and D). Visible on the figure are areas and spot of EDS analysis (see Table 4.4 and Table 4.5) ...... 127

Figure 4.6: EDS spectrum of the oxalate precipitate obtain at pH 1.1 (A) and 2.5 (B). The spectrums correspond to the field of view in images A and C of Figure 4.5...... 128

Figure 4.7: REY concentration in the leaching solution (black) and the relative elemental recovery from the sample (blue)...... 129

xv Figure 4.8: REY patterns of stability constant at infinite dilution for log HOxβ1, Oxβ1 and Oxβ2 [Schijf and Byrne, 2001]...... 131

Figure 4.9: Graphic representation of –log β (RE2Ox3.nH2O) at 25°C and infinite dilution. Note that as a precipitating phase, the β values are negative and the graph present –log values of the constants for ease of comparison with aqueous complexation constants...... 133

Figure 4.10: Linear free-energy relationships for RE2Ox3.nH2O oxalate complexes with various organic acids. Displayed on graphs are equation of linear regression with R² values and 95 % CI on the linear regression...... 135

Figure 4.11.: Patterns of –log β (RE2Ox3.nH2O) and estimations from linear free-energy relationship at 25°C and infinite dilution using linear regression equation from the carboxylic-acid (LFER CA) and the phenol (LFER P) (Figure 4.10) ...... 135

Figure 4.12: REE and Ca speciation with oxalate as a function of pH calculated by PHREEQC. All concentrations used are those found in the experimental set up. Experimental results (black line and white squares) are overlaid for comparison...... 138

Figure 4.13: PHREEQC modelling of RE-oxalate precipitation as a function of the initial concentration of REE in the initial solution. All other parameters are as in the experiment. Experimental results (black line and white squares) are overlaid for comparison...... 140

Figure 4.14: Structure of the complex [Nd2(C2O4)3.10H2O]n forming planar layers in the monoclinic system [Hansson, 1970]...... 141

Figure 5.1: Workflow of metalliferous sediment processing for the extraction of REY developed in this study...... 147

xvi

Declaration of authorship

I, Pierre Josso, declare that this thesis and the work presented in it are my own and have been generated by me as the result of my own original research.

Investigating the potential recovery of REY from metalliferous sediments in a seafloor analogue; The Troodos ophiolite, Cyprus

I confirm that:

1. This work was done wholly or mainly while in candidature for a research degree at this University; 2. Where any part of this thesis has previously been submitted for a degree or any other qualification at this University or any other institution, this has been clearly stated; 3. Where I have consulted the published work of others, this is always clearly attributed; 4. Where I have quoted from the work of others, the source is always given. With the exception of such quotations, this thesis is entirely my own work; 5. I have acknowledged all main sources of help; 6. Where the thesis is based on work done by myself jointly with others, I have made clear exactly what was done by others and what I have contributed myself; 7. [Delete as appropriate] None of this work has been published before submission [or] Parts of this work have been published as: [please list references below]:

Signed: Pierre Josso ......

Date: 16/01/2017 ......

xvii

Acknowledgements

First of all, I am most grateful to my supervisors Steve Roberts, Damon Teagle, Carlos Ponce de Leon Albarran and Richard Herrington for giving me the opportunity to realise this work. Thanks to Steve, Damon and Martin Palmer for helpful discussions, fun fieldwork expeditions and the time spent correcting my French.

Sincere gratitude to the reviewers of my PhD thesis, Dr. Alastair H. F. Robertson and Dr. Juerg Matter for their helpful comments and discussion on the improvement of this document.

I am deeply thankful to Tim Van Peer and James Fielding, for their friendship and support during hard times, the laughs, and all the stupid things we invented.

I cannot miss the opportunity to thank as well all the others students and members of the University of Southampton that made these years most enjoyable.

A special thanks to Matt Cooper and Agnes Michalik for their help and advice on my lab work.

Many thanks to Daniel Doran for his friendship and his constant reminders of how fun it was to polish umbers.

Most sincere thanks to Douglas Connelly and Daniel Jones for welcoming me in their team on the JC120 expedition in the Clarion Clipperton Zone, a most welcomed experience.

Thanks to Esther Sumner, Juerg Matter, Justin Dix and Tim Minshull, a brilliant team for some unforgettable fieldwork in Spain.

Special thanks go to Olivier Pourret, Johan Schifj, Julien Declercq and Phil Warwick for their encouraging comments and help throughout my learning of chemical modelling.

Thanks to my family and friends who have always been interested in my project.

Finally, Séverine, thank you for all the support and strength you give me since I met you.

xix

Definitions and Abbreviations

Al index Aluminium index Ce/Ce* Cerium anomaly CFB Carbonate-free basis CHARAC Charge and radius controlled process

CN Chondrite normalized

CPSW Critical point of sea water DSDP Deep Sea Drilling Project EDPA Ethane Diphosphonic Acid EDS Energy-dispersive X-ray spectroscopy EHEHPA 2-ethylhexylphosphonic acid, mono-2-ethylhexyl ester EPR East Pacific rise Eu/Eu* Europium anomaly EXAFS Extended X-ray absorption fine structure FTIR Fourier transform infra-red HDEHP Organophosphorus compound: Di-(2-ethylhexyl)phosphoric acid HDPE High density polyethylene HFSE: High field strength elements HT High temperature HREE Heavy rare earth elements (europium through lutetium) IAC Ion-adsorption clay ICP-MS Inductively coupled plasma mass spectrometry LFER Linear free-energy relationship LILE Large ion lithophile elements LLNL Lawrence Livermore National Laboratory (database) LPL Lower pillow lava LS ratio Liquid-to-solid ratio LT Low temperature LREE Light rare earth elements (lanthanum through samarium) MAR Mid Atlantic ridge MOR Mid-oceanic ridge MORB Mid-oceanic ridge basalt MREE Mid rare earth elements (samarium through holmium) MS Mother solution NASC North-American shale composite

xxi NiMH Nickel metal hydride NOCS National Oceanography Centre of Southampton

Ox Abbreviation used for “C2O4” in equations and formulas PAAS Post-Archean Australian shale PCA Principal component analysis redox reduction and oxidation REE Rare earth elements REO Rare earth oxides REY Rare earth elements and yttrium SB Sub-boiled SEM Scanning electron microscopy SLS Stock leach solution SSZ Supra-subduction zone STTFZ Southern Troodos Transform fault Zone TAG Trans-Atlantic Geotraverse TBB Organophosphorus compound: Tributyl phosphate UPL Upper pillow lava USGS United State geological survey VMS Volcanic Massive Sulphide XANES X-ray absorption near edge spectroscopy XRD X-ray diffraction XRF X-ray fluorescence

xxii Chapter 1. Introduction

Chapter 1: Introduction and context

1.1 Rare earth elements

1.1.1 Fundamental chemical properties of rare earth elements

The rare earth metals are a group of chemically coherent elements, including the lanthanide series, spanning atomic numbers from 57 (La) to 71 (Lu) together with Scandium (Sc) and yttrium (Y) which display similar geochemical behaviour. Amongst the lanthanides, Promethium is a short- lived radioactive element and its most stable isotope 145Pm, possess a half-life of only 17.7 years. Therefore, Pm does not occur in significant amounts in natural systems and is generally omitted in studies of rare earth elements.

Lanthanides are traditionally subdivided into two main categories according to their atomic weight following mineral occurrences and market applications, the light REE (LREE); La through Sm and the heavy REE (HREE) which include Eu through Lu. Yttrium, with a similar ionic radius to holmium is classified as a heavy rare earth (Table 1.1, Figure 1.1) whereas Sc and Pm are not included within this classification due to Sc mineral occurrences being rather different from other REE, and Pm being radioactive [Golev et al., 2014]. However, from a geochemical point of view, no specific classification to describe the REE group has been accepted and the term middle REE (MREE; Sm to Ho) is sometimes added to the previous categories to give more details on fractionation processes between the span of REE ionic radii [Dubinin, 2004]. The lanthanide series plus yttrium are referred to as REY and the abbreviation REE only refer to the 15 lanthanides.

The REY are widely used to describe Earth processes and in particular act as tracers of fluid geochemistry and as indicators for material sources [Hanson, 1980, Hoyle et al., 1984, Sverjensky, 1984, de Baar et al., 1988, Elderfield, 1988, Bau, 1991, Olivarez and Owen, 1991, Varentsov et al., 1991, Sholkovitz et al., 1994]. Primarily existing as trivalent ions in solution (REY3+), the progressive decrease of ionic radii from La to Lu allows the study of fractionation processes which are insensitive to redox conditions such as fluid/rock or fluid/particle interaction and fractional crystallization. The progressive filling of the inner 4f-electron shell with increasing atomic number, protected by the outer 5s electron shell, prevents important changes in chemical reactivity. The addition of an extra electron increases the nuclear charge and gradually decreases the ionic radius due to 4f shell contraction, a phenomenon referred to as the lanthanide contraction.

1 Chapter 1. Introduction

Atomic Atomic weight Electronic structure Ionic radii a Electronegativity Symbol number g/mol Valence Z = 0 Z = + 3 A Pauling scale Scandium Sc 21 44.9559 3 [Ar] 3d1 4s2 n.a.b 0.745 n.a.b Yttrium Y 39 88.90584 3 [Kr] 4d1 5s2 n.a.b 0.9 n.a.b Lanthanum La 57 138.90547 3 [Xe].5d1.6s2 [Xe].4f0 1.045 1.1 Cerium Ce 58 140.116 3,4 [Xe].4f1.5d1.6s2 [Xe].4f1 1.01 1.12 Praseodymium Pr 59 140.90765 3 [Xe].4f3.6s2 [Xe].4f2 0.997 1.13 Neodymium Nd 60 144.242 3 [Xe].4f4.6s2 [Xe].4f3 0.983 1.14 Samarium Sm 62 150.36 3 [Xe].4f6.6s2 [Xe].4f5 0.958 1.17 Europium Eu 63 151.964 3,2 [Xe].4f7.6s2 [Xe].4f6 0.947 n.a.b Gadolinium Gd 64 157.25 3 [Xe].4f7.5d1.6s2 [Xe].4f7 0.938 1.22 Terbium Tb 65 158.92535 3 [Xe].4f9.6s2 [Xe].4f8 0.923 n.a.b Dysprosium Dy 66 162.5 3 [Xe].4f10.6s2 [Xe].4f9 0.912 1.22 Holmium Ho 67 164.93032 3 [Xe].4f11.6s2 [Xe].4f10 0.901 1.23 Erbium Er 68 167.259 3 [Xe].4f12.6s2 [Xe].4f11 0.89 1.24 Thulium Tm 69 168.9342 3 [Xe].4f13.6s2 [Xe].4f12 0.88 1.25 Ytterbium Yb 70 173.04 3 [Xe].4f14.6s2 [Xe].4f13 0.868 n.a.b Lutetium Lu 71 174.967 3 [Xe].4f14.5d1.6s2 [Xe].4f14 0.861 1.27 b n.a., non available data. Table 1.1: Chemical and physical properties of rare earth, Yttrium and Scandium, modified after Henderson (1996), [Laveuf and Cornu, 2009].

The ionic radius progressively decreases from La (0.105 nm) to Lu (0.085 nm) (Table 1.1, Figure 1.1), and brings minimal but systematic changes in their chemical properties [Elderfield, 1988, Sholkovitz et al., 1994]. The confinement of electronic configuration changes for each REE to the inner shells, rather than to the outer ones, gives these elements their highly coherent chemical behaviour. This coherence explains why all REE are usually found together in any natural system, and therefore why they are so hard to process and separate [Henderson, 1996]. The REY3+ fractionation in either mineral, particle or magmatic/metamorphic fluid is thus controlled by charge and radius changes across the series, referred as the CHARAC fractionation process [Bau, 1996]. On the other hand, REY3+ aqueous geochemistry is controlled by complex stability and interactions with other particles during adsorption processes. However, some of the REE are redox sensitive and can be oxidized or reduced given the right environment leading to ionic radius and charge changes. Cerium oxidation to Ce4+ has been largely documented in seawater [de Baar et al., 1988, Elderfield, 1988] and used as a proxy for redox conditions of the environment of deposition [Elderfield and Greaves, 1982, De Baar et al., 1985, Varentsov et al., 1991]. Europium, like Ce, is redox sensitive and can take the divalent state Eu2+ under strongly reducing condition such as these encountered in high temperature hydrothermal systems [Sverjensky, 1984, Bau, 1991]. In these cases, the ionic radius of these two elements differs sufficiently from their trivalent counter-parts to produce a marked effect on their geochemistry [Henderson, 1996] (Figure 1.1). Redox chemistry may lead to their fractionation from REE3+ forming the so-called Ce and Eu anomalies. These stable states can be explained by the electronic structure of Ce4+ similar to the one of noble gas xenon, whereas Eu2+ possess a enhanced stability with a half-filled 4f shell.

2 Chapter 1. Introduction

Apart from economic considerations, REE concentrations in a material are rarely used as such. REE natural abundances show a rhythmic odd/even abundance variation with atomic number as a function of neutron and proton number. This is caused by variations in the binding energy of the nucleus and referred to as the Oddo-Harkins rule [Elderfield, 1988]. The even elements have a higher number of isotopes compared to odd elements; Ce has four isotopes, Nd, Sm, Gd, Dy and Yb have seven whereas La, Eu, Lu have only two isotopes [Dubinin, 2004]. Therefore, when studying REY, a correction is necessary to exclude the influence of these natural variations in their abundances by normalizing the data. This allows for the analysis of trends of relative abundances showing REE fractionation between different samples or systems. The choice of the normalizing composition, such as chondrites, mid-ocean ridge basalt (MORB), post-Archean Australian shale (PAAS) or North-American shale composite (NASC), depends on the environment studied and presumed source material of the sample. Chondrite normalization usually applies to hypogene processes as the composition of these elements in chondrite reflects their composition in protoplanetary materials. On the other hand, shale normalization is used for supergene processes as this material reflects average REE composition of magmatic, metamorphic and sedimentary rocks [Dubinin, 2004].

Quantification of ratios such as Lan/Smn, Gdn/Ybn (subscript n represents normalisation to standard) as well as Ce and Eu anomalies, constitute straightforward ways to describe and compare sample data. In the following chapters, Ce and Eu anomalies are expressed as the ratio of the normalized values of an element by the interpolation of the adjacent elements (superscript *) such that;

3 Chapter 1. Introduction

퐶푒 퐶푒푛 Ce anomaly = ∗ = 퐶푒 √(퐿푎푛.푃푟푛) and

퐸푢 퐸푢푛 Eu anomaly = ∗ = 퐸푢 √(푆푚푛.퐺푑푛)

1.1.2 REE in the economy

1.1.2.1 Applications, market and deposits classification

Applications for rare earth metals and alloys have increased exponentially in recent years given their specific electronic, magnetic and spectroscopic properties [Abreu and Morais, 2010]. These elements play a determinant role in the development of green technologies, boost various industrial processes and allow development of specific military and aerospace applications. Indeed these metals play a predominant role in modern electronics, renewable energy generation and storage with the use of Neodymium doped magnets for wind turbines and hard drives, development of nickel metal hydride (NiMH) batteries for electric cars, energy efficient lights for screens and are used as auto-catalysts in various industries including oil refinery (Table 1.2). Their importance on the economical market is only counterbalanced by the fragility of the supply chain which underpins their criticality for our society due to the lack of reliable substitutes. The nearly total control of China over production, transformation and distribution of rare earth products coupled to a reduction of exportation quotas in 2010-2011 have resulted in skyrocketing prices for REE. USA, Japan and the EU among other industrialized countries have since looked to diversify their supply chains, exploit alternative deposits and invest in REE-bearing material recycling protocols, by investing significant capital in mining, exploration and research.

REY abundances in the continental crust contradicts their given name of “rare earth” as they are encountered in similar quantity as other industrial metals such as Ni, Cu, Zn, Sn and Pb, and enriched by more than one order of magnitude over so-called precious metals like Au, Ag, and the platinum group elements (PGE) (Figure 1.2). The main distinction between REE and these more commonly mined metals comes from geochemical characteristics that control their distributions in rocks, resulting in REE rarely being concentrated to form economically exploitable ore deposits.

4 Chapter 1. Introduction

Element Major end usage Lanthanum La Optics, batteries, catalysis, hydrogen storage, hybrid engines, metal alloys Cerium Ce Chemical applications, petroleum refining, coloring, polishing glass, catalysis, hybrid vehicles Praseodymium Pr Magnets, lighting, optics Neodymium Nd Magnets, petroleum refining, hard drives, lighting, lasers, optics, hybrid vehicle, batteries

Light-REE Samarium Sm Magnets, lasers, masers, lightweight magnets Europium Eu Lasers, lighting, medical applications Gadolinium Gd Magnets, glassware, lasers, X-ray contrast agent, computer applications, medical applications Terbium Tb Lasers, lighting, lightweight permanent magnets, phosphors Dysprosium Dy Permanent magnets, lasers, hybrid vehicle batteries Holmium Ho Lasers, glass coloring Erbium Er Lasers, medical applications, neutron-absorbing control rods in nuclear industry Thulium Tm X-ray generation Heavy-REE Ytterbium Yb Lasers, chemical industry applications Lutetium Lu Medical applications, chemical industry applications Yttrium Y Lasers, superconductors, microwave filters, lighting, ceramic Scandium Sc Alloys in aerospace engineering, lighting, fuel cells

Table 1.2: Major end uses of REY (modified after Weng et al. (2015)).

Although over 250 REE-bearing minerals have been identified [Kanazawa and Kamitani, 2006], only a handful of minerals contain high enough concentrations and are present in enough quantities to be economically exploited [Golev et al., 2014]. These include:

- Bastnäsite (Ce, La)(CO3)F

- Monazite (Ce, La)PO4

- Xenotime YPO4

- Loparite (Ce, Na, Ca)(Ti, Nb)O3

- Apatite (Ca, REE, Sr, Na, K)3Ca2(PO4)3(F, OH)

Of these 5 minerals, bastnäsite, monazite and xenotime are by far the most important source of REY with xenotime being the main source for the most valuable heavy REY (HREY). REY deposits are encountered in both primary and secondary-type deposits (Table 1.3). High concentrations of REY are found in rocks related to the late stage crystallisation of peralkaline magmas, pegmatites and carbonatites. More rarely, primary REY deposits can be formed in metamorphic or diagenetic settings. Secondary deposits constitute alteration products either in situ as laterites or in sedimentary-related deposits such as placers.

5 Chapter 1. Introduction

Deposit type Mines Igneous Hydrothermal Bayan obo (China) Mt. Pass (USA), Weshan, Maoniuping (China), Mount Weld (Australia), Carbonatites Araxa, Catalao (Brazil) Alkaline rocks Khibiny, Lovozeiro (Russia), Posos de Caldas (Brazil) Alkaline grantites Strange Lake (Canada) Secondary Kerala (India), Western Australia, Queesland State (Australia), Placer Richards Bay (south Africa) Conglomerate Elliot Lake (Canada) Ion-adsorption clay Longnan & Xunwu (China), Madagascar Table 1.3: Classification of rare earth ore deposits (modified after Kanazawa and Kamitani (2006)).

6 Chapter 1. Introduction

Worldwide, REE are mined from carbonatite and alkaline intrusions, either from the intrusion itself or its hydrothermally derived alteration such as in Bayan Obo, China’s biggest REY mine [British Geological Survey Report, 2011]. Alkaline and carbonatite magmas form in extensional intracontinental rifts associated with small degrees of partial melting [Goodenough et al., 2016].

Carbonatites are volatile-rich magma with more than 40 % CO2 and less than 10 % SiO2 commonly associated with peralkaline magmatism but their process of formation remains debated [Drew et al., 1990, Lottermoser, 1990, Yuan et al., 1992, Lentz, 1999, Yang et al., 2011]. The main hypothesis imply (i) segregation of the carbonate phase from the siliceous melt by immiscibility when confining pressure and temperature are insufficient to maintain the ascending magma in a homogeneous state; (ii) Carbonatitic melts formation by partial melting of CO2-rich sediment during subduction of an oceanic plate; (iii) Contamination or assimilation of large quantities of

CO2-rich fluid by an ascending magma through metasomatism of the host rocks such as marble, dolomite and limestone, producing at the same time the associated well-known skarn type deposit. It is important to consider that those key factors are not independent but rather that carbonatite formation results from a combination of them (Figure 1.3). As REE are incompatible, high concentrations (in addition to other high field strength and large ion lithophile elements (HFSE and LILE)), are found in carbonatite as they correspond to the later stages of magmatic fractionation and crystallisation. Furthermore, the high volatile content allows transportation of REE associated with chlorides, fluorides and carbonate complexes that prevent uptake of REE in early stages of crystallisation and enhance enrichment in the melt.

Figure 1.3: Schematic diagram illustrating the main environment of formation of alkaline igneous rocks and carbonatites [Goodenough et al., 2016].

7 Chapter 1. Introduction

Secondary deposits are largely dominated in numbers by placers with a main mineralogy being monazite or xenotime. However, weathered crust elution-deposited REY commonly known as ion- adsorption clay (IAC) deposits, constitute the main secondary deposit mined currently. These deposits are important in perspective of their enrichment in the most critical HREY (3% of China’s total RE reserves representing 35% of China’s total RE production in 2009 [Yang et al., 2013]). IAC are genetically related to laterites whereby the deposits form as a result of chemical weathering, biological decomposition and dissolution of REE-rich granite and porphyry with subsequent adsorption and relative enrichment on clay minerals during migration and percolation of REE solutions [Yang et al., 2013]. Conditions for formation and preservations of such deposits requires tropical meteorological conditions, limited erosion and presence of a crust of alumina-silicate to adsorb rare earth ions [Vahidi et al., 2016].

1.1.2.2 Resources and production

World production of REY has increased significantly in the 1980 as Chinese production started at Bayan Obo and increased annually by 40% until 1989 [Yang et al., 2013]. In the last 3 decades, the annual production increased from 40 kt RE oxides (REO) in the 1980s to 110 kt REO in 2010 [Weng et al., 2015] due to the increasing demand for REY in high tech applications. Over the last 15-20 years the Chinese production has increased significantly up to the point where more than 95% of REY on the market were extracted from Chinese mines. This trend is now reversing as more countries started extracting and producing REY with a stable world production of 123 kt REO in the last two years (Table 1.4).

The Bayan Obo deposit in China is the biggest operating RE deposit in the world and accounts for 83.7 % of China’s total RE reserves with 48 million tons of REO (avg. grade of 6 wt. %). In addition, more than 550 carbonatites/alkaline complexes worldwide have been identified and form the majority of the world reserves [Kanazawa and Kamitani, 2006]. Outside of China, the rest of the REY mining activity is concentrated in a few countries: India, USA, Australia, Canada, South Africa, Brazil, Russia and Commonwealth of Independent State (CIS), Malaysia (Table 1.3 and Table 1.4).

1.1.2.3 Extraction and Processing

In IAC deposits such as those of southern China, the REY are not associated with a specific mineral phase but rather adsorb as free-ions on the surface or interlayers of clay minerals. The REY content in IAC can be characterized as low grade (0.03 - 0.15 % REY [Yang et al., 2013]) but economically viable extraction is achieved as no physical beneficiation is required and direct hydrometallurgical treatment can be applied. The ore is treated by cation exchange in in-situ heap leaching with electrolyte solutions of sodium chloride or ammonium sulphate [Yang et al., 2013].

8 Chapter 1. Introduction

Country 2007 2008 2009 2010 2011 2012 2013 2014 2015 China 120800 124500 129400 118900 105000 100000 95000 105000 105000 Russia 2 711 2470 2500 2500 2500 2400 2500 2500 2500 Malaysia 440 150 20 471 498 100 180 240 200 Brazil 760 540 200 160 188 140 330 n.a n.a India 2000 2000 2000 2000 2000 2900 2900 n.a n.a Australia 0 0 0 0 2 188 3200 2000 8000 10000 USA 0 0 0 0 0 800 5500 5400 4100 Thailand n.a n.a n.a n.a n.a n.a 800 2100 2000 Vietnam n.a n.a n.a n.a n.a n.a 220 n.a n.a Total 126711 129660 134120 124031 110186 109540 109430 123240 123800 China % 95% 96% 96% 96% 95% 91% 87% 85% 85% n.a data not available Table 1.4: World annual production of rare earth oxides (including yttrium) in tonnes. Data for 2007 – 2011 from Golev et al. (2014), data for 2012 -2015 from USGS reports on mineral commodities [U.S.G.S, 2013, U.S.G.S, 2014, U.S.G.S, 2015, U.S.G.S, 2016].

The solutions are then collected in streams downhill. Although effective, this process is extremely polluting and environmentally devastating as only the first 5m of soil are treated requiring deforestation on extensive surfaces prior to operation. These large scale operations of vegetation removal and hill leaching destabilize soil and contaminate streams with toxic waste. It is estimated that the production of 1 ton of REO from IAC ore requires removal of 300 m² of vegetation and topsoil, produces 2000 t of tailings and 1000 t of contaminated wastewater [Yang et al., 2013]. In the Ganzhou region more than 150 km² of forest were destroyed for REE extraction in 10 years and now abandoned. By contrast, copper production from one of the world’s biggest open pits (Escondida, Atacama desert, Chile) has a ground area of roughly 140 km² including open pit, processing plant and tailings for an estimated production of 1.2 Mt per year over the next decade [Rode, 2015].

Mines targeting bastnäsite, monazite and xenotime from hard-rock deposits are far more costly and complicated to process although the average grades are much higher; 6 % REO in Bayan Obo, 5 - 10 % REO in Mountain Pass and 12 % REO at Mount Weld [Kanazawa and Kamitani, 2006]. The main industrial process for the separation and purification of rare earths from such deposits is solvent extraction, involving multiple stage of stripping from loaded solvent extractants using aqueous solutions of inorganic acids [Konishi and Noda, 2001]. Initial stage of comminution (except for placer deposits) and pre-concentration of REE-bearing minerals is required through physical methods (gravity concentration, flotation, magnetic and electrostatic separation) to progressively concentrate monazite, xenotime and bastnäsite and separate these minerals from gangue. Afterwards, the ore is commonly treated with an acid or alkali leach (sulphuric acid and sodium hydroxide being the most common) at temperatures of 200-400°C in a process commonly

9 Chapter 1. Introduction referred to as cracking [Habashi, 2013, Golev et al., 2014]. The cracking usually produces REE-rich

-1 leach liquor in the range of 1 to 40 g.L RE2O3 with small amounts of impurities [Ru'an et al., 1995, Abreu and Morais, 2010]. Following this stage, REE are separated from co-leached elements by hydrometallurgical techniques such as solvent extraction and ion exchange using selective precipitation as carbonates or oxalates by pH adjustment. Rare earth oxalates and carbonates are then calcinated to form a mixed RE oxide product ready for sale as a RE alloy known as mischmetal or sent to specialized factories for further separation treatment into individual high purity RE oxides or reduced to pure metal products depending on the required end use [Christie et al., 1998, Golev et al., 2014]. The separation of a mixed product into individual elements is inherently difficult due to the very similar chemical behaviour of REY. Multiple methods exist such as selective oxidation/reduction, fractional crystallisation-precipitation, and ion-exchange, though the most effective approach is solvent extraction using organophosphorus compounds (e.g, EDPA, HDEHP, EHEHPA, or TBP). All methods require multiple iterations to obtain a high purity individual REO. The production of REY from carbonatites and other alkaline rocks suffers the same issues as IAC processing, with pollution on site from the hazardous chemical used and major land allocation required for both mining and processing operations. In addition, REY in magmatic systems are commonly associated with high concentrations of U and Th which are both concentrated during beneficiation of REY. These U and Th wastes are of serious concern because of the high levels of radioactivity of the tailings after cracking although the co-production of radioactive elements is feasible for the highest U-Th concentrations albeit with significant additional costs [Golev et al., 2014]. In most cases, industrial processing is followed by costly remediation and disposal in long term storage. Significant radioactive content constitutes an important economical drawback for any new REY exploitation project.

1.2 REE in the ocean: sources and behaviour

Oceans constitute an important reservoir for REE which records a constant rebalancing between major inputs from landmass erosion and weathering, hydrothermal springs, aeolian fall out, and outputs by sedimentation and (bio-) mineralization [Elderfield, 1988]. The geochemical cycle of REE within the ocean can be simplified as a two steps process of transportation and phase association/complexation (Ohta 1999). First, riverine waters transport dissolved REE originating from the weathering of crustal rocks into oceans via estuarine areas. Estuaries play a major role at the boundary between those two environments as radical changes in the chemistry of seawater occur when riverine water mixes with oxidized oceanic waters [Elderfield, 1988]. Secondly, REE enter the ocean water circulation and bound with various complexes for vertical transport that are incorporated in fine into marine authigenic phases, including deep-sea main scavenger:

10 Chapter 1. Introduction ferromanganese mineralisation such as nodules and encrustations. Although hydrothermal fluids are enriched in REE over seawater, their influence in the oceanic budget is limited due to rapid and effective scavenging after fluid emission in seawater. REE derived from hydrothermal fluids thus play a minor role into REE distribution and fractionation in the water column but constitute a major source for proximal hydrothermal sediments that form a net sink for oceans [Rhulin and Owen, 1986b, Rhulin and Owen, 1986a, Olivarez and Owen, 1989].

1.2.1 Rivers input and estuarine mixing

Evaluation of the influence of riverine input on the oceanic REE budget has been investigated since the 70’s [Hogdahl, 1970, Varshal et al., 1975, Martin et al., 1976]. First estimates on the Gironde estuary (France) covering a salinity range from 0.1 – 28 ‰ indicated that nearly 50 % of dissolved REE input from the river are removed in the estuarine mixing zone [Martin et al., 1976]. Further lab experimentations by Hoyle et al. (1984) demonstrated that these initial results constitute an underestimation, as 65 to 95 % REE removals were observed between initial fluid solution and the precipitates produced. Further experiments highlighted the strong variability of these removal rates as a function of the Fe and organic content of riverine waters [Goldstein and Jacobsen, 1988]. Indeed, formation of Fe colloids which constitutes an effective scavenger for REE in aqueous solutions, is catalysed by organic material in presence of sea salt cations [Hoyle et al., 1984].

Additionally, beyond salinity’s influence, fractionation along the REE series has been described as a result of the kinetics of colloidal iron aggregation, impacting the rate of formation and size of particles [Mayer, 1982]. Fractionation between LREE and HREE has been observed in most of the rivers worldwide [Sholkovitz et al., 1986, Upstill-Goddard et al., 1986, Goldstein and Jacobsen, 1988, Sholkovitz and Elderfield, 1988] with removal ranging from about 70 % for the light REE to no more than 40 % for the heavy REE at low salinity. Indeed, LREE are rapidly integrated within the structure of colloids whereas HREE remain preferentially into the solution or are associated with the smallest particles.

Estimations of REE riverine inputs to the oceans have been centred on obtaining average global REE concentrations in rivers. Although there is a general agreement that rivers constitute the main source for REE, estimations on the overall riverine REE budget and flux to oceans vary considerably from one study to the other [Piper, 1974, Goldstein and Jacobsen, 1988, Olivarez and Owen, 1991]. Inherent limitations to such budget assessment derive from (i) the span of REE concentrations observed in rivers across the world (average Nd in Gironde River = 260 pmol/l [Hogdahl, 1970] whereas in the Russian River Oka Nd = 8500 pmol/l [Varshal et al., 1975]) as a

11 Chapter 1. Introduction result of the composition of terrane drained, (ii) the lack of a complete database for rivers across the world incorporating seasonal variability on composition and fluxes, and (iii) the limited understanding and global assessment of REE trapped in estuarine sedimentary cells. In addition, the lack of knowledge for modelling the inner circulation path of particles within the ocean prevents estimation of a representative average content of REE inputs in oceans by rivers.

Finally, quantification of atmospheric fluxes, even if marginal contributors to the total oceanic REE budget, brings further uncertainty to estimates the amounts of those elements in the ocean and their behaviour by the lack of studies on the kinetic solubilisation of such aerosols [Alibo and Nozaki, 1999].

1.2.2 REE in seawater

1.2.2.1 REE fractionation and particle associations

Partitioning of trace elements between particles and seawater is controlled by the interplay of surface and solution chemistry. Concentrations variations encountered in the different layers of the water column reflect different removal rates influenced by adsorption to different types of inorganic and biogenic particles [Sholkovitz et al., 1994]. Sinking particles constitute an important trap for trace elements as scavenging occurs by adsorption on the surface of colloidal particles. This sorption process is strongly linked to the formation of complexes with functional groups tightly bound to particle surfaces such as hydroxyls groups for Si, Fe and Mn oxides or carboxyl and phosphoryl groups for organic complexes [Schijf et al., 2015]. Fractionation among REE occurs as a result of the differences in relative affinity of each trivalent REE for surface adsorption to particles and for complexation with ligands in seawater [Sholkovitz et al., 1994]. With respect to

+ 2- seawater, carbonates form the main complexing agent for REE, with REE-CO3 and REE-CO3 complexes accounting for at least 85 % of REE-ligand in seawater [Elderfield, 1988]. Other important ligands complexing with REE involve sulphates, fluorides and hydroxides. The fractionation of REE between water and particles in the ocean shows a similar behaviour to those observed within riverine waters. A preferential uptake of LREE by biogenic particles relative to HREE is observed in surface seawater whereas the heavier REE reveal a greater tendency to complex with seawater ligands and be retained in solution [De Baar et al., 1985, Elderfield, 1988, Sholkovitz et al., 1994, Alibo and Nozaki, 1999].

1.2.2.2 REE distribution in the water column

The source of REE, nature and effectiveness of the scavenging agents in the oceans vary both vertically in the water column and horizontally with respect to the proximity of terrestrial masses

12 Chapter 1. Introduction and seawater current. In the upper water column trace elements are dominantly influenced by biologically-derived particles whereas the influence of sedimentary processes is dominant in deeper waters [Elderfield, 1988]. Both influences are important when considering oceanic margins.

Studies of the Atlantic Ocean [Elderfield and Greaves, 1982, De Baar, 1983], Pacific Ocean [Klinkhammer et al., 1983, De Baar et al., 1985, Zhang and Nozaki, 1996] and Mediterranean Sea [Alibo and Nozaki, 1999] display similar vertical distribution profile for REE (Figure 1.4) with various degree of enrichment relative to differences in terrestrial and aeolian input, stronger in the Atlantic Ocean as a direct influence of Saharan dust. All REE3+ exhibit the same concentration vs depth profile characterized as nutrient-like profile; depleted in superficial waters and rapidly increasing with depth, their concentration stabilize between 2 and 3 km beneath sea level [Elderfield, 1988, Schijf et al., 2015]. Although REY are poorly solicited by biogenic processes as micronutrient unlike Fe, Mo, S, P, their depletion in surficial waters has more to do with their incorporation in carbonate and silica tests and strong reactivity whereby highly reactive metals (Mn, Sn) strongly scavenged dissolved anionic complexes [Schijf et al., 2015]. This is followed by particle settlement and regeneration at depth by oxidation or dissolution of organic tests, similarly to the cycle of silica (Figure 1.4). In this view, REE trends as a function of depth highlight these interactions showing a progressive enrichment in total REE content with depth and a more developed Ce negative anomaly (Figure 1.5). Also, the trends highlight the different particle associations as the upper organic-rich layer of the water column (0 – 150 m) show flat HREE patterns whereas deeper trends (> 500 m) exhibit HREE enrichment relative to shale. REEN patterns for oceanic waters show a constant increase from light to heavy REE and a marked negative Ce anomaly (Figure 1.5). In contrast to other REE, Ce behaves conservatively in the first 150 -200 m with a smooth decreasing tendency. This opposite behaviour finds its source in the

4+ tetravalent oxidation state of Ce removing insoluble Ce from seawater under the form CeO2 or

Ce(OH)4 by autocatalytic process. Ce increasing concentration in the upper layer suggests this oxidation reaction and then subsequent scavenging without releases at depth [Elderfield, 1988]. Subsequent process of adsorption and release at depth of REE3+ by desorption will then increase progressively the uptake on particles of insoluble Ce from seawater developing an increasing Ce anomaly with depth [De Baar et al., 1985, Alibo and Nozaki, 1999]

As previously noted REE fractionation occurs during adsorption according to the nature and size of particles. This process consequently affects their distribution at depth. Although REE (except for Ce) possess a similar distribution in the water column with a nutrient-like profile, the surficial seawater signature is slightly depleted in LREE compared to HREE as they are more susceptible to form complexes with organic matter or other ligands. The reverse tendency is observed at depth,

13 Chapter 1. Introduction as proportionally more LREE are released during restructuration or dissolution of sinking particles. Hence, the preferential removal of LREE in the upper water column is followed proportionally by their preferential release at depth [Sholkovitz et al., 1994] exemplified by variations in LREE/HREE ratios such as La/Yb or Nd/Er for oceanic water columns (Figure 1.4).

Figure 1.4: Vertical profiles for REE (pmol/kg), Nd/Er, La/Yb ratio and for main

nutrients (PO4 and SiO2 in µmol/kg) in filtered seawater in total acid digestion; triangle (East Caroline Basin), grey square (Coral Sea Basin) and black diamonds (Tasman Sea Basin) from Zhang and Nozaki (1996), black asterisk (western North Pacific ocean near Japan) from Alibo and Nozaki (1999).

14 Chapter 1. Introduction

1.3 REE in hydrothermal systems

1.3.1 REE in hydrothermal fluids in the oceanic crust

The behaviour of REE in hydrothermal fluids has been investigated for a variety of submarine vent fluids emanating in mid-ocean ridge (MOR) settings and back-arc basins. Numerous studies have enhanced the knowledge on REE speciation and mobility in aqueous phases in various geological settings and under a large range of pressure, temperature and chemical compositions. Hydrothermal fluids, despite some large variations in their concentrations over the globe, are enriched in REE by several orders of magnitude (10 to 1000 times) in comparison with seawater [Mitra et al., 1994, Schmidt et al., 2010].

1.3.2 REE patterns in hydrothermal fluids

The majority of published REE analysis are derived from hydrothermal sources at oceanic ridges [Michard, 1989, Klinkhammer et al., 1994, Mitra et al., 1994, Douville et al., 1999, Douville et al., 2002, Schmidt et al., 2010], and thus involve a common geological context and substratum for the greater part of them. Therefore, studies carried out on the Mid-Atlantic Ridge (MAR), East Pacific Rise (EPR) and the Manus back-arc basin, despite changes in host rock composition (mafic to ultramafic), salinity (190 to 713 mM Cl) and temperature of emanation in MOR-settings, have strikingly similar chondrite-normalized patterns (Figure 1.6 A). The patterns exhibit light rare

15 Chapter 1. Introduction earths enrichment with a smooth and regular decrease from Ce to Lu, a positive La anomaly is often present and a strong positive Eu anomaly (Eu/Eu* = 5 – 46) [Schmidt et al., 2010]. All these fluids possess common characteristics such as high temperature, acidic pH and high Cl- content.

16 Chapter 1. Introduction

to Ca evolution directly depending on anhydrite precipitation [Schmidt et al., 2010]. (D) end-member filtered hydrothermal fluids from Two Boats vent; Variable amounts of particles (mainly anhydrite) in collected samples are thought to explain variation in REE enrichment and the negative Eu anomaly in the most REE enriched sample [Schmidt et al., 2010]. (E) end-member hydrothermal fluids from MAR and the Manus Basin [Douville et al., 1999, Douville et al., 2002]: strong LREE enrichment and Eu anomaly influenced by ultramafic substratum in Snake pit and Rainbow vents fluid, signatures from PacManus show LREE depletion with relatively flat MREE-HREE trends and pattern from Desmos vent show LREE depletion and HREE enrichment without Eu anomaly highlighting influence on REE fractionation of magmatic volatiles input in hydrothermal fluids. (F) hydrothermal anhydrite sampled from the active chimney in Two Boats [Schmidt et al., 2010]. Trends display flat LREE to MREE patterns, variable Eu anomaly and HREE depletion. The size of the Eu anomaly appear to be inversely related to the total REE enrichment

Other studies on the Manus back-arc basin [Craddock et al., 2010] and East Scotia subduction zone [Cole et al., 2014] provide contrasting data including Chloride-poor hydrothermal fluids within host-rocks no longer basaltic but rather dacitic to rhyolitic. Such fluids display REE signatures with LREE/HREE ratio < 1 with LREE depletion and flat HREE trends and both negative and positive Eu anomaly. Schmidt et al. (2010) reported fluid samples from MAR (5°S), with pressures and temperatures above the critical point of seawater (CPsw = 298 bars/407°C) exhibiting a concave-downward distribution with a MREE enrichment and weakly positive or even negative Eu anomaly (Figure 1.6 C). Similar results were observed at Menez Gwen in the Manus basin [Douville et al., 1999, Craddock et al., 2010] and Lucky Strike [Von Damm et al., 1998] where the shallow depth induces phase separation. These results imply multiple controls on REE mobility and speciation during hydrothermal transport and point toward several questions; What are the dominant processes controlling the REE content of hydrothermal fluid during fluid-rock interactions to induce such enrichment in comparison with sea-water? And what is the influence of kinetic parameters like temperature, pH, ligand concentration and precipitation of hydrothermal phases on REE fractionation in hydrothermal fluids?

1.3.3 Fluid-rock interactions and REE enrichment in hydrothermal fluids

Host-rock influence on hydrothermal fluid composition has been tested in several studies comparing analysis carried on fluid interacting with either ultramafic rocks or peridotites (Logatchev and Rainbow vents on MAR, Douville et al. (2002)) and also andesitic to rhyolitic host- rocks in the Manus basin [Craddock et al., 2010]. It appears that host-rock composition has a limited impact on the overall fluid chemistry for metal complexation. However, there is a noticeable enrichment in transition metals (Ni, Co) and trace elements (REE, Sr, Y and In)

17 Chapter 1. Introduction observed in sulphides and fluids from Rainbow vents [Douville et al., 2002]. The main variation at this site corresponds to an acidification of the fluid caused by the serpentinisation of the ultramafic host rocks. This specific fluid-rock interaction involving olivine and orthopyroxene

+ alteration enhances H2S, Si, Ca and Al release in the fluid with liberation of H ions by breakdown of the H2S. The acidification of the fluid causes increased breakdown of magnetite, allowing transition metals contained within the magnetite to be dissolved. Hence, involvement of ultramafic rocks does not influence REE fractionation, but instead allows a higher extraction efficiency [Douville et al., 2002]. Results from Craddock et al. (2010) on hydrothermal fluids interacting with host rocks ranging from basalts (Manus Spreading Centre) to dacites (Eastern Manus Basin) do not show correlation between primary whole-rock REE abundances and REE distribution patterns measured in vent fluids.

The REE enrichment of a hydrothermal fluid is therefore independent from host-rock bulk REE composition but present remarkable similarity with patterns measured in plagioclase phenocrysts [Douville et al., 1999]. REE enrichment could then be explained by inheritance of intensive fluid/rock interactions involving crystal-fluid exchanges [Klinkhammer et al., 1994]. In this view, the resultant fluid composition would reflects the strong relation between concentration, ionic radii and electronic charge, known as CHARAC exchange equilibrium [Bau, 1996], with the hydrothermally altered host rock. More evidence from the Onuma diagram [Onuma et al., 1968] plotting MORB-normalized fluids data against ionic radii highlights the correlation of REE fractionation trends mimicking those of plagioclase phenocrysts [Douville et al., 1999]. This suggests control by disequilibrium dissolution and ion exchange reactions between fluids and crystals whereby REE3+ and Eu2+ substitute for Ca2+ and Sr2+ respectively in the feldspars. However, problems arise when considering analysis made on fluids from the Rainbow vent site, hosted on ultramafics which provide little support for plagioclase alteration as a major source for REE in the fluid [Douville et al., 2002]. The authors suggest as an alternative that the fluid at this site, richer

2- 3+ 2+ in SO4 and H2S via magmatic input, favoured REE solubility and reduction of Eu to Eu in the fluid, therefore putting the REE content of the fluid as the result of a less selective mineral leaching process.

In addition, some argue that substantial variations in REE concentrations and REEN patterns found within individual vent areas on a small spatial scale suggest other parameters of control, since host-rock REE composition is uniform (Figure 1.6 C and D) [Sinton et al., 2003, Miller et al., 2006, Craddock et al., 2010, Schmidt et al., 2010]

18 Chapter 1. Introduction

1.3.4 Factors of control for REE speciation in hydrothermal fluids

Although some have considered the complexation of REE as a minimal control factor on REE fractionation in hydrothermal fluids [Klinkhammer et al., 1994] considering the low concentration of the main oceanic ligands (carbonate, fluoride, sulphates) in the majority of the fluids analysed, more recent results show the importance of fluid chemistry on trace element behaviour. Several influence parameters should therefore be considered such as the concentration and nature of ligands, pH, temperature and the precipitation of mineral phases before emanation at the vent.

Most chemical analyses report high chloride concentrations for acidic hydrothermal fluids. In this

- 2- 2- view, ligands such as hydroxides (OH ), carbonates (CO3 ), bicarbonates (HCO3 ) and phosphates 3- (PO4 ) will have a negligible influence on REE aqueous behaviour [Douville et al., 1999]. Therefore, in the absence of stronger complexing agents, REE fractionation in hydrothermal fluids could be expressed as a function of chloride and fluoride stability [Schmidt et al., 2010]. Calculations of REE speciation [Haas et al., 1995] show that LREE are mainly and equally combined with both fluoride and chloride ions whereas HREE have more stability while complexed with fluoride ions.

The stability of chloride-REE complexes increases with temperature but decreases from light to heavy REE [Douville et al., 1999, Mayanovic et al., 2008, Migdisov and William-Jones, 2008, Schmidt et al., 2010]. Therefore, as temperature increases, the LREEs become increasingly concentrated in the fluid over the HREE, which explains the observed fractionation and varying slopes of REEN trends for most high temperature vents (Figure 1.6A). Similarly, the degree of the positive Eu anomaly could be explained by Eu redox sensitivity. Under its divalent form, Eu-Cl complexes are more stable than other REE. This increased stability favours preferential retention in the fluid and enhance the decoupling and subsequent fractionation of Eu from the other trivalent REE [Schmidt et al., 2010].

Examples of unusual REE patterns (Figure 1.6 B, C and E) can be explained by abnormal chemistry of the hydrothermal fluids and/or precipitation of mineral phases that act as sinks or supplementary sources for REE in case of dissolution. Results from the PACMANUS, Desmos sites, and East Scotia subduction zone [Douville et al., 1999, Craddock et al., 2010, Cole et al., 2014]

(Figure 1.6 B and E) all consider unusual hydrothermal fluids enriched in Mg, SO4, and F due to magmatic contributions by fluid exsolution incorporated into the hydrothermal fluid. The input of magmatic volatile-rich fluid during hydrothermal circulation provides an additional source of anionic ligands capable of complexing with the REE [Cole et al., 2014]. The increasing sulphate content may lead to LREE depleted fluid signature (Figure 1.6 B) as these are preferentially removed during barite and/or anhydrite precipitation [Douville et al., 1999]. Alternatively,

19 Chapter 1. Introduction anhydrite REE patterns from chimney pieces from Two Boats vent field (TAG mound) are L-MREE- rich with a pronounced depletion in HREE and variable Eu anomaly [Schmidt et al., 2010]. The influence of this mineral on hydrothermal fluids has been evoked to explain depletion in REE for the TAG hydrothermal mound and the Kemp submarine volcano where sulphate mineral acts as a major sink [Mills and Elderfield, 1995, Bach et al., 2003, Cole et al., 2014]. Formation of anhydrite also enhances formation of a positive Eu anomaly in the fluid as the ionic radius of Eu2+, larger than Eu3+, does not fit as well as other REE into the anhydrite crystal lattice [Schmidt et al., 2010]. Also, the competition for preferred REE complexation either with chloride or sulphate complexes influences the anhydrite REE signature depending on fluid temperature. As REE chloride complex stability decreases with temperature, the control exercised by Cl complexes on anhydrite forming in low-temperature vents (ca. 200°C) will be less important. LREE depletion and negative Eu anomaly in anhydrites are also produced by crystal re-equilibration induced by changes in fluid composition over time. Therefore, during recrystallization and partial dissolution, preferential mobilization of LREE and Eu by the fluid could occur and explain the highly variable range of REE signatures encountered in some TAG vent fields [Schmidt et al., 2010].

1.3.5 Conclusion on hydrothermal REE signature

Aqueous REE concentration and fractionation do not relate directly or show a clear relationship with the REE pattern of the host-rock with which the hydrothermal fluids interact, despite crustal rock being the primary source of REE in hydrothermal fluids. Instead, the variety of REE signatures encountered across vent fluids from the MAR, EPR and Manus back-arc basin points toward a control on REE fractionation dominated by fluid pH, nature of ligand (chloride, sulphate and fluoride), and their concentration. In turn, variations in these dominant parameters are related to fluid temperature, phase separation and mineral precipitation.

However, the influence of some of these parameters are still not well constrained and could bring major variations. For example, brine phases, formed by condensation from gas-rich vapour phases produced during supracritical phase separation have specific physico-chemical properties. Indeed, small changes in P-T conditions have drastic impact on ion association and complexation. In this view, the process of phase separation above the critical point of sea water (CPsw) on REE mobilization, solubility, transportation and fractionation is poorly constrained [Douville et al., 1999, Schmidt et al., 2010].

20 Chapter 1. Introduction

1.4 REE behaviour during mixing of hydrothermal solution with seawater

1.4.1 The hydrothermal REE budget to open ocean

Hydrothermal fluids have long been considered as a potential alternative source for oceanic REE, alongside river inputs [Goldstein and Jacobsen, 1988] as vents were recognized as an important source for chemical mass balances of several major seawater elements [Edmond et al., 1979]. Most hydrothermal fluids encountered in MOR settings exhibit enrichment of 10 to 104 times REE values over oceanic waters [Mitra et al., 1994] and could be readily considered as an important REE supplier to oceans. However, REE concentrations and normalized trends for seawater do not match the REE distribution in hydrothermal fluids considering both Eu and Ce anomalies as well as L/HREE ratios showing that the hydrothermal signal and REE fluxes are not transferred to open seawater.

The typical hydrothermal signature is actually manifested within proximal sediments as a result of the extremely rapid and extensive scavenging of trace metals by precipitating, highly reactive, hydrothermal oxides [Olivarez and Owen, 1991]. On the other hand, distal particles exhibit seawater-like REE distributions. Plots showing REE concentration of hydrothermal deposits against distance travelled from the paleorise crest in DSDP site 598 near the EPR [Olivarez and Owen, 1991] and TAG mound [German et al., 1990, Mitra et al., 1994] demonstrate a positive non-linear relationship between REE enrichment and the distance of deposition, with increasing REE/Fe ratio. As the time required for currents to carry hydrothermal particles to their farthest site of deposition is much less than burial time of these sediments, it is possible that this enrichment reflects continuous REE extraction from seawater during transport and their exposure once settled [German et al., 1990, Mitra et al., 1994]. In addition, analysis of REE trends from sediments more or less distant from the vent display the progressive overprint of the hydrothermal signal by seawater showing decreasing Eu positive anomaly and development of negative Ce anomaly [German et al., 1990, Olivarez and Owen, 1991]. Coherent results for hydrothermal deposits close to the point of fluid ejection thus display relatively depleted REE content compared to more distal deposits with regard to their faster burial rate limiting time of interaction between particles and seawater. Oceanic water analysed in the vicinity of vents have shown REE depletion compared to average seawater, implying the scavenging process by hydrothermal precipitates such as Fe oxyhydroxides constitute a net sink for REE (Table 1.5) [Klinkhammer et al., 1983, Olivarez and Owen, 1991, Mitra et al., 1994] explaining why the REE hydrothermal signal isn’t recorded in the open ocean.

21 Chapter 1. Introduction

Open plume Seawater water Nd 21.4 1.22 Ce 5.44 1.12 Eu 1.06 0.35 Er 5.47 0.45 REE concentration in pmol/kg Table 1.5: Comparison of selected REE concentration in open seawater and filtered water from MAR hydrothermal plume [Mitra et al., 1994].

1.4.2 Particle formation and reaction in the buoyant plume

To evaluate the importance of the scavenging process and assess the chemical continuity between hydrothermal fluids and oceanic waters mixing above the region of venting, water samples from the MAR have been collected at various points in the ascending plumes [Rudnicki and Elderfield, 1993, Mitra et al., 1994]. In the context of highly reactive trace elements like REE, three main stages of reaction could be distinguished during hydrothermal fluids mixing with seawater. The initial reactions are dominated by Fe chemistry, as its aqueous stability drops very rapidly in the early stages of mixing with near-freezing ambient seawater. During the buoyant ascension we observe (i) formation of Fe-sulphide in the first seconds of emission, then (ii) oxidation of both dissolved Fe and a portion of particulate Fe-sulphides into Fe-oxyhydroxides, and finally (iii) in the late stage of plume dispersion, when neutral buoyancy is achieved, formation of hydrothermal deposits with increasing Mn/Fe as Mn oxidation occurs at slower rates. The dilution process is extremely rapid while the fluid is ascending, with an estimated dilution factor of 10,000 during buoyant plume rise and further dilution as the neutral plume spreads laterally [Mitra et al., 1994].

In the early stage of mixing, the buoyant rise lasts less than an hour. This time is sufficiently short to consider trace metals uptake to be dominated by iron particles without influence of Mn oxidation. Fe particle formation in the ascending plume follows a two-stage process [Rudnicki and Elderfield, 1993]. Immediately after venting, about 50 % of the Fe from the high temperature fluid is removed as sulphides. REE concentration from fluid samples taken above vent throat normalized to zero Mg end-member fall within the composition range of typical hydrothermal fluid. They display linear correlation to Mg, as do Si and Mn. It may be concluded that REE are free from important particle removal within the first seconds of emission and behave conservatively until dilution factor of about 10 % [Mitra et al., 1994]. In contrast, concentrations of chalcophile elements (Fe, Cu, Zn, Pb) in the fluid decrease immediately as they are quantitatively removed through co-precipitation with Fe as polymetallic sulphides. These particulate sulphides settle rapidly on the seabed due to their high density [Mottl and McConachy, 1990]. Of the remaining

22 Chapter 1. Introduction dissolved Fe2+, ≈ 35 % is precipitated as oxide within less than one minute after emission giving an estimated half-life oxidation time of Fe2+ ≈ 2 min. During this time, ≈10% of the particulate sulphide material entrained by the plume is oxidized as well. Therefore, the two main metal- removing reactions within the buoyant plume occur extremely rapidly as sulphides incorporate the main chalcophile elements whereas Fe-oxyhydroxides uptake reactive trace metals from both the hydrothermal fluid and seawater.

1.4.3 REE fractionation by Fe particles

Observation of REE/Fe ratio in particles within the ascending plume shows a two-stage process starting by an abrupt increase followed by a continuous but slower increase of the REE content on the colloids. This observation relates to two phases of uptake whereby REE are extracted from the solution; instantaneous co-precipitation with the newly formed Fe oxyhydroxides (≈ 55 ± 20 % of dissolved REE) responsible for the strong increase in REE/Fe ratio, and further scavenging by sorption interaction during buoyant plume rise [Rudnicki and Elderfield, 1993].

Estimation from both theoretical models [Rudnicki and Elderfield, 1993] and sample measurements [Mitra et al., 1994] show that 95-99 % of available REE are removed from solution by Fe particles while reaching entrainment ratio of 350 to 570 (0.3 % to 0.18 % of hydrothermal fluid). Fe-oxyhydroxides scavenging of REE preferentially removes the intermediate REE compared to light and heavy REE [Rudnicki and Elderfield, 1993]. This increasing uptake rate of Sm and Gd by Fe oxyhydroxides is observed within REE patterns of hydrothermal particles normalised to shale (free from sulphides) displaying a convex upward pattern. Removal rates for REE in the buoyant plume range from 0.1 to 3*10-7 nmol.kg-1.s-1, whereas these rates are 300 to 2000 times slower within the neutrally buoyant plume; approximately 40 min after venting considering TAG plume thermodynamic characteristics [Rudnicki and Elderfield, 1993]. When neutral buoyancy is achieved, the plume starts to spread laterally and the reactive Fe-oxyhydroxides particles scavenge more dissolved material derived from seawater. Consequently, within a dispersing hydrothermal plume, the REE/Fe ratio progressively increases and continuously evolves to a more seawater-like signature with increasing distance from initial vent-source as the vent fluid signature is overprinted [Olivarez and Owen, 1989, German et al., 1990, German et al., 2002].

23 Chapter 1. Introduction

1.5 Hydrothermal metalliferous sediments in the ocean: diversity and mode of formation

Hydrothermally-derived sediments are mainly, but not exclusively, associated with active mid- ocean ridge settings through development of large scale hydrothermal circulation within oceanic crust as the result of sea-floor spreading. They are also found in association with submarine volcanoes along volcanic arcs and hot spot seamounts, and less commonly associated with fractures distant from oceanic ridges [Hein et al., 2008, Edwards et al., 2011]. Three main processes are globally considered regarding the genesis of metalliferous sediment. Fall-out deposit of hydrothermally-derived Fe and Mn oxyhydroxides [Rhulin and Owen, 1986b, Barret and Jarvis, 1988, Olivarez and Owen, 1989] form strata bound layers occurring as basal facies of the sediments column building up on the oceanic basement (Figure 1.7). Their lateral extent is controlled by the strength and direction of oceanic current at the emission point, and the topography of the surrounding seafloor. Major differences have been observed between metalliferous sediment extent at the MAR and EPR for instance. The relatively slow spreading speed of the Atlantic ridge forms a well-developed median rift valley confining the dispersion of hydrothermal plumes to 10’s of kilometres from the emission points [Klinkhammer et al., 1986]. On the other hand, the rapid spreading of the EPR forms a domed ridge allowing high dispersion of venting products by oceanic currents [Edmond et al., 1982]. Metalliferous sediments are also produced by diffuse hydrothermal systems within the ridge or off-axis. Typically of lower temperature (<100°C), the fluids percolate within porous formations such as sediments, pyroclasts or volcanic breccia [Fouquet et al., 1993] to precipitate layered oxide impregnations and crusts at the water/sediment interface [Alt, 1988, Mills and Elderfield, 1995, Goulding et al., 1998, Schultz, 2006]. Usually, off-axis diffuse LT hydrothermal systems produce Mn-rich encrustations and Fe-Si deposits occurring as nontronite as a result of strong gradients in Eh and pH conditions fractionating Fe from Mn [Severmann et al., 2004, Dekov et al., 2009]. Finally the last genetic process for metalliferous sediments formation corresponds to mass wasting and oxidative degradation of hydrothermal sulphide deposits and chimneys forming the so-called ocheriferous sediments [Rona et al., 1986, Thompson et al., 1988, Rona et al., 1990]. Most metalliferous sediment deposits in the TAG area are formed by a combination of those 3 components where sulphide-rich debris flows disturb hydrothermally-derived Fe-Mn oxyhydroxides settlement and get oxidized. Some more distal and later fluid circulation also introduces LT Fe oxide precipitates within the deposits [German et al., 1993, Mills et al., 1993]. The sedimentary formations associated with hydrothermal systems are highly variable in composition, depending on whether a focused or diffused discharge occurs.

24 Chapter 1. Introduction

1.6 The Troodos Ophiolite, Cyprus

1.6.1 Ophiolites: history and terminology

The recognition of oceanic rock sequences within orogenic belts showing association of tectonised peridotites, serpentinite, gabbros, dykes and lavas has been subject to multiple controversial debates. The main focus of those discussions were to understand the tectonic processes behind such occurrences and determine geochemical relations with modern seafloor analogues related to fragmentation of oceanic lithosphere, such as volcanic arc, sea-mounts and rift valleys. The appreciation of ophiolite complexes as fragments of oceanic lithosphere has

25 Chapter 1. Introduction allowed further understanding in the process of oceanic crust-making, its structure as well as the interaction of magmatic, structural and hydrothermal processes within oceanic ridges [Robinson et al., 2003].The Troodos massif played a dominant role in this process since the 60’s [Gass and Masson-Smith, 1963]. Integrating data from both ocean floor studies and investigations on the well exposed and undeformed Troodos Massif, the genetic relationship between actual oceanic rock samples from drilling the MAR and those continental sequences was uncovered.

The term “ophiolite” in its modern sense was described during the 1972 Penrose Conference as an assemblage of mafic to ultramafic rocks. A complete ophiolitic sequence exhibits from base to top an oceanic magmatic complex of ultramafic rocks representing the upper mantle, with variable amounts of harzburgite, lherzolite and dunite overlain by a gabbroic complex, layered or not, with sometimes showings of cumulates. The upper part of the sequence is mainly composed of extrusive basaltic rocks as massive flows and pillows, overlain by a cover of deep-sea pelagic sediments. A sheeted dyke complex is sometimes observable [Anonymous, 1972, Robertson, 2002]. Most of the ophiolites worldwide do not expose such complete sequences and are then characterized as dismembered ophiolites [Robertson, 2002].

1.6.2 Location and regional geology of Cyprus

The geology of Cyprus is part of a large and complex picture within the Mediterranean area. The island owes its existence to the closure of the Tethyan Ocean during the Alpine orogeny as multiple micro-plates evolved between the colliding African and Eurasian plates. The island of Cyprus covers an area of 9251 km² with a pronounced topographic high produced by the Troodos massif, an east-west trending elliptical dome rising up to 1952 m at Mount Olympus. On a regional scale, the geology of Cyprus can be divided into 4 main tectonostratigraphic units (Figure 1.8): (i) the Kyrenia range; (ii) the Mamonia complex, (iii) the association of the Late Cretaceous Troodos ophiolite sequence and in-situ sedimentary cover of the Mesaoria Plain spanning from Cretaceous to recent sediments and (iv) the Southern Troodos, or Arakapas Transform Fault Zone (STTFZ), considered to be a fossil oceanic transform fault zone separating the Troodos from the Anti-Troodos Plate [MacLeod and Murton, 1995].

1.6.2.1 The Mamonia Complex

The Mamonia Complex is cropping out in the south-eastern part of Cyprus. It is a structurally complex amalgamation of Triassic lavas to Cretaceous rocks including minor Troodos ophiolitic fragments forming an allochthonous complex of sedimentary, igneous and metamorphic rocks. This complex is juxtaposed with the Troodos massif via a combination of thrusting and strike-slip

26 Chapter 1. Introduction faults and separated by slivers of amphibolites and greenschists facies of within-plate-type basalts protoliths [Robertson, 2002].

Figure 1.8: Location map and tectonostratigraphic terranes of Cyprus [Robertson and Xenophontos, 1993].

1.6.2.2 The Kyrenia range

The Kyrenia range (Palaeozoic to recent sediments) forming the northern part of the island includes another allochthonous complex of sediment, volcanic and metamorphic rocks. These units were initially formed on a passive margin as shallow water carbonates from the Permian to the mid-Cretaceous were later deformed during the development of the Troodos ophiolite (Upper Cretaceous – Early Tertiary) with compression and thrusting during the Eocene up to now via convergence of the African and Eurasian plates (Erreur ! Source du renvoi introuvable.) [Robertson, 2000].

1.6.2.3 The Southern Transform Fault Zone

The STTFZ constitutes the southern edge of the Troodos ophiolite and represent the Anti Troodos Plate, a section of oceanic crust 3 to 5 km thick that formed coevally with the Troodos plate [Gass et al., 1994]. This section of ophiolite integrates typical oceanic crusts formations although, as opposed to the northern Troodos ophiolitic massif, they are highly deformed and chaotically juxtaposed as mantle material and gabbros can often be found lying next to extrusive lavas. Indeed, the Anti Troodos Plate lithological units suffered disaggregation, block rotation and later

27 Chapter 1. Introduction intrusion of magmatic episodes within the transform domain [Gass et al., 1994, Naden et al., 2006]. Two main sequences can be distinguished. (i) Plutonic and hypabyssal lithologies of mafic and ultramafic composition form the older axis sequence. These are tectonised harzburgites, trondhjemitic gabbros and dykes including wherlites, masses of sheared serpentinite and fragments of sheeted dykes (ii) while the second sequence is formed of pillow lavas, massive flows (boninitic lavas) and volcaniclastic sediments [Robertson and Xenophontos, 1997]. Two theories coexist at the moment as to its origin. Both have implications on the position of the Troodos in the paleo spreading environment whether the Arakapas transform fault is considered dextral or senestral. The reader is invited to look into MacLeod and Murton (1995) and Cann et al. (2001) for a discussion of the theories and their implications.

1.6.2.4 The Troodos massif

The Troodos massif was recognized as an ophiolite in the beginning of the 60’s [Gass and Masson- Smith, 1963, Gass, 1968, Moores and Vine, 1971]. It was first interpreted as mid-oceanic ridge setting thanks to broad chemical similarities of the extrusive basaltic rocks with MOR-basalts and the presence of a sheeted dyke complex known to be associated to MOR with 100 % of extension [Gass and Masson-Smith, 1963, Gass, 1968, Moores and Vine, 1971]. This interpretation was questioned when a non-negligible portion of geochemical analysis of basaltic samples fell on a calk-alkaline trend, thus suggesting an island arc origin [Miyashiro, 1973]. Although these results were contested on the basis of inappropriate methodology [Gass et al., 1975], other contemporaneous studies [Gass and Smewing, 1973, Pearce, 1975] highlighted the difference between the extrusive sequences of the Troodos ophiolite from those of MOR with the development of trace elements analysis. These studies pointed out the similarities existing between Troodos lavas and those found in a back-arc or fore-arc spreading centre where partial melts are influenced by water released from a subducting plate [Boyle, 1984, Pearce et al., 1984, Robertson, 2002, Robertson, 2004, Pearce and Robinson, 2010]. It is now accepted that magmatic formations from Troodos represent fragments of anomalous lithosphere formed in a complex environment involving spreading and subduction in small scale oceanic basins related to arc settings, often termed supra-subduction zone ophiolites (SSZ) [Pearce et al., 1984, Taylor and Nesbitt, 1988]. The extension and generation of oceanic crust was proposed to results from extension near a slab edge associated to a roll-back of the subducting African plate (Erreur ! Source du renvoi introuvable.) [Robertson, 2002, Dilek and Furnes, 2009, Pearce and Robinson, 2010]. In this view, the material from the slab was partially incorporated in the melt which influenced the magma composition from MORB to island arc tholeite and boninitic assemblage [Dilek and Furnes, 2009, Pearce and Robinson, 2010].

28 Chapter 1. Introduction

Figure 1.9: Schematic presentation of the genesis of the Troodos Ophiolite (A) and the evolution of the Island of Cyprus (B-D) [Cyprus Geological Survey, 2017]

29 Chapter 1. Introduction

The Troodos ophiolite shows no evidence of regional emplacement-related deformation but possesses an elliptical shape with concentric lithologies due to its diapiric protrusion formed by serpentinization of mantle peridotites [Robertson, 1990; Robertson and Xenophontos, 1993, Robertson, 2002] (Figure 1.8). The structure exhibits several igneous, metamorphic and sedimentary units from mantle lithologies up to sediments deposited on top of erupted volcanics. The Troodos ophiolite therefore offers a complete stratigraphic sequence from the MOHO to the paleo-seafloor. A recapitulative of the layout of Troodos lithologies with a comparison with modern oceanic crust seismic stratigraphy is presented in Figure 1.11.

The core of the antiform is an assemblage of harzburgite, lherzolites, dunites, wherlites and gabbroic rocks tectonised and altered to serpentinite via retrograde metamorphism to a depth of 20 km. The vast majority of this sequence is harzburgitic (80 %) while the remaining 20 % are dominated by pods of dunite [Robertson and Woodcock, 1980]. Approximately 1 km of the sequence is exposed at the highest levels of the Troodos massif around Mt Olympus.

This mantle sequence is overlain by a plutonic sequence of dominantly gabbroic nature favouring a multiple magma chamber spreading model alimenting volcanism at the spreading axis [Robinson and Malpas, 1990]. These chambers consist of mafic material with layered or cumulate gabbros at the base showing a transition to isotropic gabbros in the upper part of the sequence [Gass, 1980]. Presence of highly fractionated plagiogranites at the top of the sequence is thought to represent one of the final stages of the evolution of a basic magma chamber [Galley and Koski, 1999]. Isotopic U/Pb dating of zircons recovered from the plagiogranites gives an age of 90.3-92.4 Ma for this plutonic structure [Mukasa and Ludden, 1987]. A sheeted dyke complex with a dominant N-S trend 1 to 1.5 km thick is found above and consists of mutually intruding dykes, nearly vertical, suggesting formation in an extensional environment [Gass et al., 1994]. The dykes are commonly re-injected by later dykes suggesting spreading occurred either as a steady-state process [Allerton and Vine, 1987] or by formation of discrete and ephemeral seafloor grabens [Varga and Moores, 1985; Robertson, 2002]. These dykes represent the transition, or feeding system from the plutonic to the extrusive sequence of pillow basalts and flows. As such the dykes exhibit the same range of composition as the extrusive sequence [Baragar et al., 1990, Bettison-Varga et al., 1992].

The extrusive rocks are subdivided into two main units; the Lower and Upper Pillow Lavas (LPL and UPL respectively). The former is largely dominated by basaltic andesites whereas the UPL are mostly olivine basalts, though in numerous locations this boundary is transitional [Boyle, 1984]. Present within the LPL and UPL are also numerous mineralized ore bodies (massive sulphides deposits and stockwork zones, ochres and umbers) for which Cyprus is renowned, including one of the largest (Skourioutissa) located at the overlying lava-sediments interface. Indeed, the

30 Chapter 1. Introduction earliest workings for copper, gold and other metals date back to Bronze Age times, as long as 5000 years ago or longer, with later Phoenician and Roman smelting shown by the extensive slag heaps that are visible on the island today. Ancient surface and underground workings are also numerous on the island, with shafts, galleries, timbering and other manifestations of small-scale underground workings visible across the island [Naden et al., 2006]. The word “copper” may have been derived from the Greek for Cyprus, or vice versa [Bear, 1963]. As the main topic of this thesis, the ferromanganese deposits of the Troodos ophiolites are described in details later in this work.

1.6.2.5 Rotation and uplift of the Troodos massif

The Troodos sheeted dyke complex is currently aligned to a N-S orientation which is more or less at a 90° angle of what a Tethyan spreading axis would be expected (east-west). Paleomagnetic data on the sheeted dyke complex [Clube et al., 1985], umbers and Upper Cretaceous and Upper Eocene sediments [Abrahamsen and Schonharting, 1987] highlight that Cyprus underwent a 90° anticlockwise rotation since its formation in two major steps; 60 ± 10° between 90 and 50 Ma (most of it before the Lower Eocene) followed by a later episode of rotation of 20 ± 10° within the last 50 Ma [Robertson and Xenophontos, 1993]. The cause of the rotation remains debated and major hypothesis developed in Robertson and Xenophontos (1993) are summarised in Figure 1.10.

The precise timing of the Troodos emplacement is not known [Robertson, 2002] although it is suspected that it’s uplift started in the Miocene and the island have fully emerged around 1 Ma into the Pleistocene [Robertson and Xenophontos, 1993] (Erreur ! Source du renvoi introuvable.). The uplift is driven partly by the under thrusting of continental crust from the subducting African plate migrating northward [Robertson, 1990] and by serpentinization of the mantle inducing volume increase and diapirism resulting in the centred uplift of the massif and current radial layout of the ophiolite lithologies (Figure 1.8 and Figure 1.9).

31 Chapter 1. Introduction

32 Chapter 1. Introduction

1.6.3 Fe-Mn metalliferous sediments of the Troodos ophiolite

1.6.3.1 Ochre

Ochres constitute the in-place alteration products, Mn-poor and Fe-Si-rich, of previously existing massive sulphide deposits combined with hydrothermally precipitated Fe oxides [Constantinou and Govett, 1972, Robertson, 1976]. Ochre deposits are located in the Lower Pillow Lava sequence within the Troodos stratigraphy, below the unmineralized Upper Pillow Lavas, and are spatially and genetically associated with the massive cupriferous sulphide ore bodies [Constantinou and Govett, 1973, Robertson, 1976]. Studies of the structural and stratigraphic association of ochres capping SMS deposits have concluded that the ochres are the products of submarine sulphide weathering prior to further lava extrusion and sedimentation. Indeed, distinct layering observed within ocheriferous deposits, such as at Skourioutissa, indicate that they are the products of marine weathering and not the sub-aerial residue of sulphide weathering under atmospheric conditions, and are therefore distinct from gossans [Robertson, 1976, Richards and Boyle, 1986, Herzig et al., 1991]. The layering preserves various degrees of silicification and alternating Fe oxide-rich bands and lighter jarosite-rich bands, with rare gradationally bedded features, that may be the resulting alteration of original layering within sulphide debris [Robertson, 1976]. Observations from a range of sulphide deposits in Cyprus [Constantinou and Govett, 1972, Constantinou and Govett, 1973, Robertson, 1976] indicated that a wide variety of ochre occurrences can be discriminated including; conglomeratic sulphides in an ochreous matrix, finely laminated and fine grained metalliferous mudstones or tuffaceous ochres and massive siliceous limestones. The mineral composition of ochres is dominated by goethite, jarosite, and quartz with an important background of amorphous Fe oxides as well as traces of hematite and gypsum [Robertson, 1976, Herzig et al., 1991]. Important detrital pyrite and other sulphides are also found within ocheriferous sediments. Similar material was observed at the TAG vent fields [Rona et al., 1986, Thompson et al., 1988, Rona et al., 1990, Mills and Elderfield, 1995] and the EPR [Alt, 1988] on sea-floor weathered sulphide mounds.

1.6.3.2 Umbers

Umbers constitute fine-grained, brown, Fe-Mn rich mudstones of volcanic exhalative origin [Robertson, 1975], mostly amorphous in X-ray but including poorly crystallized goethite, vernadite, quartz and accessory apatite, pyrolusite, palygorskite and smectite [Constantinou and Govett, 1972, Robertson and Hudson, 1972, Boyle, 1990]. Umbers constitute a brittle and light rock with a density of 1.3 – 1.4 g.cm-3, breaking with a conchoidal fracture [Boyle, 1990]. Umberiferous sediments are found interstratified with or overlying the upper pillow lavas as lenticular bodies and in depressions of the lava surface with a massive texture. Related to

33 Chapter 1. Introduction hydrothermal solution precipitation of Fe and Mn oxides, umbers are encountered in multiple stratigraphic levels of the ophiolite, interlava and supralava umbers can therefore be distinguished.

Although far less abundant than supralava umbers, occurrences of Fe-Mn oxides can be locally abundant in the UPL [Boyle, 1984]. They present a large range of morphology as veins, horizons of ferromanganiferous mudstones and infilling space between pillows and cooling fractures [Elderfield et al., 1972, Boyle, 1984]. These sediments are largely dominated by Fe oxides with variable but generally low Mn content admixed with varying proportions of volcanoclastic and carbonate material.

Supralava umbers constitute the basal facies of the Perapedhi Formation, stratigraphically starting the sedimentary cover found on the Troodos Ophiolite rim. Outcrops mostly occur as dismembered and discrete bodies within the pillow lava sequence in particular between the basaltic basement and the other sediments of the Perapedhi formation as clay-rich (bentonite) umbers and radiolarian cherts, or the Campanian chalk of the Lefkara Formation (Figure 1.12) [Robertson, 1976]. Rarely umberiferous deposits are found in direct association with VMS deposits or ochres of the LPL like in Skourioutissa. These associations between supra-lava sediments and massive sulphides and their oceanic alteration highlight a strong tectonic structuration of the oceanic basement at this locality as the upper pillow lavas are absent. This suggests important tectonic movement post VMS formation and development of large and deep canyon, or an axial valley, to structurally tilt the lower pillow lavas containing the VMS deposits out of reach of new lava flows forming the UPL and seeing accumulation of umbers on top of this ridge. In this view, recognised large grabben structures in the Troodos ophiolite (e.g. Solea, Mitsero and Larnaca Graben) and these association of lithologies point towards an intermediate (< 10 cm/year) spreading ridge [Boyle and Robertson, 1984, Eddy et al., 1998]. This observation also explain the distribution and rather important thickness of umber deposits in Troodos as a fast spreading ridge (such as the EPR) would creates a bombed topology allowing for vast dispersion of venting product in the water column instead of being contained within the well-defined structure of axial valleys and canyons of slow to intermediate spreading ridges [Boyle and Robertson, 1984].

After allowing for differential compaction, umbers share the same dip as the overlying chalks, suggesting that extensional faulting and block rotation of the underlying basalts to form half- graben was syn-tectonic and occurred before umber accumulation [Boyle and Robertson, 1984]. Thicker deposits are most commonly related to a major tectonic structure that impacts the oceanic floor topography and provides an important depression for sediment deposition (Figure 1.12) [Robertson, 1975a]. Nowadays massive umber deposits can reach thicknesses of 4 m,

34 Chapter 1. Introduction although most are limited to 1 m. However, records indicate that outcrops up to 35 m thick existed prior to extraction in the Mangaleni quarry [Boyle, 1984]. The lateral extent of umber deposits predominantly depends on initial sea floor topography, subsequent preservation from sea floor weathering and anthropic activities.

On a smaller scale, fine laminated internal structures with graded bedding are encountered within most umber deposits when undisturbed. They display alternating dark brown and earth-brown millimetric to centimetric layers. This interlamination is usually discontinuous as the results of microfaulting with localised small-scale loading structure, ejection figures, with channelling and limited micro-cross lamination sometimes visible in the thickest umber deposits [Robertson and Hudson, 1972, Robertson, 1975a].

Models for the emplacement of umbers have evolved contemporaneously with our understanding of ophiolites, the exploration of oceanic ridge sediments and discovery of hydrothermal vents. Umbers were initially interpreted as volcanic ash fall-out [Wilson, 1959] or altered clay resulting from the weathering and leaching of underlying basalts by meteoric waters [Bear, 1960]. Later work from the Cyprus geological survey suggested precipitation in a lagoon environment [Wilson, 1959, Carr and Bear, 1960] or from open seawater [Gass, 1960, Pantazis, 1966]. It is not until the work of Elderfield et al. (1972), before the discovery of the black smokers, that a submarine volcanic influence origin was suggested. Later works dominantly from Robertson and Boyle, through extensive mapping and detailed geochemical analysis, refined the model proposing umbers as the result of Fe-Mn oxides accumulation in depressions of the lava topography and issued from high temperature vents. Therefore umbers can be considered as distal plume fall-out deposits and described as an assemblage of sedimentary components: (i) pelagic, comprising externally-derived lithogenic and biogenic sediments; (ii) a locally-derived volcanic detrital component; (iii) an Fe-Mn component resulting from oxidation of dissolved species issuing from hydrothermal vents and; (iv) a hydrogenous component by scavenging reaction of seawater dissolved trace metals [Richards and Boyle, 1986].

35 Chapter 1. Introduction

36 Chapter 1. Introduction

Figure 1.12: Reconstruction of the field relationship of a typical small umber hollow related to seafloor faulting, Troodos Massif (redrawn after Robertson and Boyle (1983)).

1.6.3.3 Umbers alteration facies

1.6.3.3.1 Mn-depleted umbers

A bright orange umberiferous layer is generally observed in the first few centimetres above the contact with pillow lavas. This is interpreted as an Fe-rich, Mn-deficient basal layer compared to the overlying umbers. Potential discrete pyrolusite concretions as nuggets or veins can be found associated with these ferruginous horizons [Robertson, 1975a]. This post depositional migration of Mn to the top of the sequence is due to changes in redox chemistry as the sediment sequence build up. The redox change is probably due to alteration of the underlying basalt making the solution in the sedimentary pile become reduced, encouraging Mn into solution, moving upwards and precipitating at the next oxidizing interface [Robertson, 1976, Boyle, 1984].

1.6.3.3.2 Silicified umbers

Early silicification of the supra-lava umbers is commonly found throughout the massif as irregular masses preserving radiolaria, umber laminations and undeformed veins of palygorskite from later compaction [Richards and Boyle, 1986]. A low temperature (6 ± 0.6°C) is inferred for this silicification stage through oxygen isotope analysis [Boyle, 1984] which mainly affected umber composition by diluting all other elements by addition of silica.

37 Chapter 1. Introduction

1.7 Thesis rational and objectives

The research presented in this thesis addresses the criticality of rare earth elements supply to industries and investigates the potential of hydrothermal metalliferous sediments as an alternative source. REY, among other elements (Co, Li, Te, Se, Nb, Ta, In, Ru, Ga, Ge), are considered critical metals in view of the high supply risk, which is linked to their production and processing in a limited number of geographical areas [European Commission, 2014]. These elements have a growing importance for the development of high technology and industry, most notably in the production of smartphones and green energy. Other applications in the chemical industry, military and aerospace technologies underpin the economic and strategic importance of these elements.

REE production in the last 30 years has been dominated by extraction from two distinct categories of deposits. Calk-alkaline rocks, such as carbonatites and pegmatites, can be considered as high grade ores but require costly extraction and hydro-metallurgical processing in addition to problematic radioactive waste management. In contrast, ion-adsorption clays (IAC) contain 40 to 500 times less REE, but economically viable extraction is achieved thanks to limited processing and the extremely low extraction costs. The Chinese production of REE in 2015 represented 85% of the world production [U.S.G.S, 2016]. The Chinese monopoly on REE production and exportation is the result of economic competition rather than resource distribution. Indeed, alternative deposits outside China are already well-known [Kanazawa and Kamitani, 2006] although most of the attention has been focused on high grade type deposits [Linnen et al., 2014, Goodenough et al., 2016]. Despite the abundance of potential primary deposits, important research has also been undertaken recently in developing REE production from alternative deposits [Kato et al., 2011] or from industrial by-products such as red-mud [Tsakanika et al., 2004, Qu and Lian, 2013, Ujaczki et al., 2015] and coal residue [Rozelle et al., 2016].

Metalliferous sediments dominantly constituted of amorphous Fe-Mn oxides share similarities with ion-adsorption clay deposits in terms of grade and potential processing workflow. In order to draw conclusions on the economic potential and feasibility of mining such deposits, it is essential to first explore this distribution, availability and continuity of these oceanic formation. Plume fall- out sediments derived from high temperature hydrothermal venting constitute mixed hydrothermal-hydrogenetic deposits via precipitation of hydrothermal Fe and Mn oxide when mixing with cold oxidizing seawater [Barret and Jarvis, 1988, Goulding et al., 1998]. The oxidation and hydration of dissolved Fe and Mn leads to the formation of high surface charge colloids:

Fe(OH)3.nH2O and MnO2.nH2O [Barret and Jarvis, 1988, Koschinsky and Halbach, 1995, Hein et al., 1997, Canet et al., 2008] that scavenge dissolved trace metals from both the hydrothermal fluid

38 Chapter 1. Introduction and seawater. The scavenging process occurs by simple electrostatic interactions [Koschinsky and Hein, 2003, Bau and Koschinsky, 2009] during transportation by bottom sea currents, and deposits could occur by settlement over thousands of kilometres. These deposits therefore present seawater-like REE signatures and enrichment in the range of concentration encountered within IAC deposits. The possibility of sea floor extraction has received widespread publicity in recent years, with the recognition of the potential of marine clays to provide an economic supply for REE [Kato et al., 2011]. However, determining the REE potential in any seafloor environment is hindered by the numerous technological challenges to investigate and evaluate these deposits. The Troodos ophiolite, Cyprus is a well-preserved fragment of Thetyan oceanic floor that preserves an excellent record of the oceanic crust stratigraphy and is readily accessible. Already well-studied for its important mineral resources of Cu-Ni-Au sulphide, chromite, asbestos and umbers, this massif constitutes an excellent area for the investigation of sea-floor processes and associated mineralisation. Therefore, the first part of this thesis examines the potential of Cyprus umbers; hydrothermal Fe-Mn oxide deposits lying on the upper pillow lava of the Troodos massif, as an economic source for REE.

The second aspect developed in this thesis is the determination of the most efficient extraction methods and commercial viability of REE production in Cyprus and ultimately the present day seafloor sediments. Techniques of simple and sequential leaching have been widely used for the study of element-phase association [Boyle, 1984, Koschinsky and Halbach, 1995] and as a means of selective extraction for mineral processing [Ru'an et al., 1995, Jun et al., 2010, Moldoveanu and Papangelakis, 2013, Qu and Lian, 2013, Ujaczki et al., 2015]. Notably REE from ion adsorption clays are extracted by in-situ leaching and optimization of the process has been largely documented [Tian et al., 2010, Yang et al., 2013, Vahidi et al., 2016]. However, similar approaches on oxide-based deposits remain largely unexplored. Here the release of REE compared to other elements, considered as impurities, is investigated under diverse leaching conditions. Given the initial low concentration of REE in umbers, cost-effective methods for the separation of REE from other elements in the leach liquor need to be implemented for further adequate processing. The commercialization of REE products ranges from mischmetal (mixed rare earth oxides, REO) to single high purity oxide. Due to their high geochemical coherence, separation of the lanthanides into single elements is a time and energy consuming process that does not fit in the scope of this project. The objective therefore lies within the development of an efficient and cheap process to form a mixed REE end-product of high purity in the least number of steps.

39 Chapter 1. Introduction

1.8 Thesis structure

The thesis is subdivided into 5 main chapters. First, the introductory chapter presents a general overview of the research goals of the project, the scientific background and prior research. Considering the objectives and the context of the project, the introduction is structured around three sections. An overview of the fundamental chemical properties of REE and the associated economic context around their exploitation is first given. An overview of the oceanic reservoirs and pathways of REE cycling within the ocean is then presented. The focus is centred on riverine inputs, the estuarine interface and the complexation and distribution of REE in the water column in a first time. A second part present REE geochemistry in hydrothermal fluids in a general sense and the parameters affecting their fractionation during fluid/rock interactions and the seawater/particle interface as the hydrothermal plume spread generating plume-fall out deposits. The last part of the introduction presents the geological context of the study area of Cyprus and the diverse metalliferous lithologies encountered in the Troodos massif.

The second chapter constitutes an investigation of umber deposits in Cyprus detailing fieldwork observations, sampling methodology, mapping and a geochemical characterisation of these deposits. Specific attention is given to stratigraphic evolution of umbers at diverse localities around the massif and how various alterations have affected element mobility. A broad comparison is made of the collected data with samples from other ophiolitic massifs and ferromanganese deposits encountered in modern oceans to explore the existing hypotheses that explain the formation of umber deposits.

The third chapter focuses on finding the optimal leaching conditions for REE extraction from umber using a series of experiments. The tested parameters include (i) the nature of the lixiviants as both acid solutions (hydrochloric, nitric and sulphuric) and ionic solutions (sodium chloride and ammonium sulphate) are tested, (ii) the molarity, or concentration in solution of the reactant in the leach preparation (0.05 – 1.75 M), (iii) the respective proportion of solid and liquid in the experiment expressed as S/L ratio (1/3 – 1/100 by weight), (iv) the time of reaction (5 min to 13 hours) and (v) the effect of temperature (20, 40 and 70°C).

The fourth chapter presents the investigation of REE selective precipitation with oxalate for the purification of the leach liquor. Although the leaching efficiency is important averaging 85% recovery, the absence of distinct mineralogy or crystalline phase in umbers prevent the physical pre-treatment for the beneficiation of a REE-rich phase. This therefore impacts the overall purity of the leach solution with an important contamination by undesired elements. Oxalate ions constitute the simplest form of organic complexes and possess high complexation constants with the lanthanides [Schijf and Byrne, 2001]. In this chapter, the selective precipitation of RE-oxalate

40 Chapter 1. Introduction complexes is investigated as a function of pH for the production of a high purity precipitate. A numerical model using PHREEQC software is used to compare experimental results and predicted fractionation of the REE with the oxalate complexes. The addition of oxalate at a controlled pH allow the precipitation of >96% of the leached REE at a purity superior to 80% by weight.

The fifth chapter is a conclusion to the study, summarising the main findings and giving an outlook for potential applications and further research to be undertaken in the field of REE extraction from metalliferous sediments.

Chapter 2. Geology and geochemistry of umbers

Chapter 2: Geology and geochemistry of umbers

2.1 Introduction

Since the recognition of ophiolite massifs as preserved fragments of oceanic crust, their study has given insightful observations and increased our understanding of mid-oceanic ridge processes. In addition, considering the economic potential of mineral resources forming in these geological contexts, the difficulty of assessing the continuity and extension of such deposits on the seafloor and costs of operating at sea, investigation on analogues preserved on-land is an alternative of major interest. Ophiolitic sequences are typically far larger than anything explored on the seafloor to date as they are more accessible. Supra-lava metalliferous sediments of the Troodos ophiolite have received important attention in the 70’s and 80’s as the whole massif was recognised as a relatively intact portion of oceanic crust. Identified as direct analogues of then newly discovered black smokers’ fall-out deposits, the study of umbers structural and geochemical characteristics brought important understanding and background to nowadays exploration of active oceanic systems. Since then, improvement and development of analytical means have allowed data production of high precision and in important quantity with the emergence of ICP-MS. This chapter therefore propose a revaluation of previously produced data and models thanks to the production of a large geochemical data set on umbers and their associated alteration facies. These new data are compared with a large range of modern data on oceanic and land-based marine ferromanganese mineralizations to highlight the mining potential of oxide-based deposits as a REY resource.

2.2 Fieldwork

2.2.1 Objectives

The fieldwork was organized as a 2 weeks campaign for the recovery of samples across the Troodos massif and mapping in May 2014. With the objective to re-evaluate the formation of umbers in the light of data available from the modern oceans, and address their potential as a new type of ore for REE, the sampling methodology focused on two main approaches: (i) a representative number of sample of umber occurrences around the Troodos massif were collected in order to evaluate the consistency of the resource on a regional scale, (ii) at well- developed and exposed outcrops of umber, stratigraphic sequences from the basalt-umber contact to the umber-deep sea sediment transition were sampled. These sequences allow an

43 Chapter 2. Geology and geochemistry of umbers assessment of the geochemical evolution of this hydrothermally-derived formation as deposition occurs and developed during diagenesis and possible further alteration.

Furthermore, two areas presenting extended outcrops of umbers were mapped in detail to provide type localities and highlight the important structural control of the basaltic basement on the dynamics of deposition and location; these areas are the east of Kampia towards Analiontas and the south-western area of Margi.

2.2.2 Sampling method

Supra-lava umbers and associated lithologies of the Perapedhi Formation were studied at 11 localities (Figure 2.1). More than a hundred samples were taken in accordance to the following methodology. Prior to sampling, outcrops were cleared of the first few surficial altered or weathered centimetres. Depending on the nature of the sample, a representative volume of fresh rock was sampled using a 20*20 cm sampling bag averaging 800 g of material. In the case of samples targeting veins within umbers, silicified nodules or alteration facies, representative material was recovered over a large portion of the outcrop as far as practicable. Each sample was headed with PJ-CY-2014 (Initial of the geologist – location – year) followed by a number constituting its reference. Complete samples ID are therefore in the form PJ-CY-2014-XXX, from 001 to 111.

2.2.3 Field observations and lithologies

The present day lateral extent of umbers is dependent on the initial sea floor topography and subsequent preservation from seafloor and terrestrial weathering and anthropogenic mining activity. However, most outcrops of umber in Troodos consist of localized bodies with a maximum lateral extension of tens of meters. At the Kampia and Margi areas, well-exposed umber sequences are preserved along horst and graben structures with outliers on topographic highs providing hundreds of meters exposure in length. In turn, good examples of depression-infilling by hydrothermal sediments are found at Asgata and Perapedhi [Boyle, 1984, Boyle, 1990] (Figure 2.2).

In good agreement with previous observations of umber deposits three main lithologies were observed and sampled during fieldwork (Figure 2.3). - The typical unaltered and massive umber lithology is present in deposits that commonly reach thickness of 4 m, although most are limited to 1 m, has a fine laminated internal structure with alternating dark brown and earth-brown millimetric

44 Chapter 2. Geology and geochemistry of umbers

to centimetric layers with graded bedding such as observed in Asgata, Perapedhi, Kampia and Margi (Figure 2.3).

- The second lithology encountered correspond to the Fe-rich, Mn-poor basal diagenetic layer. Although not always present, this bright orange ferruginous layer usually develops up to the first 10 to 35 cm above the contact with pillow lavas in numerous outcrops (Figure 2.3). These layers are to distinguish from ochre; pale coloured, Mn-poor and Fe-rich sediment spatially associated with sulphide ores within the lava piles resulting from oceanic weathering of sulphides in a similar process to gossan formation in atmospheric conditions [Constantinou and Govett, 1972, Robertson, 1976, Boyle, 1990]. The contact between this orange layer and massive umber is sharp and commonly cuts across the lamination of the overlying dark brown umber assessing its secondary origin [Robertson, 1976]. In strong association with this lithology, metallic dark-grey, micro-crystalline concretions localized at the umber-basalt contact were observed. This style of mineralization was encountered at only two outcrops in Kampia where this mineralization was present but not extensively developed. Also, the spatial density of dark submetallic veins, were higher in the vicinity of this diagenetic layer.

- Thirdly, silicified umbers or cherts are extensively developed in umber outcrops at Kampia and Margi [Robertson, 1977] with three contrasting morphologies observed notably in Margi: (i) silicified nodules ranging in size from 1 to 5 cm on average and up to 30 cm in diameter. These nodules do not preserve any preferred stratigraphic position within umbers layers and their concentration varies laterally along outcrops (also observed in Perapedhi); (ii) bulbous and massive silicified layers interstratified with or directly above the umbers which vary from 5-10 cm to 50-70 cm thick. These first two morphologies constitute hard and dense rocks with a vitreous texture. (iii) Millimetric quartz veins are pervasively developed within the massive layers whereas veins up to 10 cm thick are uncommon (Figure 2.3). Silicification of umbers was

reported as simple addition of SiO2 at low temperature to the porous and not yet compacted metalliferous sediments [Boyle, 1984, Richards and Boyle, 1986]

45 Chapter 2. Geology and geochemistry of umbers

yprus yprus

(2006)

et et al.

Naden

in in

Geological map of the eastern part of the Troodos Massif and sampling location. (Modified after 1:250 000 Geological map ofC map Geological 000 1:250 after (Modified location. sampling and Massif Troodos the of part eastern ofthe map Geological

: :

Figure Figure

46 Chapter 2. Geology and geochemistry of umbers

hic hic

r also show a a show also r

of the outcrop is 11 m. 11 is outcrop the of

column. Lateral extent Lateral column.

outcrop near Asgata, with the umber infilling a depression in the lava topography. The umbe The topography. lava the in a depression infilling the umber with Asgata, near outcrop

good example of layering with an alternation of massive beds and thinner more friable ones as schematised in the stratigrap the in schematised as ones friable more thinner and beds massive of alternation an with oflayering example good

Photograph of an umber ofan Photograph

: :

2 Figure Figure

47 Chapter 2. Geology and geochemistry of umbers

48 Chapter 2. Geology and geochemistry of umbers

2.2.4 Mapping

Two areas on the northern rim of the Troodos massif with extended umber outcrops were mapped in details during fieldwork (Figure 2.4).

2.2.4.1 Kampia area

The area of Kampia is located on the northern border of the Troodos massif in close vicinity to the Theotokos Monastery and the Kampia sulphide mine to the south (Figure 2.4, Figure 2.5). Here regular discrete outcrops of umber are exposed but are rarely more than 1m thick (Figure 2.5 A). Most umber outcrops show dip and strike readings from 10° to 45° west, although the southern outliers present shallow dip and strikes ranging from south to east. Umbers in the northern part of the locality occur as discontinuous lenses along the main basalt-sediment contact, whereas outcrops of the southern half occur as outliers. Here, umbers define topographic highs due to differential erosion compared to the surrounding sediments, with silicified umbers providing the more competent horizons. Outcrops in the NW display a good example of the basement influence on umbers occurrences, with NE-SW trending faults initiating small-scale repetition of the sequence basalt-umbers-sediment (Figure 2.5 B). It has been suggested that the similar fault orientation in this area with the strike of dykes near Kampia suggests some of these normal faults developed during graben formation, potentially parallel to the paleo ridge axis [Varga and Moores, 1985]. However, the faults present in the mapped area clearly displaced the Maastrichtian chalk deposits therefore suggesting a later formation. Also no observed variation of dipping direction of umber deposits in this area suggests their deposition filled a pre-existing depression. In detail the exact setting of the umbers varies including infilling of fault bounded half graben, hollows in pillow lava topography and as “drapes” over basalts on the NW western ridge (Figure 2.5 C). The highest point on the NW ridge is dominated by a massive body of silicified umber sitting on top of pale green altered pillow lavas and hyaloclastites. Considering the localised and intense alteration of volcanic rocks and the presence of massively silicified umbers, this fault may have formed a preferential pathway for a later stage of hydrothermal alteration [Prichard and Maliotis, 1998].

2.2.4.2 Margi area

The area SW of Margi is marked by a N-S faulting sequence dipping west which repeats the upper part of the Upper Pillow Lavas extrusive sequence and sediments (Figure 2.6). These normal faults developed parallel to the spreading axis at a time estimated between eruption of the LPL and UPL initiating the formation of small grabens and half-grabens [Boyle and Robertson, 1984]. Detailed mapping of an area located 500 m west of Kampia focuses on one of these major faults with

49 Chapter 2. Geology and geochemistry of umbers excellent exposure of umberiferous deposits (Figure 2.7). Extensive mapping of the faults and lava succession present in the area [Boyle and Robertson, 1984] have demonstrated the original tectonic structuration of the paleo seafloor prior to umber deposition with one transform fault and at least two generation of normal faults present. The umbers filling existing half-grabens display a North-South outcrop extension of nearly 800 m along the basalt/chalk contact and unconformably overlie the Upper Pillow Lavas sequence, dipping 30-50° ENE. These outcrops have been exposed by the N-S fault downthrowing umbers and chalk. Extended anthropogenic developments for crops and animal farming have levelled the chalk formation and mask much of the contact between umbers and the sediments. Unlike at Kampia, the umbers in Margi are not developing outliers in the sedimentary sequence but rather display good examples of the metalliferous sediments filling lava hollows and depressions. This is highlighted by a significant variation of umber thickness along the exposure commonly ranging between 30 cm to 2 m, and up to 7 m in the northern part (sample 37-42). By comparison, the stratigraphic sequence locality, approximatively 30 m to the north shows a maximum thickness of 2.5 m although the outcrop is continuous between the two sampling points. Similar controls on umber thickness were observed in outcrops near Asgata and Perapedhi where relatively thick umber deposits fill in depressions in lava topography.

Figure 2.4: Geological map showing location of detailed mapping near Kampia (1) and Margi (2). Data from Naden et al. (2006).

50 Chapter 2. Geology and geochemistry of umbers

Figure 2.5: (A) Geological map of the Kampia area, Cyprus. (B) Close up on the North West of the geological map of Kampia displaying location of cross- section transects referred to as a, b and c. (C) Schematic assemblage of cross- section transects showing layout of umbers.

51 Chapter 2. Geology and geochemistry of umbers

Chapter 2. Geology and geochemistry of umbers

depleted umbers with intial umber layering. umber intial with umbers depleted

of the outcrop presenting lithologies layout. Locally, umbers umbers Locally, layout. lithologies presenting outcrop ofthe

Note the cross cutting relationship ofMn relationship cutting cross the Note

B cross section of the northern part part the northern of section cross B

thickness up to 7 meter was observed. was meter 7 to up thickness

Geological map of the Margi area, Cyprus. A Cyprus. area, Margi ofthe map Geological

: :

. 2

Figure

53 Chapter 2. Geology and geochemistry of umbers

2.3 Geochemical characterization of umber deposits

2.3.1 Methods

2.3.1.1 Mineralogy

The mineralogy of the umbers was investigated using X-ray diffraction (XRD) with a Panalytical Catel X/P Pro at Ocean and Earth Science at the National Oceanography Centre Southampton (NOCS). Samples were grounded under methanol, pipetted onto a 32 mm round glass, allowed to dry and then analysed for approximatively one hour using Cu-Kα radiation over 2θ ranging from 2° to 75° at 35 kV and 40 mA. Minerals were identified using X’pert 9 search-match software. Scanning electron microscopy (SEM) was performed to study the detailed structure, composition of micro-laminations and veins found in umbers.

2.3.1.2 Major elements analysis

Major element analyses were undertaken by X-ray Fluorescence (XRF) with a Philips MagiX Pro spectrometer at NOCS. Analysis on major elements (SiO2, Fe2O3, MnO, TiO2, Al2O3, MgO, CaO, K2O,

P2O5) were made using glass-beads prepared with a dilution factor of 20:1 mixing 0.25 g of sample powder with 5 g of lithium tetraborate (Li-T). Accuracy of internal calibration and measurements was determined against standards run as unknown. Standards FeR1, FeR2, FeR3, FeR4 (iron formations), Nod-P-1, Nod-A-1, FeMn-1 (polymetallic nodules), GXR-1 (jasperoid) and Mn Ore were selected for their high concentration in Fe, Mn and Si and similar oxide matrix, analysed and compared with their recommended and published values [Jochum et al., 2005] (Figure 2.8). Good results are observed for SiO2, MnO, Fe2O3, CaO, P2O5 and K2O with less than 10% deviation observe between the measurement and recommended value. On the other hand, some standard values for Al2O3, MgO and TiO2 depart from their predicted position due to important matrix effect from the dominant iron oxides matrix and initial low concentrations of the measured elements amplifying errors. Considering that Al2O3 and MgO are of minor importance in this study, no corrections were applied to their concentrations.

All concentration thereafter expressed as % represent weight %. In addition, as geochemical concentrations for each element do not follow a normal distribution, results are later reported as the median with a 95 % confidence level interval (95 % CI).

54 Chapter 2. Geology and geochemistry of umbers

Figure 2.8: Measured versus recommended values for rock standards measured by XRF.

2.3.1.3 Trace elements

Trace and rare earth elements were analysed by inductively coupled plasma mass spectrometry (ICP-MS) using an Element X-series 2 at NOCS. The digestion procedure was as follows: 50 mg of powdered sample was treated in Teflon beakers with 4 mL of Aqua Regia: 3mL of 6 mol/L hydrochloric acid (HCl) and 1 mL of sub boiled (SB) nitric acid (HNO3), for 24 h on a hot plate at 130°C. Samples were then dried out and redigested with a mixture of hydrofluoric acid (HF) and nitric acid on hot plate for 24h. Samples were dried down again to incipient dryness and sufficient SB 6 mol/L HCl was added to dissolve the residue for 4 hours. Dissolved samples were then carefully transferred to pre-weighed high density polyethylene (HDPE) bottles and made up to appropriate volume with HCL and Milli-Q® water forming the mother solutions (MS). Solutions for

55 Chapter 2. Geology and geochemistry of umbers measurements by ICP-MS were produced at appropriate dilution after drying out with 3% Nitric acid solutions containing internal spike (Ru-Re (5 ppb) and Be (20 ppb)) and made up to an appropriate volume of 5 mL. Assessment of protocol procedure, cleanliness of material and measurement stability was made by insertion of standard duplicates and blanks every 40 samples. Data were calibrated using combination of international standard JB-1a, BIR-1, BHVO-2, JB3, JGb- 1, Ru-Re-Be spike and internal standard JA2, BRR1 and BAS206. Rock standards BHVO2, BIR1 and JB3 were prepared in triplicate and run as unknown at different stage of the run for monitoring accuracy and reproducibility along the analysis. All trace elements analysed here fall within the range of accepted values for the standards (Table 2.1). Reproducibility and consistency are also excellent with a standard deviation less than ± 3 % of the average concentration measured by ICP for all elements except for Ni, Cu and U (< ± 5.5 %). A standard deviation of ± 10.1 % for Th was calculated for standard BIR1 were the concentration is 0.03 ppm, relatively close to the detection limit (Table 2.1).

All values expressed in ppm average (n = 3) BHVO2a average (n = 3) BIR1a average (n = 3) JB3a Sc 33.5 ± 0.2 32 48.8 ± 0.2 41 - 45 35.6 ± 0.2 32 - 35.9 Co 42.7 ± 0.7 34 - 56.2 49.3 ± 1.4 41.3 - 64.2 36.4 ± 1.4 24 - 43.3 Ni 124.5 ± 4.1 97.9 - 186 174.8 ± 5.4 116.7 - 249.5 39.8 ± 2.2 32 - 49.1 Cu 132.1 ± 3.7 102 - 168 123.2 ± 4.1 97.3 - 152 186.1 ± 8.4 142 - 196 Sr 386.4 ± 5.5 317 - 438.3 111.7 ± 1.6 90 - 130 393.3 ± 4.9 392 - 452 Y 26.9 ± 0.2 18.8 - 30.4 17.0 ± 0.2 11.28 - 18 26.3 ± 0.1 21.7 - 28.6 Zr 180.6 ± 2.5 124.5 - 195.3 16.3 ± 0.2 4.87 - 32.87 96.3 ± 0.7 86.2 - 110.43 Ba 133.4 ± 0.6 99 - 150 7.34 ± 0.14 0.635 - 17.2 240.5 ± 0.4 217 - 271 La 15.71 ± 0.11 13 - 16.9 0.70 ± 0.02 0.36 - 3.6 8.58 ± 0.04 7.65 - 10.18 Ce 38.76 ± 0.19 32.41 - 45.3 2.07 ± 0.02 1 - 4.44 21.60 ± 0.03 19.5 - 56 Pr 5.56 ± 0.03 4.72 - 5.93 0.42 ± 0.01 0.10 - 0.46 3.32 ± 0.01 3 - 5 Nd 25.18 ± 0.15 22.36 - 28.1 2.61 ± 0.03 1.6 - 3.3 15.91 ± 0.12 14.5 - 17.03 Sm 6.27 ± 0.07 5.14 - 6.9 1.20 ± 0.01 0.8 - 1.24 4.31 ± 0.09 2.95 - 4.83 Eu 2.11 ± 0.02 1.71 - 2.7 0.57 ± 0.00 0.3 - 0.6 1.32 ± 0.02 1.24 - 1.6 Gd 6.40 ± 0.05 4.68 - 7.34 2.02 ± 0.01 0.36 - 3.2 4.70 ± 0.03 3.9 - 5.2 Tb 0.96 ± 0.01 0.72 - 1.08 0.39 ± 0.00 0.22 - 2.28 0.74 ± 0.01 0.65 - 0.94 Dy 5.38 ± 0.06 4.03 - 6.05 2.75 ± 0.02 2.5 - 4 4.57 ± 0.06 3.9 - 4.98 Ho 1.01 ± 0.01 0.7 - 1.12 0.62 ± 0.01 0.34 - 0.64 0.95 ± 0.01 0.82 - 1 Er 2.55 ± 0.01 1.7 - 2.82 1.83 ± 0.02 1 - 1.89 2.68 ± 0.03 2.39 - 2.88 Tm 0.34 ± 0.00 0.26 - 0.42 0.27 ± 0.00 0.14 - 0.3 0.39 ± 0.01 0.34 - 0.59 Yb 2.07 ± 0.03 1.59 - 2.38 1.82 ± 0.03 1.36 - 1.8 2.57 ± 0.05 2.2 - 2.83 Lu 0.29 ± 0.01 0.2 - 0.38 0.28 ± 0.00 0.18 - 0.31 0.38 ± 0.01 0.32 - 0.5 Th 1.26 ± 0.03 1.03 - 3.51 0.05 ± 0.01 0.03 1.31 ± 0.05 1 - 1.42 U 0.44 ± 0.02 0.32 - 0.51 0.01 ± 0.00 0.031 0.49 ± 0.03 0.47 - 0.54 a Standard , published values from Jochum et al., [2005] Table 2.1: Average ± 95 % CI of measured trace element concentration of rock standards run as triplicates and their recommended published values for accuracy and reproducibility checking.

2.3.1.4 Sr isotopes

87Sr/86Sr isotopes analyses were measured to analyse the influence of fluids involved (seawater vs hydrothermal) in the deposition of umbers and formation of their alteration facies. Analyses were

56 Chapter 2. Geology and geochemistry of umbers performed at NOCS on a TRITON thermal ionization mass spectrometer (TIMS). Samples were prepared for Sr extraction following standard protocol for Sr column separation and loaded on a Ta ionizing filament. Accuracy was checked against measurements on NBS987 standard with value of 0.710239 ± 7 (n = 7) showing excellent reproducibility across measurement and well within the error of recommended value: 0.71034 ± 26 (Figure 2.9). Measurements for BHVO-2, JB1-A and JB3 standards fall within the range of 87Sr/86Sr published values [Jochum et al., 2005] (Table 2.2).

Measured Published JB1A 0.704117 ± 7.7 0.704100 - 0.70412 BHVO2 0.703470 ± 8.1 0.703404 - 0.70370 JB3 0.703411 ± 7.9 0.703396 - 0.70351

Table 2.2: Comparison of measured and published 87Sr/86Sr values for rock standards [Jochum et al., 2005].

2.3.1.5 Principal component analysis

The Pearson coefficient matrix was calculated on the data set via a principal component analysis (PCA) to observed correlated geochemical variations and extract information on possible phase associations during formation of umber deposits. SiO2, TiO2, Al2O3, Fe2O3, MnO, MgO, CaO, K2O,

Na2O, P2O5, Co, Ni, Cu, Cr, V, Zn, Ba, Mo, Sc, Rb, Sr, Y, Zr, Nb, Th, U, La, Gd, Yb, ΣREY were taken into account for this statistical analysis.

57 Chapter 2. Geology and geochemistry of umbers

2.3.1.6 Isocon diagrams

Elemental fluxes and mass changes in umbers alteration facies were investigated using the immobility isocon diagrams method [Gresens, 1967, Grant, 1986]. This method predicts that elements identified as immobile during alteration, can be used to estimate gains and losses of other components through equations including chemical analyses and specific gravities of altered and unaltered specimens [Gresens, 1967, Humphris et al., 1998]. This relationship was later modified [Grant, 1986] to incorporate a mass change term based on immobile elements concentrations allowing abstraction of density such that:

A O A O Ci = M /M (Ci +ΔCi)

A O A where Ci = concentration of component i in the altered rock; M = mass of the protolith; M = O mass of the altered rock; Ci = concentration of component i in the original rock; and ΔCi = change in concentration of component i. Plausible immobile elements in umbers were identified using X-Y concentration plots for all unaltered and altered samples which are required to define a straight line passing by the origin of the plot with significant positive correlation (r > 80 %). In that view, ratios for immobile elements between unaltered and altered samples must remain constant. Indeed, this method follow the logic that during alteration or mineralization events, the concentrations of immobile elements will be either diluted by mass gain, getting closer to the origin as infinite dilution ultimately decreases the concentration to zero, or increase if mass is removed [Cail and Cline, 2001]. The immobility of one element has to be inferred by comparison

A O of all samples (altered and unaltered) when plotted against their least altered protolith (Ci vs Ci ). These plots must produce consistent behaviour for a similar type of alteration.

Although one immobile element is sufficient to calculate mass balance, immobility in intense alteration zones could result from a variety of processes such as the nature of the protolith and its heterogeneity, analytical errors and departure from ideal immobility of the assumed elements [Mukherjee and Gupta, 2008]. To minimize inherent errors to those parameters, the determination of the isochemical line, or line of no-mass transfer, must be based on as many geochemically dissimilar species as possible, though it might results in some scatter [Grant, 1986].

Furthermore, to avoid bias in isocon determination induced by the arbitrary scaling of elements to make them fit on a convenient diagram [Mukherjee and Gupta, 2008], the isocon diagrams in this study present data plotted at equal distance from the graph origin such that the sum of the square concentrations equals 1 [Humphris et al., 1998]. By evaluating consistency of a group of immobile elements, the average mass change term (MO/MA) is then used to correct for altered

A O concentrations (Cci ) to calculate elemental gains and losses. The choice of the protolith (Ci ) for

58 Chapter 2. Geology and geochemistry of umbers each altered sample was determined by spatial proximity of one of the unaltered sample rather than using an average composition of the unaltered samples as regional variability on umbers composition is observed around the Troodos massif. This would have resulted in statistical bias rather than real gains and losses observations.

2.3.2 Results

2.3.2.1 Mineralogy

Most results from our mineralogical investigation are consistent with previous observations [Constantinou and Govett, 1972, Robertson and Hudson, 1972, Robertson and Fleet, 1976, Robertson, 1977, Boyle, 1990].

Neither optical or X-ray studies show evidence of a discrete crystalline Mn phases in umber samples. On all XRD spectra, amorphous Fe-Mn oxides develop a strong background noise (Figure 2.10). The predominant umber mineralogy is pseudo-isotropic but includes poorly crystallized goethite and Mn oxides. Boyle (1990) inferred that Mn oxides in umbers might be present as δ-

4+ MnO2 (birnessite) based on similar Mn /Mntotal ratio and concomitant XRD peak with goethite potentially masking its identification. Minor mineral such as quartz and aluminum-silicate phases identified as part of the smectite group (montmorillonite and saponite [Boyle, 1990]) are identified by XRD. Investigation on sample 74 with a high calcium and phosphorus concentration

(14.9 % CaO and 9.9 % P2O5) indicates the additional presence of hydroxyapatite (Figure 2.10) in good agreement with Robertson and Fleet (1976). X-ray diffraction of a dark grey metallic sample (sample 63) recovered at the basalt / Mn-depleted umber interface near Kampia match the nodular pyrolusite identified by Boyle (1984). Finally, a sample of radiolarian chert (Sample 104) which overlies the umbers is dominated by an association of SiO2 polymorphs; quartz and trydimite, with presence of zeolite identified as phillipsite. No cristobalite was identified in our analysis but disordered cristobalite and opal C-T were previously detected [Robertson, 1977].

2.3.2.2 Scanning electron microscopy

The detailed structure of umbers matrix was investigated by SEM whereby a full scan of a carbon coated thin section (Figure 2.11) highlighted the presence of millimetric veins commonly running perpendicular to bedding with a small-scale meandering pattern, also referred to as ptygmatical folding, with fold axis lying within the plane of umbers bedding [Boyle, 1984]. Multiple aggregates and larger grains, either black or white on the electron backscattered image, emerge from the otherwise very fine grained matrix.

59 Chapter 2. Geology and geochemistry of umbers

Figure 2.10: X-ray diffraction patterns of apatite-rich umber, pyrolusite concretion and radiolarian cherts from the Perapedhi Formation.

Energy-dispersive X-ray spectroscopy (EDS) for elemental mapping was produced on an area of 4.5 * 7.5 mm of the thin section, selected after the visual observations of layers on the SEM imaging and aggregates of distinct morphologies. Elemental maps were produced for Fe, Mn, Si, Al, Ca, P, Cl, Mg, V, Na, Cu and K (Figure 2.12) and allow the following observations: (i) An important grain control is observed for Fe and Mn distribution in the slide. Although the matrix is amorphous, this distribution may represent amalgamation of similar oxides during deposition in amorphous state resulting

60 Chapter 2. Geology and geochemistry of umbers

in this granular texture. Umber’s matrix therefore comprises fine grained homogenous, amorphous Fe and Mn oxide [Robertson, 1975b] with larger Fe concretions (up to 0.5 mm) of goethite. No discrete Mn-rich grains, identified as equivalent of micro-nodules in oceanic sediments [Boyle, 1990] were found in the analysed thin sections. (ii) Veins are clearly visible, containing most of the Si, Mg, Al and Na. Their composition was investigated by EDS and relates these veins to the broad family of the smectite group and probably to Fe-rich palygorskite (Table 2.3). No Mn-rich veins were identified in the sample, though the assumption that Mn could be remobilised during diagenesis would produce such features [Robertson, 1975a, Robertson, 1976, Richards and Boyle, 1986]. (iii) Identification of discrete Ca-P-rich grains confirm the presence of discrete apatite grains which could either be fish bones debris, phosphatized ostracods or the results of early diagenetic process [Robertson and Fleet, 1976, Boyle, 1990, Ravizza et al., 1999]. (iv) The concomitant enrichment of some grains in Al and K might represent the clay detrital component included during oxide deposition.

The horizontal layered structure visible on hand sample and thin section is not observed on these elemental maps, however the element profile produced at higher magnification displays some variations (Figure 2.11). A negative correlation is observed between variations of MnO and Fe2O3 with small scale peaks that suggest changes in the composition of the matrix of ± 1 % MnO and ±

3 % Fe2O3. Discrete variations in the image of the thin section can be linked to changes in the Fe/Mn ratio and consequently represent a compositional layering within the section. Absence of such variations on the larger scale elemental mapping is probably due to detection limit of the SEM on a larger field of view.

Spot X-ray analysis for the mineralogical phases identified in thin section confirm the composition of both apatite grains and Si-rich veins and describe the composition of different aggregates in the matrix; either Si- (black bleb) or Fe- (white bleb) rich (Figure 2.14). The so-called “matrix area” in this figure corresponds to a fine grained zone with homogenous repartition of light and dark grains. When compared with the composition of the blebs and grains, this spectrum suggest that Mn is mainly located in the very fine grained portion of the matrix material while iron and silica can form discrete concretions. It is not excluded that Si-rich grains could correspond to detrital quartz grain incorporated during Fe and Mn oxyhydroxides deposit, though no magnification on the SEM enabled direct observation of any characteristic and recognizable crystal structure.

61 Chapter 2. Geology and geochemistry of umbers

profile considering the following oxides: Fe2O3, SiO2, MnO, Al2O3, MgO, CaO and CoO (Figure 2.11Erreur ! Source du renvoi introuvable.).

PJ-CY-2014-36 Palygorskite Analysis 1 Analysis 2 (a) n=10 (b) 160-3 (b) 160-4 Na2O 0.13 0.00 0.20 0.23 0.24 MgO 8.47 9.09 8.70 10.64 9.96 Al2O3 5.82 6.20 12.20 7.94 7.84 SiO2 52.40 55.76 54.00 63.31 59.73 CaO 0.56 0.62 1.10 0.19 0.2 MnO 0.37 0.19 0.50 0.11 0.11 FeO 10.00 8.76 2.15 7.99 7.97 H2O 18.50 Total 77.76 80.62 97.35 90.41 86.05 Table 2.3: SEM quantitative analysis on veins contained in umbers from sample PJ-CY-2014-36 compared with palygorskite data from (a) Newman (1987) and (b) Boyle (1984).

62 Chapter 2. Geology and geochemistry of umbers

Figure 2.12: X-ray element map by SEM of sample PJ-CY-2014-36. In these maps, the lighter shades indicate higher concentrations of the considered element.

63 Chapter 2. Geology and geochemistry of umbers

. Each square on the thin section section thin onthe square Each .

(± 2σ) (±

ray for 3 min. 3 for ray

2014

area scanned by by X scanned area

represents the the represents

ray quantitative analysis on sample PJ onsample analysis quantitative ray

Geochemical profile by X by profile Geochemical

: :

Figure Figure

64 Chapter 2. Geology and geochemistry of umbers

2014

matrix of sample PJ sample of matrix

ray spectrum for geochemical identification of the various types and morphologies of grain identified by SEM in the the in SEM by identified grain of morphologies and types various the of identification geochemical for spectrum ray

: localised X :localised

2 Figure Figure

65 Chapter 2. Geology and geochemistry of umbers

2.3.2.3 Geochemical characterization of umbers and associated lithologies

Fieldwork established three principal umberiferous lithologies: massive unaltered umber; a bright-orange layer typically developed at the basalt-umber interface; and silicified umber. Clay- rich umbers were not considered here as an independent facies as these sediments record the evolution from massive umbers to the overlying clays and radiolarian cherts and are therefore transitional. Plotting whole rock geochemical data of samples recovered during fieldwork on a Fe- Mn-Si ternary diagram actually provide a good discrimination of these three lithologies (Figure 2.15). Complete geochemical data are presented in Appendix A.

Figure 2.15: Fe-Si-Mn ternary diagram displaying the three main umberiferous lithologies encountered in the Troodos Ophiolite.

2.3.2.4 Massive umber geochemistry

2.3.2.4.1 Whole rock geochemistry

The composition of the massive umber (n = 59) is dominated by Fe and Mn oxides with significant content in total iron, 52.8 ± 1.9 % Fe2O3 and 9.4 ± 0.7 % MnO (up to 19.5 %). These elements are present in relatively constant proportions with Mn/Fe ratios typically between 0.1 and 0.5.

Together with SiO2 (16.6 ± 1.2 %), these elements dominate umber’s geochemistry; 69 to 94 % of their total composition in good agreement with previously published data [Robertson and Fleet, 1976, Boyle, 1990, Ravizza et al., 1999] (Table 2.4). The remaining fraction comprises variable amounts of CaO and P2O5 up to 14.9 % and 9.9 % respectively, showing that massive umbers were subject to various stages of phosphatisation. On that basis, a sub-group of umbers is created and referred to as apatite-rich umbers (n = 11). Dissociating samples containing an important Ca and P

66 Chapter 2. Geology and geochemistry of umbers content from other unaltered samples of umbers explain the discrepancies observed between our data and published average values for umbers [Boyle, 1984]. Elements associated with the aluminosilicate fraction (Al2O3, TiO2, MgO, K2O) are present in relatively constant concentrations 6.3 ± 0.5 % cumulated concentrations. The base metal concentrations of unaltered umbers decrease in the following order of abundance: 736 ± 216 ppm Cu, 310 ± 54 ppm Zn, 220 ± 57 ppm Ni and 150 ± 37 ppm Co. Noticeably high concentrations of V (964 ± 220 ppm), Ba (1080 ± 553 ppm) and Sr (1038 ± 626 ppm) are encountered in unaltered umbers. The radioactive content is low and remains below 10 ppm of cumulated U and Th.

The low Fe/Mn ratios, low transition metal concentrations (Cu, Ni, Co) and trace elements concentration suggest a hydrothermal origin for umbers deposits in contrast to other oceanic ferromanganese deposits (hydrogenetic crusts and polymetallic nodules) which show much higher Mn, Cu, Ni and Co contents. Classification of umbers within a ternary diagram for ferromanganese mineralization [Bonatti et al., 1972] positions umbers in the hydrothermal field with some hydrogenetic influence (Figure 2.16). A more recent classification scheme using REE and HFSE data [Josso et al., 2016] further highlights the influence of a seawater-derived component within umber geochemistry as samples range from the hydrothermal field towards the hydrogenetic in good agreement with [Boyle, 1984, Boyle, 1990]

Figure 2.16: Ternary discriminative diagrams for the genetic classification of oceanic ferromanganese deposits [Bonatti et al., 1972, Josso et al., 2016].

67 Chapter 2. Geology and geochemistry of umbers

Published This study data on Supra lava Apatite-rich Silicified Mn-dep. Supra lava Umber Umber Umber Umber Umber n = 59 n = 11 n = 8 n = 8

SiO2 19.3 a 17.5 ± 4.5 14.3 ± 3.6 80.5 ± 8.5 23.2 ± 9.6

TiO2 0.2 a 0.15 ± 0.05 0.11 ± 0.03 0.03 ± 0.01 0.16 ± 0.05

Al2O3 4.1 a 3.7 ± 1.3 2.6 ± 1.1 0.5 ± 0.3 4.1 ± 1.3

Fe2O3 44.2 a 52.1 ± 7.4 49.9 ± 5.8 12.4 ± 5.8 59.0 ± 7.5 MnO 9.1 a 10.1 ± 2.6 8.2 ± 2.2 1.4 ± 1.1 1.4 ± 0.4 MgO 1.7 a 1.8 ± 0.5 1.5 ± 0.4 0.3 ± 0.2 1.7 ± 0.5 CaO 2.5 a 1.7 ± 0.5 8.0 ± 3.5 1.3 ± 0.8 1.2 ± 0.7 K2O 0.8 a 0.6 ± 0.3 0.4 ± 0.3 0.1 ± 0.0 0.7 ± 0.6

P2O5 1.4 a 0.8 ± 0.3 4.8 ± 2.4 0.6 ± 0.6 0.7 ± 0.2

Co 125 a 150 ± 37 149 ± 36 58 ± 62 209 ± 49 Ni 254 a 220 ± 57 186 ± 42 63 ± 36 223 ± 70 Cu 846 a 736 ± 216 610 ± 69 130 ± 86 367 ± 144 Cr 42 a 20 ± 9 19 ± 8 46 ± 54 128 ± 29 V 1106 a 964 ± 220 976 ± 171 348 ± 288 1626 ± 856 Mo 38 ± 20 32 ± 19 16 ± 13 21 ± 12 As 366 ± 84 372 ± 69 74 ± 67 249 ± 101 Zn 361 a 310 ± 54 246 ± 35 98 ± 63 319 ± 114 Ba 1071 a 1080 ± 553 946 ± 283 219 ± 196 42 ± 12 Rb 22 a 21 ± 9 13 ± 4 4 ± 1 26 ± 15 Sr 1044 a 1038 ± 626 1171 ± 364 188 ± 210 49 ± 18 Y 92 a 98 ± 24 201 ± 100 28 ± 21 85 ± 16 Zr 183 a 90 ± 23 62 ± 10 14 ± 11 75 ± 10 Pb 265 a 179 ± 69 116 ± 26 35 ± 34 108 ± 50 U 2.4 ± 1 3.5 ± 1.3 1.4 ± 0.9 1.6 ± 0.8 La 102.5 b, c 106 ± 29 198 ± 91 29.0 ± 23 67 ± 18 Ce 28.6 b, c 41 ± 18 24 ± 7 6.5 ± 5.2 27 ± 13 Pr 21.5 b, c 27 ± 12 43 ± 21 5.9 ± 4.9 18 ± 4 Nd 89.9 b, c 103 ± 30 174 ± 90 23.6 ± 20 75 ± 15 Sm 18.8 b, c 22 ± 6 33 ± 16 4.8 ± 4.0 16 ± 2.9 Eu 4.7 b, c 5.7 ± 2.6 8.3 ± 3.8 1.2 ± 1.0 4.2 ± 0.8 Gd 19.0 b, c 23 ± 6 38 ± 19 5.3 ± 4.3 18 ± 3.1 Tb 2.8 b, c 3.7 ± 2.4 5.5 ± 2.4 0.8 ± 0.6 2.7 ± 0.5 Dy 16.9 b, c 20 ± 5 33 ± 14 4.6 ± 3.6 16 ± 2.7 Ho 3.3 b, c 4.1 ± 1.3 6.9 ± 3.0 1.0 ± 0.7 3.3 ± 0.6 Er 9.1 b, c 10 ± 3 19 ± 8 2.6 ± 2.0 9.1 ± 1.5 Tm 1.3 b, c 1.6 ± 1.2 2.6 ± 1.0 0.4 ± 0.3 1.3 ± 0.2 Yb 7.8 b, c 9.0 ± 2.1 16 ± 5.8 2.3 ± 1.7 8.4 ± 1.4 Lu 1.1 b, c 1.3 ± 0.2 2.2 ± 0.8 0.3 ± 0.2 1.2 ± 0.2

68 Chapter 2. Geology and geochemistry of umbers

2.3.2.4.2 Statistical analysis

The Pearson coefficient correlation matrix (Table 2.5) highlights element co-variation in the massive umbers (not including apatite-rich umber, n = 59). In addition, the principal component analysis defines 2 dominant axes of co-variations representing respectively 39.4 and 19.2 % of the variance in the dataset (Figure 2.17). The first axis displays the opposition of the aluminosilicate fraction and the Fe component. Si, Al, Ti, Mg, K, Sc, Rb, Cr, Zr and Th co-vary (0.7 < r < 0.98) and are negatively correlated to Fe2O3 (r < -0.7) which is solely positively correlated with V (r = 0.73).

The excellent covariance between Al2O3 with TiO2 (r = 0.98) reflects the clastic inputs within umber deposits. The sources for these elements are multiple considering externally derived bentonitic silty clays progressively diluting the metalliferous sediments towards the top of the sequence and the potential influence of volcanic fragments within the basal layers. The second axis of covariations best represent changes in CaO, P2O5, ΣREY, La, Gd and Yb concentrations independently from variations in the aluminosilicate phase or the Fe oxide phase. Despite not including apatite-rich samples from this statistical analysis, CaO and P2O5 co-vary strongly (r = 0.9).

No covariation is observed between the REY and CaO-P2O5. However, including apatite-rich samples in the principal component analysis (n = 70) display the strong association of REY and apatite. This observation therefore allows the distinction between two modes of REY incorporation within the deposit, either with the apatite or the oxide phase. A week covariance is observed between TiO2 and ΣREY (r = 0.62), but this association is not significant for any individual

REY with TiO2. Mn, Cu, Ni, Co, Ba and Sr do not show any statistically significant co-variation with other elements included in this analysis.

Chapter 2. Geology and geochemistry of umbers

0.88

0.94

0.96

0.88

0.94

-0.11

-0.48

-0.40

-0.29

0.58

0.40

0.44

0.51

0.00

0.74

0.53

0.64

0.69

0.00

0.73

0.88

0.79

0.81

0.79

0.01

0.40

0.46

0.15

0.12

0.21

0.28

0.00

0.05

-0.09

-0.30

0.54

0.36

0.40

0.47

0.03

0.97

0.69

0.35

-0.27

0.52

0.40

0.43

0.44

0.79

0.65

0.33

0.84

-0.02

-0.20

0.42

0.29

0.43

0.46

0.06

0.44

0.31

0.48

0.07

0.08

-0.03

0.05

0.14

0.06

-0.03

-0.02

-0.06

-0.04

-0.10

-0.02

-0.03

-0.20

massive umber samples (n = 59). = (n samples umber massive

0.48

0.30

0.34

0.37

0.40

0.67

0.43

0.36

0.33

0.35

-0.05

-0.10

0.20

-0.30

-0.12

-0.19

-0.28

-0.03

-0.67

-0.48

-0.23

-0.64

-0.48

-0.13

-0.09

-0.18

0.02

0.45

0.16

0.52

0.47

0.04

-0.03

-0.05

-0.09

-0.05

-0.14

-0.34

-0.28

-0.11

-0.20

0.06

0.07

0.02

0.03

0.00

0.06

0.02

0.12

0.35

0.24

0.00

0.26

-0.02

-0.10

-0.24

-0.02

0.40

0.28

0.27

0.18

0.53

0.43

0.42

0.48

0.52

0.16

0.54

0.07

0.15

-0.34

-0.03

-0.28

0.07

0.51

0.44

0.25

0.12

-0.20

-0.06

-0.14

-0.18

-0.05

-0.37

-0.35

-0.15

-0.35

-0.20

-0.41

-0.12

-0.15

0.08

0.19

0.11

0.12

0.31

0.28

0.03

0.30

0.17

-0.05

-0.35

-0.36

-0.37

-0.39

-0.01

-0.13

-0.33

-0.11

-0.14

0.32

0.26

0.28

0.41

0.35

0.33

0.34

0.36

0.09

0.05

0.17

0.30

-0.07

-0.27

-0.40

-0.08

-0.03

-0.19

-0.11

0.61

0.43

0.47

0.55

0.01

0.90

0.68

0.42

0.93

0.89

0.13

0.11

0.32

0.42

0.22

0.54

0.31

-0.09

-0.59

-0.27

-0.34

Pearson coefficient correlation matrix for unaltered unaltered for matrix correlation coefficient Pearson

0.17

0.26

0.21

0.23

0.30

0.46

0.17

0.02

0.27

0.11

0.90

CaO

-0.12

-0.28

-0.23

-0.27

-0.21

-0.10

-0.23

-0.05

-0.18

-0.11

-0.18

: :

0.39

0.26

0.35

0.32

0.57

0.45

0.28

0.63

0.72

0.28

0.18

0.30

0.04

0.28

0.16

0.66

-0.05

-0.02

-0.04

-0.43

-0.39

-0.22

-0.01

MgO

Table Table

0.05

0.01

0.03

0.16

0.03

0.10

0.09

0.59

0.13

0.42

0.19

0.28

0.19

-0.02

-0.18

-0.29

-0.07

-0.37

-0.25

-0.16

-0.24

-0.03

MnO

0.15

0.32

0.73

0.41

0.30

0.19

-0.53

-0.42

-0.46

-0.48

-0.81

-0.61

-0.40

-0.81

-0.76

-0.13

-0.07

-0.18

-0.30

-0.11

-0.44

-0.51

-0.80

-0.68

-0.30

0.58

0.39

0.46

0.50

0.02

0.88

0.70

0.37

0.92

0.94

0.16

0.03

0.34

0.48

0.27

0.46

0.28

0.96

0.74

0.09

-0.12

-0.58

-0.32

-0.41

-0.22

-0.79

0.62

0.42

0.49

0.56

0.02

0.91

0.74

0.40

0.94

0.91

0.17

0.01

0.37

0.48

0.23

0.44

0.31

0.96

0.69

0.11

0.99

TiO

-0.12

-0.62

-0.37

-0.38

-0.21

-0.79

0.34

0.22

0.26

0.28

0.00

0.74

0.49

0.16

0.82

0.08

0.15

0.68

0.19

0.27

0.11

0.79

0.83

0.86

0.83

SiO

-0.19

-0.16

-0.42

-0.38

-0.37

-0.16

-0.19

-0.67

REY

CaO

SiO

TiO

MgO

MnO

Na Al Fe

70 Chapter 2. Geology and geochemistry of umbers

2.3.2.4.3 Rare earth elements

The total REE content of the massive umbers averages 377 ± 95 ppm; excluding apatite-rich samples 5, 16, 24, 29, 41, 68, 70 and 74 where presence of apatite results in significant REE enrichment (550 ± 181 ppm including up to 1390 ppm in sample 74). The massive umbers display homogenous upward convex REE trends when normalized to PAAS values [Taylor and McLennan,

1985] with LaN/SmN mostly bracketed between 0.56 and 0.95 though some normalized trends exhibit ratios up to 1.17 correlating with the light REE enrichment of apatite. GdN/YbN ratios remain constantly > 1 (1.13 – 1.86). Pronounced negative Ce* anomalies (Ce/Ce* = CeN / √

(LaN*PrN) = 0.02 – 0.31) and slight positive Eu* anomalies (Eu/Eu* = EuN / √ (SmN*GdN) = 1.07 – 1.19) are observable in normalized trends of umbers (Figure 2.18). These results display a strong homogeneity in umbers REE signature from all samples location across the Troodos massif in good agreement with data previously obtained by neutron activation analysis [Robertson and Fleet, 1976] and ICP-MS [Ravizza et al., 1999]. Mn-depleted umber fall within the range of REE concentration of unaltered umber with similar trends of fractionation. They are in average less enriched (267 ± 57 ppm) and present more developed negative Ce anomalies. Silicified umbers display similar trends to unaltered umbers (except sample 54) with various degree of dilution. These signatures are discussed in more details later.

Figure 2.18: PAAS-normalized REE plot for unaltered umbers, Mn-depleted umbers and silicified umbers (PAAS values from Taylor and McLennan (1985)).

71 Chapter 2. Geology and geochemistry of umbers

2.3.2.5 Alteration of Umber

The umbers typically show two distinct alteration facies, Mn-depleted and silicified. Their geochemistry compared to unaltered umbers is described in detail below.

2.3.2.5.1 Mn-depleted umbers

The orange layer commonly found at the base of umbers outcrops present by comparison with unaltered umbers higher SiO2 (23.2 ± 9.6 %) and Fe2O3 (59.0 ± 7.5 %) content, compensating for their depletion in MnO (1.4 ± 0.4 %) on bulk analysis (Table 2.4). However, this apparent enrichment might be relative to mass loss related to post-depositional migration of other element (Mn) therefore impacting normalization values during measurement by XRF. Calculating mass changes and elemental fluxes during the development of Mn-depleted umbers is achieved using isocon diagrams [Gresens, 1967, Grant, 1986]. Immobile elements in umbers were identified using X-Y concentration plots for all unaltered and altered samples (Figure 2.19).

Figure 2.19: Immobile elements 2D plots for unaltered umbers, silicified umbers and Mn-depleted umbers. Legend similar to Figure 2.16.

72 Chapter 2. Geology and geochemistry of umbers

In Mn-depleted umbers, Al2O3, TiO2, Sc ± Nb are immobile and define an isochemical line, or line of no-mass-transfer. The mass change term (MO/MA) for Mn-depleted samples show a range of element losses and gains ranging from a 55 % mass loss (sample 86) to a 31 % gain (sample 91). The Mn-depleted umbers are consistently depleted in MgO, Sr, Ba, Cu, Ni, Co, REE and U with gains in Fe2O3, SiO2, V, and Cr (Figure 2.20). Remaining elements possess variable behavior showing depletion and/or enrichment depending on samples. Anomalous gains in CaO, P2O5 and REE are related to localized higher concentration of apatite in the Mn-depleted sample compared to the unaltered sample.

Redox sensitive species show different fractionation patterns with Cr and V are consistently strongly enriched, Mo oscillates between enrichment and depletion depending on the sample, whereas Mn is always depleted. Mn removal from umbers implies Mn transportation and passage into solution under reducing conditions [Robertson, 1976], such that:

2+ - Mn(OH)2 (s. am.)  Mn (aq.) + 2OH

The absence of remobilization of Fe in this facies defines a window in Eh-pH space (Figure 2.21), characterizing sediments conditions during this remobilization. Mo, V and Cr are oxyanions scavenged by Fe-Mn oxyhydroxides during deposition of hydrothermal particles in an oxic environment. Although their budget in umbers is potentially influenced by the presence of volcanoclastic material of the substratum in the basal layer, the consistent element concentration in most massive umber samples points toward a dominant hydrogenetic source. During diagenesis, these elements are sensitive to changes in the oxidation state of sediments pore waters [Schaller et al., 2000]. The breakdown of Mn oxyhydroxides releases adsorbed species that can utilize available hydroxyl groups to oxidize. The observed positive anomalies in Cr, V and ± Mo describes the differential susceptibility of these species to become oxidized and fixed into the sediments during redox events and pH changes. Considering redox potential (Eh) and pH conditions, most transformations will occur as a function of OH- availability. Considering the range in Eh-pH conditions constrained by the absence of Fe remobilization, released chromium can readily form amorphous oxides stable in pH > 5 thus maintaining its concentration in the Mn- depleted umber.

3+ - Cr + 3OH  Cr(OH)3 (s. am.),

Correlation between the Cr enrichment and mass loss in the Mn-depleted umber highlights a process of relative enrichment and no influence of external inputs; samples without mass changes (MO/MA = 1) show conservative concentration otherwise.

73 Chapter 2. Geology and geochemistry of umbers

Figure 2.20: Examples of isocon diagrams plotting concentration of Mn- depleted umbers sample 88 (upper) and 98 (lower) rescaled against unaltered umbers for the analysis of element mobility during alteration. The grey line, determined by immobile elements, forms the line of no mass transfer; elements falling above constitute elemental gains while the zone of elemental loss is under the isochemical line. The concentration data have been normalized so that the sum of squares equal 1 [Humphris et al., 1998] avoiding errors in graphic interpretation due to arbitrary scaling

74 Chapter 2. Geology and geochemistry of umbers

Alternatively, vanadium does not form a single stable hydroxide and experiences a series of reducing reactions depending on the range of pH forming vanadate (IV) oxyhydroxides such as

2- - - VO3OH , VO2(OH)2 and (VO)2(OH)5 which show increasing stability with decreasing pH respectively [Takeno, 2005]. The Mn/V ratio in Mn-depleted umbers ranges from 1.8 – 15.5 (both element expressed in wt. %) whereas the average umber has Mn/V ratio of 67.1 typical of more oxic conditions. Consequently, V will tend to oxidize and remain in the facies showing strong relative enrichment.

Mn oxyhydroxides constitute the most effective scavenger for Mo by adsorption of molybdates

2- (MoO4 ), a stable Mo complex under a large range of pH and Eh in oceanic conditions [Goldberg et al., 2012]. Released molybdates during diagenetic dissolution of Mn oxides are likely to behave similarly to Mn2+ and be remobilized to more oxidizing conditions where they will be fixed by co- precipitation with Mn-oxides. However, if other electron receptors are available, scavenging of molybdates will be challenged by Fe-oxyhydroxides [Goldberg et al., 2012] which may explain the erratic behavior of Mo during isocon studies of the Mn-depleted umbers.

The pyrolusite concretion (sample 63) found in Kampia at the contact between the basaltic basement and Mn-depleted umber displays a contrasting pattern with enrichment in MnO (84.4 %), Ba (13370 ppm), Sr (2252 ppm), Cr (132 ppm), Mo (34 ppm) and V (834 ppm) when compared to the closest sample of unaltered umber (Sample 89) with 11.3 % MnO, 1428 ppm Ba, 312 ppm Sr, 18 ppm Cr, 17 ppm Mo and 753 ppm of V. These results suggest that the pyrolusite sample forms the end-member of the remobilization that originally formed the Mn-depleted umbers, concentrating remobilized Mn and associated element in a concretion where more oxidizing conditions were found.

Diagenetic remobilization, whereby interstitial waters become more acidic, may explain the post- depositional migration of Mn, Ba, Sr and other redox sensitive species in the basal layer of some umber outcrops. The presence of pyrolusite concretions at the margins of Mn-depleted zones suggests that this alteration formed preferentially through a lateral flow with the basalt/umber contact offering a preferred pathway, rather than an upward flow, even though discrete subvertical pyrolusite veins were observed in other localities. The origin of the acidification however remains uncertain. Mn-depleted umber could have formed through (i) a redox event with acidity being released from the alteration of underlying basalt or locally deposited sulphides, (ii) later hydrothermal circulation, (iii) and/or redox changes of pore fluids occurring in the sedimentary pile as it develops [Elderfield et al., 1972, Robertson, 1976, Boyle, 1984].

75 Chapter 2. Geology and geochemistry of umbers

Figure 2.21: pH vs Eh diagram presenting stability field for Fe and Mn as oxides and oxyhydroxides. Mn species left of the bold line are in the ionic form Mn2+, the grey area represent the Eh-pH window for diagenetic remobilisation in Mn- depleted umbers. The two diagonals long dotted lines define the stability field of water (298.15 K, 105 Pa) (modified from Takeno (2005)).

2.3.2.5.2 Silicified umbers

Umbers show variable degrees of silicification including localised and discrete layers, nodules, massive bulbous bodies and veins which suggests variation in the processes responsible for and timing of silicifcation. The strong SiO2 enrichment (64.1 – 90.5 % SiO2) makes the identification of immobile elements through conservative behavior impracticable as the high SiO2 content extensively dilutes all other components and therefore significantly increases error margins.

Nevertheless, Al2O3 and Sc exhibit similar mass changes and can be considered as immobile to calculate the mass gain (190 to 2679 %) and mass change terms (MO/MA = 2.9 – 26.3). However, estimation of gains and losses of other elements through isocon diagrams are not reliable due to the SiO2 dilution which induce a dense pack of elements on a graphic representation in the depletion area. Consequently, elements mobility and the different silicification types are studied after a two-stage normalization process. Firstly, composition of silicified samples are normalized to the average SiO2 ± 2σ concentration of unaltered umbers to correct the dilution induced by the silicification. The value of any element X is reclaculated as follows;

[X]N = [X]I + ([SiO2]I – avg. [SiO2]US)*([X]I / (Sum ([X]I;[Y]I;…;[Z]I) – [SiO2]I)

76 Chapter 2. Geology and geochemistry of umbers

Where the subscript “N” represents the normalized value of any element X, subscript “I” is the initial concentration in the silicified sample, and “avg. [SiO2]US” is the average SiO2 concentration of unaltered samples (n = 59). Once every element is recalculated on average SiO2 ± 2σ concentrations, samples are then normalized to the average concentration of unaltered umber. This process allows direct comparison of silicified samples with an unaltered protolith to study element mobility during silicification events. The resulting trends are grouped in 3 categories (Figure 2.22).

The first group (Figure 2.22 A) contains sample 7 (silicified nodule) and 8 (bulbouous mass) as visible on photographs E and F from Figure 2.3 recovered near Margi. Sample 9 taken in the same bulbous mass display the same trend as sample 8 and is consequently not represented. These samples display relatively close fractionation trends to the unaltered umbers once silicification dilution is corrected. Relative enrichment in CaO and P2O5 in sample 7 highlights that phosphatisation and formation of apatite in umbers occurred prior to silicification.

Group B includes samples 56 and 99 from Kampia and Perapedhi respectively, which were collected from discret silicified beds in umber outcrops (Figure 2.22 B). These samples are characterised by depletion in MnO, Ba and Sr and enrichment in Cr, Mo and As relative to unaltered umbers characteristic of Mn-depleted umbers. The preservation of this signature in silicified samples suggests that the silicification post-dates the formation of Mn-depleted umbers.

Samples from the group C were also collected in the Kampia area, although they sampled a massive silicified outcrop directly overlying a fault zone south of the Teothokos Monastery (Figure 2.22 C). These three samples possess the highest mass change term (MO/MA = 22.3, 25.5 and 26.3 for samples 51, 50 and 56 respectively) forcing an extreme dilution of other elements and highlighted by depletion in Al2O3, Zr, and REE usually considered as immobile. The increase in the mass change term visibly effects the REE with greater susceptibility for the LREE to be removed. Positive anomalies in Ce and Eu relative to the fractionation of other lanthanides are possibly due to their different oxidation states. These samples also present positive anomalies in CaO and K2O not observed in other silicified samples without enrichment in P2O5 and thus not related to a mineralogical effect with the presence of apatite pre-silicification. Sample 54 differs from all the other silicified samples by depletion in Cr and Ni; usually strongly enriched for Cr (together with U), or showing conservative behavior for Ni, relative to unaltered umbers after correcting for the

SiO2 dilution. This last group of silicified umber stands out by the intensity of the silicification encountered and by its distinct signature which can’t be related to a simple dilution by addition of

SiO2 of either unaltered umber (sample 8), apatite-rich umber (sample 7), or Mn-depleted umbers

77 Chapter 2. Geology and geochemistry of umbers

(sample 56 and 99). These samples thus record a different type of silicification, probably hydrothermal in origin and favored by the proximity of the fault (Figure 2.5).

Figure 2.22: Silicified umbers recalculated on an average SiO2 concentration normalized to average unaltered umber composition.

As exemplified by the normalization of silicified umbers to average umber composition, at least two different types of silicification can be identified. Nodules, discrete beds and bulbous masses of silicified umbers, after compensating for SiO2 dilution, exhibit similar compositional trends to either massive umbers, apatite-rich umbers or Mn-depleted umbers. These results suggest the silicification dominantly produced a geochemical dilution of pre-existing umberiferous lithologies and is likely to be the result of the diagenetic remobilization of silica incorporated during deposition of umbers. It is generally established that biogenic opal or siliceous microfossils are the principal source of silica for the formation of cherts in sediments [Hein et al., 1981, Adachi et al.,

78 Chapter 2. Geology and geochemistry of umbers

1986] and the presence of radiolarian in umbers has been assessed and widely used as a means of dating the sedimentary sequence of the Troodos Ophiolite [Bragina and Bragin, 2006]. In that view, the radiolarian cherts overlying umbers could be considered a source of biogenic silica for the silicification of the metalliferous sediments. However detailed study of the matrix of these cherts highlights the preservation of radiolarian shells without signs of dissolution [Robertson, 1977] ruling out the biogenic source hypothesis. In addition, many cherts have formed in clay or zeolite-rich sediments which would be problematic for the remobilization of biogenic silica and therefore inadequate to explain their formation, suggesting rather potential devitrification of volcanic glasses [Robertson, 1977]. Umbers are dominated by hydrothermal over detrital or biogenic material and it is therefore necessary to consider hydrothermal amorphous silica colloids as a dominant component of the total silica budget.

By contrast, the silicified umbers observed near Theotokos Monastery in Kampia (Samples 50, 51 and 54), are quite likely to be the result of hydrothermal silicification with fluid circulation favored along the fault identified in this area. Distinct REE fractionation trends support this distinction (Figure 2.18). Detailed analysis of this silicification by Prichard and Maliotis (1998) further demonstrated the later, “off-axis”, origin of this mineralization probably of low temperaature. Although this kind of silicification is important considering its gold potential, it appears to be really localized to some areas and fault zones which remain minor at the scale of the Troodos massif.

2.3.2.6 Sr isotopes

87Sr/86Sr ratios, age corrected to 90 Ma for all analyzed samples, are listed in Table 2.6. Unaltered umbers group around an average value of 0.70767 ± 0.00005 which is significantly more radiogenic than 90 Ma Cretaceous seawater [Mc Arthur et al., 2001] and in agreement with previous data by Gale et al. (1981) (Figure 2.23). 87Sr/86Sr ratios for silicified umbers range from 0.70725 to 0.70784 whereas Mn-depleted umbers are more radiogenic with 87Sr/86Sr ratios of 0.70802 to 070906 much closer to the value of present day seawater. Radiolarian cherts of the Perapedhi Formation that stratigraphically cover umbers have 87Sr/86Sr isotopic ratios ranging from 0.70751 to 0.70843. The umbers and silicified umbers have a strontium signature intermediate between Cretaceous seawater and the hemi-pelagic, detrital sediments of the Perapedhi formation without any influence of hydrothermal signature. These more radiogenic ratios result from the proportion of Sr derived from the terrigenous fraction incorporated within umbers pre-silicification. These results corroborate the REE seawater-like signature observed in umbers and the process of enrichment of this formation in trace elements through extensive contact with seawater overprinting the original hydrothermal signature of the particles. The 87Sr/86Sr ratios observed in Mn-depleted umbers, more radiogenic than the Perapedhi formation,

79 Chapter 2. Geology and geochemistry of umbers cannot be explained by a similar mixing process between 90 Ma seawater signature and detrital particles in umbers. In a similar way, the influence of the substratum can be ruled out as fresh basaltic glasses of the Troodos extrusive suite have 87Sr/86Sr < 0.704 whereas altered basalts have isotopic signatures ranging from 0.704 to 0.707 [Bickle and Teagle, 1992]. The pyrolusite sample, the other end-member of the post-depositional remobilization of Mn, Sr and Ba possess 87Sr/86Sr ratio extremely close to Cretaceous seawater. These could highlight that most of the seawater- derived Sr incorporated in umbers was remobilized within the pyrolusite leaving the Mn-depleted umbers with a residual, more radiogenic pelagic signature. A contamination by recent seawater also constitutes a solution to explain the signature of this facies, whereas a hydrothermal fluid or contamination from the underlying altered basalt would have produced a much lower isotopic

87 86 signature. No significant correlation between Sr/ Sr signatures and any of the major (Al2O3,

Fe2O3, MnO, CaO, P2O5) or trace elements (Sr, Rb, Ba) is observed for all facies.

87 86 Sr/ Sr Sr Rb Ba TiO2 Al2O3 Fe2O3 MnO CaO P2O5 ppm ppm ppm wt. % wt. % wt. % wt. % wt. % wt. % Massive Umbers 14 0,707623 ± 7,4 1471 19,8 2238 0,17 4,3 56,5 13,3 1,4 0,6 46 0,707666 ± 8,4 313 14,4 814 0,11 2,5 50,4 14,7 1,7 1,0 65 0,707721 ± 8,2 642 12,5 851 0,11 2,8 52,1 10,4 1,6 0,6 74 0,707625 ± 7,6 628 9,7 736 0,08 2,0 49,6 6,2 14,9 9,9 90 0,707732 ± 7,7 303 31,3 1237 0,20 5,1 41,3 12,6 1,2 0,6 Silicified Umbers 7 0,707250 ± 7,8 414 2,5 339 0,04 0,5 12,4 2,4 1,5 0,8 8 0,707410 ± 8,2 462 3,6 377 0,04 0,8 15,6 2,9 0,5 0,2 9 0,707389 ± 7,7 496 3,0 447 0,04 0,7 14,3 2,9 1,6 0,8 50 0,707648 ± 6,5 45 5,0 478 0,02 0,1 11,8 2,0 2,2 0,1 51 0,707843 ± 9,6 13 3,6 65 0,02 0,1 5,9 0,6 0,2 0,1 56 0,707829 ± 7,5 34 6,4 12 0,06 1,1 24,3 0,5 1,4 0,9 99 0,707463 ± 8,0 36 1,8 9 0,03 0,4 10,9 0,2 2,7 1,9 Mn-depleted Umbers 12 0,708024 ± 8,1 62 18,2 37 0,18 4,9 64,8 0,7 1,7 1,1 53 0,709030 ± 9,6 40 20,5 47 0,14 3,6 63,1 1,6 0,6 0,6 57 0,708684 ± 8,8 58 16,1 37 0,11 3,1 60,1 1,3 0,7 0,6 87 0,708937 ± 7,4 28 37,0 49 0,14 3,6 58,9 2,0 0,4 0,5 88 0,709064 ± 8,5 40 21,0 49 0,15 3,7 63,2 1,5 0,6 0,6 98 0,708062 ± 8,6 88 11,2 20 0,12 2,8 68,0 1,5 2,4 0,9 Pyrolusite Concretion 63 0,707376 ± 10,1 2252 1,5 13370 0,04 0,5 1,4 84,4 0,8 0,2 63 (dpl) 0,707379 ± 7,7 Radioalrian Cherts from the Perapedhi Formation 80 0,707867 ± 9 368 34,5 65 0,28 7,2 5,1 0,7 2,1 0,3 80 (dpl) 0,707879 ± 8,1 81 0,707513 ± 7,8 81 15,6 38 0,12 2,8 1,9 0,5 0,8 0,1 104 0,707896 ± 8,1 171 33,8 64 0,27 6,1 4,5 0,2 1,0 0,1 104 (dpl) 0,707906 ± 8,6 105 0,708432 ± 8,1 103 32,3 63 0,21 5,1 3,6 0,3 0,7 0,1 Table 2.6: Sr isotope data for the different umber lithologies facies and sedimentary cover of the Perapedhi Formation of the Troodos Ophiolite (dpl = duplicate).

80 Chapter 2. Geology and geochemistry of umbers

Figure 2.23: 87Sr/86Sr isotopic ratio for unaltered umbers, silicified umbers, Mn- depleted umbers, pyrolusite concretion and sediments of the Perapedhi formation compared with data for late Cretaceous and Modern Seawater [Mc Arthur et al., 2001].

2.3.2.7 Stratigraphic profiles

Geochemical stratigraphic profiles for umbers have been reported already from Mangaleni, Drapia [Robertson, 1975b] Skourioutissa [Robertson, 1976] and Margi [Boyle, 1984] and traditionally show an upward evolution from hydrothermally derived sediments to more pelagic terrigenous sediment. Three umber sequences at Margi, Kampia and Asgata were sampled in detail to investigate geochemical variations through stratigraphic profiles and compare with previously published work. Locations of stratigraphic sequences from Kampia and Margi are presented on the geological maps (Figure 2.5 and 2.7). The sequence in Margi is composed of 23 samples (PJ-CY-2014-12 to 34) taken regularly over a 225 cm thick umber deposit. The outcrop evolves from a reddish basal layer in contact with the basaltic basement to unaltered umber in the central part of the sequence, with a gradational transition to clay-rich umbers in the upper 20- 30 cm. The sequence from Kampia includes nine samples (88 to 96) taken regularly along a 120 cm thick umber outcrop showing a similar lithologies succession. Two samples of bentonitic clays (samples 104 and 105) overlying the umber outcrop were sampled to characterize the background pelagic sediment and investigate if a remnant hydrothermal signal is present. The third stratigraphic sequence was recovered from Asgata which presents an important umber accumulation (> 4 m thickness) in a basaltic depression (Figure 2.2). This outcrop shows well defined beds, alternating massive layers of umber with more friable, thin layers of umberiferous material with a slate-like habit. Samples 64 to 75 were taken along the main exposure taking care to recover the various lithologies (n = 12). Geochemical profiles for selected elements on the stratigraphic sequences from Kampia and Margi are quite similar and compare well with previously reported profiles [Robertson, 1976, Boyle, 1984]. They display a progressive decrease

81 Chapter 2. Geology and geochemistry of umbers in Fe2O3 concentration upwards through the sequence in favor of increasing SiO2 (Figure 2.24 and Figure 2.25). MnO is constant through the central portion of the sequence but progressively decreases in the uppermost part of the profile as umbers become increasingly clay-rich with increasing Al2O3, MgO, K2O and TiO2. Trace metals (Cu, V, Co, Ni, Zn and Sr) tend to follow the iron contents in the sequence, showing greater decrease in abundance in the sequence from Kampia as Fe2O3 < 20% with greater detrital dilution. The basal layer is Fe-V-rich, Mn-depleted, with associated depletions in Sr, Ba, Cu, P2O5, CaO and REE while Cr and V are enriched relative to main umber. Locally, apatite-rich layers demonstrate covariation of CaO and P2O5 and display increased REY content. The two samples from the sedimentary cover in the stratigraphic sequence from

Kampia are typical of radiolarian cherts; SiO2 > 78 %, 5.1 – 6.1 % Al2O3, low metal content; Fe2O3 < 4.5 %, MnO < 0.3 % and concentrations of Co, Ni and Cu ranging from 150 to 365 ppm. This part of the profile clearly marks the transition into background sedimentation. The aluminum index, defined as 100*Al / (Al + Fe + Mn) [Boström and Peterson, 1968], for this stratigraphic sequence shows a trend from low ratios in the main part of the profile towards a more “pelagic” end- member at the top of the sequence. When the Al-index is lower than 40, such as in umbers, it suggests either an hydrothermal origin of the particles or later influence and alteration of a matrix by hydrothermal fluids [Boström, 1973]. The shift to an Al index ≈ 50 within the two radiolarites samples marks the transition from a largely hydrothermally-derived system towards a pelagic and detrital end-member. On the other hand, the aluminum index remains below 10 in the sequence from Margi showing less evidence of dilution by background sedimentation. This absence of marked trends relates probably to an incomplete sequence, eroded at the top.

By contrast, the sequence from Asgata does not show a clear evolution in major and trace elements trends from the base to top of the profile as observed in profiles from Kampia and Margi (Figure 2.26). Rather, the sequence depicts alternation between massive umber and finely layered umber. Layers of massive umber (0-175 cm, 250-350 cm and 375-425 cm, Figure 2.26) show increasing Fe2O3 concentrations from 52 to 67 % upward, homogenous SiO2 content between 12 and 16 %, decreasing MnO concentration from 10 to 5 % and constant concentration of Al2O3,

MgO, K2O, and Na2O. The intercalated smaller layers display an alternation of facies between finely layered, highly fissile umber and slightly silicified umber. The finely fissile umbers show greater concentrations of CaO and P2O5 (up to 14.9 % and 9.9 % respectively), that are balanced by lower Fe2O3 content (48 - 52 % instead of 54 - 67 %). The apatite identified by XRD in these layers controls the rare earth element and yttrium content with an enrichment factor 3 to 7 fold compared to the massive umbers of this sequence. The partially silicified umbers (20 % SiO2 compared to a rather constant 12 – 16 %) display enrichment in MgO relative to the massive umber and Cr values twice those of the massive umber.

82 Chapter 2. Geology and geochemistry of umbers

Figure 2.24: Geochemical profile for selected major and trace elements from Margi outcrop.

83 Chapter 2. Geology and geochemistry of umbers

Figure 2.25: Geochemical profile for selected major and trace elements from Kampia outcrop. On each plot, the horizontal dotted line indicates the separation between lithologies.

84 Chapter 2. Geology and geochemistry of umbers

Figure 2.26: Geochemical profile for selected major and trace elements from Asgata. On each plot, the horizontal dotted line indicates the separation between lithologies.

85 Chapter 2. Geology and geochemistry of umbers

The absence of trends in the geochemistry of this outcrop near Asgata combined with the presence of multiple layers rich in apatite suggests that multiple episodes of sedimentation of hydrothermal particles happened. The good correlation between the concentration of dissolved phosphate in ambient seawater and hydrothermal plume P/Fe ratios, together with transmission electron microscopy analysis of such particles [Feely et al., 1990, Feely et al., 1998], suggests that the original P2O5 content incorporated in hydrothermal sediments derives from the co- precipitation of seawater P2O5 with Fe oxyhydroxides rather than adsorption processes during hydrothermal plume dispersal [Edmonds and German, 2004]. During diagenesis, the recrystallization of ferrihydrite into goethite induces the release of scavenged P2O5 in the sediments due to the partial dissolution of the Fe amorphous phase before nucleation and recrystallization [Poulton and Canfield, 2006]. This P2O5 fraction is then remobilized towards the sediment-seawater interface and eventually released into the water column to be ultimately reincorporated onto newly depositing Fe-oxyhydroxides. As the availability in Fe-oxyhydroxides decreases, the formation of authigenic apatite becomes the next major sink for P2O5 fixation. Such remobilization has been documented in continental margin areas and deep sediments [Ruttenberg and Berner, 1993, Eijsink et al., 2000, Van der Zee et al., 2002], however significant phosphatized horizons like those observed in umbers from Asgata, Cyprus remain quite rare in hydrothermally-derived sediments [Poulton and Canfield, 2006]. Ultimately the phosphate rich horizons in umbers may represent periods of quiescent hydrothermal activity enabling authigenic mineral formation, as well as biogenic apatite accumulation from fish debris [Robertson and Hudson, 1972, Boyle, 1990]. However, the detrital fraction incorporated during these intervals should correlate and be proportionally higher, which is not observed as Al2O3 and TiO2 remain fairly constant over the stratigraphic sequence from Asgata. Given the above, the presence of multiple apatite-rich layers intercalated with massive beds of unaltered umbers most likely describes the remobilization of P2O5 during early diagenesis within the deposit. Still, the reason why Ca and P remobilized to form such layers remains uncertain. As a consequence these results also invalidate the use of umbers of the Troodos Ophiolite as a potential record of paleo seawater phosphate concentrations. The absence of evolution in the sequence of the hydrothermal/detrital signal towards the top of the sequence suggests that the initial outcrop was initially much thicker than what has been preserved up to now. Deposits as thick as 7 meters were observed in some location near Kampia in association with lava depressions which constitute a low estimate considering no decrease in the hydrothermal signal is observed in the geochemical profile.

86 Chapter 2. Geology and geochemistry of umbers

2.4 Discussion

2.4.1 Comparison of umbers with other metalliferous sediments

2.4.1.1 Major elements

Given similar geological settings, mineralogy and association with massive sulphide, the metalliferous sediments from the EPR and TAG constitute some of the best potential modern analogues for umbers of the Troodos Ophiolite [Robertson, 1975a, Robertson and Boyle, 1983,

Boyle, 1990]. When plotting on a Fe2O3 vs MnO diagram (data recalculated on a carbonate-free basis (CFB)) the geochemical data from past and modern oceanic metalliferous sediments, polymetallic nodules and hydrogenetic crusts together with our data set from Cyprus (Figure 2.27), samples from the EPR offer the best comparison with Cyprus umbers based on Mn/Fe ratios and comparable total Fe2O3 + MnO content [Robertson and Hudson, 1972, Robertson and Boyle, 1983].

Alternatively, the majority of samples from the Atlantic and from Arc settings are scattered along the Fe2O3 axis and correlate with ocheriferous lithologies from Cyprus and/or proximal Fe oxides- rich type deposits. Metalliferous sediments from TAG and the eastern Pacific originally possess a maximum content of ~50 % Fe2O3 with MnO values ranging from 0.1 to 1 % on average and up to

10 % in Mn-rich layer concomitant with low Fe2O3 values [Boström and Peterson, 1968, Piper et al., 1975, Marchig and Gundlach, 1982, Jarvis, 1984, Metz et al., 1988, Toyoda et al., 1990] these data contrast with the average umber composition of 45 – 65 % Fe2O3 and 7 – 15 % MnO, a significantly higher concomitant concentration of Fe and Mn compared to deep-sea metalliferous sediments. These similarities are only permissible with the normalization to a carbonate-free basis as otherwise the ferromanganese fraction of sediments from the EPR is mostly below 20 % cumulated Fe2O3 and MnO.

Multiple possibilities can be invoked to explain divergence between the rock record from Cyprus and sediments forming around modern active vent sites. First of all, umbers from Cyprus are relatively low in carbonates whereas most data from the literature contain important portions of carbonate from cores retrieved at depth, relatively diluting the metalliferous fraction. It would be possible to explain this divergence by envisaging umbers formation below the carbonate compensation depth; however, the presence of coccoliths and carbonate in intra-pillows sediments discredit this hypothesis and implies supra-lava umbers became CaCO3-poor compared to intra-lava oxides during later diagenesis [Boyle, 1984, Boyle, 1990]. By comparison, the important dilution of actual metalliferous sediments by the dominant carbonate fraction act as a fixing agent preventing remobilization and further accumulation of hydrothermal oxides in

87 Chapter 2. Geology and geochemistry of umbers hollows of the seafloor topography. The accumulation of carbonate, by the same manner, acts as a buffer limiting further interaction between Fe dominated oxide deposits and seawater. This barrier then inhibits Mn enrichment through mixed hydrogenetic-hydrothermal precipitation within oxide deposits by preventing remobilization and further interaction with seawater.

88 Chapter 2. Geology and geochemistry of umbers

Figure 2.27: Fe2O3 vs MnO plot for the Cyprus umbers recalculated on a carbonate-free basis (CFB) relative to (A) modern submarine ferromanganese precipitates, note that our samples are relatively poor in carbonates (av. < 5%

CaCO3) while most data from cores incorporate a large portion of carbonates. Data sources: Back-arc Fe-rich sediments [Mc Murtry et al., 1991, Murphy et al., 1991, Binns et al., 1993, Sun et al., 2011]; arc Fe-rich sediments [Smith and Cronan, 1983, Savelli et al., 1999, Dekov et al., 2011]; metalliferous sediments from the Atlantic ridge [Metz et al., 1988, German et al., 1993, Mills et al., 1993]; metalliferous sediments from the East Pacific ocean [Piper, 1973, Heath and Dymond, 1977, Marchig and Erzinger, 1986, Rhulin and Owen, 1986b, Hrischeva and Scott, 2007]; Hawaii metalliferous sediments [Edwards et al., 2011]; metalliferous sediments from Japan [Kato et al., 2005a]; polymetallic nodules [Calvert and Price, 1977, Calvert and Piper, 1984, Dymond et al., 1984, Ohta et al., 1999, Baturin, 2009, Wegorzewski and Kuhn, 2014]; Hydrogenetic crusts [Bau, 1996, Kuhn et al., 1998, Hein et al., 2005, Asavin et al., 2010, Muiños et al., 2013]. (B) Fe2O3 vs MnO plot relative to on-land Fe-Mn deposits [Robertson and Hudson, 1972, Robertson and Fleet, 1976, Boyle, 1984, Robertson and Fleet, 1986, Karpoff et al., 1988, Robertson and Degnan, 1998, Kato et al., 2005a]

Best comparison and support to this hypothesis come from ferromanganese deposits encountered in the Japan Cretaceous Accretionary Complex. These stratiform deposits are found in the Northern Chichibu Belt, central Shikoku, Japan, directly overlying greenstones of MORB origin and underlying red cherts [Kato et al., 2005a] which have been interpreted as ancient counterpart of MOR-type hydrothermal sediments evidenced by similarities of REE signature and X/Fe ratios [Kato et al., 2005a]. These formations display CaO concentrations inferior to 5.8 %, comparable SiO2 concentrations to umbers (13.6 – 24 %) and cumulated Fe2O3 and MnO content similar or even higher (up to 86 %) to samples from this study largely due to higher MnO content (Figure 2.27 B). As a result, their Mn/Fe ratio is higher and some deposits present a positive Ce anomaly which is interpreted as strong hydrogenetic influence within the hydrothermal deposit in a more oxidizing oceanic environment (Panthalassic ocean, Permian era) than modern oceans [Kato et al., 2005a, Kato et al., 2005b].

Comparatively, sediments from the late Cretaceous Semail ophiolite, Oman show similar stratigraphic sequences covering the extrusive or plutonic units. There, metalliferous oxide sediments similar to umbers evolve to more argillaceous lithologies, radiolarian cherts and then pelagic chalk [Robertson and Fleet, 1986]. Major geochemical distinction with oxide sediments from Oman is their silica content, usually 10 to 40 % higher than the unaltered umbers from Cyprus. Although silicified umbers as bedded cherts or nodules are present in Troodos, the silicification results from diagenesis and localised SiO2 remobilisation. In contrast, the silicification

89 Chapter 2. Geology and geochemistry of umbers in Oman has been demonstrated to be hydrothermal in origin, happening post-emplacement of the oxide in hollows and affecting them pervasively which explains sample scattering due to silica dilution (Figure 2.27 B) [Robertson and Fleet, 1986]. Otherwise, metalliferous sediments from Oman are genetically and temporally similar to Cyprus umbers although more cherty.

2.4.1.2 Evidence from REE

The process of REE enrichment within hydrothermal particles and metalliferous sediment occurs through the electrostatic interaction between dissolved species from both hydrothermal fluids and oceanic waters on surface-charged Fe colloids; Fe(OH)3.nH2O [Koschinsky and Halbach, 1995, Koschinsky and Hein, 2003]. This continuous uptake of REE by Fe-oxyhydroxides from seawater results in hydrothermal particles exhibiting REE trends evolving from an early hydrothermal signature towards a more seawater-like signature. This evolution has been well documented through analysis of particles sampled in the buoyant and neutrally buoyant plume as well as in fall-out sediment traps around hydrothermal vents [Sherrell et al., 1999, German et al., 2002, Edmonds and German, 2004] (Figure 2.28). Hydrothermal particles show an evolution from more vent fluid-influenced REE trends exhibiting larger Eu anomalies to more seawater-like distribution with increasing negative Ce anomalies as the plume spreads [Edmonds and German, 2004] (Figure 2.28) consistent with data of Mitra et al. (1994), German et al. (1997) and German et al. (2002) from MAR, the Juan de Fuca Ridge and EPR respectively. Similarly metalliferous sediments from core retrieved during DSDP Leg 92 in the vicinity of the EPR display similar trends to hydrothermal particles of the plume, without the Eu anomaly, due to the extensive contribution of REE derived from seawater [Barret and Jarvis, 1988].

A recent discovery relates formation of umber-like material to the ultra-diffuse flow of LT hydrothermal systems at the South of Loihi Seamount (Hawaii) [Edwards et al., 2011] and present an interesting alternative for the formation of Fe-Mn oxides rich deposits showing a similar structure. This amorphous deposit, formed through bacteria mediation, is dominated by Fe oxides

(32 – 58 % Fe2O3) with localized Mn enrichment up to 8.3 % MnO in crust-like structures, negatively correlated to the Fe2O3 content (Figure 2.27). However, the REE signature of the Loihi Seamount oxides (Figure 2.28) clearly indicates a proximal LT hydrothermal origin with a strong positive Eu anomaly, no Ce anomaly and extremely low REE content one to two orders of magnitude below PAAS values, quite distinct to the REE signature of oxide particles observed in HT hydrothermal plumes and umbers from Troodos [Olivarez and Owen, 1989, German et al., 1990, German et al., 2002].

Umbers from Cyprus display similar trends to these reported from the EPR [Robertson and Fleet, 1976], with similar levels of enrichment (Figure 2.28). Although seawater constitutes the main

90 Chapter 2. Geology and geochemistry of umbers source of REE for hydrothermal particles, sediments formed by hydrothermal activity do not usually exhibit similar HREE enrichment. This HREE flattening was first attributed to a diagenetic effect whereby HREE are preferentially removed from the crystal lattice of Fe-oxyhydroxide during recrystallization in Fe-smectite [Barret et al., 1986].

91 Chapter 2. Geology and geochemistry of umbers

However, data on suspended and newly deposited hydrothermal particles recovered in sediment traps, free of any diagenetic effect, already display this decrease in HREE leading to the view that this flattening of the HREE is a primary feature of REE uptake by Fe oxides [Edmonds and German, 2004]. This divergence between particles and seawater is probably due to the differential incorporation, or fractionation, of REE seawater complexes by Fe oxyhydroxides with preferential fractionation of the LREE (60 – 70%) compared to the HREE (14 – 40%) [Klinkhammer et al., 1983, Rudnicki and Elderfield, 1993] which is in agreement with the decreasing particle reactivity associated with increasing atomic number [Elderfield, 1988]. Similar REE signatures to Cyprian umbers can be found in other ophiolitic sequences preserved on-land with excellent comparison for instance with the ferromanganese deposits of the Japan Accretionary Complex [Kato et al., 2005a, Kato et al., 2005b].

2.4.2 Model of emplacement

The latest model describing the origin of umber deposits in Cyprus correlate their geochemistry, irregular spatial distribution and proximity with massive sulphide deposits to the oxidation of hydrothermal effluents influenced by local lithogenous, and homogenous terri- bio- and hydrogenous components [Boyle and Robertson, 1984, Richards and Boyle, 1986, Boyle, 1990]. These non-hydrothermal components are transcribed differently in the records; (i) Local lithogenous material include hyaloclastite debris and alteration of lava flows mostly found at the base of deposits, (ii) carbonates, clays and biogenic material are evenly supplied over the active ridge and form the background sedimentation, (iii) the hydrogenous component is transcribed as the trace metal enrichment of umbers, their characteristic REE seawater signature and the Mn content.

Potential alternative to this model would invoke formation via low temperature hydrothermal discharge on the ocean floor, either off-axis or associated to local heat anomalies within a seamount. Low temperature hydrothermal systems usually produce crust-like formations at the sediment-sea water interface or mineralization of a porous matrix with important fractionation between Fe and Mn due to strong pH and Eh gradient [Moore and Vogt, 1976, Alt, 1988, Bodeï et al., 2008, Hein et al., 2008]. This type of fractionation and morphology are not observed in umbers from Cyprus, rather they display a homogenous texture without any complex mineralogy developing for the Fe and Mn phases. Although LT deposits from the Lohoi seamount, Hawaii, possess these later characteristics [Edwards et al., 2011], their REE signature rule-out these deposits as the most likely analogues for umbers.

92 Chapter 2. Geology and geochemistry of umbers

Since the 1980’s, when the last comprehensive studies on the genesis of umber deposits were published, a large volume of data and discoveries have since shed light on black-smokers vent and their associated deposits. By comparing our new data (alteration of umbers, stratigraphic sequences, 87Sr/86Sr isotopes) with existing data on umbers and the growing data on modern deposits, our results corroborate the previous model of emplacement: Cyprus umbers are the result of oxide accumulation in hollows of the paleotopography from plume fall-out formed by a high-temperature hydrothermal field within a background of pelagic sedimentation.

The influx of terrigenous clastic material into umbers deposit as observed in stratigraphic sequences from this study are well recorded by the Fe/Ti ratio vs Al index which displays a log normal relationship reflecting the balance between a decreasing supply in a hydrothermally derived element and greater content in detrital clays (Figure 2.29). This reduction of the hydrothermal contribution to the deposit could be related to two factors: (i) an increasing distance between the source vent and place of deposition with greater dilution of the hydrothermal plume and therefore an exponential density drop of related fall-out [German et al., 1990] and, (ii) diminution or cessation of the vent activity, associated with the waning stages of a hydrothermal system against a constant background of pelagic sedimentation.

Figure 2.29: Al index (100*Al)/(Al+Fe+Mn) vs Fe/Ti ratio displaying evolution of the umber deposit of Kampia from a hydrothermal end-member (sample 88- 89) at the base of the profile towards a pelagic and detrital end-member (samples 104 and 105) at the top of the stratigraphic sequence.

From our observation, a sequence of events can be constructed to summarize previous conclusions and include our field observations and geochemical results (Figure 2.31). Initial precipitation of Fe and Mn oxides occurs from a HT hydrothermal plume emanating from some local vent site. Following initial precipitation, accumulation within topographic lows developed on the sea floor with continuous background sedimentation accumulating progressively and diluting the hydrothermal signature. Umbers are then subject to multiple mineralization events. The

93 Chapter 2. Geology and geochemistry of umbers creation of a plausible sequence of event starts by considering that silicified umbers formed by the diagenetic remobilization of Si. The preservation in these silicified bodies of the geochemical signature of apatite-rich umbers and Mn-depleted umbers consequently post-date the remobilization of P2O5 in umbers as well as the formation of the Mn-depleted facies. Furthermore, subvertical palygorskite veins identified by SEM, were emplaced when the umbers were sufficiently competent to sustain vein formation while compacting and acquired their deformation during later compaction of umbers [Boyle, 1984]. These veins and the initial umber lamination are preserved within silicified layers and some nodules, though the veins are not deformed. Chert formation therefore occurred early enough in the deposition history of umbers to overprint the veined umbers and protect them from later alteration or deformation. Detailed analysis of the silicified outcrop and quartz veins running through umber, silicified umber and ochres near Kampia [Prichard and Maliotis, 1998] highlights the presence of 2 further silicification events, supposedly off-axis. Firstly, fluid circulation along the faulted basement produced the massively silicified outcrop presently studied. This outcrop preserves a bright orange layer at its base, possibly of Mn-depleted umber, consequently post-dating it. Secondly, gold-bearing quartz veins cross-cutting umbers and silicified umbers without deformation are present in this area and therefore represent a late stage post compaction event affecting the upper sequence of the Troodos Massif.

2.4.3 Umbers in Cyprus, resources and limits

Umbers from Cyprus have been quarried extensively for pigments since ancient time and more recently for cement with a production decreasing from 30,000 tons per year in the late 70’s to an average 6,000 tons per year in the last decade [Morse and Stevens, 1979, Cyprus Geological Survey, 2006]. This diminution in umber exploitation is related to both decrease of the use of umber as a natural pigment in paint or cement and to the great reduction of mineable umbers accumulation. Most umber deposits in Cyprus now form outcrops limited in size to tens of meters length for 1 or 2 meters thickness following the original paleotopography of the oceanic floor. Such outcrops are too dismembered to present any economic potential due to limited availability of the resource. Nevertheless, Kampia and Margi constitute two of the largest area where umbers still crop out on large surfaces that could justify eventual exploration and resource estimation. However, numerous facies variation including transition to clay-rich umbers and alteration to either silicified umbers or Mn-depleted umbers result in grade uncertainty and decrease any economic potential for the exploration of these areas. The exploration for new umber deposits in Cyprus seems unlikely considering these concerns. However, incorporating REY production to the existing processing of umbers could bring an important value to such exploitation and reduce

94 Chapter 2. Geology and geochemistry of umbers economic vulnerability by diversifying end products. This seems to be the strategy employed in a number of alternative deposits whereby REY extraction could be cost-effective from tailings as a by-product after main ore treatment such as processed bauxite [Tsakanika et al., 2004, Qu and Lian, 2013, Ujaczki et al., 2015] or in coal residue [Rozelle et al., 2016].

Although umbers deposits in Cyprus are now too scarce to be economically viable, similar metalliferous sediment deposits can be found throughout most ophiolitic sequences preserved on land with significant tonnage to be considered of potential interest. Such potential deposits include notably the multiple occurrences of umberiferous deposits in Japan; Mineoka Hills (Kenzai Industrial Company), the Kunimiyama deposits in the Chichibu Belt and the Mugi and Tyujin umbers in the Shimanto belt [Kato et al., 2005a, Kato et al., 2005b]. These deposits are part of an accretionary complex composed mainly of Carboniferous to Jurassic allochthonous blocks of MORB-like oceanic sequences metamorphosed to green schist facies with Middle Jurassic to Early Cretaceous argillaceous matrices emplaced in the centre of Shikoku Island (Japan) [Kato et al., 2005a, Kato et al., 2005b]. The ferromanganese deposits were primarily deposited in a trench as products of MOR hydrothermal activity in the context of the Kula-Pacific ridge-forearc collision (Panthalssa Ocean) and then accreted onto the proto-Japanese island arc [Kato et al., 2005a, Kato et al., 2005b]. These umber deposits are in average 4m thick and commonly reach up to 12m. Within the frame of the Tethyan related ophiolites, metalliferous ferromanganese sediments are found throughout most massifs but remains discrete bodies comparable to the actual extent of umbers in Troodos or smaller, such as in the Othris and Pindos ophiolite (Greece) [Robertson and Varnavas, 1993], the Kizildag (Hatay) ophiolite (Arabian platform) [Robertson, 2002], or in the Oman ophiolite though, here metalliferous sediments are heavily silicified [Robertson and Fleet, 1986]. Also, ocheriferous and umber-like deposits outside of the context of ophiolites, such as those found in the Appalachian mountains actively exploited by New Riverside Ochre in Georgia (USA) with reserves of 1 Mt form, readily available resources with similar mineralogical structure and RE content.

95 Chapter 2. Geology and geochemistry of umbers

Figure 2.30: Schematic diagram outlining the formation and evolution of umber deposits in the Troodos Massif, Cyprus

96 Chapter 2. Geology and geochemistry of umbers

is modified from Robertson and Boyle (1983) to highlight detailed field relations between massive umbers and alteration facies. The illustration shows: the irregular basal Mn-depleted umbers cross-cutting umber layering with associated pyrolusite concretion; phosphate-rich horizons; the various morphologies of silicified umber as nodules, layers and bulbous masses conserving undeformed palygorskite veins and overprinting phosphate-rich umbers and Mn-depleted umbers. The top of the outcrop shows the variation from massive umber layers to more clay-rich umbers and the transition to radiolarian cherts of the Perapedhi Formation. Late stage silicification along fault zones as quartz veins are best preserved with the more competent silicified masses.

2.5 Conclusion

The presented geochemical analysis and field observations reinforce the previous model dealing with the genesis of umbers. These deposits constitute hydrothermal Fe-Mn-Si deposits, low in carbonate with localized apatite enrichment, basal depletion in Mn and overprinted by various silicification events during diagenetic post-depositional remobilization and hydrothermal events. The excellent comparison of rare-earth element signatures between umbers and high temperature hydrothermal plume particles related to flank-ridge sediments affiliate umbers to MOR hydrothermal activity. In addition, this signature and composition can be achieved in any oceanic environment where hydrothermal discharge is realised in a relatively open environment, wether it is in back-arc, fore-arc or hot spot volcanism as long as an important transport allows for hydrogenetic enrichment as opposed to localized and distal low temperature hydrothermal

- formations. Comparison with other CaCO3 poor metalliferous sediments found in ophiolitic sequences best explain that yet, no deposit similar to umbers from Cyprus, Oman or Japan in terms of composition and thickness have been observed on the oceanic floor to date.

Chapter 3. REY extraction by leaching experiments

Chapter 3: REY extraction by leaching experiments

3.1 Introduction

Alternatives to alkaline rock and IAC mining for REY production have already been investigated with much attention given to processed bauxites also known as red mud [Klauber et al., 2011] and coal residue [Rozelle et al., 2016]. These formation contains REE-bearing minerals and/or adsorbed REE which are transferred to the wastes after processing that constitute a low-grade resource with important tonnage due to the importance and spread of coal and aluminium industry [Goodenough et al., 2016]. Sea-floor deposits including ferromanganese nodules [Fujimoto et al., 2016], hydrogenetic crusts, and marine clays [Kato et al., 2011] have also been known for containing significant REY concentrations [Hein et al., 2013]. Although of economic interest, their relative inaccessibility and need for advanced mining technologies make their exploitation challenging. A first approach to estimating and processing such marine deposit would be to investigate preserved equivalents of these deposits on land within ophiolitic structure, granting ease of access and sampling. In that view the metalliferous sediments of the Troodos ophiolite in Cyprus known as umbers probably constitute one of the best choices for this purpose. These deposits represent direct analogue of actual oceanic metalliferous sediments formed in the vicinity of hydrothermal vents and have already been regarded as mining material in the past.

Umbers are brown Fe-Mn-rich mudstones, commonly carbonate-free formed by accumulation of hydrothermal plume fall-out of Fe-Mn oxyhydroxides. Once Fe and Mn issued at hydrothermal vents oxidize in contact with seawater to form colloids, their strong surface charge allows effective scavenging of dissolved species [Koschinsky and Halbach, 1995, Koschinsky and Hein, 2003] during extensive exposure of oxides with seawater while the plumes spread. Umbers thus acquired typical REE seawater signature with concentrations up to ~500 ppm total REY (0.06 %

REO as RE2O3) with only low concentrations of radioactive elements (average U = 2.4 ppm and Th = 2.0 ppm (n=58)). These Fe-Mn oxide formations thus fall in the range of concentrations extracted from the IAC deposits (ΣREY = 300 – 1500 ppm, [Yang et al., 2013]). The REY content is 3 to 8 times lower than some of the lower concentration range of mines already in activity (outside China) like the Brockman and Narraburra deposits in Australia, respectively 0.2 % and 0.3 % TREO, or the Norra Kiirr deposit in Sweden with 0.45 % TREO. However, umbers possess a dominated amorphous oxide mineralogy and low content in radioactive elements which has a significant impact on the overall extraction and treatment costs of the ore considering low hardness and no need for any pre-concentration treatment of a REY-bearing phase. Given the extremely low content of targeted elements, it is of prime importance to investigate the processing conditions

99 Chapter 3. REY extraction by leaching experiments under which the release of REY from umbers is maximal. We therefore explore in this chapter the influence of major kinetic parameters (lixiviants, concentration, liquid-to-solid ratio, time of reaction and temperature) on the recovery of REY to develop a constrained recipe for best extraction.

3.2 Material and methods

Leaching experiment were undertaken to investigate the potential recovery of REY from the metalliferous sediments. Leaching were produced testing common inorganic acid; nitric, sulphuric and hydrochloric acid as well as ion exchange solutions of sodium chloride and ammonium sulphate. The experiments were designed to test the influence of solution concentration, liquid- to-solid (LS) ratios, time of reaction and effect of temperature on the efficiency of REY extraction in the liquid phase as opposed to the release of other elements in the solution, considered as impurities. Each factor was studied independently by keeping other parameters constant (Table 3.1).

Tested parameter Lixiviants Molarity LS ratio Time Temperature Ionic liquid Acid Min 0.05 M 2 2 min 20°C NaCl HNO Range 3 Max 1.75 M 100 13 h 70°C (NH₄)₂SO₄ H2SO4 When constant 1 M 25 120 min 20°C HCl Table 3.1: Range of experimental parameters and type of lixiviant used in the leaching experiments.

Samples used in this study were recovered south-west of Margi on the north-eastern rim of the Troodos massif, Cyprus. Sample PJ-CY-2014-19 and 91 (Table 3.2) were chosen from a pool of other samples as representative of the mean umber composition (n = 58). These two samples are devoid of traces of alteration, veins or diagenetic remineralisation that involves depletion in Mn and formation of apatite known to be enriched in REE. The samples were first crushed using a manual press and dried for 48h at 65°C. Once dried, the crushed material was milled in agate pestle and mortar using a centrifuge for 3 minutes at 350 rotations per minute. Quantitative grain size data of powdered umbers were obtained using a Malvern Mastersizer laser particle size analyser at the National Oceanography Centre of Southampton (NOCS). Two aliquots of sample PJ-CY-91 were measured in triplicate with good reproducibility (Table 3.3). After grinding, 90% of the sample size fraction is inferior to 370 µm which constitute the working material for the following leaching experiments.

100 Chapter 3. REY extraction by leaching experiments

115

ΣREY

523,7

519,6

1,3

1,2

244

1474

± 83.8 ±

95%

577,7

592,5

596,1

539,5

525,6

724,5

473,8

486,9

568.9 568.9

9,0

9,3

982

888

± 37.5 ±

90%

1,5

314

220

336,7

380,9

330,2

319,4

391,2

432,1

368,7

387,9

369.9 369.9

801

600

10,3

11,3

± 27.2 ±

84%

256,8

265,2

250,5

255,3

308,1

317,8

292,1

314,7

3,9

4,3

308

180

282.6 282.6

± 25.2 ±

119

129

20,0

21,8

80%

223,6

229,3

217,1

224,8

271,5

277,4

258,0

279,4

247.7 247.7

3,5

3,7

6,2

11,6

± 20.1 ±

56,9

59,0

53,0

58,7

95,8

99,1

87,9

50%

101,3

0,5

0,6

23,3

26,0

per fraction ofthesamplefractionper

76.5 76.5

µm)

5,6

6,0

0,7

0,1

6,9

7,2

6,6

6,8

8,3

8,7

7,6

8,6

± 0.8 ±

20%

7.6 7.6

1,2

0,7

elements in ppm in elements

grain sizegrain (

22,5

23,8

5,3

5,5

5,1

5,3

5,7

5,9

5,2

5,9

± 0.3 ±

16%

1,24

1,45

CaO

110,1

117,1

5.5 5.5

2,4

1,7

27,0

28,9

MgO

3,6

3,7

3,6

3,5

3,3

3,5

± 0.1 ±

10%

3.4 3.4

66,3

33,1

12,2

13,0

MnO

Grain size measurement (in µm) for two aliquots of sample PJ ofsample aliquots two for µm) (in measurement size Grain

2,5

2,4

2,5

2,2

2,1

2,3

± 0.1 ±

.3: .3:

38,1

54,6

122,5

121,9

2.3 2.3

6,0

4,0

Table Table

97,0

109,4

1,5

2,8

0,2

11,2

11,49

11,2

10,92

13,89

13,99

13,88

13,8

Obscuration

TiO

4,6

1,3

24,1

15,7

SiO

Chemical composition of selected samples for leaching experiments. Major elements are expressed in wt. % and trace trace % and wt. in expressed are elements Major experiments. leaching for samples ofselected composition Chemical

± σ ±

.2: .2:

A2-3

A2-2

A2-1

A1-3

A1-2

A1-1

PJ-CY-91

PJ-CY-19

PJ-CY-91

PJ-CY-19

Average

Avg. 2nd aliquot 2nd Avg.

Avg. 1st aliquot 1st Avg.

SamplePJ-CY-91 Table Table

101 Chapter 3. REY extraction by leaching experiments

Major, trace and rare earth element concentrations were determined by inductively coupled plasma mass spectrometry using an Element X-series 2 at NOCS. Solutions for measurements by ICP-MS were produced at appropriate dilution with 3% Nitric acid solutions containing internal spike (In-Re (5 ppb) and Be (20 ppb)). Artificial element standard were produced at 2, 5, 10, 25, 50, 75, 100, 125, 150, 175 and 200 ppb and used together with the internal spikes for the calibration of the instrument and sample drift correction. All element standard calibration curves display less than 3.5 % analytical error with excellent linearity. International rock standard BHVO2, BIR1 and JB3, were run in triplicate as unknown to monitor accuracy (Table 3.2). Reproducibility between triplicate is good with a standard deviation less than 7 % of the mean for all elements apart K (up to 12 % deviation in standard BIR1 where K concentration are less than 250 ppm). In all measured rock standards, concentrations fall within the published range of recommended values [Jochum et al., 2005] with the exception of Fe that is always higher in our measurement by 2.5 to 5.2 % compared to standard values. Potassium was found also to have an error of + 7 % compared to JB3 and + 18.6 % compared to BIR1 where concentrations are low. Mn was measured in excess of 4.8 % of the upper range value for this element only in standard JB3.

3.3 Results of the leaching experiments

3.3.1 Ion exchange solution

Leaching experiments using ammonium sulphate ((NH₄)₂SO₄) or sodium chloride (NaCl) at different concentration (0.05 – 1.75 M) and at a LS ratio high enough for the electrolytes not to be considered as a limiting reactant (up to 100:1) have not proved to be efficient for the leaching of REY from umbers (Figure 3.1). The use of these solutions targets easily exchangeable cations and the maximum cumulated REY concentration in the leachate was measured at 1515 ppb, equivalent of 0.3 % recovery of the initial REY content of the sample. Na shows the most efficient recovery within these experiments with 54 to 77 % recovery when leaching with ammonium sulphate, whereas Na was not measured in the sodium chloride leaching experiments. Considering all experiments using ion exchange solutions, recovery was in the range 25 - 40 % for Ca, 0 – 15 % for Sr, 3 - 7 % for Mg and 2.5 - 6 % for K. The recovery of Mn, Fe, Al, Ti, V, Co, Ni, Cu, Y, Mo, Ba was extremely low in all experiments with a maximum of 1.5 % of the initial sample content leached in most favourable leaching conditions, therefore most of these elements are not represented in Figure 3.1.

102 Chapter 3. REY extraction by leaching experiments

All values expressed in ppm average (n = 3) BHVO2a average (n = 3) BIR1a average (n = 3) JB3a Na 15588 ± 381 15262 - 16100 12513 ± 522 13550 19498 ± 958 20400 Mg 42878 ± 2125 38853 - 43600 57440 ± 1480 54752 - 60300 30818 ± 821 20321 - 34431 Al 69812 ± 3701 70062 - 71600 81122 ± 2005 79244 - 85434 90122 ± 2690 85381 - 94215 K 4696 ± 480 4280 - 5080 266.6 ± 31 198 - 250 6735 ± 489 6000 Ca 80522 ± 2796 78087 - 82440 96502 ± 3421 94000 - 97281 68505 ± 4087 70000 Ti 16405 ± 227 15113 - 27300 5736 ± 74 4875 - 6923 8460 ± 98 8842 Mn 1338 ± 42 1020 - 1432 1383 ± 41 1278 - 1414 1473 ± 49 1263 - 1355 Fe 88263 ± 286 71293 - 83865 84672 ± 4468 68668 - 79900 87855 ± 3379 82500 - 83689 Sc 31.0 ± 0.8 n.a.b 41.7 ± 0.7 n.a.b 33.1 ± 0.4 n.a.b Co 43.1 ± 0.6 34 - 56.2 50.0 ± 1.2 41.3 - 64.2 34.5 ± 1.3 24 - 43.3 Ni 118.5 ± 3.3 97.9 - 186 168.2 ± 6.6 116.7 - 249.5 37.6 ± 2.1 32 - 49.1 Cu 130.6 ± 3.6 102 - 168 119.0 ± 4.2 97.3 - 152 193.4 ± 9.4 142 - 196 Sr 427.3 ± 9.7 317 - 438.3 105.7 ± 5.1 90 - 130 442.1 ± 6.2 392 - 452 Y 23.6 ± 1.2 18.8 - 30.4 14.1 ± 0.7 11.28 - 18 23.6 ± 1.1 21.7 - 28.6 Zr 161.2 ± 12.8 124.5 - 195.3 13.4 ± 0.9 4.87 - 32.87 87.5 ± 5.8 86.2 - 110.43 Ba 131.0 ± 5.2 99 - 150 6.4 ± 0.2 0.635 - 17.2 239.8 ± 7.6 217 - 271 La 15.00 ± 0.84 13 - 16.9 0.60 ± 0.03 0.36 - 3.6 8.37 ± 0.40 7.65 - 10.18 Ce 37.80 ± 1.94 32.41 - 45.3 1.90 ± 0.09 1 - 4.44 21.44 ± 0.97 19.5 - 56 Pr 5.27 ± 0.35 4.72 - 5.93 0.37 ± 0.03 0.103 - 0.46 3.21 ± 0.20 3 - 5 Nd 25.11 ± 1.05 22.36 - 28.1 2.46 ± 0.10 1.6 - 3.3 16.17 ± 0.59 14.5 - 17.03 Sm 6.34 ± 0.25 5.14 - 6.9 1.15 ± 0.05 0.8 - 1.24 4.42 ± 0.17 2.95 - 4.83 Eu 2.14 ± 0.11 1.71 - 2.7 0.54 ± 0.03 0.3 - 0.6 1.37 ± 0.07 1.24 - 1.6 Gd 6.41 ± 0.34 4.68 - 7.34 1.94 ± 0.10 0.36 - 3.2 4.81 ± 0.24 3.9 - 5.2 Tb 0.97 ± 0.06 0.72 - 1.08 0.38 ± 0.02 0.22 - 2.28 0.76 ± 0.05 0.65 - 0.94 Dy 5.61 ± 0.31 4.03 - 6.05 2.74 ± 0.14 2.5 - 4 4.85 ± 0.25 3.9 - 4.98 Ho 1.03 ± 0.06 0.7 - 1.12 0.61 ± 0.03 0.34 - 0.64 0.99 ± 0.05 0.82 - 1 Er 2.69 ± 0.16 1.7 - 2.82 1.84 ± 0.11 1 - 1.89 2.86 ± 0.16 2.39 - 2.88 Tm 0.35 ± 0.02 0.26 - 0.42 0.27 ± 0.02 0.14 - 0.3 0.41 ± 0.02 0.34 - 0.59 Yb 2.13 ± 0.13 1.59 - 2.38 1.80 ± 0.10 1.36 - 1.8 2.68 ± 0.14 2.2 - 2.83 Lu 0.30 ± 0.02 0.2 - 0.38 0.28 ± 0.02 0.18 - 0.31 0.40 ± 0.02 0.32 - 0.5 Th 1.30 ± 0.06 1.03 - 3.51 0.05 ± 0.00 0.03 1.34 ± 0.05 1 - 1.42 U 0.45 ± 0.02 0.32 - 0.51 0.02 ± 0.00 0.031 0.52 ± 0.03 0.47 - 0.54 Standarda, published values from Jochum et al., [2005] b n.a. , non available data Table 3.2: Elements concentration as a mean of triplicate ± 1 standard deviation measured by ICP-MS for international rock standards BHVO2, BIR1 and JB3 compared to their published values from Jochum et al. (2005).

103 Chapter 3. REY extraction by leaching experiments

3.3.2 Acid concentration effect

The effect of acid concentration on the leaching of REY is studied by preparing Erlenmeyer flasks with a liquid-to-solid ratio of 25:1 by weight and solutions concentrations ranging from 0.05 M to 1.75 M. The flasks were then put on a shaking table for 1h45 min at laboratory temperature. The resulting suspensions were centrifuged and the liquid phase sampled after a leaching time of 120 min. The following results present the analysis of the pregnant solution extracted after centrifugation (Figure 3.2).

The experiments display hyperbolic curves approaching 80-85 % recovery of the initial REE content in the acid leach after levelling off. A threshold concentration is achieved for molarity superior to 0.4 M H2SO4, 0.75 M HNO3 and 0.8 M HCl which is consistent with the diprotic nature of H2SO4. Together with the lanthanides, Y shows a similar pattern achieving recovery of 82 % in

H2SO4, 85 % in HCl and 89 % in HNO3 at the asymptote. It can be seen that further increases of the

104 Chapter 3. REY extraction by leaching experiments molarity improves the recovery of REE by only a few % while increasing the concentration of contaminant in the leach, notably Th, V and to some extent Ba (in nitric acid only) which display a continuous release in the solution as the molarity increases. Considering the major elements, Al, K, Ti, Mn and Fe are poorly mobilized by the acid leach with less than 10 % of the solid passing into solution. Na, Ca and Mg show higher degree of recovery in the solution though these elements remain minor phases in the original sample. The profit of increasing the molarity beyond the threshold concentration is clearly outweighed by the diminishing purity of the leach liquor considering the minor benefits obtained: + 1.2 % REY recovered by increasing molarity from 1 M to 1.75 M HCl whereas the increased concentration in nitric or sulphuric acid does not appear to bring any improvement for REY recovery. Also, in practice, one will consider the increase in cost and safety of using more concentrated acid inappropriate when transferring such process to an industrial scale.

Figure 3.2: Effect of acid molarity on the yield in the leach solution at 20°C for 120 min and liquid to solid ratio of 25:1 by weight.

105 Chapter 3. REY extraction by leaching experiments

3.3.3 Liquid-to-solid ratio effect

The influence of the liquid-to-solid (LS) ratio was determined using Erlenmeyer flasks by mixing powdered sample and acid solution at LS ratios ranging from 2:1 to 100:1 by weight. A concentration of 1 M was chosen for HCl and HNO3 and 0.5 M for H2SO4, as above the threshold concentration (Figure 3.2) to observe the influence of liquid-to-solid ratio at optimum acid efficiency. Similarly, the flasks were put on a shaking table for 1h45 min at room temperature and samples of the liquid phase taken after centrifugation.

The LS ratio doesn’t appear to have much influence on the recovery of most major elements showing less than 5 % difference in the yield between LS ratio of 3 and 100 for Na, Mg, Al, K, Ti, Mn and Fe (Figure 3.3). On the other hand, Ca shows greater variability between low and high ratios in sulfuric and hydrochloric acids. Experimental results demonstrate that LS ratios greater than 15 are necessary to maximize recovery and reach the asymptotic values for most trace elements with the exception of Th and V that continue to increase as the liquid to solid ratio rises. The yield for REE reaches 72%, 83% and 82% for nitric, sulfuric and hydrochloric respectively at SL ratio = 20. The gain in REE recovery is only of a few % for higher LS ratio whereas a gain of nearly 10% is achieved with sulfuric acid between SL ratio of 20 and 100. Retrospectively, this justifies the choice of LS ratio = 25 for the test on molarity’s effect. Levels of recovery for Y, Co, Ni, Cu, Zn, Zr are comparable across the range of lixiviants whereas the nitric leach appears to give a much lower yield for U compared to sulfuric and hydrochloric leaches whereas the recovery of Th was stronger in the sulfuric leach.

3.3.4 Time of reaction

The time of reaction was analysed using optimized parameters for liquid-to-solid ratio and solution concentration derived from asymptotic values observed in the previous experiments.

Erlenmeyer flasks were prepared at LS ratio of 25:1 with 0.5M H2SO4 and 1M HCl then placed on a shaking table. These parameters were chosen in accordance to the fact that neither stronger molarity nor higher LS ratios improve significantly the recovery of REE considering the increase in volume and concentration of lixiviants used. Each flask was then sampled at a different time of reaction from 2 min to 9h for sulfuric acid and up to 13h for HCl.

Experimental results display hyperbolic recovery trends for most elements with a threshold reaction time greater than 100 min (Figure 3.4). The total recovery increases in the first 2 hours of reaction to level off afterward with only 4 % gain in the recovery of REE for 11 h of additional time of reaction when considering HCl solution. Thus, after a really short period of time, most of the

106 Chapter 3. REY extraction by leaching experiments reaction has already occurred with more than 75 % of the total recovery observed after only 15 minutes.

Figure 3.3: Effect of liquid to solid ratios on the yield in the leach solution using 1M HCl, 1M HNO3, 0.5M H2SO4 at 20°C for 120 min.

107 Chapter 3. REY extraction by leaching experiments

Figure 3.4: Effect of the time of reaction on the yield in the leach solution.

3.3.5 Effect of Temperature

The kinetic influence of temperature on leaching efficiency was investigated at controlled temperature (40 and 70°C) by mixing powdered sample and acid solution (0.5 M H2SO4 and 1 M

HCl and HNO3) at 25:1 liquid-to-solid ratio for 2 hours on a heating and stirring plate. The acid solutions were first placed on the hot plate for 1h to thermally equilibrate prior to insertion in the flask. As acid activity increases with temperature, elements recovery in the leach increases as a function of temperature (Figure 3.5). For all major elements, the increase in recovery is quite constant over the different type of lixiviants used with increasing gain from a few % up to 26 % for Na using HCl. Average gains in the recovery across the range of temperature 20-70°C for the 3 acids increase in the following order: Ti, Fe, Mn, Al, Ca, Mg, K, Na. Recovery for REY increases by 4 and 19 % when using HNO3 and HCl respectively when leaching temperature rises from 20 to 70°C

108 Chapter 3. REY extraction by leaching experiments reaching 84 and 93 % recovery of the initial REY content. No beneficiation in REY recovery is observed when increasing the T° for the sulphuric acid leach, which is unexplained as all other elements increase with temperature with the exception of Ba. These results show that temperature is a key parameter and that REY recovery can be increased significantly by adjusting the temperature to 70°C and HCl appears the most efficient of the 3 acid tested at higher temperature. Increasing the temperature also increases impurities concentration in the leach with a net gain of 2768, 5133 and 11447 ppm cumulated impurities (non-REY elements) in the leach when increasing temperature from 20 to 70°C for HNO3, HCL and H2SO4 respectively. This increase is largely dominated in mass by rising concentration in Fe, Mn, Mg and Na.

Figure 3.5: Effect of leaching temperature on element recovery from umbers using 1M acid solution at 25:1 liquid to solid ratio for 2h without stirring at 40 and 70°C.

109 Chapter 3. REY extraction by leaching experiments

3.3.6 REE fractionation in the recovery

The recoveries for each rare earth elements (Y included between Ho and Er) during leaching experiments testing molarity, liquid-to-solid ratio and time of reaction are presented in Figure 3.6. In both experiments testing influences of molarity and LS ratio (Figure 3.6 A and B), the use of sulphuric and hydrochloric acid produces similar trends of recovery. For all trends, the recovery increases as a function of atomic number reaching its maximum for Ho and decreasing after, with the exception of Ce showing a really low recovery in all experiments. The limited yield for Ce reflects its divalent oxidation state that enables for the formation of acid-resistant Ce oxide. On the other hand, nitric acid favours the recovery of the light and middle RE while heavy RE are poorly mobilized by the leaching. Similar trends to HNO3 are observed for the HCl time series favouring light to middle REE recovery (Figure 3.6 C). The recovery trends for H2SO4 time series stands out in the sense that lanthanum recovery is unusually low in these experiments; 42 - 48 % when compared to the 65 % in the equivalent conditions testing molarity. However, ignoring La, overall trends are similar to those of HCl and HNO3.

3.3.7 Multiple stage leaching

Simple leaching experiments have demonstrated their efficiency in recovering > 80 % of the initial

REY content of the sample with acid concentration greater than to 0.5 M for HNO3 and HCl and

0.2 M H2SO4. At the same time, Ca and Na also display strong leaching efficiencies (40 – 70 %) for the weakest acid concentration and constitute undesired impurities. A two-stage leaching process is designed using 0.05 M HCl for 1 hour in a first stage (L1) to investigate the possible separation of impurities contained in easily dissolved phases from rare earth and Yttrium. After centrifugation and extraction of the liquid phase, a 1 M HCl solution is introduced to the sample and let to react for another hour on a shaking table (L2).

The first leach with weak acid shows that 66 % of Ca, 64 % of Na and more than 20 % of Sr and U passes in solution in this first step (Figure 3.7). The release of REY in L1 is moderate ranging from 12 to 21 % although the most abundant RE, La and Nd, are amongst the least mobilized. In contrast, yttrium, third most abundant RE in umbers shows the greatest recovery of the RE during the first stage of leaching (~21 %). In L2, the REY recovery is greater than 65 % (up to 76 %, with the exception of Ce) for the 3 most abundant REY (La, Nd, Y) whereas the recovery continuously decreases for the HREE with only 43 % recovery for Lu. Similar trends and levels of recovery were observed for L1 with previous weak nitric acid leaching (Figure 3.6 A) whereas L2 is closer to HNO3 and HCl recovery trends at higher molarity (Figure 3.6 A and C) though the yields are lower by 10 to 20 %.

110 Chapter 3. REY extraction by leaching experiments

Figure 3.6: Detailed recovery for REY in each leaching experiment.

111 Chapter 3. REY extraction by leaching experiments

Figure 3.7: Two step leaching experiment on sample PJ-CY-91. The mass for each element (µg) is presented for the most represented leach fraction.

Therefore the two-stage leaching appears to be a viable way of increasing the purity of the leach solution containing the dominant fraction of REY by removing 60 - 65 % of Ca and Na, the main impurities which are therefore present at less than 15 % of their initial content in L2. However, 13.5 % of the sample’s REY, equivalent to 19.6 % of the overall mass of REY effectively leached out during the two experiments, go in L1 which constitutes an important loss considering the already low concentration available for recovery. Although the leaching efficiency for most major elements is < 10 % (apart Na and Ca), the mass of majors dominate largely the overall composition of the leach and highlights that further steps of purification are needed.

A leaching protocol developed for polymetallic nodules and hydrogenetic crusts [Koschinsky & Halbach, 1995] has not proven useful for REY extraction. This protocol aims at selectively digesting crystalline Fe or Mn-oxides in the L2 and L3 phases. In our case, as umbers do not present a comparable mineralogy, amorphous Fe and other easily reducible fraction including REY were recovered in L2 using hydroxylamine hydrochloride (53 % REY recovery) which is comparable but lower than our two-step leaching using HCl.

3.3.8 X-ray diffraction analysis on residues

X-ray diffraction carried on residues from the different types of leaches display no changes on the overall XRD spectrum compared to the original sample considering both mineral identification and intensity (Figure 3.8). These results indicate that the leaching process using weak acid does not affect the overall mineralogy of umbers with goethite, minor traces of quartz, zeolite and important amorphous Fe-Mn oxides. As the amorphous background can’t be quantified with precision and that goethite is still present after leaching, it appears reasonable to deduce that most of the REY recovered in the leaches derive from this amorphous phase.

112 Chapter 3. REY extraction by leaching experiments

Figure 3.8: Comparison of X-ray diffraction patterns of sample 91 and the residue collected after filtration of the leaching experiment using 1M HCl at a liquid to solid ratio of 25, 20°C for 2h.

3.4 Discussion

3.4.1 Ionic solutions

The use of ionic solutions tested here has proven to be inefficient for the leaching of REY from umbers. Although this method is widely used in China for the treatment of REY adsorbed in clay with economic yields, similar approach has only given 0.3 % REY recovery in optimal kinetic conditions. These results highlight a major difference between IAC and umber deposits considering REY location and binding within the deposit. Although REY were initially adsorbed on the surface of Fe and Mn oxides during umber deposition in the Tethyan Ocean, the REY can no longer be considered as easily exchangeable cations. This can be partly in response to modifications in the structure and mineralogy of umbers induced by later diagenesis of the deposits. Umbers show only limited mineralogy (goethite, quartz and zeolite) with dominant amorphous Fe-Mn oxides suggesting that REY are still bound by electrostatic forces to the oxides though they might be incorporated within the oxide matrix following diagenesis and cannot be substituted easily by ions of equivalent size in neutral pH.

3.4.2 REE leachability

In all acid leaching experiments, the mobilization of REY has been more efficient compared to the recovery of any other minor elements scavenged by Fe and Mn oxides. The REY, in any state of oxidation, have large ionic radii (0.86 - 1.04 Å) in comparison with other divalent cations

113 Chapter 3. REY extraction by leaching experiments potentially adsorbed like V (0.79 Å), Zn (0.74 Å), Cu (0.73 Å), Ni (0.69 Å) and Co (0.65 Å) in divalent state. REY ionic radii are by comparison 12 to 60 % larger. This difference in ionic size might account for their ease of extraction compare to these transition metals as they do not fit within the oxide structure as easily as the transition metals.

The upper limit for REY recovery, ~80-85 % of the total REY content in the original sediment, emphasizes the distinction between REY loosely bound to amorphous oxides and those incorporated within the mineral structure of minor goethite minerals formed during diagenesis, more resistant to the acid leach. Grain size of the leached material constitutes also an important parameter to take into account for this limit of recovery. This parameter can easily be adjusted in a processing plant and the crushing size will have a positive impact on the potential release of REY as the particles get smaller.

3.4.3 Optimal leaching conditions

Through the experiments carried out in laboratory conditions, all parameters tested indicates a kinetic control over the release of REY in the solution approaching at their maximum 80 – 92 % REY extracted depending on the acid used. Most recovery trends show smooth hyperbolic curves. Threshold values when approaching asymptotes can be considered as the minimal conditions to apply to optimize REY extraction for each parameter in the most cost-effective way. By implementing beyond threshold leaching settings of the n experiment within the n+1 test, optimal leaching conditions can be determined for a perfect REY extraction recipe considering molarity, LS ratio, time of reaction and temperature.

From the results reported here (Table 3.3), it can be estimated that optimal extraction of REY from umbers can be achieved with either inorganic acid concentrated at 1 M, 70 °C, SL ratio > 10 in 1 hour on a shaking table that would give a yield equivalent or superior to 90 % of the initial REY content.

Molarity LS ratio Time Temperature Treshold Yield max Yield Treshold Yield max Yield Treshold Yield max Yield max Yield M % M % - % - % minutes % hours % °C %

HNO3 0.75 73 1.75 80 10 73 100 76 X X X X 70 84

H2SO4 0.4 83 1.75 81 20 81 100 92 45 62 9 68 70 81 HCl 0.75 81 1.75 85 10 82 100 87 120 73 13 77 70 93 Table 3.3: Recapitulative table of rare earth and yttrium yields from acid leaching experiments at the threshold and maximum test parameter values.

114 Chapter 3. REY extraction by leaching experiments

This leaching process is comparatively cheap with industrial processes used for the treatment of REY ore from alkaline rocks, pegmatites or carbonatites. Indeed the presented protocol requires only low acid concentration at ambient temperature, and yields good performance considering no physical pre-concentration treatment is applied whereas cracking necessitates concentrated acid or alkaline solutions at high temperature (> 200 °C) after isolation of the REY-bearing phase. Compared to IAC heap leaching, the process applied for umbers treatment is relatively similar with an overall comparable efficiency though here acid conditions are essential but the leaching process long for only a few hours compared to the 150 - 400 days required for IACs [Vahidi et al., 2016].

3.5 Conclusions

This study shows that temperature, molarity of the lixiviant, liquid-to-solid ratio and time of reaction are significant operating control for the selective recovery of rare earth elements and yttrium from umbers. With REO concentrations reaching 0.06 wt.%, umbers are considered low grade far below concentrations encountered in main magmatic primary deposits, although they are in the range of concentrations encountered in a large proportion of active mines where REE are processed as by-products [Sinding-Larsen and Wellmer, 2012] and in IAC grades [Yang et al., 2013].

This study has demonstrated that the extraction of rare earth elements from umbers by simple leaching is effective without accumulation of any radioactive by-products. To compensate for the low grade, the overall process has been studied in a cost-effective way using weak concentration of lixiviants and short reaction time achieving recovery of 70-85 % of the initial sample content at 20°C. These results increase by nearly 10 % when temperature is raised to 70°C. The use of different acids brings comparable results for the recovery of REE in a leachate. Main impurities include presence of Ca and Na at even the lowest molarity. Two-step leaching allows for an important reduction of impurities in the REE-rich liquor though the loss approaches 20 % of the leached REY. These experimental results confirm previous views on the beneficiation potential of bauxite- or coal-processing residue [Qu and Lian, 2013, Rozelle et al., 2016] considered as raw polymetallic material and highlights deep-sea sediments potential as a REE resource [Kato et al., 2011, Fujimoto et al., 2016].

115

Chapter 4. REY selective precipitation with oxalates

Chapter 4: Selective precipitation of REY from a leach solution using oxalate: an experimental and modelling approach

4.1 Introduction

Strong complexing agents such as oxalate or carbonate have been widely used as a mean of purification of leach liquors and precipitation of REY during the chemical treatment of REE ore. This process is most often used in the treatment of the most common REE-bearing minerals such as xenotime YPO4, bastnäsite (Ce, La)CO3F and monazite (Ce, La, Nd, Th)PO4(REE)PO4 ores that -1 produce REE-rich leach liquor, usually in the range of 1 to 40 g.L RE2O3 with small amounts of impurities [Ru'an et al., 1995, Abreu and Morais, 2010]. Following heap leaching, by either acid or ionic exchange solutions, REE are separated from co-leached elements usually as carbonates or oxalates through selective precipitation by pH adjustment. Rare earth oxalates or carbonates are then calcinated to form a mixed RE oxide product ready for sale as mischmetal or sent to specialized factories for further separation treatment into individual, high purity, RE oxides [Christie et al., 1998]. Here we investigate the efficiency of this separation technique on an acid leach of umber. In contrast to leach solutions produced by the treatment of pre-concentrated REE-bearing mineral, the challenge of this study lies in the initially low REO concentration of the leach (0.1 - 0.5 g.L-1) and the high concentration of impurities imposed by the non-pre-treatment of umbers for the concentration of a REE-bearing phase.

4.2 Oxalate Chemistry

Oxalate salts of di-or tri-valent elements have long been studied for the separation of REE and actinides from leach solutions due to their low solubility [Crouthamel and Martin, 1951, Feibush et al., 1958, Bhat and Rao, 1964, Gammons and Wood, 2000]. Indeed, the electronic configurations of rare earth elements make them strong reducing agents that consequently easily

2- join with an oxygen donor such as oxalate ions C2O4 which dominate over inorganic lanthanides complexes.

The solubility of REE in an acid medium decreases as the concentration of oxalic acid increases until a peak of dissociation of the diprotic oxalic acid occurs, competing with REE binding. (Note

2- 2- that in the following sections, the oxalate ion C2O4 is abbreviated Ox ). The species distribution resulting from oxalic acid dissociation follows a two-stage equilibria such that:

117 Chapter 4. REY selective precipitation with oxalates

− + - + [HOx ][H ] H2Ox  HOx + H K1 = [H2Ox]

[Ox2−][H+] HOx-1  Ox2- + H+ K = 2 [HOx−]

With the total concentration of acid and conjugate species expressed as:

− 2− CT = [H2Ox] + [HOx ] + [Ox ]

Oxalate species distribution can therefore be calculated reorganizing K1 and K2 equation solving 2- for [H2Ox] and [Ox ]:

[ −][ +] [ −] HOx H − HOx K2 CT = + [HOx ] + + K1 [H ]

[ +] − H K2 CT = [HOx ] ∗ ( + 1 + + ) K1 [H ]

[HOx−] 1 = + CT [H ] K2 + 1 + + K1 [H ]

− + [HOx ] K1[H ] = + 2 + = a1 CT [H ] + K1[H ] + K1K2

In a similar way a0 and a2 can be determined:

+ [H2Ox] [H ]² = + + = a0 CT [H ]² + K1[H ] + K1K2

2− [Ox ] K1K2 = + + = a2 CT [H ]² + K1[H ] + K1K2

With a0 + a1 + a2 = 1.

The distribution of the oxalic, bioxalate and oxalate ions is therefore a function of pH (Figure 4.1). The bioxalate ion dominates over the range of pH 1.2 to 4.2 with a maximum at pH = 2.7 (94 % of oxalate species) whereas oxalate ions are the dominant species at pH > 4.2.

118 Chapter 4. REY selective precipitation with oxalates

Figure 4.1: Speciation of oxalic acid and conjugate oxalates as a function of pH using acid dissociation constant K1 = 5.9*10-2 and K2 = 6.4*10-5 [Chi and Xu, 1998].

4.3 Material and methods

The precipitation of REY from a pregnant leach liquor was investigated by addition of ammonium oxalate at various pH values. A stock leach solution (SLS) was first produced by the leaching of 20 g of umber by 600 mL of 1 M HCl over 2 hours at laboratory temperature (~21°C). This solution was used as a reference material for all further precipitation experiments and its composition after filtration of the solid residue is presented in Table 4.1. For each following experiment, 20 mL of the SLS was transferred into 50 ml Falcon tubes, to which was added 7 mL of a 0.32 M solution of ammonium oxalate (NH4)2C2O4.H2O. The pH of each experiment was then immediately adjusted to the desired pH value (ranging between 0.85 and 3.15) by addition of ammonia solution and measured with a Thermo Scientific Orion pH meter calibrated at laboratory temperature. All reagents used were at least analytical grade or higher purity. Solutions were left to rest for 5 hours and a white precipitate was observed in all vials after this time. Tubes were then centrifuged and the liquid phase carefully transferred into pre-weighed HDPE bottles while the precipitate was redigested with 6 M HCl and later transferred to pre-weighed HDPE bottles. Both resulting solutions were analysed by ICP-MS for major elements, trace elements and REE concentrations following the method previously presented in 3.2 Material and methods.

119 Chapter 4. REY selective precipitation with oxalates

Na Mg Al K Ca Mn Fe SLS (n=3) 67.4 44.7 24.2 21.6 121.2 100.9 71.3 Abs. Std. Dev. 1.0 0.6 0.3 0.3 1.2 0.8 0.5 Sc Ti V Co Ni Cu Sr Y Zr Ba Th U SLS (n=3) 39 684 3597 1105 548 1587 3537 2682 24 3099 32 23 Abs. Std. Dev. 0,3 4,5 9,5 2,2 3,4 2,2 7,8 4,2 0,2 2,9 0,1 0,1 La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu SLS (n=3) 3651 653 826 3582 730 186 763 108 624 117 297 37 207 28 Abs. Std. Dev. 3,6 1,3 1,0 2,1 1,0 0,3 0,6 0,2 0,8 0,2 0,3 0,2 0,5 0,2 Major elements expressed in ppm, trace elements in ppb Table 4.1: Composition of the stock leach solution (SLS) presented as the mean of triplicate and absolute standard deviation.

As the experiments carried out in this chapter involve a medium containing acid mixed with ammonium oxalates, measurement reproducibility on three international standards was checked between standards prepared with and without ammonium oxalates for any discrepancies which suggest matrix effect from the oxalates. Standards dilution to daughter solutions were therefore prepared by including ammonium oxalate matching the final concentration of (NH4)2C2O4 in samples analysed by the ICP-MS. Excellent reproducibility of results, showing less than 5 % difference between standards with and without oxalates, indicates the absence of any matrix effects (Table 4.2). Some divergences were observed for a few elements, notably Na and Ca that consistently returned concentrations 6 to 8 % lower in the presence of oxalate in 2 standards, whereas the difference was of 1.3 and 1.5 % for the third standard. Measurement difference for some REE can be high, up to 8 % in BIR1. However these errors are for elements with concentrations < 0.5 ppm and are therefore instrumental rather than linked to a matrix effect.

Because no significant and consistent matrix effect were observed, artificial standards for calibration and rock standards for accuracy checking were prepared without oxalates. International rock standard BHVO2, BIR1 and JB3, were run as unknowns for accuracy checking during sample analysis (Table 4.3) with excellent reproducibility. All analysis for trace, rare earths, Mg, Al, Fe and Mn fall within the range of published recommended values [Jochum et al., 2005]. Na was measured in excess by 6 % and 3 % in BHVO2 and BIR1 respectively, while measurement for Ca and Ti concentrations were measured as 3 % below recommended range in JB3.

Imaging of the oxalate precipitate was obtained by scanning electron microscopy (SEM). The leachate was treated with ammonium oxalates and the pH equilibrated with aqueous ammonia. The solution was filtered through 0.45 µm cellulose nitrate membrane filters and the residue washed with MQ water and then deposited onto Al-Cu pellets. The samples were left to dry in an oven at 60°C overnight and C-coated prior to analysis.

120 Chapter 4. REY selective precipitation with oxalates

Presence BHVO2 BIR1 JB3 of Ox. No Yes % diff. No Yes % diff. No Yes % diff. Na 17600 16320 7.8 14410 13160 9.5 20500 20240 1.3 Mg 44190 42410 4.2 57110 56420 1.2 30810 30220 2.0 Al 71620 69330 3.3 80040 79410 0.8 87460 87880 0.5 K 4471 4573 2.2 345.1 361 4.4 6406 6494 1.4 Ca 81016 75010 8.0 94016 87828 7.0 64974 64012 1.5 Ti 16010 16580 3.4 5613 5689 1.3 8285 8340 0.7 Mn 1319 1307 0.9 1297 1330 2.5 1377 1371 0.4 Fe 86040 84060 2.4 76930 76980 0.1 80000 80140 0.2 Sc 30.3 30.4 0.1 41.0 41.2 0.5 32.9 32.9 0.2 Co 40.7 40.6 0.1 46.8 46.7 0.3 32.6 32.3 0.8 Ni 121.6 121.3 0.2 171.6 170.6 0.6 38.6 38.3 0.7 Cu 134.2 134.3 0.1 122.2 121.3 0.7 198.3 197.2 0.6 Sr 434.4 433.4 0.2 106.8 106.5 0.3 451.9 449.2 0.6 Y 23.5 23.5 0.2 13.9 13.9 0.3 23.5 23.6 0.3 Zr 166.8 167.0 0.1 13.8 14.0 1.2 90.6 90.7 0.2 Ba 135.1 134.7 0.3 6.61 6.58 0.4 247.2 246.6 0.2 La 15.2 15.0 1.0 0.63 0.60 4.9 8.35 8.34 0.2 Ce 38.0 37.8 0.6 1.92 1.89 1.6 21.5 21.4 0.5 Pr 5.48 5.43 0.9 0.43 0.39 9.4 3.34 3.30 1.3 Nd 26.0 25.9 0.5 2.52 2.49 1.4 16.7 16.6 0.3 Sm 6.65 6.60 0.8 1.22 1.19 2.9 4.64 4.61 0.7 Eu 2.23 2.18 2.2 0.59 0.56 4.6 1.42 1.41 0.6 Gd 6.71 6.74 0.4 2.02 2.00 1.0 5.00 5.00 0.1 Tb 1.03 1.01 2.3 0.43 0.40 7.2 0.81 0.80 1.9 Dy 5.75 5.76 0.1 2.79 2.77 0.8 4.96 4.98 0.4 Ho 1.10 1.08 1.7 0.66 0.63 4.4 1.03 1.03 0.3 Er 2.80 2.78 0.6 1.89 1.88 0.7 2.93 2.96 1.1 Tm 0.42 0.40 5.9 0.32 0.29 8.4 0.45 0.44 1.4 Yb 2.20 2.18 0.9 1.83 1.82 0.4 2.76 2.75 0.2 Lu 0.36 0.34 7.2 0.32 0.29 9.5 0.44 0.44 0.9 Th 1.51 1.37 10.6 0.11 0.11 1.0 1.45 1.44 0.3 U 0.53 0.50 6.1 0.07 0.06 4.0 0.57 0.55 2.3 Table 4.2: Measurements comparison for BHVO2, BIR1 and JB3 international rock standards prepared with or without oxalates to study potential matrix effect of oxalate during ICP analysis.

121 Chapter 4. REY selective precipitation with oxalates

All values expressed in ppm BHVO2 BIR1 JB3 measured published values measured published values measured published values Na 16620 15262 - 16100 13500 13550 20280 20400 Mg 44130 38853 - 43600 57600 54752 - 60300 31740 20321 - 34431 Al 70080 70062 - 71600 81050 79244 - 85434 90780 85381 - 94215 K 4305 4280 - 5080 246 198 - 250 6429 6000 Ca 79522 78087 - 82440 93050 94000 - 97281 66838 70000 Ti 15550 15113 - 27300 5855 4875 - 6923 8752 8842 Mn 1322 1020 - 1432 1339 1278 - 1414 1388 1263 - 1355 Fe 82960 71293 - 83865 77850 68668 - 79900 80730 82500 - 83689 Sc 31.7 32 43.1 41 - 45 33.7 32 - 35.9 Co 45.6 34 - 56.2 53.1 41.3 - 64.2 35.9 24 - 43.3 Ni 116.6 97.9 - 186 166.4 116.7 - 249.5 36.3 32 - 49.1 Cu 129.3 102 - 168 119.3 97.3 - 152 187.7 142 - 196 Sr 426.1 317 - 438.3 113.0 90 - 130 439.2 392 - 452 Y 25.3 18.8 - 30.4 15.0 11.28 - 18 24.9 21.7 - 28.6 Zr 162.7 124.5 - 195.3 13.5 4.87 - 32.87 87.1 86.2 - 110.43 Ba 140.4 99 - 150 6.9 0.635 - 17.2 255.0 217 - 271 La 16.1 13 - 16.9 0.63 0.36 - 3.6 8.78 7.65 - 10.18 Ce 40.4 32.41 - 45.3 2.0 1 - 4.4 22.65 19.5 - 56 Pr 5.61 4.72 - 5.93 0.39 0.10 - 0.46 3.36 3 - 5 Nd 26.8 22.36 - 28.1 2.57 1.6 - 3.3 17.09 14.5 - 17.03 Sm 6.73 5.14 - 6.9 1.20 0.8 - 1.24 4.69 2.95 - 4.83 Eu 2.33 1.71 - 2.7 0.58 0.3 - 0.6 1.47 1.24 - 1.6 Gd 6.93 4.68 - 7.34 2.08 0.36 - 3.2 5.09 3.9 - 5.2 Tb 1.05 0.72 - 1.08 0.41 0.22 - 2.28 0.82 0.65 - 0.94 Dy 6.05 4.03 - 6.05 2.89 2.5 - 4 5.19 3.9 - 4.98 Ho 1.13 0.7 - 1.12 0.65 0.34 - 0.64 1.08 0.82 - 1 Er 2.92 1.7 - 2.82 1.95 1 - 1.89 3.08 2.39 - 2.88 Tm 0.40 0.26 - 0.42 0.29 0.14 - 0.3 0.45 0.34 - 0.59 Yb 2.32 1.59 - 2.38 1.90 1.36 - 1.8 2.89 2.2 - 2.83 Lu 0.35 0.2 - 0.38 0.30 0.18 - 0.31 0.45 0.32 - 0.5 Th 1.50 1.03 - 3.51 0.06 0.03 1.51 1 - 1.42 U 0.52 0.32 - 0.51 0.02 0.031 0.57 0.47 - 0.54 published values from Jochum et al., [2005] Table 4.3: Elements concentration measured by ICP-MS for international rock standards BHVO2, BIR1 and JB3 compared to their published values from Jochum et al. (2005).

122 Chapter 4. REY selective precipitation with oxalates

4.4 Results

4.4.1 Mass Balance

A mass balance between the solid and liquid phase was calculated to check the measurement quality by ICP-MS and assess the experimental workflow. The measured concentration of each element in the liquid phase and digested oxalate precipitate is multiplied by their respective volume to obtain the mass in both phases. The resulting masses are then compared with the initial mass of each element introduced, using the volume of SLS multiplied by its concentration in the Falcon tube for each experiment. In an ideal situation, the masses for any given element in the two daughter solutions should equal the initial mass contained in the volume of SLS introduced. Our results normalized to the concentration in the stock leach solution show that the mass balance for cumulated solution and oxalate precipitate is good (90% - 110% of initial mass) for Na, Mg, Al, Ti, Mn, Fe, V, Co and Ba (Figure 4.2). Recovery for Ni, Cu, Sr, Y, Zr and REE ranges between 75 and 100 % of the initial mass introduced depending on the element over the range of pH tested. In contrast, the masses of U and K were constantly found in higher proportions (110- 120%) when compared with the initial mass introduced. Ca shows the greatest variations potentially linked to the formation of insoluble Ca oxalate phases during the preparation of daughter solutions with 3 % nitric acid for ICP-MS analysis. Nevertheless, these data are of sufficient quality to determine the general phase distribution of each element during the precipitation of oxalates from the SLS at different pH values.

4.4.2 Elemental partitioning between solution and precipitate in various pH

The distribution of measured element concentrations between the initial leachate and subsequent precipitate is calculated as a mass percentage (Figure 4.3). Over the range of pH values considered, Na, Mg, Al, K, Ti and Fe do not appear to partition into the solid phase, with < 1 % of the measured mass retrieved in the precipitate for Mg, Al, K and Fe, and < 2 % for Na and Ti. Ca shows the greatest variation with no precipitation at pH < 1.2 and nearly complete precipitation achieved with the oxalate at pH > 1.5. The virtually complete precipitation of Ca observed in the pH window 1.2 – 1.5 appears to drive most of other major elements variations (Figure 4.3). Mn precipitation begins at pH = 1.5 and gradually increases from 0.3 % at pH = 1.5, to 8.6 % at pH = 3.15. Similar precipitation trends are observed for Ba and Sr which are completely depleted in the precipitate at pH < 1.5, and then strongly and continuously increase as pH becomes less acidic, with up to 45 % and 89 % mass fraction in the solid for Ba and Sr respectively at pH = 3.15. Sc, V, Co, Ni, Cu display a similar behaviour with important precipitation occurring at pH = 1.5, although their relative fractionation in the solid phase decreases at higher pH values.

123 Chapter 4. REY selective precipitation with oxalates

These patterns suggest metals co-precipitation with Ca oxalates, the dominant phase of complexation in relative mass.

The partition trends of the REY into the solid show that nearly complete precipitation of rare earth elements with oxalates is achieved between pH 1.3 and 2.3. Outside of these limits, the fractionation varies along the REE series. In the pH window 0.75 – 1.3, all REY show increasing affinity for oxalate complexes as a function of increasing pH. However this fractionation between the solid or liquid phase is not equal across the lanthanides. Uptake within the solid phase increases from La to Eu to then decreases until Lu. In contrast, at pH > 2.5 a decreasing gradient of affinity for the oxalate ligand is observed from light to heavy REE, which suggests a control of the ionic radius on the complexation of REE with oxalates at pH > 2.3 where light REE are preferentially incorporated over HREE.

124 Chapter 4. REY selective precipitation with oxalates

Figure 4.3: Element fractionation in the oxalate precipitate (mass percentage) as a function of pH from the stock leach solution.

4.4.3 Purity of the precipitate

As demonstrated by the mass distribution between solid and liquid phase, the precipitation of oxalates constitutes an effective stage for the purification of the leach solution. Nearly complete precipitation of REE can be realized, while most major and trace elements, considered as impurities, remain in solution. However, these trends do not address the purity of the precipitate as substantial differences in mass are not considered in the above results.

The purity of the precipitate is analysed as the ratio of the total mass of rare earth and yttrium (REY) divided by the sum of all measured masses (Figure 4.4). Disregarding the mass of the oxalates, the total mass of the precipitate expressed as the sum of all other elements measured in the oxalate precipitate range from 30 to 4900 µg with Ca making up 70 to 91 % of the precipitate at pH > 1.3. Accordingly, with more than 90 % of the total REE mass precipitated from the stock leach solution at pH > 1.1, the purity increases strongly at lower pH, where mostly REY bind with oxalates to reach a maximum at pH 1.1. The purity then decreases as Ca, the main impurity, starts to precipitate. The cumulated masses for all other elements apart Ca and REY only account for 18 wt. % of the precipitate at pH = 1.1 and less than 10 wt. % above pH 1.3, and is dominated by Mn, Cu, Ni, Na, and Fe.

125 Chapter 4. REY selective precipitation with oxalates

Figure 4.4: Element masses within the oxalate precipitate as a function of pH. The right axis represents the REY fraction or purity of the precipitate (%).

4.4.4 Scanning electron microscopy on precipitate

Duplicates of the oxalate precipitation experiments were made at pH 1.1 and 2.5 for SEM imaging. Two distinct crystal structures are observed between the two experiments which reflect their difference in composition, notably the Ca content. At pH 1.1, oxalate crystals show platy prismatic and rectangular shapes with smaller crest-like crystals covering them (Figure 4.5 A and B). Studies of Ln2(C2O4).nH2O crystal microstructure [Zinin and Bushuev, 2014] have demonstrated crystallization of RE-oxalate in the monoclinic system. Oxalate crystals formed at pH 2.5 consist predominantly of rhombic bipyramid (Figure 4.5 C & D). Energy-dispersive X-ray spectroscopy (EDS) data were acquired for bulk areas, zones with specific crystal structure as well as spot analyses. The EDS spectrum and chemical data (Figure 4.6, Table 4.5) are in good agreement with previous results on the purity of precipitates deduced from bulk ICP measurements. Oxalates being considered as the agent of selective extraction, chemical data (Table 4.5) have been measured by EDS ignoring carbon and oxygen proportions. Bulk areas always show higher Al and Cu content compared with EDS analysis on single crystals or spot analysis. These higher concentrations correspond to interference of the Al-Cu pellets used by the SEM. Otherwise Al is relatively constant and in lower concentration (0.7 – 0.9 %) in all single crystal and spot analysis. Similarly, Cu is not detected in any small scale EDS analysis suggesting its presence in the large zone only reflects signal contamination from the pellet. The EDS spectrums highlight that 9 out of the 15 REY are in detectable range within the oxalate crystal at pH 1.1 whereas only Y, La and Nd, the three most concentrated REY in the experiment are detected at pH 2.5 (Figure 4.6). The total REY content at pH 1.1 is estimated to represent 85 to 94 % of element precipitating with the

126 Chapter 4. REY selective precipitation with oxalates oxalate. Variations of composition and REY distribution between the different crystals morphologies do not appear to correlate with shape or size. Large and well-formed crystals analysed (area_Xlarge and area_Xlarge2) encompass the range of measured REY concentrations, whereas smaller crystals with a crest-like shape have an intermediate composition. At pH 2.5, Ca dominates the elements co-precipitating with oxalates (78 – 85%) and combined Y, La and Nd concentration reach a maximum of 7 %, in good agreement with ICP-MS results and purity estimates.

Figure 4.5: SEM electron back-scatter image of oxalate precipitate at pH 1.1 (A and B) and at pH 2.5 (C and D). Visible on the figure are areas and spot of EDS analysis (see Table 4.4 and Table 4.5)

127 Chapter 4. REY selective precipitation with oxalates

Figure 4.6: EDS spectrum of the oxalate precipitate obtain at pH 1.1 (A) and 2.5 (B). The spectrums correspond to the field of view in images A and C of Figure 4.5.

pH 1.1 no HF treatment Bulk area_Xlarge spot_Xlarge area_Xsmall spot_Xsmall area_Xlarge_2 Wt.% ± σ Wt.% ± σ Wt.% ± σ Wt.% ± σ Wt.% ± σ Wt.% ± σ Si 1.2 ± 0.1 0.9 ± 0.1 - 0.5 ± 0.1 0.5 ± 0.1 - Cl 1.5 ± 0.1 4.0 ± 0.2 1.7 ± 0.1 1.7 ± 0.1 5.4 ± 0.2 - K - 0.5 ± 0.1 - - - - Ca 6.9 ± 0.2 8.7 ± 0.2 8.9 ± 0.2 5.0 ± 0.1 6.2 ± 0.2 3.7 ± 0.1 Al 3.9 ± 0.3 0.8 ± 0.2 0.9 ± 0.2 0.7 ± 0.2 0.7 ± 0.2 0.7 ± 0.2 Cu 1.0 ± 0.2 - - - - 1.1 ± 0.2 Y 16.6 ± 0.6 15.3 ± 0.6 10.0 ± 0.4 10.8 ± 0.4 11.2 ± 0.5 23.2 ± 0.6 La 16.4 ± 0.5 18.5 ± 0.5 19.5 ± 0.5 19.8 ± 0.4 18.1 ± 0.5 18.0 ± 0.5 Ce 4.0 ± 0.4 4.6 ± 0.4 4.4 ± 0.4 4.9 ± 0.4 4.8 ± 0.4 4.0 ± 0.4 Pr 6.2 ± 0.5 6.1 ± 0.5 6.1 ± 0.5 7.4 ± 0.5 7.2 ± 0.5 5.9 ± 0.5 Nd 25.9 ± 0.6 26.5 ± 0.6 30.2 ± 0.6 29.2 ± 0.5 29.7 ± 0.6 26.3 ± 0.6 Sm 5.5 ± 0.5 5.5 ± 0.5 6.2 ± 0.5 6.9 ± 0.5 5.6 ± 0.5 5.9 ± 0.5 Gd 5.6 ± 0.5 5.3 ± 0.5 6.3 ± 0.5 6.4 ± 0.5 6.5 ± 0.5 5.4 ± 0.5 Dy 3.7 ± 0.5 3.4 ± 0.5 4.2 ± 0.5 5.0 ± 0.4 4.2 ± 0.5 3.9 ± 0.5 Er 1.7 ± 0.5 - 1.7 ± 0.5 1.6 ± 0.5 ± 0.0 1.8 ± 0.5 REY 85.6 ± 0.5 85.1 ± 0.5 88.5 ± 0.5 92.0 ± 0.5 87.3 ± 0.5 94.5 ± 0.5

128 Chapter 4. REY selective precipitation with oxalates

pH 2.5 no HF pH 2.5 with HF treatment Bulk spot-Xlarge spot-Xsmall Bulk spot_Xlarge Wt.% ± σ Wt.% ± σ Wt.% ± σ Wt.% ± σ Wt.% ± σ F - - - 5.7 ± 1.4 3.0 ± 0.9 Na 1.1 ± 0.2 1.2 ± 0.2 1.4 ± 0.3 0.0 ± 0.0 0.9 ± 0.2 Cl 0.8 ± 0.1 1.1 ± 0.1 1.3 ± 0.1 2.0 ± 0.1 1.0 ± 0.1 Ca 80.6 ± 0.7 83.7 ± 0.7 84.9 ± 0.7 78.0 ± 1.4 81.9 ± 0.7 Mn 7.4 ± 0.3 7.4 ± 0.3 7.6 ± 0.3 3.0 ± 0.3 7.4 ± 0.3 Al 4.9 ± 0.3 0.8 ± 0.2 0.6 ± 0.1 4.3 ± 0.2 0.9 ± 0.1 Cu 0.6 ± 0.2 - - 0.8 ± 0.1 - Sr 0.9 ± 0.3 0.7 ± 0.3 0.0 ± 0.0 - 1.0 ± 0.3 Y 1.0 ± 0.4 1.8 ± 0.4 1.0 ± 0.4 1.2 ± 0.4 1.4 ± 0.4 La 1.5 ± 0.4 1.5 ± 0.4 1.7 ± 0.4 2.9 ± 0.4 2.8 ± 0.4 Nd 1.2 ± 0.4 1.7 ± 0.4 1.4 ± 0.4 2.2 ± 0.5 2.8 ± 0.4 REY 3.7 ± 0.4 5.0 ± 0.4 4.2 ± 0.4 6.3 ± 0.5 7.0 ± 0.4 Table 4.5: Energy-dispersive X-ray spectroscopy analysis of the oxalate precipitates at different pH. Note that EDS analysis were made on a free C and O basis. Area and crystals analysed are displayed in Figure 4.5.

4.4.5 Overall REY recovery and distribution

The stock leach solution used for the oxalate precipitation experiment, corrected for appropriate dilution factor regarding sample PJ-CY-91, contain 406 ppm REY consisting of 25.2 % La, 24.7 % Nd and 18.5 % Y. The recovery rate for each rare earth during the leaching procedure is > 95 % for Pr, Nd, Sm, Eu, Gd, Y and Tb, 90 % for La and constantly decreasing for the heavy REE with increasing atomic number from Dy (92 %) to Lu (65 %) (Figure 4.7). The total REY content of the leach consequently represents a yield of 85 % of the initial sample REY content with the greatest loss occurring for the recovery of Ce representing 62 % of the total REY content not leached out. By comparison with other REE, Ce resistance to the leaching process is due to its different oxidation state as Ce4+, which allows it to form acid-resistant Ce oxides complexes.

Figure 4.7: REY concentration in the leaching solution (black) and the relative elemental recovery from the sample (blue).

129 Chapter 4. REY selective precipitation with oxalates

The formation of an oxalate precipitate has proven to be an efficient way of selectively precipitating REE from the solution and separating them from other impurities with an efficiency highly dependent on pH. The minimal uptake during precipitation is observed for pH = 0.89 (61.5%) and rapidly increases up to pH = 1.3 where precipitation efficiency is > 96 % for all REY. Adding the efficiency of oxalate precipitation on the leaching process (85 % recovery), the REY recovery in the oxalate precipitate relative to the original sample increases from 51 % of the initial REY content at pH = 0.89 to 82 % for pH = 1.3 – 2. Relative to the impurities that are co- precipitating, the optimal purity for the precipitate is achieved for pH = 1.1 where 76 % of the initial sample REY content is recovered. Based on these values following optimal leaching and precipitation conditions, the processing of 1 ton of umber with an average 500 ppm REY would produce an oxalate precipitate containing 380 g of mixed rare earth elements.

4.5 Modelling REY oxalate precipitation from umber leach solution.

Detailed recovery trends within the oxalate precipitate (Figure 4.3) have highlighted that strong REY fractionation along the lanthanide series occurs at pH values < 1.5. In addition, the decreasing recovery observed at pH > 2 as a function of atomic number constitutes another unexplained observation. The heavy REY have the greatest commercial values, so it is important to understand the reason for their less efficient recovery. Furthermore, speciation of oxalic acid as a function of pH (Figure 4.1) shows that the presence of oxalate ions is minimal in the pH window considered here, although precipitation of REE oxalate is observed at pH 0.8 and decreases at higher pH when the activity of Ox2- increases. To explore these questions, a computer model using the PHREEQC software was developed to reproduce as closely as possible the chemistry involved in the precipitation experiments.

4.5.1 Solution modelling

4.5.1.1 Stability constant of aqueous REY-oxalate complexes.

Under the condition of the precipitation experiments, the REE contained in the leachate are partitioned between the solution and a precipitate forming after the addition of ammonium oxalate. However, not all oxalate ions will be in the correct ionic form to bind with REY to precipitate and will therefore remain in solution. Potential aqueous complexes of REY3+ and oxalates can be described by the following formulas taking into account the bioxalate (HOx-) and oxalate (Ox2-) ions:

130 Chapter 4. REY selective precipitation with oxalates

3−푚 퐻푂푥훽푚 = [푅퐸푌(퐻푂푥)푚 ]

3−2푛 푂푥훽푛 = [푀푂푥푛 ]

th - Where HOxβm is the m (order of complexation) stability constant of the bioxalate (HOx ) ion with 3+ th 2- 3+ any REY and Oxβn the n stability constant of oxalate ion (Ox ) with any REY . Schijf and Byrne (2001) have shown through modelling and experimental results, that one orders of complexation for bioxalate (m = 1) and 2 for oxalate ions (n = 2) are satisfactory for modelling REY binding behaviour with oxalates in aqueous solution as further orders of complexation remain minor. A large number of studies have explored the binding behaviour of REY and oxalates. However, most are limited to the analysis of one or two individual REE. Schijf and Byrne (2001) were the first to present a valid and complete set of stability constants for these complexes which compare well with previous studies despite different media, ionic strength and techniques (see reference in

Schijf and Byrne (2001)). The validity of these stability constants (K1 and K2) is inferred from the analysis of the pattern of stepwise stability constant ratios K2/K1 = Oxβ2/(Oxβ1)². As these two values are strongly anti-correlated, variations in the resulting ratio reflect errors in experimental determination of β1 and β2. With K2/K1 = 0.05 ± 0.02 (excluding La and Ce, 0.23 and 0.36 respectively) the error is minimal and the data are valid to include in our model.

Figure 4.8: REY patterns of stability constant at infinite dilution for log HOxβ1,

Oxβ1 and Oxβ2 [Schijf and Byrne, 2001].

131 Chapter 4. REY selective precipitation with oxalates

The stability constants Oxβ1 and Oxβ2 (Figure 4.8) present increasing trends with atomic numbers that reflects increasing stability of the REY-oxalate complex as a function of ionic radii following the lanthanide contraction (Y introduced between Dy and Ho) in simple electrostatic interactions.

With a more or less constant difference of 4 log units, the double RE-oxalate complex (Oxβ2) is dominant over the formation of single (Oxβ1) complexes. In contrast, logarithmic values for the binding of bioxalate ion with REY (HOxβ1), oscillate between 1.9 and 2.45, without showing any specific trends. Complexation with bioxalate ions is therefore 3 to 5 orders of magnitude lower than complexation with oxalate ions which will be the dominant complex for REY in aqueous solutions.

4.5.1.2 Stability constant of solid REY oxalate complexes.

The RE oxalate precipitate has been reported to be a hydrated REY salt of the form REY2Ox3.nH2O such that:

3+ 2- 2REY + 3OX + nH2O  REY2Ox3.nH2O

For most REY, the hydration degree (nH2O) is 10, apart for Ce with a hydration degree of 9 [Crouthamel and Martin, 1951]. Although the formation of solid REY oxalates complexes has attracted the attention of many since the 1950’s [Crouthamel and Martin, 1951, Feibush et al., 1958, Bhat and Rao, 1964, Grenthe et al., 1969, Chi and Xu, 1998, Chung et al., 1998, Schijf and Byrne, 2001, Xiong, 2011] a complete set of solubility constants for solid rare earth oxalate remains elusive with no study presenting results for all lanthanides and Y simultaneously under identical experimental conditions. The most complete data set on REY2Ox3.nH2O complexes can be found in Bhat and Rao (1964), Chung et al. (1998) and Xiong (2011) (Table 4.6, Figure 4.9). Although diverging by two log units, two of the sets are consistent in showing increasing solubility constants from La to Gd, which then decrease towards the HREE. This behaviour contrasts with the variations in log Oxβ1-2 for aqueous REY oxalates complexes.

Although data from Xiong (2011) lack information on HREE, they present a better comparison with those of Chung et al. (1998) with similar constants for La, Ce and Sm, whereas data from Bhat and Rao (1964) for La, Nd and Tb are low compared to their direct neighbours suggesting the presence of potential errors.

132 Chapter 4. REY selective precipitation with oxalates

log β (REY2Ox3.nH2O) Bhat and Rao Chung et al, Xiong (1964) (1998) (2011) Y -28.91 -29.29 x La -26.91 -29.22 -29.15 Ce -28.79 -30.40 -30.18 Pr -29.64 x -30.91 Nd -29.69 -30.89 -31.57 Sm -30.30 -31.35 -31.59 Eu x -31.38 x Gd -29.49 -31.37 -32.31 Tb -29.14 x x Dy -29.44 -30.70 x Ho x x x Er -28.82 -30.05 x Tm x x x Yb -28.25 -30.02 x Lu x x x

Table 4.6: Data of log β (RE2Ox3.nH2O) at 25°C and infinite dilution.

133 Chapter 4. REY selective precipitation with oxalates

4.5.1.3 Estimation of missing constant via linear free-energy relationship

Linear free-energy relationships (LFER) allow calculation of an unknown energy change for a reaction by linear interpolation with data on energy change in a reference reaction for the same material. With relatively complete data over the lanthanide series and consistency with other studies, data from Chung et al. (1998) are used to estimate missing constants by linear free- energy relationship in combination with the NIST databases for critical stability constant [Martell and Smith, 1977, Martell and Smith, 1982, Smith and Martell, 1989]. To find the best match between oxalates and another acid, organic acids in the NIST database were filtered by logβLa/logβSm < 1 and logβGd/logβYb > 1 ratios to match the incomplete convex upwards trends formed by oxalates data (Figure 4.9). Accordingly, REE stability constants of 14 acids have been compared with data from Chung et al. (1998). The 6 best-fit are presented in Figure 4.10, with an excellent correlation (R² > 0.9) observed for a carboxylic acid (abbreviated thereafter “CA”, Figure 4.10 A), a phenol (“P”, Figure 4.10 B) and a naphthol (“N”, Figure 4.10 C). The correlation coefficients rapidly decrease afterward to R² values of 0.75 - 0.84 (Figure 4.10 D, E, F) and to insignificant R² values for other organic acid tested. A similar process was applied to data of Bhat and Rao (1964). When compared with the 14 organic acids of the NIST databases previously selected, optimal correlation coefficient of 0.846, 0.839 and 0.813 are found for the phenol P, an amino-carboxylic acid and the carboxylic acid CA respectively. The difference in best correlation coefficient between the two dataset therefore justify the use of data from Chung et al. (1998).

Missing values in the oxalate series are estimated using an equation of linear regression (Figure 4.10). The correlation coefficients for CA and P are similar therefore two LFER calculations are calculated (Figure 4.11). The two LFER show < 0.6 % difference in stability constant estimation for the missing elements Pr, Tb, Ho, Tm and Yb. The patterns formed combining the values of Chung et al. (1998) and LFER estimations gives close trends although calculation using the phenol (P) data gives a smoother trend and will therefore be used to complete the data set for the computer model.

134 Chapter 4. REY selective precipitation with oxalates

Figure 4.11.: Patterns of –log β (RE2Ox3.nH2O) and estimations from linear free- energy relationship at 25°C and infinite dilution using linear regression equation from the carboxylic-acid (LFER CA) and the phenol (LFER P) (Figure 4.10)

135 Chapter 4. REY selective precipitation with oxalates

4.5.1.4 Parameters of the model

Thermodynamic data for oxalates complexes have not yet been included in any database available for calculations in PHREEQC and therefore needs to be defined in the script directly (Appendix B). Following the work of Schijf and Byrne (2001) the REY-oxalate complexes considered in the model are the following:

RE3+ + HOx- = REHOx2+ (aq.) RE3+ + Ox2- = REOx+ (aq.)

3+ 2- - RE + 2Ox = RE(Ox)2 (aq.) 3+ 2- 2RE + 3Ox + nH2O = RE2Ox3.nH2O (s)

In a similar way to REE, only partial data exists to describe oxalate complexation with any other major or trace elements. Some data exists for Ca oxalates complexes, as these have been studied extensively in medicine as a dominant constituent of kidney stones. This is convenient for this study as Ca constitutes the dominant impurity in the REY liquor. Therefore, only the complexation of Ca and the rare earths are taken into account in this simplified model of previous experiments. The model was set up to calculate speciation of each rare earth with the 3 aqueous oxalate complexes and include the kinetic effect of REY-oxalate solid precipitation if saturation was reached (Appendix B).

A single solution model containing all species was preferred to a mixing model between a solution of leach and one of ammonium oxalate as the changing parameter in the experiment was the pH, fixed after mixing. The effect of the [Ox]/[REY] mole ratio was not experimentally tested as oxalates are constantly in excess. The initial solution composition is therefore calculated from the concentration measured in the stock leach solution (Table 4.1), including dilution induced by the addition of ammonium oxalate solution. All calculations were made using the Lawrence Livermore National Laboratory (LLNL) database [Delany and Lundeen, 1990] presenting the most complete thermodynamic data for REY for other complexes.

4.5.1.5 Inherent limits of the model.

Caution should be taken when interpreting results from the numerical model and comparing them with those of the experiments. Indeed, the attempt to numerically represent complex chemical reaction suffers from major limitations. It is therefore important to consider the following: - All RE-oxalate stability constants used in this model and described above result from single element-oxalate equilibration experiments. Consequently these constants might

136 Chapter 4. REY selective precipitation with oxalates

not be representative of the real elemental behaviour during ligand binding in the presence of other competing ions. - As a consequence, only single element-oxalate complexes are considered, although REY

oxalate of mixed composition are likely to form such that XYOx3.nH2O where X and Y are two different RE. In addition and although minor, complex hydrated oxalate salts

including Na and NH4 with REY have been reported to form such as: NaREOx2, NH4REOx2,

Na4Ce2Ox5 [Grenthe et al., 1969, Gammons and Wood, 2000] but no constants are available. - For ease of comparison in the literature, all constants are extrapolated from their experimental conditions to infinite dilution and ionic strength of 0, which is not representative of the leaching conditions in terms of concentrations and ionic strength. - The model is a simplification of the real experiment because of the lack of data on oxalate complexation for other elements (other than REY and Ca). - The amount of REY precipitated in the model accounts only for the formation of oxalate crystals. However, in addition to co-precipitation (as explained above), Na- or Ca-oxalate may also scavenge REY from the solution by simple electrostatic interactions and this is not taken into account within the model.

4.5.2 Results

Speciation calculated by the PHREEQC model using REE concentrations similar to those in the precipitation experiments display important differences along the lanthanide series (Figure 4.12). The 5 most abundant complexes present in the results of the model have been selected for graphic representation. Combined they account for > 94 % of REY complexes at the lowest pH and > 99% at pH > 1. Using experimental conditions and complexes calculated from the LLNL database, oxalate-binding dominates REE complexation. Other complexes considered in the

3+ 2+ + - calculation decrease in the following order of abundances: M , MOH , MO , MO2H, MO2 (where M stands for any RE) and account for less than 1% at pH >1.

Important differences between the model and the experiment can be observed. In the range of pH and concentrations explored here, the Ca-oxalates complexes considered never reach saturation and remain in solution. In contrast with the experiments, no steep precipitation curve is observed between pH 1.1 and 1.5. Another major difference in the results from the model concerns the total lack of precipitation for Tb, Ho, Er, Tm, Yb and Lu. For these elements, complexation is dominated by aqueous single oxalates at pH < 0.8 and then by increasing proportions of double oxalate at more than 90 % at pH > 1.6. It is important to note that these 6 elements have some of the lowest concentrations in solution ([RE] < 220 ppb). Interestingly, Eu

137 Chapter 4. REY selective precipitation with oxalates precipitates although its concentration is only 138 ppb. All elements precipitating display convex upward trends of precipitation as a function of pH with a maximum reached for pH 0.9-1.1. All these bell-shape curves show diverse degrees of slope inversely related to the concentration of RE in the stock leach solution. These observations suggest a control of both concentration and atomic number in the speciation of REE-oxalates.

138 Chapter 4. REY selective precipitation with oxalates

Accordingly, the PHREEQC modelling was adjusted to test these hypotheses by calculating the speciation of all lanthanides in equal concentration at 0.1, 0.5, 1, 5 and 10 ppm. Complete speciation results for each set of concentration are available in Appendix C and Figure 4.13 presents the formation of RE-oxalates precipitated at different concentration for 6 rare earth elements selected along the lanthanide series. From the following model the formation of solid RE-oxalate increases up to pH 0.9-1.1 to decreases thereafter. The position of the maximum precipitation is not constant and changes from pH 1.1 to 0.9 as a function of increasing atomic number. Also, the formation of a precipitate is directly affected by the initial rare earth concentration in solution. It appears that the formation of middle rare earth oxalate is favoured against light- and even more against heavy-rare earth oxalate at equal concentration. This distribution can be considered as a direct representation of the stability constant β (RE2Ox3.nH2O) where constants increase from La to Gd and decrease again towards the heavy rare earth elements (Figure 4.11). However, although the stability constant of La is the lowest, modelling shows formation of La-oxalate at potentially lower concentrations than is possible for the heavy rare earths to form solid complexes. This suggests that the HREE are affected by another parameter that inhibits their precipitation. Species distribution (Appendix C) highlights the

- increasing dominance of REOx2 aqueous complexes as pH increase up to the point of inhibiting formation of solid oxalates. Although the stability constant Oxβ2 increases as a function of atomic number (log K2 = 10.5 - 11.7 from La to Lu) and could account for the decreasing formation in solid oxalates for HREE by competing in REE binding, this interpretation is counter intuitive considering

RE2Ox3.nH2Oβ >> Oxβ2 by at least 18 log units.

4.5.2.1 Comparison between experimental and modelled results

Although the modelled and experimental trends for the precipitation of REE oxalates do not agree exactly, the model reproduces some of the observations from the experiments, which rules out errors related to the experimental or instrumental setup. These similarities bring some answers to our initial questions. - The fractionation observed in the recovery from the experiments and the model at pH

< 1.1 follows closely the bell-shape of the RE2Ox3.nH2O stability constant. - The modelling shows that at equal concentrations, the formation of solid RE-oxalate is favoured in the following order: middle > light >> heavy. These trends result from (i)

the positive gradient of log Oxβ2 along the lanthanide series coupled to (ii) the increasing competition of non-REE (e.g dominantly Ca) species uptaking oxalate ions. Accounting for the initial differences in concentration in the experiment, this explains the decreasing recovery for the HREE observed in the experiment at pH > 2.

139 Chapter 4. REY selective precipitation with oxalates

In the model conditions, the maximum formation of a REE oxalate precipitate occurs at pH 1 and similar values have been previously reported for the purification of monazite leach liquor [Salman et al., 2014]. The species distribution following oxalic acid dissociation shows that Ox2- ions represent less than 0.03 % of the total oxalate species distribution at pH 1. On the other hand at pH 3 and beyond, the Ox2- concentration rises steeply and dominates species distribution at pH >

4.2. However the precipitation in the model is minimal. The dominance of Re2Ox3 complexes over aqueous complexes at pH 0.9 - 1.1 can be understood as a constant balance of the system. Free trivalent REE partitioning between the different oxalate complexes follows the magnitude of the stability constants. Therefore RE3+ will bind dominantly with Ox2- over HOx- to form a precipitate until [Ox2-] become the limiting reactant. The system therefore needs to balance the equilibrium induced by the decrease of [Ox2-] at a fixed pH by dissociating more oxalic acid into HOx- and Ox2-. The newly formed Ox2- ions are therefore readily available for complexation with RE3+ in the solution and to precipitate. The cycle carries on balancing until one of the reactants becomes

140 Chapter 4. REY selective precipitation with oxalates limiting. The decrease in Re2Ox3 at higher pH in favour of aqueous RE-oxalate complexes relates to the increasing competition in oxalate binding from Ca which starts to precipitate at pH > 1.1 and is largely in excess over [HOx-] and [Ox2-]. As pH rises, more oxalate and bioxalate ions bind with Ca over REY regarding its larger abundance in the solution; 25-3375 fold REY concentration. Oxalate constitutes a chelating agent for REE and usually binds forming bidentate ligand with a ring structure to form precipitates [Hansson, 1970] (Figure 4.14). The decreasing availability of Ox2- at higher pH due to Ca binding prevent the formation of these ring structure and explains the dominant complexation of RE3+ as aqueous complexes.

Figure 4.14: Structure of the complex [Nd2(C2O4)3.10H2O]n forming planar layers in the monoclinic system [Hansson, 1970].

The absence of Ca oxalate precipitation in the numeric model explains the observed difference between REY recovery trends from the model and the experiment at pH > 2. In the experiment, Ca oxalate is the dominant precipitate at pH > 1.5 and REY can be integrated as co-precipitating elements within the newly formed crystals. This explains why REY recovery trends do not decrease as steeply as in the numerical model. Furthermore, a level of supersaturation of oxalate ion might be needed for the precipitation of a solid phase which is not considered in this model.

This parameter might also contribute to the underestimation of the model for RE2Ox3 precipitation run at the experimental concentrations.

141 Chapter 4. REY selective precipitation with oxalates

4.6 Conclusion

The use of oxalate has proven to be an efficient way of precipitating REY from acid leach liquor with more than 96 % of the total REY content precipitated between pH 1 and 2. The strong dependence on pH for precipitation of diverse species allows for the selective precipitation of REY from other impurities. The purity is optimal at pH 1.1 before Ca oxalate precipitation starts. Fractionation observed between the different rare earth elements in the experiment have been successfully explained via numeric modelling using PHREEQC software: (i) the increasing recoveries from L- to M-REE and decreasing trends towards the HREE at pH < 1.5 follow the solid

RE-oxalate solubility constant distribution -log β (RE2Ox3.nH2O), (ii) the decreasing recovery trends at pH > 2 results from competition with Ca oxalate formation, (iii) the decrease in the recovery is not as steep in the experiment as it is in the model due to co-precipitation of REY with the Ca- oxalate phase or other phases not taken into account in the model. In addition, the model predicts an optimal pH window for the precipitation of REY between 0.9 and 1.1 as a result of oxalic acid dissociation, availability of Ox2- and lack of competing ions in agreement with experimental results.

142 Chapter 5. Conclusion

Chapter 5: Conclusion

5.1 General conclusions of the study

The objective of this study was to investigate the potential of the supra-lava metalliferous sediments of Cyprus, known as umbers, as a resource for REY. This goal encompass multiple aspects such as the understanding of (i) the genesis and mode of emplacement of these deposits and the associated geological context, (ii) the metallogenesis and enrichment processes leading to REY enrichment, (iii) umber deposits distribution and variability. The culmination of this research has therefore allowed for (iv) the exploration of efficient techniques for the recovery and (v) separation of REY from umbers.

Data presented in chapter 2 address these first three points. Field observations and geochemical characterisation of samples collected in Cyprus display a large range of composition and morphologies for the supra-lava metalliferous sediments. Three distinct facies have been identified: massive unaltered umbers; Mn-depleted, Fe-rich umbers; and silicified umbers. From geochemical characterization and mineralogy, a sub-group of apatite-rich umbers was recognised within the massive unaltered umbers. Umbers constitute fine-grained mudstones of Fe-Mn oxyhydroxides dominantly x-ray amorphous. REE fractionation trends and Sr isotopes ratios close to seawater signatures highlighted seawater as the main source for trace elements in these hydrothermal deposits. Fractionation of seawater dissolved trace metals by Fe and Mn oxyhydroxides have led to umbers acquiring a total RE content of 350-500 ppm dominated by La, Nd and Y.

Study of the two alteration facies; Mn-depleted umbers and silicified umbers, demonstrated that umber deposits were subject to multiple post-deposition geochemical remobilisation events, as diagenesis and alteration of underlying basalts induced changes in Eh and pH conditions. These conditions led to the remobilisation of Mn, Sr and Ba dominantly from the basal layer of the deposit to form pyrolusite concretions and veins accumulating these elements. Detailed geochemical analysis of the silicified umbers found throughout the Troodos massif have shown silicification post-dates the formation of Mn-depleted umbers, the formation of phosphatized horizons and compaction of deposits.

Comparison of these new data with modern oceanic metalliferous sediments highlighted the similarity between umbers and deposits resulting from geothermal activity occurring along oceanic ridge axes. Major and trace elements support previous hypotheses on the formation of umber deposits as resulting from the accumulation of oxidized high-temperature hydrothermal

143 Chapter 5. Conclusion plume particles. An important distinction exists between modern and past records of hydrothermal sediments as a result of different carbonate precipitation rate. Umber’s strong enrichment in most seawater trace metal, REE fractionation trends and high Mn concentration highlights the important hydrogenetic overprint of these hydrothermally-derived particles. In contrast, no similar concomitant levels of Fe and Mn enrichment have been found to date in modern oceanic deposits. These dissimilarities are to relate to the importance of carbonate precipitation associated with Fe and Mn oxides, buffering further enrichment via hydrogenetic precipitation or seawater scavenging for modern distal hydrothermal deposits.

Based on the understanding of REY incorporation within umber deposits through scavenging from seawater by Fe oxides during simple electrostatic interactions and the absence of complex and well crystallised mineralogy, the work presented in chapter 3 explored the release of these elements by simple leaching experiments. Main leaching agents used in the industry for rare earths extraction were used and their efficiency compared throughout the test of the influence of molarity, temperature, pulp density and time of reaction. This comparative study established the efficiency of acid-leaching as opposed to ionic solutions of sodium chloride and ammonium sulphate, widely used in the treatment of ion adsorption clays in China. The extremely low REY recovery from umbers when using these later solutions demonstrated REY could not be considered as easily exchangeable cations in umbers and that stronger reducing conditions were necessary for their extraction. Acid-promoted release of REY displays hyperbolic trends of recovery for all kinetic parameters tested. REY constitute the most susceptible elements to the leaching conditions tested with recovery reaching 80-90% of the sample content in optimized conditions. Main impurities included in the leach solutions are Mn, Ca, Fe and Na by weight but Ca and Na show proportionally the highest recovery rate. The leaching stage therefore produces an enrichment factor ranging from 50 to 75 for REY from sample to leach solution.

A two-stage leaching process with HCl proved to efficiently separate most of the contaminant (Ca, Na) in the first leach. However, around 20 % of the total REY content of the sample was also leached out which represents an important loss given the low purity of the leach. Therefore, the valuable purity beneficiation of the second leach was balanced by an important concentration reduction of the targeted elements in the second leach.

Chapter 4 presented results on the purification of the leach liquor via selective precipitation of REY as an oxalate precipitate. The study of pH influence on REY-oxalate binding has demonstrated that maximal precipitation (96-99 %) is obtained for pH conditions between 1 and 2. Consideration of impurities fractionation between the leach solution and the precipitate determined that pH 1.1 was optimal for the precipitate purity considering mainly Ca-oxalate

144 Chapter 5. Conclusion formation. These results were confirmed by further EDS analysis by SEM showing oxalate crystal composition ranging from 85 to 95 % REY at pH 1.1. In contrast, oxalate crystals precipitated at pH = 2.5 contain less than 7 % REY. Important fractionation along the REY series was observed as a function of atomic number and pH: at pH < 1.1 the uptake of MREE by oxalates is favoured against LREE and HREE whereas at pH > 2.5 the binding affinity of REY for oxalate decreases as a function of atomic number. Insights from PHREEQC modelling of the experiment showed that the uptake of REY by oxalate closely follow the bell-shape distribution of REY-oxalate solid complexes stability constant (–log β (RE2Ox3.nH2O)) justifying fractionation trends observed at pH < 1.1. In addition, the model demonstrated that at equivalent REE concentration in solution, oxalates precipitates will fractionate REY in the following order: MREE > LREE >> HREE. This ordering and the various degree of difference relates to the interplay of aqueous REY-oxalate complexes (log HOxβ1, Oxβ1 and

Oxβ2) with the natural fractionation induced by solid REY-oxalate stability constant distribution. The decreasing trends of recovery in the experiment for the heavier REE at pH > 2 highlight this process. However, the model and the experiment diverge significantly regarding the importance of that decrease as the experiment still display an important uptake of HREE at high pH values. It is inferred that this divergence emerge from REY co-precipitation and adsorption onto the surface of other Ca and Na oxalate complexes forming increasingly as pH values rise that haven’t been included in the numeric model.

At pH 1.1 the oxalate precipitation therefore produces an enrichment factor of REY concentration between 28 and 32 from the leach solution to the precipitate. Overall, the combined leaching process and selective oxalate precipitation therefore produces a total enrichment factor ranging between 1400 and 2400 for REY from the sample to the oxalate precipitate in a simple two-step process forming a high-purity end-product of mixed REY.

5.2 New knowledge gained from this study

 Extensive data set of whole rock geochemistry of umbers and their two major alteration facies with good spatial coverage of the Troodos massif.  First 87Sr/86Sr isotopic data produced on umbers and their alteration facies highlighting seawater as a dominant component over hydrothermal sources for the origin of trace metals in umber deposits.  In-depth analysis of umbers alteration facies considering elements mobility in the context of changes in Eh and pH conditions during diagenesis for Mn-depleted umbers. The high

analytical precision of whole rock composition also allowed for the correction of SiO2- induced dilution to identify timing and origins of silicification affecting umbers deposits in the Troodos ophiolite, either of diagenetic or hydrothermal origin.

145 Chapter 5. Conclusion

 Based on this founding, we produced a refined model of the emplacement and evolution of umber deposits in the Troodos ophiolite building on existing scheme, comparing these deposits with the numerous available data on recent and fossilized oceanic metalliferous deposits across actual oceans and other ophiolitic sequences.  First extensive and comparative study on the leaching of umbers as a potential mineral resource for rare earth elements.  Refining of the knowledge associated with the use of oxalate as a purification agent for the production of a mixed REY solid phase in industrial processes.  This study has demonstrated through experiments and numerical model the effect of pH on the precipitation efficiency and fractionation of REY from an acid leach solution.  First numerical model to integrate all lanthanides at once in the study of oxalate-REE fractionation.

5.3 From laboratory to industrial scale.

This study has demonstrated the technical efficiency of REY extraction from Cyprian supra-lava metalliferous sediments at laboratory scale. The projection of these results to an industrial scale should consider the following adjustment for processing costs reduction and optimization of the plant economic viability.

Optimal conditions for the leaching of umbers have shown that large volumes of acid are needed as 1:10 solid-to-liquid ratios are used. Implementing a distillation step following the leach would allow recovering hydrochloric acid after condensing to recycle in the next batch (Figure 5.1). Here, the additional cost of heating up for distillation is counterbalanced by the significant reduction of volume of reactant needed on a long term period. Furthermore, leach efficiency is improved at higher temperature and could therefore benefit from the temperature of distilled HCl.

PHREEQC modelling has demonstrated the importance of the initial leach REE concentration for the precipitation of oxalates. The higher the initial REE concentration the higher is the precipitation rate. Therefore, the distillation and associated evaporation would induce an overconcentration of elements contained in the leach liquor and significantly increases oxalate precipitation efficiency that in turn would lead to a diminution in oxalate consumption.

Similarly to HCl recycling, consumption of ammonium oxalate can be seriously reduced if the oxalate cake is digested by sodium hydroxide. This reaction allows the conversion of REY oxalate into hydroxides and formation of sodium oxalate salts [Habashi, 2013] such that:

- RE2Ox3 + 6NaOH  2RE(OH)3 + NaxOxy

146 Chapter 5. Conclusion

The hydroxides are then calcined to form a mix REY oxide product and the sodium oxalate reintroduced as a reactant in the precipitation step (Figure 5.1).

Figure 5.1: Workflow of metalliferous sediment processing for the extraction of REY developed in this study.

5.4 Economic feasibility and future of REY extraction in Europe

The economic feasibility of such project at an industrial scale depends on socio-economic parameters difficult to estimate. Indeed, no analogous extraction and processing plant for a comparable type of resource exist in Europe at this time to stand as a comparison for cost estimation and evaluate economic viability. Projections could be assessed, but unknown factors such as costs of labour, extraction of deposit, transport, comminution, conception and instalment of technical installations, operating costs of the processing plant, life time of operation and remediation, vary intrinsically with the market and location of plant that make any economic projection hard to estimate at this stage. However, the cost of consumables used in this protocol could be estimated and normalized for the production of 1 kg of mixed REY from umbers. A yield of prices for each product for industrial applications (HCl: 150 – 260 $/ton at 31 – 37 % conc., ammonium oxalate: 1000 – 5000 $/ton at 99.8% purity, ammonia: 850 – 920 $/ton at 99.6 % purity) was considered to give a low and high estimate. Similarly, recovery rate for the leaching was set ranging from 80 to 90 % to estimate the required volume of umbers while the initial REY concentration was considered between 350 and 500 ppm, S/L ratio was fixed at 10 and precipitation of REY-oxalates at pH 1.1. In these conditions the production of an oxalate precipitate containing 1 kg of REY costs between 960 and 2110 $. Although this range of price would certainly decrease on a long-term basis by beneficiating from the recycling scheme presented before, the selling price of La-Ce mischmetal remains below 10 $/kg on the

147 Chapter 5. Conclusion international market making this production not economically viable just by considering consumable costs.

Price Conditions Molarity (M) density volume (L) M needed dilution factor volume produced $/ton mol/L L mol/L L HCl 150 - 260 31-37 % Conc. 12.08 1.155 865.8 1.0 12.1 10459 Am. Ox. 1000 - 5000 99,8 % purity 0.3 25000 Ammonia 850 - 920 99,6 % purity 0.6 1666.7 0.3 4.0 6667

Table 5.1: Calculation of produced volume of reactant in experimental settings using prices and conditions of industrial products.

Consumable Price to buy $ ton low high production of 1kg of mix REY from umbers @ 500g/ton Recovery rate kg of rock needed Unknow costs for extraction, transport and 0.8 2500.0 comminution 0.9 2222.2 S/L ratio = 1/10 for HCl leaching Kg of rocks Volume of HCl needed (L) 2500.0 25000.0 2.4 358.5 621.5 2222.2 22222.2 2.1 318.7 552.4 Addition of 1 volume of ammonium Ox for 10 volume of leach Volume of HCl needed (L) Vol of Am. Ox. to add (L) volume total 1 (L) 25000.0 2500.0 27500.0 0.1 100.0 500.0 22222.2 2222.2 24444.4 0.1 88.9 444.4 Addition of Ammonia 1 for 7 by volume volume total 1 Vol of ammonia to add (L) volume total 2 (L) 27500.0 3928.6 31428.6 0.6 500.9 542.1 24444.4 3492.1 27936.5 0.5 445.2 481.9 Total low ($) Total high ($) 852.8 1663.6

Despite the high efficiency and simplicity of the process developed in this study, the extraction of REY from umbers, either as primary ore or as a by-product of an existing plant, appears unlikely in the near future. Beyond geological limitations on deposit distribution and continuity to form a steady supply of ore material, the actual economic context is not favourable for this mining activity. In the current market situation, prices for REE products have never been so low (6-12 $/Kg for mixed REO), decreasing continuously since the prices skyrocketed in 2011. Umber deposits from Cyprus can be consequently considered as a sub economic resource given the current market condition. These low prices directly result from China’s economic strategy with ultra-low production costs and a strongly devaluated currency that combine to prevent

148 Chapter 5. Conclusion development of alternative means of production outside its territory. However, the development of environmental concerns over the last decades has pushed strategic political decisions to

st drastically reduce CO2 emissions and move towards more sustainable economy. The recent 21 Conference of the Parties (COP21) treaty signed in 2015 constitutes one example, legally-binding for participants, to start a significant transition towards the development of green energies. As a result, the demand for high-tech metals essential for the production of wind turbines, photovoltaic cells and other batteries is expected to increase strongly in the coming years. The European Union will have to adapt its political and economic views to favour extraction and production of critical metal in its territory to reduce its vulnerability to the international market. Only after this turnover with significant support from governments will umbers from Cyprus and other industrial waste such as red mud and coal residue be regarded as viable deposits to produce locally REY and other high-tech metals.

5.5 Prospective research

This study has deepened the knowledge on umber deposits and in the light of the promising results obtained here more profound research would be beneficial. A follow-up project on umber using Extended X-ray Absorption Fine Structure (EXAFS) analysis or X-ray Absorption Near Edge Spectra (XANES) using synchrotron radiation could highlight iron and manganese oxidation state and how neighbouring trace metals fit in their frame. Similarly, Fourier Transform Infra-Red (FTIR) analysis on umbers would give more details on the fate of amorphous iron oxyhydroxides evolution and recrystallization into goethite after diagenesis and compaction of deposits. Information extracted from these analyses could provide valuable information on REY binding specificities with major components of metalliferous sediments and explain observed fractionation and preferential incorporation of MREE in hydrothermal particles. In addition, the feedbacks from EXAFS, XANES and FTIR measurements could narrow optimal leaching conditions presented here and orientate future research towards new means of extraction with novel material. Many technics for metal extraction from ferromanganese deposits exist, most of them focusing mainly on chemical leaching, this one included. While the efficiency is important, these processes consume large volume of lixiviants and energy. Of major interest are alternative bioleaching’s using fungi (Mehta et al., 2010) or siderophores (Mohwinkel et al., 2014) that have proven to be efficient for the extractions of critical metals from oxidized PGE ores and marine ferromanganese deposits such as crusts or nodules, but still requires research for improvement.

In an era of extending market and applications for REY, the importance of diversifying and expanding our deposit record is of prime importance. This research project has brought promising results on the potential of low-grade deposits for REY extraction similarly to recent studies on REY

149 Chapter 5. Conclusion extractive metallurgy from red mud and coal residue. These land-based deposits form attractive alternative as valorising already existing industrial waste and important research needs to be undertaken for the extraction of valuable metals from mining by-products enriched during primary ore processing. Future work should also target the recycling of REY-containing product that would improve the internal life-cycle of these elements and lower the supply risk. Only one industry in Europe, Solvay, has developed such recycling processes at the industrial scale for luminophores, batteries and magnets. In terms of volume, important demand lay on neodymium for the production of permanent magnet used in wind turbines and hard-drives. With an estimated product life-time of 15 to 20 years, the expanding market of wind turbines in Europe will produce increasing volumes of waste material already rich in high-tech metals; for instance a 3 MW wind turbine uses magnets up to 2700 kg containing around 15 % Nd [Brown et al., 2002]. However, at the knowledge of the author, no process exists at the moment for the recycling of such material.

The results presented here also connect to the future of marine minerals extraction. Multiple deposits are attracting acute attention for their high grade in Au, Cu (seafloor massive sulphide; SMS); Te, Co, Mo, REY (hydrogenetic crusts) and Cu, Ni, REY (nodules) compared to their land- based deposits counterparts. In addition to these, deep-sea mud has recently been the centre of attention for its potential as a REY resource (Kato et al., 2011). Our investigation on Cyprus umbers has demonstrated the feasibility of REY extraction from hydrothermal sediments analogues to modern plume fall-out of FeMn oxides encountered in the vicinity of deep-sea vents. Although a large number of studies have reported occurrences and grade for deposits of nodules, crusts and SMS, ocean resources remain largely unexplored and their distribution and relation to specific ecological habitat is poorly understood. Future research should not only focus on reporting new discoveries of marine resources and their genesis but rather develop multivariate approach (geology, ecology, oceanography, biology) at the local and basin scale for identifying factors of influence for the formation of marine deposits before any mining activity can be considered. While the impact of terrestrial mining is rather well understood and remediation plans implemented, the economic feasibility and environmental effects of oceanic extraction have yet to be demonstrated and mitigation plans to be designed.

150 Appendices

Appendices

Appendix A: Geochemical data for sample PJ-CY-2014 used in this study

Appendix B: Code used in the PHREEQC modelling approach to REY-oxalate precipitation

Appendix C: Supplementary figures issued from PHREEQC modelling

Appendix D: Other project and published work

151

Appendix A. Supplementary geochemical data on sample PJ-CY-2014-XXX

Appendix A

790

477

ppm

1169

1208

1588

1623

1530

1638

1464

1627

1688

1684

1038

1639

1526

1603

1836

1474

1933

2386

1466

1471

1204

1859

1527

1966

1152

1133

1132

ppm

6,6

6,8

6,6

7,5

7,0

8,7

7,3

6,3

6,7

7,0

5,4

6,2

6,0

5,4

6,0

6,2

5,5

5,2

7,7

8,2

9,9

8,1

5,9

5,8

8,4

9,6

6,5

12,0

ppm

357

365

338

357

352

318

350

332

391

406

408

369

378

388

361

512

375

379

376

394

382

370

326

382

375

349

346

317

338

ppm

342

338

321

288

347

308

346

310

362

354

363

317

324

288

241

259

220

210

231

367

348

282

356

257

324

336

339

288

332

ppm

625

633

624

545

662

620

711

628

717

691

786

691

653

602

825

765

600

638

657

773

984

656

659

682

629

696

742

ppm

2130

1074

184

182

190

170

179

177

176

166

185

274

214

193

188

194

197

206

180

203

209

240

254

247

233

203

198

291

321

174

254

ppm

146

141

126

100

108

120

105

142

145

137

145

167

126

117

149

129

189

215

240

245

230

196

144

117

156

218

109

104

ppm

expressed as oxides in wt.%, trace elements in ppm. in elements trace wt.%, in oxides as expressed

ppm

939

853

978

808

905

783

871

812

947

983

991

834

941

885

854

865

916

849

780

945

ppm

1055

1094

1081

1186

1108

1051

1252

1463

1423

4,2

4,4

4,0

5,4

4,9

6,1

4,8

4,5

3,8

3,7

2,7

3,4

3,7

2,8

3,2

3,0

4,2

2,8

2,6

3,4

4,3

6,1

4,7

3,1

3,2

5,1

5,0

8,5

4,0

Index

0,15

0,17

0,20

0,21

0,17

0,18

0,19

0,16

0,18

0,19

0,22

0,19

0,26

0,28

0,19

0,26

0,27

0,21

0,18

0,16

0,15

0,18

0,26

Mn/Fe

XXX. Major elements are elements Major XXX.

0,7

0,5

0,6

0,5

1,9

1,0

0,5

0,8

1,2

1,3

1,8

1,3

0,6

0,7

1,2

1,6

0,6

0,7

0,6

1,2

1,7

1,0

1,1

0,5

0,6

wt. % wt.

2014

0,2

0,9

0,3

0,2

0,4

0,5

0,3

0,1

0,3

0,6

0,2

0,1

0,2

0,1

0,2

0,1

0,4

0,2

0,1

0,2

0,0

wt. % wt.

0,5

0,6

0,5

0,7

0,8

0,7

0,6

0,5

0,4

0,5

0,6

0,4

0,6

0,5

0,7

0,5

0,8

1,0

0,7

0,4

0,7

0,9

1,1

0,5

wt. % wt.

1,6

1,5

1,3

1,7

1,2

3,5

1,8

1,1

1,4

2,2

2,3

3,1

2,1

1,0

1,5

1,4

2,2

2,8

1,4

2,0

1,8

2,5

3,1

2,0

2,3

1,4

CaO

wt. % wt.

1,8

1,9

2,1

2,2

1,7

2,1

1,7

1,4

1,9

1,3

1,6

1,3

1,7

2,3

1,7

1,2

1,9

1,6

1,7

2,4

2,0

2,9

2,4

2,6

2,0

MgO

wt. % wt.

7,2

7,7

8,5

9,1

8,6

7,9

7,5

8,6

9,1

8,3

8,7

9,1

9,3

9,9

9,7

9,3

9,4

7,5

7,2

7,6

10,9

10,5

13,0

13,8

13,3

12,2

12,4

wt. % wt.

MnO

53,3

53,4

50,7

47,3

50,3

45,3

51,3

49,5

53,8

52,8

58,2

52,6

53,1

53,4

54,4

60,6

54,6

59,4

53,1

57,1

56,5

49,5

52,3

56,9

59,0

50,9

53,2

46,2

53,5

wt. % wt.

3,6

3,8

3,3

4,3

4,1

4,7

4,0

3,6

3,3

3,2

2,4

2,9

3,2

2,4

2,9

4,0

2,9

2,4

3,2

4,3

5,4

4,1

2,8

3,0

4,2

4,3

6,7

3,7

wt. % wt.

0,1

0,2

0,1

0,2

0,1

0,2

0,1

0,2

0,1

0,3

0,2

TiO

wt. % wt.

: Table of geochemical data for samples PJ samples for data geochemical of :Table

16,4

17,6

20,2

17,8

21,3

17,5

16,6

14,8

15,5

11,8

14,5

14,3

13,6

17,6

14,5

15,7

13,6

10,3

15,8

15,7

20,0

19,9

16,9

15,6

23,1

21,3

26,8

19,1

SiO

wt. % wt.

Appendix Appendix

PJ-CY-2014-38

PJ-CY-2014-37

PJ-CY-2014-36

PJ-CY-2014-34

PJ-CY-2014-33

PJ-CY-2014-32

PJ-CY-2014-31

PJ-CY-2014-30

PJ-CY-2014-29

PJ-CY-2014-28

PJ-CY-2014-27

PJ-CY-2014-26

PJ-CY-2014-25

PJ-CY-2014-23

PJ-CY-2014-21

PJ-CY-2014-20

PJ-CY-2014-19

PJ-CY-2014-18

PJ-CY-2014-17

PJ-CY-2014-15

PJ-CY-2014-14

PJ-CY-2014-13

PJ-CY-2014-11

PJ-CY-2014-10

PJ-CY-2014-6

PJ-CY-2014-4

PJ-CY-2014-3

PJ-CY-2014-2 PJ-CY-2014-1 Unaltered umbersMassive

153

Appendix A. Supplementary geochemical data on sample PJ-CY-2014-XXX

Ch.

0,72

0,74

0,72

0,73

0,72

0,73

0,72

0,77

0,73

0,72

0,73

0,75

0,73

0,75

0,74

0,75

0,76

0,74

0,73

0,74

0,73

0,74

Eu/Eu*

Ch.

0,17

0,18

0,19

0,21

0,20

0,21

0,16

0,12

0,16

0,13

0,20

0,16

0,11

0,13

0,16

0,13

0,22

0,13

0,19

0,18

0,33

0,20

0,12

0,13

0,15

0,13

0,25

0,23

Ce/Ce*

REE

417

373

353

424

435

434

433

436

638

467

399

376

430

399

341

340

410

215

348

382

372

494

319

421

369

439

445

340

334

ppm

1,3

1,2

1,0

1,3

1,2

1,3

1,5

2,0

1,3

1,2

1,0

1,2

0,7

1,1

1,3

1,4

1,1

1,3

1,6

1,7

1,1

ppm

9,8

8,9

7,7

9,5

9,4

9,0

9,5

9,7

9,6

9,0

9,9

9,7

8,5

7,7

9,3

5,2

8,4

9,6

9,7

8,0

9,9

9,5

8,1

7,9

11,2

14,8

10,7

11,5

12,3

ppm

1,6

1,4

1,2

1,6

1,5

1,8

2,5

1,6

1,5

1,6

1,4

1,2

1,5

0,8

1,3

1,6

1,8

1,3

1,6

1,9

2,0

1,3

ppm

8,7

5,4

9,4

9,0

9,2

8,9

11,5

10,2

11,2

10,7

10,6

10,7

12,9

17,8

11,4

10,9

10,3

11,5

11,3

10,0

11,3

11,0

10,9

12,6

12,0

11,0

13,3

14,4

ppm

4,4

3,9

3,4

4,2

4,1

4,8

6,7

4,4

4,1

4,3

4,2

3,8

3,4

4,3

2,0

3,6

4,2

4,1

4,8

3,4

4,6

4,2

5,0

5,3

3,5

3,4

ppm

22,3

19,8

18,1

21,4

21,1

21,5

21,7

22,7

33,6

23,9

20,8

19,8

21,7

21,0

18,9

17,8

21,8

10,1

17,9

21,5

20,6

24,8

17,0

23,5

20,6

25,3

26,6

18,0

17,1

ppm

3,8

3,4

3,2

3,7

3,8

5,7

4,2

3,6

3,7

3,6

3,2

3,1

3,7

1,7

3,1

3,7

3,5

4,2

2,9

4,0

3,5

4,3

4,5

3,1

2,9

ppm

26,4

23,2

21,2

25,6

25,0

25,7

25,3

26,2

39,6

28,7

24,4

22,6

25,5

24,6

22,1

21,0

26,0

11,8

21,2

24,5

23,0

26,7

19,8

27,8

23,5

28,9

29,6

20,3

19,7

ppm

5,9

5,3

5,2

5,9

6,0

6,1

5,8

9,1

7,0

5,6

5,7

6,0

5,6

4,9

5,0

6,0

2,9

4,9

5,8

5,4

6,5

4,7

6,4

5,5

6,7

6,8

4,8

4,6

ppm

23,8

21,3

21,6

24,2

24,7

24,9

25,6

22,8

37,1

29,6

23,2

21,6

24,1

23,0

19,2

19,8

23,8

11,5

19,6

23,5

21,5

25,2

18,9

25,8

21,6

26,9

26,8

19,9

18,6

ppm

96,1

93,1

53,1

96,5

99,5

87,1

92,3

88,6

ppm

115,5

102,4

101,4

114,9

117,2

121,7

119,9

116,2

187,5

138,7

113,3

102,7

119,8

113,3

117,1

107,4

121,0

125,1

106,5

128,8

129,3

28,7

25,4

28,4

29,8

30,7

30,3

29,1

47,1

35,1

28,7

27,2

30,0

28,6

23,6

23,1

28,9

13,4

24,1

26,6

25,1

31,7

21,5

30,7

26,3

31,9

32,0

22,9

21,9

ppm

Appendix A continued… A Appendix

40,7

39,7

39,5

51,7

50,8

51,5

52,2

40,9

46,1

44,7

31,6

40,8

40,7

26,6

25,8

32,1

33,1

27,1

27,8

40,5

38,1

88,8

36,7

30,2

28,1

38,1

33,7

46,7

43,8

ppm

94,9

69,0

88,0

88,5

94,1

ppm

121,4

107,2

120,8

129,5

122,0

121,3

136,0

188,4

126,1

120,9

105,5

129,9

124,3

102,6

121,9

109,3

100,8

108,1

134,3

118,0

106,0

114,3

119,8

2,8

3,0

2,7

2,9

2,6

3,1

3,0

3,3

3,0

3,3

2,4

3,3

3,0

2,8

3,0

2,8

3,0

3,5

2,9

2,8

1,9

2,0

3,9

2,4

1,6

2,3

ppm

1,9

2,0

1,6

2,3

1,7

1,9

1,5

1,3

2,0

1,9

1,0

1,1

1,4

1,3

1,1

1,2

1,7

1,9

1,7

1,3

1,4

1,7

1,5

2,8

2,4

ppm

183

173

167

244

212

211

199

173

199

133

179

150

124

131

148

133

127

131

154

153

373

144

135

132

151

131

189

180

ppm

884

733

846

841

760

744

829

701

799

865

875

857

846

965

888

929

899

936

638

ppm

1080

1113

1641

2025

2114

2238

1973

1255

1012

1783

4,4

4,3

4,2

5,7

5,3

5,5

5,3

4,3

5,0

5,5

4,5

4,2

4,6

3,1

2,8

3,4

3,5

2,8

2,9

4,1

3,9

4,3

3,7

3,3

3,0

3,9

3,5

6,1

4,0

ppm

100

111

118

109

123

114

124

ppm

115

102

104

133

167

100

108

115

105

109

100

115

108

125

130

ppm

PJ-CY-2014-38

PJ-CY-2014-37

PJ-CY-2014-36

PJ-CY-2014-34

PJ-CY-2014-33

PJ-CY-2014-32

PJ-CY-2014-31

PJ-CY-2014-30

PJ-CY-2014-29

PJ-CY-2014-28

PJ-CY-2014-27

PJ-CY-2014-26

PJ-CY-2014-25

PJ-CY-2014-23

PJ-CY-2014-21

PJ-CY-2014-20

PJ-CY-2014-19

PJ-CY-2014-18

PJ-CY-2014-17

PJ-CY-2014-15

PJ-CY-2014-14

PJ-CY-2014-13

PJ-CY-2014-11

PJ-CY-2014-10

PJ-CY-2014-6

PJ-CY-2014-4

PJ-CY-2014-3

PJ-CY-2014-2 PJ-CY-2014-1 Unaltered umbersMassive

154

Appendix A. Supplementary geochemical data on sample PJ-CY-2014-XXX

586

223

257

309

244

303

313

494

763

694

740

924

615

602

642

174

129

526

580

821

324

266

272

313

362

332

ppm

2610

1788

1089

1144

9,5

8,7

ppm

6,9

8,2

6,4

8,6

6,8

6,7

4,9

5,2

5,7

4,7

3,7

4,9

7,2

5,6

9,1

8,1

7,4

4,9

6,5

8,2

7,2

5,9

6,7

5,7

7,2

12,3

10,1

11,6

10,5

10,0

ppm

339

280

345

157

225

241

220

246

397

470

536

614

548

432

595

487

518

350

271

329

325

323

368

302

336

352

335

360

282

341

ppm

460

353

372

255

361

404

314

379

347

293

238

254

245

237

238

237

243

349

266

413

343

212

355

327

321

317

273

252

326

316

ppm

732

590

683

593

757

791

801

946

634

701

714

692

714

688

734

794

616

706

805

593

632

590

620

663

803

675

746

635

ppm

1010

1096

295

161

175

263

338

345

308

334

196

151

152

156

154

160

179

184

199

247

167

358

342

189

259

219

224

228

215

206

321

161

ppm

136

145

129

100

142

149

119

151

141

126

193

191

179

183

177

187

173

141

174

159

196

161

180

158

132

141

140

ppm

925

472

625

676

642

717

753

846

594

947

852

920

822

825

962

987

904

925

710

801

ppm

1116

1078

1272

1475

1477

1346

1183

1311

1173

1112

3,9

5,6

4,1

8,4

6,5

8,1

6,5

3,7

2,9

2,0

1,9

2,5

1,9

1,6

2,0

3,3

3,2

7,4

5,0

4,0

2,6

4,1

5,5

3,9

2,7

3,0

7,4

5,2

13,4

Index

0,18

0,21

0,16

0,37

0,32

0,35

0,34

0,24

0,13

0,12

0,13

0,11

0,12

0,16

0,22

0,17

0,24

0,29

0,30

0,47

0,21

0,28

0,32

0,30

0,29

0,37

0,16

Mn/Fe

0,7

0,6

0,8

0,7

0,6

0,5

0,6

1,0

0,6

1,0

0,8

0,7

0,8

1,3

1,2

0,6

0,5

0,3

0,7

0,6

0,7

0,8

0,6

1,0

0,7

0,8

0,7

0,5

wt. % wt.

Appendix A continued… A Appendix

0,2

0,1

0,5

0,4

0,7

1,6

0,3

0,2

0,1

0,0

0,1

0,2

0,4

0,0

0,2

0,1

0,4

0,6

0,3

0,4

0,8

0,5

0,9

0,1

wt. % wt.

0,7

1,1

0,6

1,7

1,3

1,1

1,2

0,9

0,5

0,3

0,2

0,4

0,3

0,2

0,3

0,5

0,8

0,5

0,6

0,8

0,5

0,3

0,4

1,3

0,6

wt. % wt.

1,1

1,6

2,1

1,5

1,3

1,1

1,2

1,4

1,2

1,6

1,3

1,4

1,8

1,9

2,0

1,6

1,1

0,9

1,4

1,2

1,3

1,2

1,7

1,3

1,7

1,4

CaO

wt. % wt.

1,6

1,8

2,0

3,1

2,3

2,0

2,4

2,0

1,7

1,2

1,1

1,0

1,2

1,4

1,0

1,1

1,6

1,2

2,8

1,6

1,5

1,7

1,5

2,1

2,0

1,5

1,4

2,2

2,5

MgO

wt. % wt.

9,8

8,2

9,5

7,0

7,2

7,8

6,4

6,8

8,3

8,7

8,2

9,9

6,8

10,2

12,3

12,4

12,2

12,6

11,3

10,4

12,2

14,7

19,5

11,5

13,8

14,7

13,7

12,7

wt. % wt.

MnO

60,2

54,2

58,3

28,2

36,6

42,9

38,1

41,3

51,8

57,2

67,1

67,9

64,6

63,4

60,6

57,6

52,1

57,2

38,3

47,1

54,9

46,1

52,4

44,9

47,8

50,4

50,8

52,1

38,3

48,3

wt. % wt.

3,9

5,1

3,8

7,9

6,1

5,2

6,0

5,1

3,3

2,6

2,0

1,9

2,5

1,8

1,4

1,8

2,8

2,9

5,0

4,2

3,9

2,4

3,6

4,5

3,3

2,5

2,7

2,8

5,6

4,0

wt. % wt.

0,2

0,3

0,2

0,1

0,2

0,1

0,2

0,1

0,2

TiO

wt. % wt.

17,5

20,6

18,7

34,3

24,8

21,1

24,1

20,2

14,8

14,7

14,3

12,8

16,0

20,5

12,0

13,0

15,7

15,1

30,4

17,1

16,9

12,0

14,6

19,1

15,3

12,2

13,0

12,6

21,7

21,1

SiO

wt. % wt.

PJ-CY-2014-111

PJ-CY-2014-110

PJ-CY-2014-109

PJ-CY-2014-94

PJ-CY-2014-93

PJ-CY-2014-92

PJ-CY-2014-91

PJ-CY-2014-90

PJ-CY-2014-89

PJ-CY-2014-75

PJ-CY-2014-73

PJ-CY-2014-72

PJ-CY-2014-71

PJ-CY-2014-69

PJ-CY-2014-67

PJ-CY-2014-66

PJ-CY-2014-65

PJ-CY-2014-62

PJ-CY-2014-61

PJ-CY-2014-59

PJ-CY-2014-58

PJ-CY-2014-52

PJ-CY-2014-49

PJ-CY-2014-48

PJ-CY-2014-47

PJ-CY-2014-46

PJ-CY-2014-45

PJ-CY-2014-44

PJ-CY-2014-42 PJ-CY-2014-40 Unaltered umbersMassive

155

Appendix A. Supplementary geochemical data on sample PJ-CY-2014-XXX

Ch.

2,99

0,74

0,73

0,72

0,73

0,75

0,74

0,75

0,74

0,76

0,74

0,73

0,75

0,73

0,74

0,73

0,74

0,73

Eu/Eu*

Ch.

0,74

0,22

0,17

0,33

0,26

0,32

0,28

0,23

0,19

0,17

0,10

0,13

0,16

0,12

0,10

0,13

0,20

0,30

0,29

0,19

0,23

0,22

0,19

0,18

0,17

0,16

0,20

0,22

0,23

Ce/Ce*

REE

269

490

427

586

560

447

427

480

408

273

303

203

241

223

244

212

245

250

237

517

305

249

360

496

408

325

343

269

471

324

ppm

1,3

1,4

1,5

1,7

1,5

1,3

1,6

1,5

1,2

1,1

0,8

0,9

1,0

0,9

1,0

0,9

1,7

1,1

1,0

1,3

1,4

1,1

1,2

1,4

1,1

1,4

1,0

ppm

1,4

9,0

8,6

8,4

5,6

6,9

6,4

7,5

6,3

7,0

7,1

6,6

7,8

7,3

9,5

8,4

8,6

7,7

7,3

10,5

10,4

11,8

12,2

10,1

11,2

11,0

12,3

10,7

10,0

10,4

ppm

1,7

2,0

1,6

1,5

1,8

1,3

0,9

1,1

1,0

1,2

1,0

1,1

2,0

1,2

1,5

1,8

1,4

1,6

1,2

1,7

1,2

10,3

ppm

1,6

9,2

6,2

7,3

7,0

8,5

6,9

7,7

7,6

7,2

8,6

7,9

9,6

9,9

8,4

8,1

12,1

12,2

14,0

14,4

10,9

10,3

13,1

12,6

14,4

10,7

12,6

11,5

12,0

ppm

4,6

5,3

5,5

4,1

3,9

4,9

4,7

3,4

3,5

2,2

2,7

2,5

3,1

2,5

2,9

2,8

2,6

5,5

3,2

2,9

4,0

4,8

3,7

3,6

4,2

3,1

4,5

3,1

11,2

ppm

4,1

24,3

23,1

26,7

27,9

20,9

20,0

25,2

23,6

17,3

18,2

11,0

14,2

12,3

14,6

12,1

14,9

14,3

13,7

27,9

15,9

14,1

20,0

25,5

19,8

18,0

20,8

15,5

23,1

16,4

ppm

4,3

4,0

4,6

4,8

3,6

3,5

4,3

4,0

2,9

3,1

1,8

2,4

2,0

2,4

2,0

2,5

2,4

2,3

4,8

2,7

2,3

3,4

4,5

3,5

3,0

3,5

2,6

4,0

2,8

21,1

ppm

3,5

28,2

26,7

31,4

32,5

24,1

23,3

29,0

26,2

17,7

19,3

12,1

14,5

12,9

16,0

13,0

15,7

15,3

14,6

32,0

18,4

15,3

23,0

29,9

23,4

20,7

23,4

17,4

27,5

18,9

ppm

6,9

6,2

7,2

7,6

5,7

5,6

6,9

6,2

4,3

4,8

2,9

3,7

3,0

3,6

2,9

3,8

3,6

3,5

7,5

4,3

3,5

5,3

7,2

5,8

4,7

5,3

4,1

6,5

4,6

22,9

ppm

5,4

29,3

25,3

29,2

31,2

23,7

22,5

27,2

25,0

17,7

20,0

11,5

14,8

12,1

13,8

11,4

15,5

14,4

30,5

17,2

14,0

21,4

30,3

25,3

18,9

20,6

16,2

26,1

19,3

ppm

21,5

76,4

91,4

53,8

67,4

57,7

66,3

55,3

67,5

63,3

62,5

78,7

65,2

99,7

89,8

96,8

74,2

88,8

ppm

136,2

119,2

147,0

150,5

112,1

110,1

130,6

114,7

145,6

143,1

119,1

127,7

99,8

34,1

28,9

36,3

36,8

27,5

27,0

31,6

27,8

19,5

23,2

13,2

17,4

14,6

16,5

13,9

17,0

15,5

15,6

36,0

19,1

16,1

24,1

35,7

30,4

21,8

23,1

18,0

31,7

22,4

ppm

Appendix A continued… A Appendix

24,6

60,6

41,9

80,3

76,9

66,3

60,9

42,9

25,6

18,1

15,6

22,4

15,9

15,0

16,7

26,4

38,8

35,7

57,1

38,9

29,9

38,6

51,8

41,0

29,6

30,4

29,7

59,4

42,5

ppm

104,8

40,0

68,0

81,8

65,5

65,4

75,2

75,0

67,1

61,8

63,1

56,8

87,7

68,6

97,3

93,8

90,7

69,5

87,4

ppm

136,0

121,1

164,5

153,0

123,9

122,5

131,6

105,7

139,6

136,7

115,2

135,3

1,7

3,4

4,1

1,4

1,6

2,0

1,5

1,9

2,3

1,0

1,2

1,3

1,1

1,0

1,3

1,1

0,9

3,4

1,9

4,3

3,6

1,7

2,3

2,0

2,1

2,6

2,3

2,1

2,3

2,5

ppm

2,1

3,4

2,3

6,5

4,9

4,5

4,6

3,8

2,0

1,2

0,8

0,6

1,2

0,7

0,8

1,1

2,3

2,7

2,9

2,0

1,3

2,4

2,9

1,7

1,5

3,6

2,0

ppm

2,4

249

205

328

322

327

296

294

242

142

103

102

138

191

159

275

199

160

212

230

191

155

151

153

264

166

ppm

965

872

982

788

990

773

754

948

822

839

851

245

202

997

814

893

875

693

ppm

1283

1024

1103

1277

1237

1428

1302

1214

4058

1111

1027

1275

4,1

6,0

4,5

8,0

6,9

7,5

6,0

5,5

4,0

3,0

2,2

2,0

2,8

1,8

1,7

2,0

2,9

3,8

4,8

5,6

4,2

2,7

4,4

5,5

5,3

3,3

3,1

6,3

4,4

ppm

128

105

122

119

139

115

118

129

120

114

118

102

ppm

121

142

141

102

131

120

134

102

118

106

127

114

ppm

PJ-CY-2014-111

PJ-CY-2014-110

PJ-CY-2014-109

PJ-CY-2014-94

PJ-CY-2014-93

PJ-CY-2014-92

PJ-CY-2014-91

PJ-CY-2014-90

PJ-CY-2014-89

PJ-CY-2014-75

PJ-CY-2014-73

PJ-CY-2014-72

PJ-CY-2014-71

PJ-CY-2014-69

PJ-CY-2014-67

PJ-CY-2014-66

PJ-CY-2014-65

PJ-CY-2014-62

PJ-CY-2014-61

PJ-CY-2014-59

PJ-CY-2014-58

PJ-CY-2014-52

PJ-CY-2014-49

PJ-CY-2014-48

PJ-CY-2014-47

PJ-CY-2014-46

PJ-CY-2014-45

PJ-CY-2014-44

PJ-CY-2014-42 PJ-CY-2014-40 Unaltered umbersMassive

156

Appendix A. Supplementary geochemical data on sample PJ-CY-2014-XXX

496

462

414

628

907

728

730

ppm

1089

1371

1355

1570

1615

1681

1212

1,8

6,4

2,4

3,6

5,0

3,0

3,6

2,5

8,9

8,6

ppm

0,9

2,7

0,2

0,3

1,7

2,0

1,5

7,5

6,1

8,6

9,1

7,2

7,9

5,1

4,6

5,1

7,4

5,3

5,0

4,9

5,2

5,9

6,2

13,8

11,8

13,9

ppm

4,4

8,9

6,9

8,5

ppm

231

100

155

227

282

133

124

415

365

294

422

387

436

500

255

267

389

334

400

374

335

ppm

207

175

102

121

182

203

419

235

216

456

467

371

214

196

199

223

240

243

274

268

297

258

300

ppm

159

185

208

240

178

260

152

434

247

260

519

514

548

522

569

576

659

689

468

610

644

705

666

607

ppm

112

107

110

108

237

275

260

224

271

301

112

133

136

214

201

177

202

191

207

210

265

ppm

214

145

212

261

207

123

235

275

217

165

177

143

161

205

100

122

102

150

206

111

ppm

176

109

174

134

126

116

173

ppm

407

193

350

430

333

892

993

768

608

829

562

862

875

ppm

1004

2473

2261

2024

2992

1074

1063

1086

1038

1060

1081

1207

3,0

3,3

2,2

1,7

0,7

2,9

3,0

2,6

5,1

3,0

4,1

4,3

3,7

4,0

5,3

2,5

1,9

2,2

3,3

8,4

4,3

2,4

2,7

2,3

3,0

3,7

10,8

Index

0,02

0,03

0,11

0,19

0,22

0,20

0,21

0,04

0,02

0,03

0,04

0,03

0,02

0,03

0,01

0,14

0,15

0,12

0,22

0,21

0,17

0,18

0,25

0,17

0,26

0,14

Mn/Fe

1,9

0,9

0,0

0,1

0,8

0,2

0,8

0,9

0,6

0,5

0,7

0,6

1,1

9,9

7,5

6,4

3,4

1,7

7,0

3,0

4,8

2,7

2,4

4,2

wt. % wt.

Appendix A continued… A Appendix

bdl

0,1

0,0

0,1

0,2

0,1

0,0

0,4

0,2

0,3

0,1

0,2

0,1

0,03

wt. % wt.

0,1

0,2

0,1

0,4

0,2

0,5

1,0

2,2

0,3

0,5

0,7

0,2

0,3

1,2

0,5

0,3

0,4

0,3

0,5

wt. % wt.

2,7

1,4

0,7

0,2

2,2

1,6

0,5

1,5

1,7

2,4

0,6

0,4

1,2

0,7

0,6

1,7

5,9

4,7

4,9

8,1

4,4

4,2

7,1

CaO

14,9

11,7

10,2

11,8

wt. % wt.

0,0

0,3

0,0

0,1

0,2

0,5

0,7

0,4

1,9

1,2

1,5

1,2

2,2

2,6

1,4

1,7

1,0

1,7

1,9

1,3

1,6

1,4

1,6

1,4

2,1

MgO

wt. % wt.

0,2

0,5

0,1

0,6

2,0

2,9

2,4

1,7

1,5

2,0

1,2

1,3

1,6

0,7

6,2

7,0

5,5

7,2

6,5

8,3

9,1

6,7

10,5

13,0

wt. % wt.

MnO

4,1

5,9

10,9

24,3

11,8

14,3

15,6

12,4

49,6

68,0

63,2

58,9

44,3

60,1

63,1

64,8

49,8

52,4

48,7

53,7

38,0

42,0

52,3

45,8

59,3

55,4

51,4

wt. % wt.

0,4

1,1

0,1

0,7

0,8

0,5

3,7

2,8

3,7

3,6

7,3

3,1

3,6

4,9

1,9

1,6

3,0

5,6

2,9

2,0

2,1

2,2

2,9

3,0

wt. % wt.

0,0

0,1

0,0

0,2

0,1

0,3

0,1

0,2

0,1

0,2

0,1

TiO

wt. % wt.

88,1

64,1

90,5

87,6

75,8

81,7

72,6

83,9

38,8

21,3

15,3

21,5

39,7

16,8

15,1

17,0

11,3

13,7

12,2

16,2

24,2

13,2

12,7

11,4

12,7

12,2

17,3

SiO

wt. % wt.

PJ-CY-2014-99

PJ-CY-2014-56

PJ-CY-2014-54

PJ-CY-2014-51

PJ-CY-2014-50

PJ-CY-2014-9

PJ-CY-2014-8

PJ-CY-2014-7

Silicified umbers

PJ-CY-2014-100

PJ-CY-2014-98

PJ-CY-2014-88

PJ-CY-2014-87

PJ-CY-2014-86

PJ-CY-2014-57

PJ-CY-2014-53

PJ-CY-2014-12

Mn-depleeted umbers

PJ-CY-2014-74

PJ-CY-2014-70

PJ-CY-2014-68

PJ-CY-2014-64

PJ-CY-2014-43

PJ-CY-2014-41

PJ-CY-2014-39

PJ-CY-2014-24

PJ-CY-2014-22

PJ-CY-2014-16 PJ-CY-2014-5 Apatite-rich umbers

157

Appendix A. Supplementary geochemical data on sample PJ-CY-2014-XXX

Ch.

0,70

0,72

0,99

0,71

0,72

0,73

0,72

0,71

0,72

0,75

0,74

0,76

0,69

0,70

0,72

0,74

0,70

0,72

0,73

Eu/Eu*

Ch.

0,13

0,14

0,56

0,21

0,26

0,08

0,17

0,08

0,22

0,19

0,15

0,21

0,30

0,09

0,15

0,13

0,02

0,04

0,03

0,10

0,18

0,07

0,08

0,07

0,09

0,08

Ce/Ce*

REE

114

213

149

117

341

227

226

338

285

177

229

317

550

635

349

369

719

477

561

441

544

596

ppm

1390

0,3

0,8

0,0

0,1

0,5

0,3

0,4

1,0

1,1

1,3

1,0

1,2

1,6

4,3

2,0

2,1

1,2

1,7

3,1

1,6

2,0

1,6

1,8

2,2

ppm

2,3

5,5

0,2

0,4

0,8

3,7

2,1

3,0

7,3

7,8

8,7

9,0

7,1

8,8

9,0

11,5

30,9

14,5

15,6

11,9

22,0

11,9

14,9

11,3

13,1

16,0

ppm

0,4

0,9

0,0

0,1

0,6

0,3

0,5

1,2

1,3

1,4

1,1

1,3

1,8

5,4

2,4

2,7

1,5

1,9

3,6

2,0

2,5

1,8

2,2

2,6

ppm

2,9

6,3

0,2

0,4

0,7

4,4

2,3

3,5

8,7

8,0

7,3

8,7

10,1

12,6

41,1

17,9

19,7

10,6

13,6

24,5

14,1

17,6

13,1

15,6

18,5

ppm

1,1

2,3

0,0

0,1

0,2

1,6

0,9

1,3

3,3

3,1

3,7

3,0

2,6

3,1

4,6

6,6

7,5

4,0

4,9

8,5

5,2

6,4

4,8

5,8

6,7

15,6

ppm

5,1

0,2

0,7

1,0

7,9

4,4

6,2

11,2

16,5

15,3

18,7

15,0

12,8

15,2

22,4

74,0

29,9

36,2

20,0

23,6

38,3

25,3

30,3

23,5

28,8

32,4

ppm

0,8

1,9

0,0

0,1

1,3

0,8

1,0

2,9

2,5

3,1

2,6

2,1

2,5

3,7

4,8

6,1

3,4

3,8

6,0

4,2

4,9

3,9

4,9

5,4

12,7

ppm

6,0

0,1

0,8

0,9

9,1

5,2

7,1

13,1

20,0

15,6

16,1

19,8

17,1

13,9

16,1

24,4

94,6

34,0

43,2

23,0

24,4

41,1

29,5

34,4

27,4

34,1

36,4

ppm

1,2

3,0

0,0

0,2

2,0

1,2

1,6

4,5

3,4

3,8

4,8

4,0

3,2

3,7

5,7

6,9

9,3

5,3

5,5

8,5

6,6

7,5

6,1

7,7

8,3

19,8

ppm

4,9

0,1

0,8

0,6

8,1

4,9

6,3

12,4

18,8

13,6

15,1

19,9

16,5

12,3

14,8

20,8

80,6

26,2

36,7

21,1

20,8

33,2

26,5

29,0

24,2

30,9

33,2

ppm

0,2

3,6

2,8

26,6

59,1

41,8

22,5

32,4

93,0

60,0

65,6

94,5

76,4

53,3

64,8

92,2

97,0

ppm

442,4

144,9

190,5

100,8

188,4

135,6

158,0

122,9

161,2

172,8

6,7

0,1

0,9

0,6

5,6

8,0

15,0

10,4

23,0

14,5

15,6

23,5

18,6

12,3

15,6

21,5

35,6

46,1

24,6

23,7

47,5

34,1

39,4

30,9

39,9

43,7

ppm

105,7

Appendix A continued… A Appendix

9,3

0,2

1,6

1,3

7,3

8,6

6,0

7,9

17,3

43,3

22,6

17,3

39,0

44,2

18,4

21,5

16,1

13,7

14,0

20,9

35,9

30,8

23,7

23,9

24,2

29,7

31,2

ppm

0,2

3,9

2,5

46,0

64,1

50,0

25,6

39,7

97,4

57,9

52,2

89,1

70,2

40,2

54,9

73,2

ppm

446,9

210,7

205,2

103,5

100,0

263,1

156,9

190,2

145,5

168,2

186,1

1,1

1,0

0,1

1,1

0,5

2,3

2,7

2,3

0,7

0,9

2,1

0,8

0,7

2,7

2,0

2,8

7,1

2,9

3,1

1,6

2,2

3,7

3,5

3,7

3,4

ppm

0,3

0,7

0,0

0,1

0,0

0,3

0,2

2,3

1,8

2,2

2,6

4,4

1,6

2,0

1,4

0,7

0,6

0,8

1,4

1,9

0,9

1,0

0,8

1,3

1,4

ppm

2,8

6,4

1,0

112

109

146

185

172

104

110

146

164

114

129

116

124

129

ppm

478

447

377

339

736

824

535

979

917

838

864

863

ppm

1130

1010

1713

0,6

1,5

0,0

0,2

0,1

0,8

0,9

0,7

3,3

3,5

4,3

3,9

5,6

3,4

4,1

3,3

2,0

1,6

1,8

2,4

3,2

2,9

3,1

3,5

ppm

0,4

2,7

1,2

ppm

1,5

3,6

6,1

126

484

234

227

112

129

255

142

180

128

151

175

ppm

PJ-CY-2014-99

PJ-CY-2014-56

PJ-CY-2014-54

PJ-CY-2014-51

PJ-CY-2014-50

PJ-CY-2014-9

PJ-CY-2014-8

PJ-CY-2014-7

Silicified umbers

PJ-CY-2014-100

PJ-CY-2014-98

PJ-CY-2014-88

PJ-CY-2014-87

PJ-CY-2014-86

PJ-CY-2014-57

PJ-CY-2014-53

PJ-CY-2014-12

Mn-depleeted umbers

PJ-CY-2014-74

PJ-CY-2014-70

PJ-CY-2014-68

PJ-CY-2014-64

PJ-CY-2014-43

PJ-CY-2014-41

PJ-CY-2014-39

PJ-CY-2014-24

PJ-CY-2014-22

PJ-CY-2014-16 PJ-CY-2014-5 Apatite-rich umbers

158 Appendix A. Supplementary geochemical data on sample PJ-CY-2014-XXX

Ch.

103

171

0,72

ppm

Eu/Eu*

Ch.

0,83

0,79

ppm

Ce/Ce*

REE

6,3

8,6

112

ppm

0,2

0,3

1,0

0,5

ppm

1,6

2,2

ppm

0,3

0,4

2,6

2,4

ppm

1,7

2,4

ppm

0,6

0,9

ppm

3,3

4,3

ppm

0,6

0,7

ppm

3,8

4,8

ppm

0,9

1,1

ppm

4,1

4,8

ppm

49,6

49,7

Index

19,1

22,5

ppm

0,09

0,05

Mn/Fe

4,8

5,7

0,1

0,2

ppm

wt. % wt.

Appendix A continued… A Appendix

0,2

0,4

32,0

38,0

ppm

wt. % wt.

0,8

0,9

17,8

23,6

ppm

wt. % wt.

0,7

0,5

0,7

1,0

CaO

ppm

wt. % wt.

3,2

2,8

1,4

1,8

ppm

MgO

wt. % wt.

0,3

0,2

ppm

wt. % wt.

MnO

3,6

4,5

ppm

wt. % wt.

3,7

3,6

5,1

6,1

ppm

wt. % wt.

0,2

0,3

ppm

TiO

wt. % wt.

ppm

80,8

78,3

SiO

wt. % wt.

PJ-CY-2014-105

PJ-CY-2014-104

Radiolarian cherts

PJ-CY-2014-105 PJ-CY-2014-104 Radiolarian cherts

159

Appendix B. PHREEQC code for oxalate precipitation experiment Appendix B

PHREEQC code for oxalate precipitation experiment

SOLUTION_MASTER_SPECIES Ox Ox-2 0.0 Ox 88.03 #

SOLUTION_SPECIES Ox-2 = Ox-2 log_k 0

H+ + Ox-2 = HOx- log_k 3.92

H+ + HOx- = H2Ox log_k 1.09

Ca+2 + Ox-2 = CaOx log_k 2.46

Ca+2 + H2O + Ox-2 = CaOx:H2O log_k 8.6

Ca+2 + 3H2O + Ox-2 = CaOx:3H2O log_k 8.14

#REOx+ La+3 + Ox-2 = LaOx+ log_k 5.87 #Schijf and Byrne 2001

Ce+3 + Ox-2 = CeOx+ log_k 5.97 #Schijf and Byrne 2001

Ox-2 + Pr+3 = PrOx+ log_k 6.25 #Schijf and Byrne 2001

Nd+3 + Ox-2 = NdOx+ log_k 6.31 #Schijf and Byrne 2001

161 Appendix B. PHREEQC code for oxalate precipitation experiment

Ox-2 + Sm+3 = SmOx+ log_k 6.43 #Schijf and Byrne 2001

Eu+3 + Ox-2 = EuOx+ log_k 6.52 #Schijf and Byrne 2001

Gd+3 + Ox-2 = GdOx+ log_k 6.53 #Schijf and Byrne 2001

Ox-2 + Tb+3 = TbOx+ log_k 6.63 #Schijf and Byrne 2001

Dy+3 + Ox-2 = DyOx+ log_k 6.74 #Schijf and Byrne 2001

Ho+3 + Ox-2 = HoOx+ log_k 6.77 #Schijf and Byrne 2001

Er+3 + Ox-2 = ErOx+ log_k 6.83 #Schijf and Byrne 2001

Tm+3 + Ox-2 = TmOx+ log_k 6.89 #Schijf and Byrne 2001

Yb+3 + Ox-2 = YbOx+ log_k 6.95 #Schijf and Byrne 2001

Lu+3 + Ox-2 = LuOx+ log_k 6.96 #Schijf and Byrne 2001

#REOX2- La+3 + 2Ox-2 = La(Ox)2- log_k 10.47 #Schijf and Byrne 2001

Ce+3 + 2Ox-2 = Ce(Ox)2- log_k 10.86 #Schijf and Byrne 2001

Pr+3 + 2Ox-2 = Pr(Ox)2- log_k 10.82 #Schijf and Byrne 2001

Nd+3 + 2Ox-2 = Nd(Ox)2- log_k 10.82 #Schijf and Byrne 2001

162 Appendix B. PHREEQC code for oxalate precipitation experiment

Sm+3 + 2Ox-2 = Sm(Ox)2- log_k 11.08 #Schijf and Byrne 2001

Eu+3 + 2Ox-2 = Eu(Ox)2- log_k 11.09 #Schijf and Byrne 2001

Gd+3 + 2Ox-2 = Gd(Ox)2- log_k 11.1 #Schijf and Byrne 2001

Tb+3 + 2Ox-2 = Tb(Ox)2- log_k 11.27 #Schijf and Byrne 2001

Dy+3 + 2Ox-2 = Dy(Ox)2- log_k 11.35 #Schijf and Byrne 2001

Ho+3 + 2Ox-2 = Ho(Ox)2- log_k 11.41 #Schijf and Byrne 2001

Er+3 + 2Ox-2 = Er(Ox)2- log_k 11.51 #Schijf and Byrne 2001

Tm+3 + 2Ox-2 = Tm(Ox)2- log_k 11.65 #Schijf and Byrne 2001

Yb+3 + 2Ox-2 = Yb(Ox)2- log_k 11.75 #Schijf and Byrne 2001

Lu+3 + 2Ox-2 = Lu(Ox)2- log_k 11.77 #Schijf and Byrne 2001

#REHOx+2 La+3 + HOx- = LaHOx+2 log_k 1.92 #Schijf and Byrne 2001

Ce+3 + HOx- = CeHOx+2 log_k 2.42 #Schijf and Byrne 2001

Pr+3 + HOx- = PrHOx+2 log_k 2.09 #Schijf and Byrne 2001

Nd+3 + HOx- = NdHOx+2 log_k 2.16 #Schijf and Byrne 2001

163 Appendix B. PHREEQC code for oxalate precipitation experiment

Sm+3 + HOx- = SmHOx+2 log_k 2.35 #Schijf and Byrne 2001

Eu+3 + HOx- = EuHOx+2 log_k 2.2 #Schijf and Byrne 2001

Gd+3 + HOx- = GdHOx+2 log_k 2.03 #Schijf and Byrne 2001

Tb+3 + HOx- = TbHOx+2 log_k 2.28 #Schijf and Byrne 2001

Dy+3 + HOx- = DyHOx+2 log_k 1.96 #Schijf and Byrne 2001

Ho+3 + HOx- = HoHOx+2 log_k 2.17 #Schijf and Byrne 2001

Er+3 + HOx- = ErHOx+2 log_k 2.09 #Schijf and Byrne 2001

Tm+3 + HOx- = TmHOx+2 log_k 2.18 #Schijf and Byrne 2001

Yb+3 + HOx- = YbHOx+2 log_k 2.41 #Schijf and Byrne 2001

Lu+3 + HOx- = LuHOx+2 log_k 2.28 #Schijf and Byrne 2001

PHASES CaOx:H2O CaOx:H2O = Ca+2 + Ox-2 + H2O log_k -8.60 #Martell and Smith

CaOx:3H2O CaOx:3H2O = Ca+2 + Ox-2 + 3H2O Log_k -8.14 #Martell and Smith

LaOxalate La2Ox3:10H2O = +2La+3 +3Ox-2 + 10 H2O log_k -29.22 #Chung et al 1998

164 Appendix B. PHREEQC code for oxalate precipitation experiment

CeOxalate Ce2Ox3:9H2O = +2Ce+3 + 3Ox-2 + 9 H2O Log_k -30.39 #Chung et al 1998

PrOxalate Pr2Ox3:10H2O = +2Pr+3 +3Ox-2 +10 H2O log_k -30.66 #LFER

NdOxalate Nd2Ox3:10H2O = +2Nd+3 +3Ox-2 + 10 H2O log_k -30.89 # Chung et al 1998

SmOxalate Sm2Ox3:10H2O = +2Sm+3 +3Ox-2 +10 H2O log_k -31.35#Chung et al 1998

EuOxalate Eu2Ox3:10H2O = +2Eu+3 +3Ox-2 + 10 H2O log_k -31.37#Chung et al 1998

GdOxalate Gd2Ox3:10H2O = +2Gd+3 +3Ox-2 + 10 H2O log_k -31.37#Chung et al 1998

TbOxalate Tb2Ox3:10H2O = +2Tb+3 +3Ox-2 + 10 H2O log_k -30.77 #LFER

DyOxalate Dy2Ox3:10H2O = +2Dy+3 +3Ox-2 +10 H2O log_k -30.699 #Chung et al 1998

HoOxalate Ho2Ox3:10H2O = +2Ho+3 +3Ox-2 +10 H2O log_k -30.26 #LFER

ErOxalate Er2Ox3:10H2O = +2Er+3 +3Ox-2 +10 H2O log_k -30.05 #Chung et al 1998

165 Appendix B. PHREEQC code for oxalate precipitation experiment

TmOxalate Tm2Ox3:10H2O = +2Tm+3 +3Ox-2 +10 H2O log_k -29.97#LFER

YbOxalate Yb2Ox3:10H2O = +2Yb+3 +3Ox-2 +10 H2O log_k -30.02 #Chung et al 1998

LuOxalate Lu2Ox3:10H2O = +2Lu+3 +3Ox-2 +10 H2O log_k -30.07#LFER

END

SOLUTION temp 20 pH 1.1 pe 4 redox pe units mmol/kgw density 1 Ca 67508 ug/kgw Ce 483 ug/kgw Cl 1 Na 1 Charge Dy 462 ug/kgw Er 220 ug/kgw Eu 138 ug/kgw Gd 565 ug/kgw Ho 86 ug/kgw La 2700 ug/kgw Lu 20 ug/kgw Nd 2653 ug/kgw Ox 0.0829 Mol/kgw Pr 611 ug/kgw Sm 541 ug/kgw Tb 80 ug/kgw Tm 28 ug/kgw

166 Appendix B. PHREEQC code for oxalate precipitation experiment

Yb 153 ug/kgw water 1 # kg

EQUILIBRIUM_PHASES 1 NdOxalate 0 0 GdOxalate 0 0 SmOxalate 0 0 PrOxalate 0 0 LaOxalate 0 0 EuOxalate 0 0 CeOxalate 0 0 DyOxalate 0 0 TbOxalate 0 0 CaOx:H2O 0 0 CaOx 0 0 CaOx:3H2O 0 0 ErOxalate 0 0 YbOxalate 0 0 CaOx:3H2O 0 0 HoOxalate 0 0 TmOxalate 0 0 LuOxalate 0 0

SELCTED_OUTPUT ***

167

Appendix C. Supplementary figures from PHREEQC modelling

Appendix C

Supplementary figures from PHREEQC modelling

PHREEQC modelling of the 5 dominant REE complexes during precipitation experiments as a function of pH testing the influence of REE concentration. The following figures are presented in increasing order of REE concentration: 0.1, 0.5, 1, 5, and 10 ppm, specified in the y axis of the figure.

169 Appendix C. Supplementary figures from PHREEQC modelling

170 Appendix C. Supplementary figures from PHREEQC modelling

171 Appendix C. Supplementary figures from PHREEQC modelling

172 Appendix C. Supplementary figures from PHREEQC modelling

173 Appendix C. Supplementary figures from PHREEQC modelling

174 Appendix D. Ore Geology Review publication Josso et al., (2016)

Appendix D

The following work was realised in parallel of the presented PhD thesis on results obtained during a 6 month internship at IFREMER, Brest, France in 2013 on the geochemical characterisation of oceanic ferromanganese mineralisation.

Title of publication

A new discrimination scheme for oceanic ferromanganese deposits using high field strength and rare earth elements.

Authors

Pierre Josso, Ewan Pelleter, Olivier Pourret, Yves Fouquet, Joel Etoubleau, Sandrine Cheron, Claire Bollinger.

Published in

Ore Geology Reviews special edition on Marine Minerals

Josso, P., Pelleter, E., Pourret, O., Fouquet, Y., Etoubleau, J., Cheron, S. & Bollinger, C. (2016). A new discrimination scheme for oceanic ferromanganese deposits using high field strength and rare earth elements. Ore Geology Review (accepted).

175 Appendix D. Ore Geology Review publication Josso et al., (2016)

OREGEO-01927; No of Pages 13 Ore Geology Reviews xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Ore Geology Reviews

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / o r e g e o r e v

A new discrimination scheme for oceanic ferromanganese deposits using high field strength and rare earth elements

a,d, a b a a a c P. Josso , E. Pelleter , O. Pourret , Y. Fouquet , J. Etoubleau , S. Cheron , C. Bollinger a Ifremer, Geochemistry and Metallogeny Laboratory, Plouzané, France b HydrISE, Institut Polytechnique LaSalle Beauvais, 60026 Beauvais cedex, France c Institut Universitaire Européen de la Mer, UMS 3113, Plouzané, France d University of Southampton, National Oceanography Centre, European Way, Southampton S014 3ZH, UK

a r t i c l e i n f o a b s t r a c t

Article history: Ferromanganese (Fe-Mn) deposits constitute a ubiquitous mineral type in oceanic settings, with metal (Cu, Ni, Zn, Co, Pt) and Received 8 February 2016 rare earth element (REE) enrichments of potential economic interest. Routine analysis of trace el-ements by ICP-MS has Received in revised form 25 August 2016 advanced our understanding of the impact of hydrogenetic, diagenetic and hydrothermal processes on the mobility and Accepted 1 September 2016 interaction of high field strength elements (HFSE: Zr and Ti) and REE and yttrium (REY) with Fe-Mn oxyhydroxides. Recent Available online xxxx discoveries in the French exclusive economic zone (EEZ) of Wallis and Futuna (southwest Pacific Ocean) have brought new Keywords: insight into the formation of low temperature (LT) hydro-thermal Mn deposits and lead us to reconsider the classification and Ferromanganese mineralization discrimination diagrams for Fe-Mn deposits and ore-forming processes. Using a suite of LT hydrothermal Fe-Mn crusts from Hydrogenetic crusts Wallis and Futuna, we investigate how contrasting genetic processes influence the distribution of metals (Mn, Fe, Cu, Ni, and Nodules Hydrothermal Co), HFSE and REY in hydrogenetic, diagenetic, hydrothermal and mixed-type deposits from different environments in the deposits Rare earth global ocean. The interaction of the different metal oxide-forming processes indicates that: (i) enrichment of Co, HFSE and elements REY is favored by hydrogenetic precipitation, (ii) diagenetic processes produce higher Mn, Cu, and Ni concentrations with High field strength elements oxic remobilization in the sedimentary column, while suboxic conditions promote greater Mn and Fe remobilization that Classification competes with the incorporation of Cu and Ni ions in nodules. HFSE and REY de-rived from seawater are usually low in diagenetic precipitates, which discriminate between hydrogenetic and diagenetic inputs within nodules, (iii) hydrothermal Fe- Mn deposits show strong depletion in HFSE and REY due to rapid formation and high contents of either Fe or Mn oxides. We present a new discrimination scheme for the genetic types of Fe-Mn deposits using a 10 ∗ (Cu + Ni + Co) − 100 ∗ (Zr + Y + Ce) − (Fe + Mn) / 4 ter-nary diagram. The use of HFSE and REY in the classification allows for a more robust discrimination of: (i) each ore-forming process with well-delimited fields, without overlap of metal-rich hydrothermal samples and hydrogenetic samples, (ii) oxic and suboxic diagenesis within nodules, (iii) trends between hydrogenetic and diagenetic end- members forming a continuum, and (iv) mixed genetic types such as the presence of hydrother-mal particles within hydrogenetic crust layers. Alternatives are also explored to adapt our discriminative diagram to elements measurable by on- board instruments to aid in exploration at sea.

1. Introduction processes and distributions exist (Halbach et al., 1981; Aplin and Cronan, 1985; Halbach et al., 1988; Hein et al., 1997, 2000, 2013; Hein and Ferromanganese (Fe-Mn) deposits reflect common forms of miner- Koschinsky, 2013). These deposits are classified in three main cate-gories: alization in the modern ocean. They may contain significant metal (Cu, Ni, hydrogenetic, diagenetic and hydrothermal (Bonatti et al., 1972; Hein et al., Zn, Co) and rare earth elements plus yttrium (REY) enrichments. Oceanic Fe- 1997; Bau et al., 2014; Schmidt et al., 2014) and are common-ly Mn deposits include ferromanganese crusts, also referred to as cobalt-rich discriminated by using characteristic variations of Fe, Mn, Cu, Ni, and Co crusts, polymetallic nodules, and hydrothermal crusts and impregnations. contents. Numerous reviews of Fe-Mn deposit genetic Crusts form by hydrogenetic precipitation of colloidal particles of Fe and Mn oxyhydroxides onto rock substrates (Dymond et al., 1984; Bau et al., 1996; Hein et al., 2000; Schultz, 2006). Hydrogenetic crusts con-centrate Corresponding author at: University of Southampton, National Oceanography Centre, some trace metals (Co, Ni, Ti, HFSE and REY) at many orders of magnitude Ocean and Earth Sciences, Waterfront Campus, European Way, Southampton S014 3ZH, Hampshire, UK. above their concentration in seawater (Bau et al., 1996; Hein and Koschinsky, E-mail address: [email protected] (P. Josso). 2013), which is promoted by the high reaction

Please cite this article as: Josso, P., et al., A new discrimination scheme for oceanic ferromanganese deposits using high field strength and rare earth elements, Ore Geol. Rev. (2016), http://dx.doi.org/10.1016/j.oregeorev.2016.09.003

176 Appendix D. Ore Geology Review publication Josso et al., (2016)

2 P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx

2 surface area of Fe-Mn oxyhydroxides (average of 325 m /g for bulk δ-MnO2 multiple volcanic edifices. Three atypical low-temperature Fe-Si-Mn de-posits and FeOOH) (Hein and Koschinsky, 2013; Pourret and Davranche, 2013), were discovered during the cruise aboard R/V Atalante, Utu Uli, Anakele, and slow oxidation reaction kinetics, and slow growth rates (Bau et al., 1996). Utu Sega (Fig. 1) from which samples were collected by dredging and HOV Polymetallic nodules constitute mixed-source deposits formed by dive operations. The deposits found at Utu Uli and Anakele occur as massive, hydrogenetic and/or diagenetic precipitation of Fe-Mn colloids around a dense and laminated crusts of Mn oxyhydroxides from the summit of volcanic nucleus on the surface of soft sediments. Inter-actions of both hydrogenetic edifices composed of pyro-clastic material, pillow lavas and more rarely and diagenetic processes lead to a range of bulk Mn/Fe ratios from 1 to 2.5 sediments. Locally, the Mn crusts are crosscut by Fe-Si precipitates for oxic diagenesis and up to 50 when the sedimentary column is suboxic near displaying ridge or vein-like structures up to 1 m high (Pelleter et al., 2016). the seabed (Calvert and Piper, 1984; Schultz, 2006). Metals (Ni, Cu and Co), The mineralization ex-tends below these crusts as Mn oxyhydroxides, Fe REY and HFSE in nodules are mainly derived from seawater through oxides and nontronite cementing basaltic pyroclastic facies and brecciated hydrogenetic precipitation and further enrichment in metallic elements (Mn, lavas (Pelleter et al., 2016). Crusts have been observed covering three distinct Cu, Ni and Zn) occurs via diagenetic precipitation (Ohta et al., 1999). volcanic edifices in the Utu Uli area representing an approximate area of 1.5 2 km with thicknesses up to 5 cm, whereas sediment accumulation at the other Hydrothermal Fe-Mn deposits form under various conditions of sites prevented estimation of the extent of the deposits. No hydrothermal temperature and geological settings; (i) plume fall-out deposits associ-ated discharge was observed in the vicinity of the three sites. The present study with high temperature venting (Bonatti, 1975; Corliss et al., 1978; Barret et focuses on 16 samples recovered from dredge hauls showing a range of al., 1987; German et al., 2002) form when particles derived from composition from nearly pure Mn oxide to mixed Mn-Fe ± Si compositions. hydrothermal solutions precipitate during mixing with cold oxi-dizing Sample FU-DR01-03 was later subdivided in three subsamples of oxides from seawater and settle on the seafloor forming the well-known met-alliferous the impregnated host sediment, an in-ternal layer of the crust and the surface sediments encountered near hydrothermal fields such as along the East Pacific layer. See Pelleter et al. (2016) for bulk composition. rise (EPR) or the Mid-Atlantic Ridge (MAR) (Barret and Jarvis, 1988; Goulding et al., 1998). Enrichment of metals, REY and other critical elements in oxyhydroxides occurs by sorption from seawater onto colloids in seawater and on the surface of crusts and nodules after accretion of the colloids. (ii) 2.2. Methods Diffuse hydrothermal systems are usually associated with off-axis hydrothermal circulation along MOR and within submarine volcanoes along Mineralogical identification was made through X-ray diffraction (XRD) at volcanic arcs and hot spot seamounts, or less commonly associated with Ifremer using a BRUCKER AXS D8 Advance (with Bragg-Bentano fractures distant from oceanic ridges (Hein et al., 2008b; Edwards et al., goniometer and VANTEC-1 positive sensitive detector — PSD) and 2011). They form stratabound layers and crusts, and cement the sediment BRUCKER AXS D2 Phaser. Prior to analysis, samples were dried at 60 °C column with either Fe or Mn oxides by fluids percolating within porous and milled with agate mortar then deposited on a XRD sample holder and sediment, py-roclastic deposits, and volcanic breccias (Fouquet et al., 1993). flattened with a glass slide. Samples were analyzed using Cu-Kα radiation These de-posits show a distinct mineralogy and texture from hydrogenetic over 2θ ranging from 2° to 70° at 40 kV and 30 mA. Ad-ditional analyses crusts (Burgath and Von Stackelberg, 1995; Schultz, 2006). Fe-Mn hydrother- were made to characterize clay minerals (Pelleter et al., 2016). Minerals were mal deposits exhibit a wide range of Mn/Fe ratios from 0.001 (nearly no Mn) identified using Eva search-match software. up to 4000, as a function of fluid temperature and redox conditions (Burgath Bulk chemical composition of each sample was determined. Samples were and Von Stackelberg, 1995; Schultz, 2006). The rapid precipi-tation of these ground to a powder (90% of particles b80 μm) using an agate pes-tle and hydrothermal oxides commonly forms deposits with low minor metal mortar. Major elements and selected trace elements were ana-lyzed by X-ray contents; those influenced by hydrogenetic precipita-tion show an fluorescence with a BRUCKER AXS S8 Tiger automated XRF spectrometer intermediate composition. on pressed pellets (Pelleter et al., 2016). Additional trace elements (Sr, Y, Zr, Nb, Th, and REE) were analyzed by In 2010 an Ifremer mission exploring the mineral resources potential of inductively coupled plasma mass spectrometry using an ELEMENT II mag- the Wallis and Futuna archipelago (SW Pacific) dredged Fe-Mn hy- netic field ICP-MS at Institut Universitaire Européen de la Mer (IUEM) in drothermal samples showing the strongest metal enrichments recorded in Fe- Brest. The dissolution procedure was as follows; 0.1 g of sample powder was Mn oceanic deposits (Pelleter et al., 2016). Hence the discovery of these digested in a Teflon bottle with 4 mL of 6 mol/L hydrochloric acid for 24 h on metal-rich low temperature (LT) hydrothermal deposits requires explanation a hot plate (120 °C) with a Tm spike. If present, the residual phase composed of (i) their mode of formation, (ii) the associated fluid geo-chemistry, and (iii) of mostly silicates and refractory minerals was then ex-tracted by centrifuge their economic potential as an oceanic resource, as well as (iv) the way we and digested by a mixture of hydrofluoric and hydro-chloric acid (3:1) for 48 classify and discriminate Fe-Mn oceanic Fe-Mn deposits. Although we h on a hot plate (120 °C), evaporated and then remixed with the previously present bulk geochemical and mineralogical results for these unusual digested phase. The solution (0.5 μL) was then evaporated on a hot plate and deposits, the first questions are presented else-where (Pelleter et al., 2016). the residue was made up to 10 mL with a 2% nitric and 0.05% hydrofluoric We develop here how these deposits chal-lenge our way of classifying Fe-Mn acid solution for trace element anal-ysis by ICP-MS. Data were corrected oceanic deposits and explore new geochemical tools for their discrimination. using internal calibrations, BHVO-2 standard, and a Tm spike correction (Barrat et al., 1996). Concentrations are expressed as % for weight %.

2. Material and methods The rare earth and yttrium (REY) are discussed separately from the HFSE group for ease of description of their geochemical behavior and fractionation 2.1. Material into Fe-Mn oceanic deposits although most lanthanides are HFSE from a geochemical point of view apart for Eu in its divalent state which is Hydrothermal samples were recovered during an Ifremer research cruise considered a large ion lithophile element (LILE). The Ce and Eu anomalies in the French exclusive economic zone (EEZ) of Wallis and Futuna in August (Ce/Ce and Eu/Eu ) are calculated as the ratio of the normalized values 2010. The area studied is bordered by the North Fiji fracture zone, the active (subscript n) of an element by the interpolation of the adjacent elements Tonga and Vanuatu subduction zones and associated Lau and North Fiji back- (superscript *) such that; arc basins, and the currently active Samoan hotspot. Several extensional zones were recognized (Pelletier et al., 2001; Fouquet et al., 2015) including Ce=Ce ¼ Cen pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi the Futuna and Alofi ridges and an area formed by a complex system of Lan Prn ð diffuse magmatism with Þ

177 Appendix D. Ore Geology Review publication Josso et al., (2016)

P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx 3

Fig. 1. Wallis and Futuna map showing the position of the main hydrothermal mounds (stars) discovered during Ifremer's cruises.

and CaO (1.75 ± 1.0%), Al2O3 (0.64 ± 0.5%) and TiO2 (0.05 ± 0.04%). Focus metal contents of some samples are anomalously high in compar-ison with Eu Eu=Eu n typical hydrothermal Fe-Mn deposit metal concentrations (Hein et al., 1997), ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi SmGd with 0.72 ± 0.7% Ni, 0.35 ± 0.3% Co and 0.15 ± 0.2% Cu, with maximum ð nnÞ values of 4.7% Ni, 2.2% Co and 1.5% Cu (Table 1). This enrichment in Note that all following data of this study have been calculated using nor- potential economic metals is unusual for LT hy-drothermal deposits and malization to chondritic values, and normalization to Post-Archean Aus- suggests the absence of sulphide precipitation at depth at higher pressure and tralian Shale (PAAS, Taylor and Mclennan (1985)) is used for calculation of temperature conditions, which usually scavenge most of the Ni and Cu. These normalized Y/Ho ratio (Bau et al., 2014). deposits might constitute the dis-tal end-member of high temperature systems Pearson coefficient correlation matrix was calculated on the data set (n = whereby metals were transported in a fluid with high oxygen fugacity. 18) to observe correlated geochemical variations and extract infor-mation on Continuous mixing with seawater lead to the precipitation of Fe and Mn possible phase associations during precipitation. SiO2, Al2O3, Fe, Mn, CaO, oxides, rapidly scavenging transported metals (Pelleter et al., 2016). Other trace ele-ments fall into the expected range of concentrations for LT Fe-Mn MgO, TiO2, P2O5, Co, Ni, Cu, Y, Zr, La, Ce, Eu, Yb, Ce/Ce , Eu/ Eu and total REE content (ΣREE) were taken into account. As geochem-ical hy-drothermal deposits; Sr (329 ± 37 ppm), Y (18 ± 4 ppm) and Zr (5.3 ± 3 concentrations for each element do not follow a normal distribution, results ppm, excluding sample FU-DR15-06 with 85 ppm Zr). are later reported as the median with a 95% confidence level. REE concentrations in these hydrothermal samples are low (ƩREE = 26.1 3. Results ± 11.5 ppm), with a maximum content of 126 ppm REE in sample FU-DR15- 06 associated with the highest concentration in P2O5 (0.43%) and Zr. 3.1. Mineralogy Normalized REE trends (Fig. 2) show two distinctive signatures. One group includes all samples from dredge FU-DR22, while the second group includes Bulk X-ray diffraction on the massive surface layer displaying a me-tallic the rest of the hydrothermal sample set. This latter group displays marked grey luster shows birnessite and todorokite, 7 Å and 10 Å manga-nates light-REE (LREE) enrichment with regards to middle-REE (MREE); respectively (Pelleter et al., 2016). The presence of vernadite (poorly LaCH/SmCH varies from 2.98 to 7.02 and GdCH/ YbCH ratios from 0.88 to 1.20. Most patterns display a negative Eu anom-aly (0.67–0.98) and Ho and crystalline δ-MnO2) is suspected but could not be confirmed due to the overlap of reflections with birnessite and todorokite. Iron ox-ides are Er enrichment forming the top of a smooth convex-up pattern on the HREE. On the other hand, samples from core FU-DR22 have approximately flat dominated by amorphous oxyhydroxides, with the presence of discrete goethite grains. Nontronite, the ferrous end-member of dioctahedral smectite, LREE-MREE patterns (LaCH/SmCH = 0.83–1.52) with distinct heavy-REE forms the cement for and replaces pyroclastic rocks onto which the Mn oxide (HREE) enrichment; average GdCH/YbCH = 0.47. A marked positive Eu layers developed. anomaly is present as well as both positive and negative Ce anomalies for different samples. The YPAAS/HoPAAS ratio for these hydrothermal samples 3.2. Bulk geochemistry varies from 0.66 to 1.37, regardless of the REE pattern, showing a depletion in Y relative to shale for many of those samples. This signature is thought to Bulk sample composition is dominated by Mn (40.5 ± 3.6%), SiO2 (3.5 ± repre-sent precipitation of oxides from a fluid close to seawater composition 2.6% including 5 outliers ranging from 12.4 to 16.7%) and Fe (3.0 ± 2.2% (Bau et al., 2014; Schijf et al., 2015). High temperature hydrothermal fluids including 4 outliers ranging from 10 to 16.5%), with Mn/ Fe ratio ranging associated with vent systems usually show well-developed posi-tive Eu from 1.6 to 3929 (Table 1). Other major elements de-crease in abundance as anomalies (Michard, 1989; Bau, 1991; Douville et al., 1999; follows; Na2O (4.4 ± 0.4%), MgO (2.5 ± 0.4%),

178 Appendix D. Ore Geology Review publication Josso et al., (2016)

0.5

0.4

0.5

0.6

0.3

1.2

1.8

5.8

1.7

0.7

1.5

1.7

1.2

–

0.9

0.6

1.8

ICP

ppm

0.92

0.66

0.83

0.91

0.87

0.84

0.82

0.87

0.84

1.03

1.25

1.06

1.26

–

1.03

1.37

1.02

1.14

(Y/Ho)PAAS

4.6

1.6

4.6

5.7

1.4

2.7

6.9

24.3

5.1

3.3

1.9

0.9

4.1

–

4.4

0.9

1.4

2.5

ICP

ppm

0.43

0.47

0.39

0.72

0.35

0.88

0.96

1.20

0.92

1.12

1.14

0.89

–

1.10

0.91

0.95

1.05

(Gd/Yb)n

2.0

1.6

1.7

1.5

1.1

5.2

8.3

32.7

11.6

3.5

10.8

13.5

7.8

–

5.3

4.6

3.6

10.1

ICP

ppm

.81

1.45

1.07

1.32

0.83

1.52

3.21

3.17

3.93

4.43

2.98

6.01

7.02

4.54

–

4.18

4.55

4.26

(La/Sm)n

255

262

434

104

309

740

816

637

1121

2509

905

651

230

–

301

518

141

448

ICP

ppm

126

–

ppm

REE

0.05

0.12

0.13

0.04

0.09

0.04

0.02

0.21

–

0.07

0.05

0.07

0.04

ICP

ppm

1.47

1.72

1.56

1.18

1.48

0.66

0.70

0.71

0.69

0.90

0.69

0.71

0.70

–

0.72

0.69

0.70

0.79

Eu/Eu Eu/Eu

0.34

1.28

4.71

0.64

0.18

0.42

0.44

1.33

–

0.42

0.23

0.29

0.52

ICP

ppm

1.06

0.47

1.23

1.47

0.59

0.26

0.43

0.27

0.49

0.11

0.04

0.33

–

0.48

0.12

0.23

0.14

Ce/Ce

6.0

3.2

5.5

16.0

5.0

–

ICP

ppm

2.33

7.66

36.54

3.89

57.14

2.20

1.40

6.31

–

4.77

2.38

1.90

2.69

ICP

ppm

22.4

22.5

21.1

17.5

13.5

–

ICP

ppm

–

1.40

1.74

1.19

6.36

0.42

0.85

1.08

0.49

–

0.08

0.60

0.49

0.68

ICP

ppm

421

360

399

285

362

267

308

405

416

536

289

279

236

–

329

347

278

197

ICP

ppm

–

101

166

133

–

ICP

ppm

–

ICP

ppm

–

772

3944

4595

703

328

347

2143

–

0.3

0.2

424

398

ICP

ppm

129

–

ICP

ppm

–

512.5

208.3

272.9

441.6

29.3

559.8

368.9

401.9

–

53.1

345.1

46.2

270.3

ICP

ppm

506

112

256

468

724

414

120

135

113

150

206

XRF.....

ppm

0.0

0.2

0.5

0.1

0.0

0.2

–

0.3

0.0

0.1

ICP

ppm

4475

4933

4806

3066

4333

15,305

11,650

2013

15,329

163

1041

249

115

200

XRF

ppm

–

0.0

0.1

0.0

–

0.0

–

0.0

ICP

ppm

2175

6569

2298

1297

2909

6297

6380

10,184

7866

44,705

20,942

46,666

17,435

12,084

1962

8204

1687

46,129

XRF

ppm

–

0.1

0.4

1.1

0.1

0.0

0.3

–

0.1

0.0

0.1

ICP

ppm

11,254

8331

12,304

6800

9686

9040

1309

488

13,267

4431

22,371

2428

494

1086

2232

1191

1510

1552

XRF

ppm

0.5

0.6

0.3

0.4

0.2

0.3

0.7

0.4

0.1

0.2

0.3

0.2

–

0.1

0.3

ICP

ppm

1.79

1.98

1.94

1.12

1.69

3.06

1.93

1.27

3.65

4.93

4.44

4.93

1.80

1.32

0.43

0.95

0.32

4.79

wt.%

Cu Cu + Co +

3.0

3.6

3.2

2.0

2.1

1.4

1.9

4.6

2.4

0.8

1.1

1.9

1.5

–

0.8

0.7

0.6

1.7

ICP

ppm

63.41

3929.1

121.64

17.44

196.38

12.03

2.33

1.57

13.94

13.45

12.74

80.00

2.34

3.12

38.41

40.93

14.56

12.49

Mn/Fe

2.7

3.5

2.8

1.9

1.8

1.5

1.9

4.8

2.4

0.8

1.5

2.2

1.6

–

0.9

0.8

0.7

1.9

ICP

ppm

0.03

0.02

0.03

0.04

0.02

0.04

0.11

0.19

0.05

0.04

0.03

0.12

0.08

0.03

0.04

0.05

XRF

wt.%

0.8

1.1

0.8

0.6

0.5

0.6

1.6

0.8

0.3

0.5

0.7

0.5

–

0.3

0.2

0.6

ICP

ppm

0.04

0.01

0.05

0.29

0.03

0.06

0.11

0.02

0.03

0.06

0.04

0.02

0.0

XRF

wt.%

3.32

3.33

3.3

2.94

3.14

3.52

2.28

2.18

2.88

4.19

3.68

4.19

2.83

2.20

3.19

2.34

3.97

3.99

XRF

wt.%

2.9

4.0

2.9

2.6

1.7

1.6

2.4

7.1

3.1

1.1

1.7

2.5

1.9

–

1.2

1.0

0.9

2.4

ICP

ppm

0.32

0.66

0.38

0.46

0.68

0.78

0.99

0.85

0.71

0.76

0.83

0.7

0.90

0.24

0.63

0.35

0.63

XRF

wt.%

0.3

0.5

0.3

0.2

0.4

1.1

0.5

0.2

0.3

–

0.2

0.1

0.3

ICP

ppm

1.47

1.79

1.65

2.64

1.80

1.29

1.66

1.71

1.53

1.03

1.06

1.03

1.54

0.62

1.66

0.58

1.21

XRF

5.53

1.46

4.31

5.76

2.69

1.11

1.22

1.44

1.29

0.77

1.00

0.83

1.02

0.99

1.81

1.15

1.78

0.83

XRF

wt.%

1.6

2.0

1.5

1.8

0.9

1.5

2.2

6.8

2.8

1.1

1.6

2.1

1.7

–

1.1

0.8

0.7

2.2

ICP

ppm

36.35

39.29

38.26

29.38

41.18

42.06

27.93

25.80

39.74

43.07

44.33

52.01

27.07

31.31

49.91

48.65

45.18

42.42

XRF

wt.%

0.6

0.8

0.6

0.5

0.3

0.4

1.4

0.5

0.3

0.4

0.3

–

0.2

0.1

0.5

ICP

ppm

0.57

0.01

0.31

1.68

0.21

3.50

12.01

16.48

2.85

3.20

3.48

0.65

11.57

10.03

1.30

1.19

3.10

3.40

XRF

wt.%

0.9

1.0

0.8

1.1

0.5

1.0

1.6

5.2

1.6

0.7

1.1

1.2

1.1

–

0.8

0.5

1.5

ICP

ppm

0.67

0.35

0.70

2.62

0.45

0.21

0.82

0.87

0.15

0.19

0.18

0.12

0.42

0.33

0.09

0.10

0.12

0.44

XRF

wt.%

Wallis Futuna & Wallis

0.70

0.10

0.78

7.80

0.33

1.63

6.15

5.79

1.64

1.59

1.83

0.41

6.33

7.81

1.20

1.07

2.39

1.75

XRF

wt.%

—

2.8

2.6

2.5

3.2

1.5

5.4

8.2

25.3

8.0

3.4

7.0

7.6

5.5

–

4.0

2.8

2.6

8.4

ICP

ppm

)

03 C

continued

(

DR22

DR15

DR01

DR22

DR15

DR01

Imp

Hydt

int

Imp

Hydt

int

Hydrothermal crusts Hydrothermal

Sample

Hydrothermal crusts Hydrothermal

Sample

eochemical data for hydrothermal Wallis samples and from Futuna.

Table 1

179 Appendix D. Ore Geology Review publication Josso et al., (2016)

P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx 5

0.22656

−

Eu/Eu

0.08988

Ce/Ce 0.668989 −

Yb 0.379188 0.517461 0.67695

Eu 0.921624 0.26731 0.269953 0.859331

0.14567

Ce 0.83314 0.639921 0.156456 0.931109 −

Fig. 2. Chondrite-normalized REE patterns of hydrothermal samples from Wallis and

Futuna EEZ.

0.49765

0.37431

0.799314 0.646561 0.425281 0.935918 − −

Craddock et al., 2010). Consequently, REE trends from Utu Sega samples

(FU DRR 22) with small positive Eu anomalies reflect a greater influ-

0.2093

Zr 0.822937 0.973839 0.794824 0.59099 0.051195 0.926843 −

ence of a HT fluid than at the two other sites.

The correlation coefficients (Table 2; n = 18) can be explained in terms of mineralogical control for most major elements. Iron and Si

0.741298 0.758344 0.755922 0.917156 0.860125 0.09114 0.116292 0.87873

are strongly fractionated from Mn during precipitation as Fe and Si pre- Y

cipitate under slightly reducing conditions to form nontronite, or amor-

phous silica and Fe oxyhydroxides as highlighted by the strong

0.03913

0.0002

0.009769 0.049042 0.080356 0.220391 0.052749 0.003643 0.053767 − correlation of Fe and SiO (0.75). On the other hand, Mn forms − 2 Cu

oxyhydroxides at the seawater-sediment interface or impregnations

within the volcaniclastic-sedimentary pile under oxidizing conditions,

showing a negative correlation with Fe (−0.72) and Si (−0.80). Mn is

0.1282 0.15874 0.1056 0.22235 0.46071 0.31895

0.35753

0.08207 0.244949 0.051819 − − − − − −

− Ni

negatively correlated to MgO, P2O5, Zr and Ce (−0.63 to −0.72). Alumi-

num is strongly positively correlated to TiO2 (0.96), CaO and MgO

(r N 0.7). Copper, Ni and Co do not show significant correlations

0.00369 0.32517 0.21258 0.22813 0.03183 0.17535

0.17982

0.346705 0.117808 0.196981 0.330377 − − − − − − − (−0.5 b r N 0.3) with elements considered here and behave indepen- Co

dently from each other in this sample set. P 2O5 correlation with Fe

(0.96) is consistent with phosphate sorption onto Fe oxyhydroxides in

0.00873 0.0065 0.13081 0.3969 0.40203 hydrothermal systems (Edmonds and German, 2004). Among HFSE 2

0.62505 0.901268 0.78364 0.837461 0.600485 0.394881 0.819019 − − − − − P

and REY, all are positively correlated to P2O5 (0.62 to 0.90), except Yb

which does not show significant correlations (0.39). The total REE con-

= 18). =

tent (ΣREE) correlates positively with P O and Fe (0.82 and 0.69 re- 2

2 5 n

0.16238 0.04129

0.06573

0.215378 0.242684 0.345846 0.061285 0.348961 0.331438 0.207125 0.636615 0.110786 0.207963 − − −

spectively) and not significantly to Mn (−0.48), suggesting that

Futuna Futuna (

minor scavenging from open seawater occurred during formation of TiO

MgO nd the Mn crusts with REE mainly derived from the LT hydrothermal fluids. Wallisa

0.33983 0.05875

0.694079 0.155431 0.081866 0.237914 0.354644 0.280181 0.252436 0.437783 0.503891 0.593823 0.457778 0.167058 − −

CaO 4. Discussion samplesfrom

0.25872 0.50157

0.04674 0.38722 0.14926

0.503977 0.576407 0.29228 0.041681 0.074076 0.052211 0.175507 0.319181 0.916921 0.631802 − − − −

4.1. Classification and alternatives −

hydrothermal Fe-Mn deposits have typically been classified using the Bonatti ter- Mn

0.63126 0.53246 0.72197 0.18411

0.25225 0.49285 0.64316 0.27257 0.60877 0.57419 0.52591 0.42883 0.13368 0.48263

0.157463 0.303373 − − − − − − − − − − − − −

nary diagram (Bonatti et al., 1972) using major chemical characteristics −

(Fe, Mn, Cu-Ni-Co) to discriminate them. Our hydrothermal samples

scatter on the Mn-(Co + Cu + Ni) side of the diagram as a result of

0.72652 0.34959 0.43934 0.18772 0.44734

matrixfo r

0.111644 0.150492 0.960537 0.004443 0.006447 0.472955 0.797846 0.682528 0.717067 0.455706 0.256953 0.693932 − − − − the unusual metal enrichments which places them outside of the origi- −

nally identified hydrothermal field (Fig. 3) (Bonatti et al., 1972). Some

samples have even higher minor metal contents than the polymetallic n

Al correlati o

nodules. Nevertheless, apart from the atypical Ni-Co-Cu enrichments 2

0.27626

0.58987 0.00576 0.04971

0.100589

0.698972 0.767908 0.967027 0.161775 0.0394 0.250393 0.309583 0.321525 0.372293 0.317537 0.760726 0.298811 0.158285 − − −

of these deposits, samples from Wallis and Futuna are similar in compo- −

sition and mineralogy to LT hydrothermal deposits collected in the Gulf

coefcient

SiO fi

0.06702 0.26706

of Aden (Cann et al., 1977), the Mediterranean Sea (Dekov et al., 2009), 0.14988

0.79544 0.39322

0.747663

0.568735 0.051305 0.436206 0.623521 0.664924 0.221157 0.528992 0.28386 0.461612 0.257957 0.084553 0.227307 0.345145 − − − − −

the Mid-Atlantic Ridge (Mills et al., 2001; Severmann et al., 2004) and

REE

Al Mn Ca O Mg O TiO P Co Ni Cu Y Zr La Ce Eu Yb

C e/ C e E u/ E u Σ

Pear

3 5

the Pacific Ocean (Corliss et al., 1978; Alt, 1988; Fouquet et al., 1993;

Sun et al., 2012; Zeng et al., 2012). son

Table2

O O 2

2 Hein et al., 1996, 2008b; Kuhn et al., 2003; Fitzgerald and Gillis, 2006; 2

Please cite this article as: Josso, P., et al., A new discrimination scheme for oceanic ferromanganese deposits using high field strength and rare earth elements,

180 Appendix D. Ore Geology Review publication Josso et al., (2016)

6 P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx

towards hydrogenetic ratios is observed representing the evolution of the exchange equilibrium between seawater and scavenging Fe-Mn ox-ides (Bau et al., 2014). Although really robust, this criterion cannot be used when considering hydrogenetic samples affected by phosphatisation due to a shift of the Y /Ho ratio to values as high as 1.8 (Asavin et al., 2010) PAAS PAAS (Fig. 4). We will therefore try to accom-modate this characteristic of hydrogenetic samples within our discrim-inative diagrams including other common discriminative features of Fe-Mn deposits.

4.2. New genetic discrimination diagrams

The objective of a new discrimination scheme therefore lies on the incorporation of new elements allowing the discrimination of hydro-thermal samples from hydrogenetic and diagenetic deposits without re-lying solely on the metal content. Fe-Mn deposits have long been known for their economic potential and accordingly most studies focused on the geochemistry of Fe, Mn, Cu, Ni and Co or a specific aspect of their trace element content with too few studies publishing a complete dataset including major, minor and trace elements. The Fig. 3. Ternary discrimination system of hydrothermal samples from Wallis and Futuna. Note published data used in the following plots were chosen to be representative of the abnormal distribution of samples away from the hydrothermal field due to unusual metal the diversity of Fe-Mn occurrences found throughout the deep ocean. enrichments. Therefore data for polymetallic nod-ules include mixed hydrogenous- After Bonatti et al. (1972). diagenetic samples from the central and north eastern part of the Pacific Ocean (n = 149) (Calvert and Price, 1977; Calvert and Piper, 1984; Dymond In addition to the problem posed by this set of samples from the Wallis et al., 1984; Hein et al., 1997; Ohta et al., 1999; Wegorzewski and Kuhn, and Futuna EEZ, the ternary classification of Bonatti et al. (1972) does not 2014), diagenetic nod-ules from the Peru basin (n = 31) (Von Stackelberg, clearly discriminate Fe-Mn deposits formed by mixed genetic processes such 1997), as well as samples from the Indian Ocean (n = 18) (Pattan and as nodules with various hydrogenetic/diagenetic proportions, hydrothermally Banakar, 1993; Balaram et al., 1995; Baturin and Dubinchuck, 2010). Data derived particles incorporated during hydrogenetic crust growth, and our used for hydrogenetic crusts come from samples of the open Pacific Ocean (n hydrothermal samples. The issue for nodules was discussed previously = 72) (De Carlo and Mc Murtry, 1992; Bau et al., 1996; Bau and Koschinsky, (Halbach and Puteanus, 1988) and hydrogenetic, diagenetic, and mixed-typed 2009; Asavin et al., 2010), including averaged compositions for diverse areas were identified within the existing diagram. of the Pacific (n = 4) (Hein et al., 1997), and samples from the Atlantic Ocean (n = 31) (Baturin and Dubinchuck, 2011; Muiños et al., 2013) and the Indian Alternative classification systems have been proposed and discussed Ocean (n = 5) (Surya Prakash et al., 2012). In addition to our own samples recently for the discrimination of Fe-Mn deposits. Other ways have been from Wallis and Futuna Archi-pelago, hydrothermal Fe-Mn samples from the explored using exclusively major (Conly et al., 2011), or minor and trace literature include data from the Pacific Ocean: Mariana-Izu-Bonin Arc System elements (Choi and Hariya, 1992; Nicholson, 1992; Bau et al., 2014). Within (n = 6) (Hein et al., 2008b), East Diamante Caldera (n = 6) (Hein et al., 2014), a MgO ∗ 10 − Fe2O3t − MnO2 ternary diagram, Conly et al. (2011) described Valu Fa Ridge system (n = 13) (Sun et al., 2011), PACMANUS hydrothermal discriminating fields for oceanic and terrestrials Fe-Mn deposits. Despite field (n = 18) (Zeng et al., 2012), Baby bare seamount (n = 9) (Fitzgerald and some overlaps, this diagram allows sample separa-tion due to the use of Mg Gillis, 2006), Okhotsk Sea (n = 5) (Baturin et al., 2010), metalliferous as a discriminative feature for fluid chemistry; oceanic waters contain ~ 1200 sediment and silica-hematite deposits of the Blanco Fracture Zone (n = 12) ppm of Mg while end member hydro-thermal fluids are considered Mg-free (Hein et al., 2008a), ironstones from Cen-tral Pacific Seamounts (n = 3) (Hein (Conly et al., 2011). However, considering LT hydrothermal and distal HT et al., 1994), and bacterially medi-ated metalliferous sediment from Loihi hydrothermal deposits, fluid chemistry for major elements is close to seawater Seamount, Hawaii (n = 13) (Edwards et al., 2011). Data for hydrothermal composition. Hence this classification scheme is not well-suited for the samples from the Atlantic Ocean include the ultramafic hosted deposits from compositional range found for oceanic Fe-Mn deposits, although it does the Mid Atlantic Ridge (MAR) (n = 8) (Dekov et al., 2011). Finally, data separate ma-rine and continental Mn-oxide deposits that are not mixed with from the Eolo Seamount in the Tyrrhenian sea (n = 22) (Dekov et al., 2009) Fe oxides. and the Indian ocean (n = 3) (Surya Prakash et al., 2012) are considered. Mixed type samples of hydrogenous and hydrothermal origin from the Using trace-element mobility in hydrothermal fluids and seawater, Central Indian Ridge (n = 17) (Kuhn et al., 1998; Takahashi et al., 2007) and Nicholson (1992) proposed a (Co + Ni) vs (As + Cu + Mo + Pb + V + Zn) the mercury- and silver-rich encrustation off the coast of Cal-ifornia (n = 11) binary diagram with distinction be-tween hydrogenetic, diagenetic, or (Hein et al., 2005) are included as well. supergene and hydrothermal pre-cipitation. Nodules and hydrogenetic crusts are well-discriminated on this diagram, however the unusually high Co and Ni contents of samples from Wallis and Futuna plot away from the designated hydrothermal field. A similar conclusion is reached for the classification Major differences occur among Fe-Mn deposits when considering proposed by Choi and Hariya (1992) using a Ni-Zn-Co ternary diagram. The concentrations in high field strength elements such as Zr, Ti and the REY. binary discrimination diagrams of Bau et al. (2014) are based on Nd Geochemical divergences for each mineral-forming process are il-lustrated in concentra-tions and fractionation of the geochemical twin Y and Ho against cross-plots (Fig. 4). Using characteristics of REY fraction-ation such as the the Ce anomaly. These diagrams demonstrate the efficiency of REY as a dis- Ce anomaly, one of the most distinctive features encountered in the various criminative feature for the various genetic types of marine Fe-Mn de-posits. types of Fe-Mn deposits, the YPAAS/HoPAAS ratio (Bau et al., 2014), Zr or Ti One of the most efficacious is the YPAAS/HoPAAS ratio, which if b1 content plotted against Mn/Fe ratio, and metal enrichments provide fairly characterizes a hydrogenetic or diagenetic process. Hydrothermal deposits good discrimination for each type of deposit. Relationships in Figs. 4 and 5 exhibit a wider range of ratios, mostly N1 though a continuum reflect the control of growth rates, precipitation kinetics, and pH conditions during

181 Appendix D. Ore Geology Review publication Josso et al., (2016)

P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx 7

Fig. 4. Discrimination diagrams displaying HFSE concentrations, Ce anomalies and YPAAS/HoPAAS ratios (Bau et al., 2014) against changes in Mn/Fe ratios and potentially economic metals for oceanic Fe-Mn deposits. mineral-forming processes and overall metal concentrations and trace element comparison with hydrogenetic deposits due to low contents of such el-ements budget in the source fluid. The HFSE and REY are highly deplet-ed in in pore fluids. Hydrothermal deposit depletion in HFSE relates to the poor hydrothermal deposits (by one order of magnitude) and richer in hydrogenetic capacity of these elements to bind with the dominant chlo-ride complex in deposits, while polymetallic nodules of mixed hydrogenetic-diagenetic origin hydrothermal fluids therefore implying a HFSE-deplet-ed source fluid for Fe- display more limited enrichments in Mn hydrothermal deposits (Douville et al., 2002).

182 Appendix D. Ore Geology Review publication Josso et al., (2016)

8 P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx

essential not to combine Co, HFSE and REY to separate metal-rich hydrothermal deposits from crusts and nodules.

4.3. Incorporating REE and HFSE in a new classification scheme

The choice of elements to incorporate in a new discrimination dia-gram should accommodate variations in hydrogenous or diagenetic in-puts to nodules, oxic/suboxic diagenesis in nodule bulk compositions, mixed hydrothermal-hydrogenetic formation, and metal-rich Fe-Mn hydrothermal deposits. We propose to use a first apex representing combined Fe and Mn concentrations (Fe + Mn)/4 (wt.%), an index that represents the main constituent of Fe-Mn samples. The second apex represents the various metal enrichments produced by the three precipitation processes: 10 ∗ (Cu + Ni in wt.%). Cobalt is deliberately left out of this apex as a common marker of hydrogenetic precipitation although its addition does not fundamentally change the positioning of samples, and could be considered in the case of extreme Co enrichment. HFSE and REY allow for the separation of hydrothermal samples from hydrogenetic and diagenetic deposits (Fig. 4). The last apex should therefore be formed by a combination of these elements. As studies of Fe-Mn deposits have not always presented complete minor and trace el-ement data, we will focus on the most commonly published HFSE (e.g. Ti and Zr) and REY. The enrichment in REY and other HFSE is a function of growth rate, fluid source and chemistry, and also represents uptake Fig. 5. Co/(Ni + Cu) vs Mn/Fe ratio diagram highlighting the relationship of hydrogenetic and of seawater complexes by Fe and Mn oxyhydroxides/oxides. Carbonates are diagenetic inputs to the composition of polymetallic nodules. See text for details on data used + 2 – the main complexing agent for REE; REE-CO 3 and REE-CO 3 account for for hydrogenetic crusts. Oxic and suboxic diagenetic fields follow Dymond et al. (1984). at least 85% of REE ligands in seawater (Schijf et al., 2015), while Zr or Ti are dominantly complexed with hydroxides (Bruland, 1983; Bau et al., 1996) and enriched by surface complexation. Combining these ele-ments that are fractionated by sorption on either Fe- or Mn-oxides has the objective of Furthermore, the fast precipitation of hydrothermal Fe-Mn deposits limits creating a restricted field for hydrogenetic precipitates in the diagram by hydrogenetic enrichment of Zr, Ti and REY (Schmidt et al., 2014). In including both scavenger phases and eliminates the influence of Mn/Fe ratios. contrast, crusts and nodules growing at extremely slow rates benefit from Cerium constitutes probably the most effi-cient element among the REE series extensive contact with seawater and pore fluids to scav-enge dissolved trace to use as a discriminative feature for Fe-Mn deposits. Indeed, Ce is easily metals. Of these two deposits, nodules display the largest range in hydrolyzed and continuously and irreversibly scavenged from seawater on the composition considering Mn/Fe ratios and minor and trace element contents surface of Mn oxides (Takahashi et al., 2007) and positive Ce anomalies in due to the influence of both diagenetic and hydrogenetic precipitation. As Fe-Mn deposits are therefore regarded as typical of hydrogenetic exemplified in Fig. 4, phosphatized samples of hydrogenetic crusts (Asavin et precipitation. Owing to the differences in precipitation kinetics among the al., 2010) do not follow the discriminative features of the YPAAS/HoPAAS different Fe-Mn deposits and irreversible Ce uptake from seawater, this ratio (Bau et al., 2014) and these samples could not be distinguished from element will be most enriched in hydrogenetic deposits, lower in diagenetic non-phosphatized crusts using other criteria (Ce, REE, and Zr). deposits, and depleted in hydrothermal deposits.

The ratio of diagenetic to hydrogenetic input to nodules, and their associated minor metals, can be shown qualitatively by the Mn/Fe ratio. The Yttrium and La were also considered as these elements show strong higher the Mn relative to Fe, the higher is the diagenetic input. Nodules enrichment in Fe-Mn deposits and are commonly analyzed in published analyzed from the global ocean show that there is a con-tinuum between the studies. Combinations using Zr and Ce with either La or Y give similar re- diagenetic and hydrogenetic end-members (Fig. 5). Metals (Ni, Cu, and Co), sults owing for the consistent fractionation of Y and La across the di-verse HFSE and REY in Fe-Mn deposits are mainly derived from seawater with environments of formation of Fe-Mn deposits. On the other hand, using an hydrogenetic precipitation in Fe-Mn crusts and nodules. Further enrichment apex combining Zr, Y and La brings scatter in sample dis-tribution. Notably, in elements such as Mn, Cu, Ni and Zn occurs during diagenesis whereby mixed-type deposits deviate from their predicted po-sition between metals are released to pore fluids through the dissolution of the Mn hydrothermal and hydrogenetic deposits. The absence of discrimination using oxyhydroxide fraction of the sedi-ment releasing carried metals which La and Y relates to their common levels of enrich-ment in hydrogenetic and become available to migrate up to the nodules at the oxic-suboxic interface mixed hydrogenetic-hydrothermal deposits whereas using Ce brings more (Ohta et al., 1999; Koschinsky et al., 2001). The diagenetic process does not weight to the HFSE-REY apex. Indeed Ce will be strongly enriched (positive link to Fe, Co, REY and HFSE enrichment in polymetallic nodules as Fe is Ce anomaly) in purely hydrogenetic precipitate whereas the faster growth rate mainly re-precipitated in the sediment as Fe oxyhydroxides or associated with of mixed hydrogenetic-hydrothermal deposits (up to 29 mm/Ma) will prevent sil-ica to form nontronite (Calvert and Piper, 1984). Moreover, the limited the formation of a positive Ce anomaly due to slow Ce oxidation kinetics mobility of Co, REY, and HFSE in interstitial waters compared to Cu, Ni, and (Kuhn et al., 1998). Consequently, using a combination of the most dis- Zn explains their relative enrichment in the most hydrogenetic samples and tinctive features of REE and HFSE behavior highlighted here, an apex with depletion with increasing diagenetic input (Jung and Lee, 1999; Ohta et al., 100 ∗ (Zr + Y + Ce in wt.%) is a viable discriminant (Fig. 6). 1999; Schultz, 2006; Baturin and Dubinchuck, 2009). In the resulting ternary diagram, hydrogenetic crust samples define a well- Therefore, Co, HFSE and REY enrichment in Fe-Mn deposits charac- delineated group close to the HFSE and REY apex (Fig. 6A). These elements terize hydrogenetic precipitation and it would be tempting to associate them can be used as a marker for hydrogenetic input over diagenet-ic input, in a new diagram to discriminate diagenetic and hydrothermal formation. allowing discrimination of these contributions for polymetallic nodules that However with unusual Co concentration averaging 0.6% and up to 2.2% in are otherwise ambiguous. Following the genetic classifica-tion of polymetallic hydrothermal samples from Wallis and Futuna, it is nodules in Fig. 5 (Dymond et al., 1984), we observe

183 Appendix D. Ore Geology Review publication Josso et al., (2016)

P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx 9

Fig. 6. Alternative discriminative diagram for oceanic Fe-Mn deposits using high field strength and rare earth elements. A. Hydrogenetic crusts, B. polymetallic nodules, C. hydrothermal deposits, D. mixed deposits and outliers.

2+ 2+ a partitioning of nodule data (Fig. 6B) in accord with the position of organic carbon. This abundance of Mn and Fe therefore compete with 2+ 2+ hydrogenetic crust data. The transition from hydrogenetic to diagenetic favors Cu and Ni for incorporation in the todorokite lattice promot-ing the shift high concentration in Ni and Cu at the expense of HFSE and REY defining a observed in the ternary diagram (Halbach et al., 1981; Hein and Koschinsky, trend from the potential economic-metal apex with diage-netic nodules 2013). Detailed analysis of nodule microlayers from the Peru Basin towards the REY-HFSE apex with mixed hydrogenetic-diagenetic samples. (Wegorzewski and Kuhn, 2014) have shown that these trends are only valid Data for diagenetic suboxic samples from the Peru Basin (Von Stackelberg, when considering bulk composition and that oth-erwise nodules are made of 1997) lack Y concentrations. Concentrations in-cluded in the calculation of alternating layers of hydrogenetic and dia-genetic composition in various the HFSE apex therefore used the average Y concentration of 69 ppm for Peru proportions. In the scope of this new discrimination diagram, only bulk Basin samples from Hein and Koschinsky (2013) including a standard compositions are considered. deviation of ±30 ppm calculat-ed from the nodule dataset used in this study (n In Fig. 6C, the use of REY and HFSE separates the overlap of metal-rich = 184). These samples define here a line moving towards the Fe-Mn apex hydrothermal samples with those of hydrogenetic or diagenetic or-igins as with a consistent HFSE-REY index at the expense of the metal content. This compared to other schemes. Metal-rich samples from Wallis and Futuna are specific nod-ule trend is consistent with the evolution of the bulk Ni + Cu spread between the Fe-Mn and Cu-Ni apex. Most hydro-thermal samples fall content in polymetallic nodules described by Halbach et al. (1981). Indeed, otherwise close to the Fe-Mn apex in accordance with their typical depletion transi-tion from hydrogenetic to oxic diagenetic nodules favors metal enrich- in HFSE and minor metals (Cu, Ni, Co, and Zn). ment with high grades of Ni + Cu with Mn/Fe ratios as high as 5. However, transition to suboxic diagenetic samples from the Peru Basin (Von Stackelberg, 1997) with a bulk Mn/Fe ratio of up to 100 displays decreasing 4.4. Limits of the discrimination diagram metal contents (Halbach et al., 1981; Dymond et al., 1984). Dominant suboxic conditions favor formation of a strong gradient of Mn and Fe in the pore Samples which do not fit the major trends observed for the majority of the waters by dissolution of Mn oxides and reduction of ferric iron in the dataset are presented in Fig. 6D. Most of these samples present specific sediment in the presence of required quantities of atypical characteristics or constitute mixed-type deposits and allow for the exploration of the diagram's limits as to which samples

184 Appendix D. Ore Geology Review publication Josso et al., (2016)

10 P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx can be correctly classified using this discrimination scheme. Mixed hy- 4.5. New discrimination diagram for Fe-Mn oceanic deposits drothermal-hydrogenetic crust samples (Hein et al., 2005; Muiños et al., 2013) result from the incorporation of hydrothermally derived ele-ments into The data distribution in Fig. 6 supports the efficacy of HFSE and REY as hydrogenetic crusts either as fall-out particles or acquire a hydrothermal a discriminant for Fe-Mn oceanic deposits, as introduced by Bau et al. (2014). signature through the input of LT hydrothermal fluids. Hydrogenetic samples As highlighted here, previously proposed discrimination schemes do not from the Indian Ocean Ridge display such a char-acteristic with a faster completely distinguish metal-rich hydrothermal sam-ples from the two other growth rate supported by a distal hydrothermal supply which in turn impacts dominant types of Fe-Mn deposits, as well as separation of mixed deposit the time available for HFSE and REY up-take from seawater (Kuhn et al., types. In addition, phosphatized hydrogenetic crust samples are positioned 1998; Takahashi et al., 2007). Samples from the Southern California within this diagram among other crusts. Consequently this new scheme Borderland present Hg-Ag enrichment, Ce positive anomalies, Co and REE provides increased sensi-tivity for discrimination among the genetic processes concentration intermediate between hydrogenetic and hydrothermal deposits that produced these various deposit types, and thus provides a powerful tool. which highlight a mixed hydrothermal-hydrogenetic geochemical signature (Hein et al., 2005). As observed on Fig. 6D, these samples plot in an Based on data distribution on this ternary diagram, we propose the intermediate position between hydrogenetic crusts and hydrothermal deposits, following identification of genetic fields for the formation of oceanic Fe-Mn supporting the use of HFSE and REY as a discriminant in this diagram. deposits (Fig. 7). This figure displays the detailed relations among the three genetic processes through well-defined fields for each deposit-forming process without overlap of metal-rich hydrother-mal samples and hydrogenetic Some hydrothermal samples from the silica-hematite deposits (Hein et al., samples. This scheme allows for the identification of fields possibly 2008a) display a nearly pure Silica composition with b3% Fe. These samples representative of oxic and suboxic dia-genesis within nodules, the trend are scattered in the middle of the diagram due to the lack of representability between hydrogenetic and diagenetic end-members that forms a continuum, of each apex considering silica dilution. Samples with Si N 70% and at least and identifies mixed genetic types such as the presence of hydrothermal 6% Fe plot well within the hydrothermal field (Fig. 6C). It is possible to particles within hydrogenetic crusts. This scheme does group hydrothermal Fe consider an apex (Fe + Mn + Si) / 5 as a sub-stitute to better accommodate oxides and Mn oxides into a single field. specific samples with extreme Si enrich-ment, usually of hydrothermal origin (Supplementary Fig. S1). Comparable mineralogical impact is observed for a hydrogenetic crust sample (Muiños et al., 2013) where δ-MnO2 is 71% and 4.6. Alternative classifications for driving exploration at sea todorokite is present, compared to 85–99% δ-MnO2 for the rest of the sample set from this area; todorokite is uncommon for open-ocean hydrogenetic Although robust, the new discrimination scheme (this study) and the crusts but common in continental margin hydrogenetic crusts (Conrad et al., discrimination diagrams proposed by Bau et al. (2014) cannot be easily 2016) which might account for the unusual signature of this sam-ple. With applied to historical data because of the lack of REE and HFSE data. regard to other samples from Muiños et al. (2013), the impor-tant presence of Additionally, analyses of REE requires land-based analytical tech-niques and non Fe-Mn phases (30% of calcite and dolomite) shift the sample position thus discrimination of Fe-Mn mineralization using REE can-not be done at towards the Fe-Mn apex. sea, which would help direct exploration. As a consequence, we investigated alternatives to the 100 ∗ (Zr + Ce + Y) apex using elements that are widely Some hydrothermal sediments from Eolo Seamount scatter along the Fe- available in the literature (e.g. Ti, Co and Ce) and/or that can be analyzed at Mn–HFSE-REY side of the diagram far from other hydrothermal samples. sea using a compact, benchtop ED-XRF spectrometer (e.g. Ti, Co, Zr and Y). This distribution highlights a strong enrichment in HFSE that is attributed to The ship-board proposed apexes for ternary diagrams include: (A) 20 ∗ Ti, (B) the remobilization and redeposition of hydrothermal particle mixing with 50 ∗ Co, (C) 50 ∗ Ce, (D) 30 ∗ (Co + Ce), (E) 50 ∗ (Ti/5 + Ce), and (F) 200 ∗ detrital material (N65% clays in bulk sample) (Dekov et al., 2009). (Zr + Y) (Supplementary Fig. S2), with all elements expressed in wt.%.

Another limit of this diagram lies in the distinction between metal-rich hydrothermal deposits and suboxic diagenetic nodules that cover similar A Ti apex brings a lot of uncertainty and scattering, notably for hy- fields. However, although these two deposit types might plot in a similar area drothermal samples, probably due to contamination of samples by the on the diagram, they can be distinguished by mineral-ogy or by plotting the protolith, which increases the HFSE elements. The presence of the protolith Mn/Fe ratio vs (Zr + Y + Ce). Besides, these two sample types would have within the mineralized sample is relatively common during sampling and its been collected from quite different environ-ments and have very different influence can be verified in this configuration by checking the amount of appearances. aluminum. Compared with the hydrothermal field identified in Fig. 7, samples The use of this diagram as a genetic discriminant therefore requires that spreading towards the HFSE apex ex-hibit increasing Al content mostly samples plotted in it contain N5% Fe and Mn when the major phase is silica above 2% and up to 7.7% whereas samples correctly positioned have b0.9% like in the case of the silica-hematite deposits. Silica dilution of the Fe-Mn Al. fraction can be significant (up to 95%) without changing much the appropriate Combinations using Co works well as a discriminant for most Fe-Mn positioning of samples because silica is not known to be associated with more deposits but fail to accommodate Co-rich hydrothermal deposits such as than trace amounts of metals, HFSE and REY. If non Fe-Mn oxide minerals samples from Wallis and Futuna. The use of only Ce as an apex also works in are present, then a N70% Fe-Mn portion is necessary to correctly discriminate the absence of Zr and Y data, however it therefore represents incorporation sample genesis; for exam-ple with N30% carbonate, clays or sulphides, dominantly through Mn oxides. In addition, discrimina-tion between samples cannot be correct-ly discriminated on this diagram. Titanium and Zr hydrothermal samples and diagenetic nodules cannot be made solely on the must be used with caution because these elements are abundant in both basis of this diagram owing to their similar contents of Ce. As exemplified in volcanic rocks hosting most Fe-Mn deposits and in detrital clay minerals that previous combinations, Ce constitutes the most effective discriminant among may be incorporated in Fe-Mn deposits. It is therefore recommended to look the REE and HFSE for deep-sea Fe-Mn de-posits. It can therefore be used as for potential contamination by mafic volcanic grains from the substratum by a viable apex in the absence of more complete data, although it is not adapted looking for high Al2O3 contents in bulk samples and determine, when for on-board exploration as in-strumental errors are important for Ce possible, the carbonate- and clay-free fraction of metalliferous sediments to measurement by ED-XRF. Conse-quently, the scheme which uses the 200 ∗ avoid the contamination by non-metallif-erous material as exemplified by (Zr + Y) apex represents the best alternative to our preferred discrimination samples from the Eolo Seamount (Dekov et al., 2009). diagram to use at sea. In the absence of data for Ce, the zone for the hydrogenetic crust is less re-strictive but interestingly, Zr and Y can be easily determined at sea using

185 Appendix D. Ore Geology Review publication Josso et al., (2016)

P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx 11

Fig. 7. (A) Ternary discriminative diagram for genetic classification of oceanic ferromanganese deposits displaying all samples used in this study and (B) corresponding genetic fields. The dashed arrows present the mixing trends between two genetic processes highlighting the continuum between hydrogenetic-hydrothermal crusts and hydrogenetic-diagenetic nodules. Solid arrows show trends of a sample set related to only one genetic process.

a compact, benchtop ED-XRF spectrometer (Fig. S3; supplementary in- polished sections. The authors would like to thank ERAMET, AREVA, formation). This easy-to-use discrimination diagram, aside from deciphering TECHNIP, BRGM, and AMP for their financial support for the FUTUNA the nature of Fe-Mn ± Si mineralization (i.e. hydrogenetic crusts, polymetallic (2010) cruise. We are thankful as well for the review and helpful discus-sion nodules and metal-rich/metal-poor hydrothermal precipitates), would be with J.R. Hein and an anonymous reviewer. useful for driving deep-sea exploration campaigns.

Appendix A. Supplementary data 5. Conclusions Supplementary data to this article can be found online at http://dx.

doi.org/10.1016/j.oregeorev.2016.09.003. Oceanic Fe-Mn deposits have major differences in HFSE enrichments and REE signatures. Such features can be used and incorporated into a new discrimination scheme. Hydrogenetic crusts possess high HFSE References concentrations with strong positive Ce anomaly, followed closely by concentrations in polymetallic nodules, whereas hydrothermal Fe-Mn Alt, J.C., 1988. Hydrothermal oxide and nontronite deposits on seamounts in the eastern Pacific. precipitates show relative depletion in HFSE, with a negative Ce anom-aly. Mar. Geol. 81, 227–239. The relation between deposit-forming processes and Fe, Mn, Cu, Ni, Co, Aplin, A.C., Cronan, D.S., 1985. Ferromanganese oxide deposits from the central Pacific Ocean, II. Nodules and associated sediments. Geochim. Cosmochim. Acta 49, 437–451. HFSE and REY concentrations has also been demonstrated for each type of Fe-Mn deposit. Trace element enrichments such as Co, Zr, Ti and REY are Asavin, A.M., Kubrakova, I.V., Mel'nikov, M.E., Tyutyunnik, O.A., Chesalova, E.I., 2010. Geo- favored by hydrogenetic precipitation, whereas diagenetic processes provide chemical zoning in ferromanganese crusts of Ita-MaiTai guyot. Geochem. Int. 48, 423–445. higher Mn, Cu, and Ni concentrations when diagenet-ic remobilization occurs Balaram, V., Anjaiah, K.V., Reddy, M.R.P., 1995. Comparative study on the trace and rare earth in the sediment column. Hydrothermal Fe-Mn deposits show strong depletion element analysis of an Indian polymetallic nodule reference sample by induc-tively coupled in Zr, Ti and REY due to rapid formation and strongly variable metal contents plasma atomic emission spectrometry and inductively coupled plasma mass spectrometry. Analyst 120, 1401–1406. of fluids. Taking these differences into account, a ternary diagram was Barrat, J.A., Keller, F., Amosse, J., Taylor, R.N., Nesbitt, R.W., Hirata, T., 1996. Determination developed that clearly discrimi-nates the various Fe-Mn deposit types. The of rare earth elements in sixteen silicate reference samples by ICP-MS after Tm addi-tion ternary diagram uses metals and ion exchange separation. Geostand. Newslett. 20, 133–139. Barret, T.J., Jarvis, I., 1988. Rare earth element geochemistry of metalliferous sediments from — HFSE–Fe-Mn that distinguish the main deposit-forming processes and the DSDP LEG 92: the East Pacific Rise Transect. Chem. Geol. 67, 243–259. interaction of those processes that form mixed-type diagenet-ic-hydrogenetic Barret, T.J., Taylor, P.N., Lugoqski, J., 1987. Metalliferous sediments from DSDP Leg 92: the nodules and hydrogenetic-hydrothermal crusts, as well as possibly East Pacific Rise Transect. Geochim. Cosmochim. Acta 51, 2241–2253. Baturin, G.N., Dubinchuck, V.T., 2009. Composition of ferromanganese nodules from Riga Bay distinguishing oxic-suboxic bulk diagenetic deposits, and variable metal (Baltic Sea). Oceanology 49, 111–120. enrichment in hydrothermal deposits. Finally, we propose an easy-to-use Baturin, G.N., Dubinchuck, V.T., 2010. On the composition of ferromanganese nodules of the version of this diagram to discriminate seafloor Fe-Mn precipitates that can be Indian Ocean. Dokl. Earth Sci. 434, 1179–1183. Baturin, G.N., Dubinchuck, V.T., 2011. Mineralogy and chemistry of ferromanganese crusts used on-board ships for driving deep-sea exploration projects using XRF from the Atlantic Ocean. Geochem. Int. 48, 578–593. measurements. Baturin, G.N., Dubinchuck, V.T., Rashidov, V.A., 2010. Composition of Fe-Mn Crusts from Okhotsk Sea. Minerals of the Ocean-5 & Deep-sea Minerals and Mining-2 Joint Inter- national Conference. St. Petersburg, Russia, 28th June–1st July, p. 2010. Acknowledgement Bau, M., 1991. Rare-earth element mobility during hydrothermal and metamorphic fluid-rock interaction and the significance of the oxidation state of europium. Chem. Geol. 93, 219– 230. We thank the crews of R/V l'Atalante and Nautile HOV (Ifremer, France) Bau, M., Koschinsky, A., 2009. Oxidative scavenging of cerium on hydrous Fe oxide: evi-dence and all participants of the FUTUNA cruise (chief scientist: Y. Fouquet) for from the distribution of rare earth elements and yttrium between Fe oxides and Mn oxides their assistance at sea. The authors would like to thank G. Bayon for his help in hydrogenetic ferromanganese crusts. Geochem. J. 43, 37–47. Bau, M., Koschinsky, A., Dulski, P., Hein, J.R., 1996. Comparison of the partitioning behav- on the method used for ICP measurements, A.S. Alix for the GIS data iours of yttrium, rare earth elements, and titanium between hydrogenetic marine fer- management and P. Fernagu for the preparation of romanganese crusts and seawater. Geochim. Cosmochim. Acta 60, 1709–1725.

186 Appendix D. Ore Geology Review publication Josso et al., (2016)

12 P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx

Bau, M., Schmidt, K., Koschinsky, A., Hein, J.R., Usui, A., 2014. Discriminating between dif- Hein, J.R., Koschinsky, A., 2013. Deep-ocean ferromanganese crusts and nodules. In: Holland, ferent genetic types of marine ferro-manganese crusts and nodules based on rare earth H.D., Turekian, K.K. (Eds.), 2nd Ed. Treatise on Geochemistry vol. 13. Elsevier, elements and yttrium. Chem. Geol. 381, 1–9. Amsterdam, pp. 273–291 (Chapter 11). Bonatti, E., 1975. Metallogenesis at oceanic spreading centers. Annu. Rev. Earth Planet. Sci. 3, Hein, J.R., Yeh, H.-W., Gunn, S.H., Gibbs, A.E., Wang, C.-H., 1994. Composition and origin of 401–431. hydrothermal ironstones from central Pacific seamounts. Geochim. Cosmochim. Acta 58, Bonatti, E., Kraemer, T., Rydell, H., 1972. Classification and genesis of submarine iron- 179–189. manganese deposits. In: HORN, D. (Ed.), Ferromanganese Deposits on the Ocean Floor. Hein, J.R., Gibbs, A.E., Clague, D.A., Torresan, M., 1996. Hydrothermal mineralization along Natl. Sci. Found., Washington, pp. 149–165. submarine rift zones, Hawaii. Mar. Georesour. Geotechnol. 14, 177–203. Bruland, K.W., 1983. Trace elements in seawater. In: RIPLEY, J.P., CHESTER, R. (Eds.), Hein, J.R., Koschinsky, A., Halbach, P., Manheim, F.T., Bau, M., Kang, J.-K., Lubik, N., 1997. Chemical Oceanography. Academic Press, pp. 157–200 (1983). Iron and manganese oxide mineralization in the Pacific. In: Nicholson, K., Hein, J.R., Burgath, K.P., Von Stackelberg, U., 1995. Sulfide-impregnated volcanics and ferromanga-nese Bühn, B., Dasgupta, S. (Eds.), Manganese Mineralization: Geochemistry and Mineral-ogy incrustations from the southern Lau Basin (southwest Pacific). Mar. Georesour. of Terrestrial and Marine Deposits, pp. 123–138. Geotechnol. 13, 263–308. Hein, J.R., Koschinsky, A., Bau, M., Manheim, F.T., Kang, J.-K., Roberts, L., 2000. Cobalt-rich Calvert, S.E., Piper, D.Z., 1984. Geochemistry of ferromanganese nodules from DOMES Site ferromanganese crusts in the Pacific. In: Cronan, D.S. (Ed.), Handbook of Marine Min-erals A, northern equatorial Pacific: multiple diagenetic metal source in the deep sea. Geochim. Deposit. CRC Press, Boca Raton, Florida, pp. 239–279. Cosmochim. Acta 48, 1913–1928. Hein, J.R., Koschinsky, A., Mc Intyre, B.R., 2005. Mercury- and silver-rich ferromanganese Calvert, S.E., Price, N.B., 1977. Geochemical variation in ferromanganese nodules and asso- oxides, southern California borderland: deposit model and environmental implica-tions. ciated sediments from the Pacific Ocean. Mar. Chem. 5, 43–74. Econ. Geol. 100, 1151–1168. Cann, J.R., Winter, C.K., Pritchard, R.G., 1977. A hydrothermal deposit from the floor of the Hein, J.R., Clague, D.A., Koski, R.A., Embley, R.W., Dunham, R.E., 2008a. Metalliferous sed- Gulf of Aden. Mineral. Mag. Lond. 41, 193–199. iment and a silica-hematite deposit within the Blanco Fracture Zone, northeast Pacif-ic. Choi, J.H., Hariya, Y., 1992. Geochemistry and depositional environment of Mn oxide de-posits Mar. Georesour. Geotechnol. 26 (4), 317–339. in the Tokoro Belt, northeastern Hokkaido, Japan. Econ. Geol. 87, 1265–1274. Hein, J.R., Schulz, M.S., Dunham, R.E., Stern, R.J., Bloomer, S.H., 2008b. Diffuse flow hydro- Conly, A.G., Scott, S.D., Bellon, H., 2011. Metalliferous manganese oxide mineralization as- thermal manganese mineralization along the active Mariana and southern Izu-Bonin arc sociated with the Boléo Cu-Co-Zn district, Mexico. Econ. Geol. 106, 1173–1196. system, western Pacific. J. Geophys. Res. 113 (BO8S14, 29pp.). Conrad, T., Hein, J.R., Paytan, A., Clague, D.A., 2016. Formation of Fe-Mn crusts within a Hein, J.R., Mizell, K., Koschinsky, A., Conrad, T.A., 2013. Deep-ocean mineral deposits as a continental margin environment. Ore Geol. Rev. (accepted for publication). source of critical metals for high- and green-technology applications: comparison with Corliss, J.B., Lyle, M., Dymond, J., Crane, K., 1978. The chemistry of hydrothermal mounds land-based resources. Ore Geol. Rev. 51, 1–14. near the Galapagos rift. Earth Planet. Sci. Lett. 40, 12–24. Hein, J.R., Ronde, C.E.J., Koski, R.A., Ditchburn, R.G., Mizell, K., Tamura, Y., Stern, R.J., Craddock, P.R., Bach, W., Seewald, J.S., Rouxel, O.J., Reeves, E., Tivet, M.K., 2010. Rare Conrad, T.A., Ishizuka, O., Leybourne, M.I., 2014. Layered hydrothermal barite-sulfide earth element abundances in hydrothermal fluids from the Manus Basin, Papua New Guin- mount field, East Diamante caldera, Mariana volcanic arc. Econ. Geol. 109, 2179–2206. ea: indicators of sub-seafloor hydrothermal processes in back-arc basins. Geochim. Cosmochim. Acta 74, 5494–5513. Jung, H.-S., Lee, C.-B., 1999. Growth of diagenetic ferromanganese nodules in an oxic deep- De Carlo, E.H., Mc Murtry, G.M., 1992. Rare-earth element geochemistry of ferroman-ganese sea sedimentary environment, northeast equatorial Pacific. Mar. Geol. 157, 127–144. crusts from the Hawaiian Archipelago, central Pacific. Chem. Geol. 95, 235–250. Koschinsky, A., Fritsche, U., Winkler, A., 2001. Sequential leaching of Peru Basin surface Dekov, V.M., Kamenov, G.D., Savelli, C., Stummeyer, J., Thiry, M., Shanks, W.C., sediment for the assessment of aged and fresh heavy metal associations and mobility. Willingham, A.L., Boycheva, T.B., Rochette, P., Kuzmann, E., Fortin, D., Vértes, A., 2009. Deep-Sea Res. II Top. Stud. Oceanogr. 48, 3683–3699. Metalliferous sedi-ments from Eolo Seamount (Tyrrhenian Sea): hydrothermal deposition Kuhn, T., Bau, M., Blum, N., Halbach, P., 1998. Origin of negative Ce anomalies in mixed and re-deposition in a zone of oxygen depletion. Chem. Geol. 264, 347–363. hydrothermal–hydrogenetic Fe–Mn crusts from the Central Indian Ridge. ett. 163, 207– Dekov, V.M., Boycheva, T.B., Halenius, U., Billstrom, K., Kamenov, G.D., Shanks, W.C., 220. Stummeyer, J., 2011. Mineralogical and geochemical evidence for hydrothermal ac-tivity at Kuhn, T., Bostick, B.C., Koschinsky, A., Halbach, P., Fendorf, S., 2003. Enrichment of Mo in the west wall of 12°50′N core complex (Mid-Atlantic ridge): a new ultramaf-ic-hosted hydrothermal Mn precipitates: possible Mo sources, formation process and phase as- seafloor hydrothermal deposit? Mar. Geol. 288, 90–102. sociations. Chem. Geol. 199, 29–43. Douville, E., Bienvenu, P., Charlou, J.L., Donval, J.P., Fouquet, Y., Appriou, P., Gamo, T., Michard, A., 1989. Rare earth element systematics in hydrothermal fluids. Geochim. 1999. Yttrium and rare earth elements in fluids from various deep-sea hydrothermal sys- Cosmochim. Acta 53, 745–750. tems. Geochim. Cosmochim. Acta 63, 627–643. Mills, R.A., Wells, D.M., Roberts, S., 2001. Genesis of ferromanganese crusts from the TAG Douville, E., Charlou, J.L., Oelkers, E.H., Bienvenu, P., Jove Colon, C.F., Donval, J.P., hydrothermal field. Chem. Geol. 176, 283–293. Fouquet, Y., Prieur, D., Appriou, P., 2002. The rainbow vent fluids (36°14′N, MAR): the Muiños, S.B., Hein, J.R., Frank, M., Monteiro, J.H., Gaspar, L., Conrad, T., Pereira, G., influ-ence of ultramafic rocks and phase separation on trace metal content in Mid-Atlantic Abrantes, F., 2013. Deep-sea Fe-Mn crusts from the northeast Atlantic Ocean: composition Ridge hydrothermal fluids. Chem. Geol. 184, 37–48. and resource considerations. Mar. Georesour. Geotechnol. 31, 40–70. Dymond, J., Lyle, M., Finney, B., Piper, D.Z., Murphy, K., Conard, R., Pisias, N., 1984. Ferro- manganese nodules from MANOP sites H, S and R — control of mineralogical and Nicholson, K., 1992. Contrasting mineralogical-geochemical signatures of manganese ox-ides: chemical composition by multiple accretionary processes. Geochim. Cosmochim. Acta 48, guides to metallogenesis. Econ. Geol. 87, 1253–1264. 931–949. Ohta, A., Ishii, S., Sakakibara, M., Mizuno, A., Kawabe, I., 1999. Systematic correlation of the Edmonds, H.N., German, C.R., 2004. Particle geochemistry in the Rainbow hydrothermal Ce anomaly with the Co/(Ni + Cu) ratio and Y fractionation from Ho in distinct types of plume, Mid-Atlantic Ridge. Geochim. Cosmochim. Acta 68, 759–772. Pacific deep-sea nodules. Geochem. J. 33, 399–417. Edwards, K.J., Glazer, B.T., Rouxel, O.J., Bach, W., Emerson, D., Davis, R.E., Toner, B.M., Pattan, J.N., Banakar, V.K., 1993. Rare earth element distribution and behaviour in buried Chan, C.S., Tebo, B.M., Staudigel, H., Moyer, C.L., 2011. Ultra-diffuse hydrothermal managanese nodules from the Central India Basin. Mar. Geol. 112, 303–312. venting supports Fe-oxidizing bacteria and massive umber deposition at 5000 m off Pelleter, E., Fouquet, Y., Etoubleau, J., Cheron, S., Labanieh, S., Josso, P., Langlade, J., 2016. Hawaii. Int. Soc. Microb. Ecol. 5, 1748–1758. Ni-Cu-Co-rich hydrothermal manganese mineralization in the Wallis and Futuna back-arc Fitzgerald, C.E., Gillis, K.M., 2006. Hydrothermal manganese oxide deposits from Baby Bare environment (SW Pacific). Ore Geol. Rev. (accepted for publication). seamount in the Northeast Pacific Ocean. Mar. Geol. 225, 145–156. Fouquet, Y., Von Stackelberg, U., Charlou, J.L., Erzinger, J., Herzig, P.M., Muehe, R., Pelletier, B., Lagabrielle, Y., Benoit, M., Cabioch, G., Calmant, S., Garel, E., Guivel, C., 2001. Wiedicke, M., 1993. Metallogenesis in back-arc environments: the Lau Basin example. Newly identified segments of the Pacific–Australia plate boundary along the North Fiji Econ. Geol. 88, 2154–2181. transform zone. Earth Planet. Sci. Lett. 193, 347–358. Fouquet, Y., Scalabrin, C., Alix, A.S., Donval, J.P., Dupré, S., Guérin, C., Guillou, M., Pourret, O., Davranche, M., 2013. Rare earth element sorption onto hydrous manganese oxide: Guyader, V., Konn, C., Patriat, M., Pelleter, E., Pierre, D., Rosazza, F., Saint-Laurent, X., a modeling study. J. Colloid Interface Sci. 395, 18–23. Dyment, J., Szitcar, F., 2015. Multi scales exploration strategy — example of research for Schijf, J., Christenson, E.A., Byrne, R.H., 2015. YREE scavenging in seawater: a new look at hydro-thermal mineralization in the French EEZ of Wallis and Futuna. Offshore an old model. Mar. Chem. 177 (Part 3), 460–471. Technology Conference, 04–07 May, Houston, Texas, USA. Schmidt, K., Bau, M., Hein, J., Koschinsky, A., 2014. Fractionation of the geochemical twins German, C.R., Colley, S., Palmer, M.R., Khripounoff, A., Klinkhammer, G.P., 2002. Hydro- Zr-Hf and Nb-Ta during scavenging from seawater by hydrogenetic ferromanganese crusts. thermal plume-particle fluxes at 13°N on the East Pacific Rise. Deep-Sea Res. I Oceanogr. Geochim. Cosmochim. Acta. Res. Pap. 49, 1921–1940. Schultz, H.D., 2006. Quantification of early diagenesis: dissolved constituents in pore water and Goulding, H.C., Mills, R.A., Nesbitt, R.W., 1998. Precipitation of hydrothermal sediments on signals in the solid phase. In: Schultz, H.D., Zabel, M. (Eds.), Marine Geo-chemistry, 2nd the active TAG mound: implications for ochre formation. In: Mills, R.A., Harrison, K. ed. Springer, Heidelberg, pp. 75–77. (Eds.), Modern Ocean Floor Processes and the Geological Record. Geological Society, Severmann, S., Mills, R.A., Palmer, M.R., Fallick, A.E., 2004. The origin of clay minerals in London, Special Publications 148, pp. 201–216. active and relict hydrothermal deposits. Geochim. Cosmochim. Acta 68, 73–88. Halbach, P., Puteanus, D., 1988. Geochemical trends of different genetic types of nodules and Sun, Z., Zhou, H., Yang, Q., Sun, Z., Bao, S., Yao, H., 2011. Hydrothermal Fe–Si–Mn oxide crusts. In: Halbach, P., Friedrich, G., Von Stackelberg, U. (Eds.), The Manganese Nodule de-posits from the central and south Valu Fa Ridge, Lau Basin. Appl. Geochem. 26, 1192– Belt of the Pacific Ocean. Ferdinand Enke, Stuttgart, pp. 61–69 ((1988) Chapter 4). 1204. Halbach, P., Scherhag, C., Hebish, U., Marchig, V., 1981. Geochemical and mineralogical Sun, Z., Zhou, H., Glasby, G.P., Yang, Q., Yin, X., Li, J., Chen, Z., 2012. Formation of Fe– control of different genetic types of deep-sea nodules from the Pacific Ocean. Mineral. Mn–Si oxide and nontronite deposits in hydrothermal fields on the Valu Fa Ridge, Lau Deposita 16, 59–84. Basin. J. Asian Earth Sci. 43, 64–76. Halbach, P., Friedrich, G., Von Stackelberg, U., 1988. The Manganese Nodule Belt of the Pa- Surya Prakash, L., Ray, D., Paropkari, A.L., Mudholkar, A.V., Satyanarayanan, M., Sreenivas, cific Ocean Geological Environment, Nodule Formation, and Mining Aspects. Enke Verlag, B., Chandrasekharam, D., Kota, D., Kamesh Raju, K.A., Kaisary, S., Balaram, V., Gurav, Stuttgart ((1988) 245 pp.). T., 2012. Distribution of REEs and yttrium among major geochemical phases of marine

187 Appendix D. Ore Geology Review publication Josso et al., (2016)

P. Josso et al. / Ore Geology Reviews xxx (2016) xxx–xxx 13

Fe–Mn-oxides: comparative study between hydrogenous and hydrothermal deposits. Chem. Mineralization: Geochemistry and Mineralogy of Terrestrial and Marine Deposits, pp. 153– Geol. 312–313, 127–137. 176. Takahashi, Y., Manceau, A., Geoffroy, N., Marcus, M.A., Usui, A., 2007. Chemical and struc- Wegorzewski, A.V., Kuhn, T., 2014. The influence of suboxic diagenesis on the formation of tural control of the partitioning of Co, Ce, and Pb in marine ferromanganese oxides. manganese nodules in the Clarion Clipperton nodule belt of the Pacific Ocean. Mar. Geol. Geochim. Cosmochim. Acta 71, 984–1008. Taylor, S.R., Mclennan, S.M., 1985. The Continental Crust: Its Composition and Evolution. Zeng, Z., Ouyang, H., Yin, X., Chen, S., Wang, X., Wu, L., 2012. Formation of Fe–Si–Mn Blackwell Scientific, Boston, Mass (312 pp.). oxyhydroxides at the PACMANUS hydrothermal field, eastern Manus Basin: mineral- Von Stackelberg, U., 1997. Growth history of manganese nodules and crusts of the Peru Basin. ogical and geochemical evidence. J. Asian Earth Sci. 60, 130–146. In: Nicholson, K., Hein, J.R., Bühn, B., Dasgupta, S. (Eds.), Manganese

188 Bibliography Bibliography

Abreu, R. D. & Morais, C. A. (2010). Purification of rare earth elements from monazite sulphuric acid leach liquor and the production of high-purity ceric oxide. Minerals Engineering, 23, 536-540. Adachi, M., Yamamoto, K. & Sugisaki, R. (1986). Hydrothermal chert and associated siliceous rocks from the northern Pacific: their geological significance as indication of ocean ridge activity. Sedimentary Geology, 47, 125-148. Alibo, D. S. & Nozaki, Y. (1999). Rare earth elements in seawater: Particle association, shale- normalization, and Ce oxidation. Geochimica and Cosmochimica Acta, 63, 363-372. Allerton, S. & Vine, F. J. (1987). Spreading structure of the Troodos ophiolite, Cyprus: some palaeomagnetic constraints. Geology, 15, 593-597. Alt, J. C. (1988). Hydrothermal oxide and nontronite deposits on seamounts in the eastern Pacific. Marine Geology, 81, 227-239. Anonymous, A. H. F. (1972). Penrose field conference on ophiolites. Geotimes, 17, 24-25. Asavin, A. M., Kubrakova, I. V., Mel'nikov, M. E., Tyutyunnik, O. A. & Chesalova, E. I. (2010). Geochemical zoning in ferromanganese crusts of Ita-MaiTai guyot. Geochemistry International, 48, 423-445. Bach, W., Roberts, S., Vanko, D. A., Binns, R. A., Yeats, C. J., Craddock, P. R. & Humphris, S. E. (2003). Controls of fluid chemistry and complexation on rare-earth element contents of anhydrite from the Pacmanus subseafloor hydrothermal system, Manus Basin, Papua New Guinea. Mineralogical Deposita, 38, 916-935. Baragar, W. R. A., Lambert, M. B. B., N & Gibson, I. L. (1990). The sheeted dyke zone in the Troodos Ophiolite. In: Malpas, J., Moorea, E. M., Panayaiotou, A. & Xenophontos, C. (eds.) Ophiolites; oceanic crustal analogues; proceedings of the symposium "Troodos 1987". Nicosia, Cyprus, Ministry of Agriculture and Natural Resources, geological Survey Department: 37-51. Barret, T. J. & Jarvis, I. (1988). Rare earth element geochemistry of metalliferous sediments from DSDP LEG 92: The East Pacific Rise Transect. Chemical Geology, 67, 243-259. Barret, T. J., Taylor, P. N., Jarvis, I. & Lugowski, J. (1986). Pb and Sr isotope and Rare Earth Element composition of selected metalliferous sediments from sites 597 to 601, Deep Sea Drilling Project Leg 92. DSDP Report. Baturin, G. N. (2009). Geochemistry of ferromanganese nodules in the Gulf of Finland, Baltic Sea. Lithology and Mineral Resources, 44, 411-426. 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, 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, 323-333. Bau, M. & Dulski, P. (1999). Comparing yttrium and rare earths in hydrothermal fluids from the Mid-Atlantic Ridge: implications for Y and REE behaviour during near-vent mixing and for the Y/Ho ratio of Proterozoic seawater. Chemical Geology, 155, 77-90. Bau, M. & Koschinsky, A. (2009). Oxidative scavenging oc cerium on hydrous Fe oxide: Evidence from the distributionof rare earth elements and yttrium between Fe oxides and Mn oxides in hydrogenetic ferromanganese crusts. Geochemical Journal, 43, 37-47. Bear, L. M. (1960). The geology and mineral resources of the Akaki-Lythrodonda area, Cyprus. Mem. Geol. Surv. Dep. Cyprus, 3, 122p. Bear, L. M. (1963). The geology and mineral resources of the Akaki-Lythrodonda area, Cyprus. Mem. Geol. Surv. Dep. Cyprus, , 3, 122p.

189 Bibliography

Bettison-Varga, L., Varga, R. J. & Schiffman, P. (1992). Relation between ore-forming hydrothermal systems and extensional deformation in the Solea Graben spreading centre, troodos Ophiolite, Cyprus. Geology, 20, 987-990. Bhat, T. R. & Rao, T. V. (1964). Solubilities of Rare Earth Oxalates in EDTA Solutions and the Determination of their Solubility Products. Zeitschrift fiir nnorganisclie nnd allgemeine Chemie, 332, 204-208. Bickle, M. J. & Teagle, D. a. H. (1992). Strontium alteration in the Troodos ophiolite: implications for fluid fluxes and geochemical transport in mid-ocean ridge hydrothermal systems. Earth and Planetary Science Letters, 113, 219-237. Binns, R. A., Scott, S. D., Bogdanov, Y., A, Lisitzin, A. P., Gordeev, V. V., Gurvich, E. G., Finlayson, E. J., Boyd, T., Dotter, L. E., Wheller, G. E. & Muravyev, K. G. (1993). Hydrothermal Oxide and Gold-rich Sulfate Deposits of Franklin Seamount, Western Woodlark Basin, Papua New Guinea. Economic Geology, 88, 2122-2153. Bodeï, S., Buatier, M., Steinmann, M., Adatte, T. & Wheat, C. (2008). Characterization of metalliferous sediment from a low-temperature hydrothermal environment on the Eastern Flank of the East Pacific Rise. Marine Geology, 250, 128-141. Bonatti, E., Kraemer, T. & Rydell, H. (1972). Classification and genesis of submarine iron- manganese deposits. In: Horn, D. (ed.) Ferromanganese Deposits on the Ocean Floor, Natl. Sci. Found. Washington, pp. 149-165. Boström, K. (1973). The origin and fate of ferromanganoan active ridge sediments. Stockholm Contrib. Geol., 27, 149-243. Boström, K. & Peterson, M. N. A. (1968). The origin of aluminium-poor ferromanganoan sediments in areas of high heat flow on the East Pacific Rise. Marine Geology, 7, 427-447. Boyle, J. F. (1984). The origin and geochemistry of the metalliferous sediments of the Troodos massif, Cyprus. Doctor of Philosophy, University of Edinburgh. Boyle, J. F. (1990). The composition and origin of metalliferous sediments from the Troodos ophiolite, Cyprus. In: Malpas, J. (ed.) Ophiolites: oceanic crustal analogues, Proceedings of the Symposium "Troodos" 1987: Nicosia, Cyprus. Geological Survey, p. 705-718. Boyle, J. F. & Robertson, A. H. F. (1984). Evolving metallogenesis at the Troodos spreading axis. Geological Society, London, Special Publications, 13, 169-181. Bragina, L. G. & Bragin, N. Y. (2006). Stratigraphy and Radiolarians of Upper Cretaceous Sedimentary Cover of the Arakapas Ophiolite Massif (Cyprus). Stratigraphy and Geological Correlation, 14, 507-523. British Geological Survey Report (2011). Rare Earth Element. In: British Geological Survey (ed.). British Geological Survey, 54p

Brown, D., Ma, B.-M. & Chen, Z. (2002). Developments in the processing and properties of NdFeb- type permanent magnets. Journal of Magnetism and Magnetic Materials, 248, 432-440. Cail, T. L. & Cline, J. S. (2001). Alteration Associated with Gold Deposition at the Getchell Carlin- Type Gold Deposit, North-Central Nevada. Economic Geology, 96, 1343-1359. Calvert, S. E. & Piper, D. Z. (1984). Geochemistry of ferromanganese nodules from DOMES Site A, northern equatorial Pacific: Multiple diagenetic metal source in the deep sea. Geochimica et Cosmochimica Acta, 48, 1913-1928. Calvert, S. E. & Price, N. B. (1977). Geochemical variation in ferromanganese nodules and associated sediments from the Pacific Ocean. Marine Chemistry, 5, 43-74. Canet, C., Prol-Ledesman, R. M., Bandy, W. L., Schaaf, P., Linares, C., Camprubi, A., Tauler, E. & Mortera-Gutierrez, C. (2008). Mineralogical and geochemical constraints on the origin of ferromanganese crusts from the Rivera Plate (western margin of Mexico). Marine Geology, 251, 47-59. Cann, J. R., Prichard, H. M., Malpas, J. & Xenophontos, C. (2001). Oceanic inside corner detachments of the Limassol Forest area, troodos ophiolite, Cyprus. Journal of the Geological Society of London, 158, 757-767.

190 Bibliography

Carr, J. M. & Bear, L. M. (1960). The geology and mineral resources of the Peristerona-Lagoudhera area. Mem. Geol. Surv. Dep. Cyprus, 2, 79p. Chi, R. & Xu, Z. (1998). A Solution Chemistry Approach to the Study of Rare Earth Element Precipitation by Oxalic Acid. Metallurgical and Materials Transactions, 30B. Christie, T., Brathwaite, B. & Tulloch, A. (1998). Rare Earths and Related Elements. Mineral Comodity Report 17. Chung, D. Y., Kim, E. H., Lee, E. H. & Yoo, J. H. (1998). Solubility of Rare Earth Oxalate in Oxalic and Nitric Acid Media. Journal of Industrial and Engineering Chemistry, 4, 277-284. Cole, C., James, R. H., Connelly, D. P. & Hathorne, E. C. (2014). Rare earth elements as indicators of hydrothermal processes within the East Scotia subduction zone system. Geochimica and Cosmochimica Acta, 140, 20-38. Constantinou, G. & Govett, G. J. S. (1972). Genesis of sulphide deposits, ochre and umber of Cyprus. Transactions of the Institute of Mining Metallurgy, 81, 34-46. Constantinou, G. & Govett, G. J. S. (1973). Geology, Geochemistry, and Genesis of Cyprus Sulfide Deposits. Economic Geology, 68, 843-858. Craddock, P. R., Bach, W., Seewald, J. S., Rouxel, O. J., Reeves, E. & Tivet, M. K. (2010). Rare earth element abundances in hydrothermal fluids from the Manus Basin, Papua New Guinea: Indicators of sub-seafloor hydrothermal processes in back-arc basins. Geochimica and Cosmochimica Acta, 74, 5494-5513. Crouthamel, C. E. & Martin, D. S. (1951). Solubility of the Rare Earth Oxalates and Complex Ion Formation in Oxalate Solution. II. Neodoymium and Cerium (III). Journal of the American Chemical Society, 72, 1382-1386. Cyprus Geological Survey (2006). Report on raw materials mineral production in Cyprus. Ministry of Agriculture Natural Resources and Environment. Cyprus Geological Survey. (2017). Geological Survey Department/Geology of Cyprus/Tectonic evolution [Online]. [Accessed 06.05.2017]. De Baar, H. J. W. (1983). The maine geochemistry of the rare-earth elements. Ph.D. Thesis, MIT/WHOI. De Baar, H. J. W., Bacon, M. P., Brewer, P. G. & Bruland, K. W. (1985). Rare earth elements in the Pacific ans Atlantic Oceans. Geochimica and Cosmochimica Acta, 49, 1943-1959. De Baar, H. J. W., German, C. R., Elderfield, H. & Van Gaans, P. (1988). Rare earth element distributions in anoxic waters of the Cariaco Trench. Geochimica et Cosmochimica Acta, 52, 1203-1219. Dekov, V. M., Boycheva, T. B., Halenius, U., Billstrom, K., Kamenov, G. D., Shanks, W. C. & Stummeyer, J. (2011). Mineralogical and geochemical evidence for hydrothermal activity at the west wall of 12°50′N core complex (Mid-Atlantic ridge): A new ultramafic-hosted seafloor hydrothermal deposit? Marine Geology, 288, 90-102. Dekov, V. M., Kamenov, G. D., Savelli, C., Stummeyer, J., Thiry, M., Shanks, W. C., Willingham, A. L., Boycheva, T. B., Rochette, P., Kuzmann, E., Fortin, D. & Vértes, A. (2009). Metalliferous sediments from Eolo Seamount (Tyrrhenian Sea): Hydrothermal deposition and redeposition in a zone of oxygen depletion. Chemical Geology, 264, 347-363. Delany, J. M. & Lundeen, S. R. (1990). The LLNL thermochemical database. Lawrence Livermore Natl. Lab. Livermore, CA, 150p. Dilek, Y. & Furnes, H. (2009). Structure and geochemistry of Tethyan ophiolites and their petrogenesis in subduction rollback systems. Lithos, 113, 1-20. Douville, E., Bienvenu, P., Charlou, J. L., Donval, J. P., Fouquet, Y., Appriou, P. & Gamo, T. (1999). Yttrium and rare earth elements in fluids from various deep-sea hydrothermal systems. Geochimica and Cosmochimica Acta, 63, 627-643. Douville, E., Charlou, J. L., Oelkers, E. H., Bienvenu, P., Jove Colon, C. F., Donval, J. P., Fouquet, Y., Prieur, D. & Appriou, P. (2002). The rainbow vent fluids (36°14'N, MAR): The influence of ultramafic rocks and phase separation on trace metal content in Mid-Atlantic Ridge hydrothermal fluids. Chemical Geology, 184, 37-48. Drew, L. J., Qingrun, M. & Weijun, S. (1990). The Bayan Obo iron-rare-earth-niobium deposits, Inner Mongolia, China. Lithos, 26, 43-65.

191 Bibliography

Dubinin, A. V. (2004). Geochemistry of Rare Earth Elements in the Ocean. Lithology and Mineral Resources, 39, 289-307. Dymond, J., Lyle, M., Finney, B., Piper, D. Z., Murphy, K., Conard, R. & Pisias, N. (1984). Ferromanganese nodules from MANOP Sites H, S and R - Control of mineralogical and chemical composition by multiple accretionary processes. Geochimica et Cosmochimica Acta, 48, 931-949. Eddy, C. A., Dilek, Y., Hurst, S. & Moores, E. M. (1998). Seamount formation and associated caldera complex and hydrothermal mineralization in ancient oceanic crust, Troodos ophiolite (Cyprus). Tectonophysics, 292, 189-210. Edmond, J. M., Measures, C., Mc Duff, R. E., Chan, L. H., Collier, R., Grant, B., Gordon, L. I. & Corliss, J. B. (1979). Ridge crest hydrothermal activity and the balance of the major and minor elements in the ocean: The Galapagos data. Earth and Planetary Science Letters, 46, 1-18. Edmond, J. M., Von Damm, K. L., Mc Duff, R. E. & Measures, C. (1982). Chemistry of hot springs on the East Pacific Rise and their effluent dispersal. Nature, 297, 187-191. Edmonds, H. N. & German, C. R. (2004). Particle geochemistry in the Rainbow hydrothermal plume, Mid-Atlantic Ridge. Geochimica et Cosmochimica Acta, 68, 759-772. Edwards, K. J., Glazer, B. T., Rouxel, O. J., Bach, W., Emerson, D., Davis, R. E., Toner, B. M., Chan, C. S., Tebo, B. M., Staudigel, H. & Moyer, C. L. (2011). Ultra-diffuse hydrothermal venting supports Fe-oxidizing bacteria and massive umber deposition at 5000m off Hawaii. International Society for Microbial Ecology, 5, 1748-1758. Eijsink, L. M., Krom, M. D. & Herut, B. (2000). Speciation and burial flux of phosphorus in the surface of the eastern Mediterranean. American Journal of Science, 300, 483-503. Elderfield, H. (1988). The oceanic chemistry of the rare-earth elements. Phil. Trans. R. Soc. Lond., 325, 105-126. Elderfield, H., Gass, I. G., Hamond, A. & Bear, L. M. (1972). The origin of ferromanganese sediments associated with the Troodos Massif of Cyprus. Sedimentology, 19, 1-19. Elderfield, H. & Greaves, M. J. (1982). The rare earth elements in seawater. Nature, 296, 6p. European Commission (2014). Report on critical materials for the EU. Report of the Ad hoc Working Group on defining critical raw materials. Brussels. Feely, R. A., Massoth, G. J., Baker, E. T., Cowen, J. P., Lamb, M. F. & Krosglund, K. A. (1990). The effect of hydrothermal processes on midwater phosphorus distributions in the northeast Pacific. Earth and Planetary Science Letters, 96, 305-318. Feely, R. A., Trefry, J. H., Baker, E. T., Paulson, A. J. & Lebon, G. T. (1998). The relationship between P/Fe and V/Fe ratios in hydrothermal precipitates and dissolved phosphate in seawater. Geophysical Research Letter, 25, 2253-2256. Feibush, A. M., Rowley, K. & Gordon, L. I. (1958). Solubility of Yttrium Oxalate. Analytical Chemistry, 30, 1610-1612. Fouquet, Y., Von Stackelberg, U., Charlou, J. L., Erzinger, J., Herzig, P. M., Muehe, R. & Wiedicke, M. (1993). Metallogenesis in back-arc environments; The Lau Basin example. Economic Geology, 88, 2154-2181. Fujimoto, J., Tanaka, K., Watanabe, N. & Takahashi, Y. (2016). Simultaneous recovery and separation of rare earth elements in ferromanganese nodules by using Shewanella putrefaciens. Hydrometallurgy. Gale, N. H., Spooner, E. T. C. & Potts, P. J. (1981). The lead and strontium isotope geochemistry of metalliferous sediments associated with Upper Cretaceous ophiolitic rocks in Cyprus, Syria, and the Sultanate of Oman. Canadian Journal of Earth Sciences, 18, 1290-1302. Galley, A. G. & Koski, R. A. (1999). Setting and characteristics of ophiolite-hosted volcanogenic massive sulphide deposits. Reviews in Economic Geology, 8, 215-236. Gammons, C. H. & Wood, S. A. (2000). The aqueous geochemistry of REE.: Part 8: Solubility of ytterbium oxalate and the stability of Yb(III)–oxalate complexes in water at 25°C to 80°C. Chemical Geology, 166, 103-124. Gass, I. G. (1960). The geology and mineral resources of the Dhali area. Mem. Geol. Surv. Dep. Cyprus, 4, 116p.

192 Bibliography

Gass, I. G. (1968). Is the Troodos Massif of Cyprus a Fragment of Mesozoic Ocean floor ? Nature, 220, 4p. Gass, I. G. (1980). The Troodos massif: Its role in the unravelling of the ophiolite problem and its significance in the understanding of constructive margin processes. In: Panayaiotou, A. (ed.) Ophiolites: Proceedings of the International Symposium, Cyprus, . Geol. Surv. Dept. Cyprus, Nicosia,: 1−10. Gass, I. G., Macleod, C. J., Murton, B. J., Panayaiotou, A., Simonian, K. O. & Xenophontos, C. (1994). The geology of the southern Troodos transform fault zone. Ministry of Agriculture and Natural Resources, Geological Survey Department, Nicosia, Cyprus. Gass, I. G. & Masson-Smith, D. (1963). The geology and gravity anomalies of the Troodos massif, Cyprus. Royal Society of London Philosophical Transactions, A255, 417-467. Gass, I. G., Neary, C. R., Plant, J., Robertson, A. H. F., Simonian, K. O., Smewing, J. D., Spooner, E. T. C. & Wilson, R. a. M. (1975). Comments on ``the troodos ophiolitic complex was probably formed in an island arc'', by A. Miyashiro and subsequent correspondence by A. Hynes and A. Miyashiro. Earth and Planetary Science Letters, 25, 236-238. Gass, I. G. & Smewing, J. D. (1973). Intrusion, extrusion and metamorphism at constructive margins: Evidence from the Troodos massif, Cyprus. Nature, 242, 26-29p. German, C. R., Bourlés, D. L., Brown, E. T., Hergt, J., Colley, S., Higgs, N. C., Ludford, E. M., Nelsen, T. A., Feely, R. A., Raisbeck, G. & Yiou, F. (1997). Hydrothermal scavenging on the Juan de Fuca Ridge: 23OThxs, 10Be, and REEs in ridge-flank sediments. Geochimica et Cosmochimica Acta, 61, 4067-4078. German, C. R., Colley, S., Palmer, M. R., Khripounoff, A. & Klinkhammer, G. P. (2002). Hydrothermal plume-particle fluxes at 13°N on the East Pacific Rise. Deep Sea Research Part I: Oceanographic Research Papers, 49, 1921-1940. German, C. R., Higgs, N. C., Thomson, J., Mills, R. A., Elderfield, H., Blusztajn, J., Fleer, A. P. & Bacon, M. P. (1993). A geochemical study of metalliferous sediment from the TAG hydrothermal mound, 26°08'N, Mid-Atlantic Ridge. Journal of Geophysical Research, 98, 9683-9692. German, C. R., Klinkhammer, G. P., Edmond, J. M., Mitra, A. & Elderfield, H. (1990). Hydrothermal scavenging of rare earth elements in the ocean. Nature, 345, 3p. Goldberg, T., Archer, C., Vance, D., Thamdrup, B., Mcanena, A. & Poulton, S. W. (2012). Controls on Mo isotope fractionations in a Mn-rich anoxic marine sediment, Gullmar Fjord, Sweden. Chemical Geology, 296–297, 73-82. Goldstein, S. J. & Jacobsen, S. B. (1988). REE in the Great Whale River estuary, northwest Quebec. Earth and Planetary Science Letters, 88, 241-252. Golev, A., Scott, M., Erskine, P. D., Ali, S. H. & Ballantyne, G. R. (2014). Rare earths supply chains: Current status, constraints and opportunities. Resources Policy, 41, 52-59. Goodenough, K. M., Schilling, J., Jonsson, E., Kalvig, P., Charles, N., Tuduri, J., Deady, E. A., Sadeghi, M., Schiellerup, H., Müller, A., Bertrand, G., Arvanitidis, N., Eliopoulos, D. G., Shaw, R. A., Thrane, K. & Keulen, N. (2016). Europe's rare earth element resource potential: An overview of REE metallogenetic provinces and their geodynamic setting. Ore Geology Reviews, 72, Part 1, 838-856. Goulding, H. C., Mills, R. A. & Nesbitt, R. W. (1998). Precipitation of hydrothermal sediments on the active TAG mound: implications for ochre formation. In: Mills, R. A. & Harrison, K. (eds.) Modern Ocean Floor Processes and the Geological Record, Geological Society, London, Special Publications, 148, 201-216 Grant, J. A. (1986). The isocon diagram: A simple solution to Gresens' equation for metasomatic alteration. Economic Geology, 81, 1976-1982. Grenthe, I., Gardhammar, G. & Rundcrantz, E. (1969). Thermodynamic Properties of Rare Earth Complexes VI. Stability Constants for the Oxalate Compelxes of Ce (III), Eu (III), Tb (III), and Lu (III). Acta Chemica Scandinavica, 23, 93-108. Gresens, R. L. (1967). Composition-volume relationship of metasomatism. Chemical Geology, 2, 47-65.

193 Bibliography

Haas, J. R., Shock, E. L. & Sassini, D. C. (1995). Rare earth elements in hydrothermal systems: estimates of standard partial molal thermodynamic properties of aqueous complexes of the rare earth elements at high pressures and temperatures. Geochimica and Cosmochimica Acta, 59, 4329-4350. Habashi, F. (2013). Extractive Metallurgy of Rare Earths. Canadian Metallurgical Quarterly, 52, 224-233. Hanson, G. N. (1980). Rare earth element in petrogenetic studies of igneous system. Earth and Planetary Science Letters, 8, 371-406. Hansson, E. (1970). Structural studies on the Rare Earth Carboxylates: The crystal and molecular structure of neodymium (III) oxalate 10.5-hydrate. Acta Chemica Scandinavica, 24, 2969- 2982. Haxel, G. B., Hedrick, J. B., Orris, G. J., Stauffer, P. H. & Hendley, J. W. (2002). Rare earth elements: critical resources for high technology. Fact Sheet. USGS. Heath, G. R. & Dymond, J. (1977). Genesis and transformation of metalliferous sediments from the East Pacific Rise, Bauer Deep, and Central Basin, northwest Nazca plate. Geological Society of America Bulletin, 88, 723-733. Hein, J. R., Koschinsky, A., Halbach, P., Manheim, F. T., Bau, M., Kang, J.-K. & Lubik, N. (1997). Iron and manganese oxide mineralization in the Pacific. In: Nicholson, K., Hein, J. R., Bühn, B. & Dasgupta, S. (eds.) Manganese mineralization: Geochemistry and mineralogy of terrestrial and marine deposits. p. 123-138. Hein, J. R., Koschinsky, A. & Mc Intyre, B. R. (2005). Mercury- and silver-rich ferromanganese oxides, southern California borderland: Deposit model and environmental implications. Economic Geology, 100, 1151-1168. Hein, J. R., Mizell, K., Koschinsky, A. & Conrad, T. A. (2013). Deep-ocean mineral deposits as a source of critical metals for high- and green-technology applications: Comparison with land-based resources. Ore Geology Reviews, 51, 1-14. Hein, J. R., Schulz, M. S., Dunham, R. E., Stern, R. J. & Bloomer, S. H. (2008). Diffuse flow hydrothermal manganese mineralization along the active Mariana and southern Izu-Bonin arc system, western Pacific. Journal of Geophysical Research, 113, BO8S14, 29p. Hein, J. R., Vallier, T. L. & Allan, M. A. (1981). Chert petrology and geochemistry, Mid-Pacific Mountains and Hess Rise, Deep Sea Drilling Project Leg 62. DSDP Report. Henderson, P. (1996). The rare earth elements: introduction and review. In: Jones, A. P., Wall, F. & William, T. (eds.) Rare earth elements. Chemistry, origin and ore deposits. 1st edn .Chapman and Hall, London: The Mineralogical Society Series, 372p. Herzig, P. M., Hannington, M. D., Scott, S. D., Maliotis, G., Rona, P. A. & Thompson, G. (1991). Gold-rich sea-floor gossans in the Troodos Ophiolite and on the mid-Atlantic Ridge. Economic Geology, 86, 1747-1755. Hogdahl, O. (1970). Distribution of lanthanides inthe waters of the river Gironde in France. IAEA Research Agreement Coordinated Program on Marine Radioactivity studies. Monaco: Int. At. Energy Agency. Houtz, R. & Ewing, J. (1976). Upper crustal structure as a function of plate age. Journal of Geophysical Research, 81, 2490-2498. Hoyle, J., Elderfield, H., Gledhill, A. & Greaves, M. (1984). The behaviour of the rare earth elements during mixing of river and sea waters. Geochimica et Cosmochimica Acta, 48, 143-149. Hrischeva, E. & Scott, S. D. (2007). Geochemistry and morphology of metalliferous sediments and oxyhydroxides from the Endeavour segment, Juan de Fuca Ridge. Geochimica et Cosmochimica Acta, 71, 3476-3497. Humphris, S. E., Alt, J. C., Teagle, D. a. H. & Honnorez, J. J. (1998). Geochemical Changes During Hydrothermal Alteration of Basement in the Stockwork Beneath the Active TAG Hydrothermal Mound. In: Herzig, P. M., Humphris, S. E., Miller, D. J. & Zierenberg, R. A. (eds.) Proceedings of the Ocean Drilling Program, Scientific Results. Jarvis, I. (1984). Geochemistry and origin of Eocen-Oligocen metalliferous sediments from the Central Equatorial Pacific: Deep Sea Drilling Project Sites 573 and 574. 781-804.

194 Bibliography

Jochum, K. P., Nohl, U., Herwig, K., Lammel, E., Stoll, B. & Horfmann, A. 2005. Geostandards and Geoanalytical Research. Josso, P., Pelleter, E., Pourret, O., Fouquet, Y., Etoubleau, J., Cheron, S. & Bollinger, C. (2016). A new discrimination scheme for oceanic ferromanganese deposits using high field strength and rare earth elements Ore Geology Review (accepted). Jun, T., Jingqun, Y., Ruan, C., Guohua, R., Mintao, J. & Kexian, O. (2010). Kinetics on leaching rare earth from the weathered crust elution-deposited rare earth ores with ammonium sulfate solution. Hydrometallurgy, 101, 166-170. Kanazawa, Y. & Kamitani, M. (2006). Rare earth minerals and resources in the world. Journal of Alloys and Compounds, 408–412, 1339-1343. Karpoff, A. M., Walter, A. V. & Pelumio, C. (1988). Metalliferous sediments within lava sequences of the Sumail ophiolite (Oman): Mineralogical and geochemical characterizartion, origin and evolution. Tectonophysics, 151, 223-245. Kato, Y., Fujinaga, K., Nakamura, K., Takaya, Y., Kitamura, K., Ohta, J., Toda, R., Nakashima, T. & Iwamori, H. (2011). Deep-sea mud in the Pacific Ocean as a potential resource for rare- earth elements. Nature Geoscience, 4, 535-539. Kato, Y., Fujinaga, K., Nozaki, T., Osawa, H., Nakamura, K. & Ono, R. (2005a). Rare earth, Major and Trace Elements in the Kunimiyama Ferromanganese Deposit in the Northern Chichibu Belt, Central Shikoku, Japan. Resource Geology, 55, 291-299. Kato, Y., Fujinaga, K. & Suzuki, K. (2005b). Major and trace element geochemistry and Os isotopic composition of metalliferous umbers from the Late Cretaceous Japanese accretionary complex. Geochemistry Geophysics Geosystems, 6, 20p. Klauber, C., Gräfe, M. & Power, G. (2011). Bauxite residue issues: II. options for residue utilization. Hydrometallurgy, 108, 11-32. Klinkhammer, G. P., Elderfield, H., Edmond, J. M. & Mitra, A. (1994). Geochemical implications of rare earth element patterns in hydrothermal fluids from mid-ocean ridges. Geochimica and Cosmochimica Acta, 58, 5105-5113. Klinkhammer, G. P., Elderfield, H., Greaves, M. J., Rona, P. A. & Nelson, T. (1986). Manganese geochemistry near high-temperature vents in the Mid-Atlantic Ridge rift valley. Earth and Planetary Science Letters, 80, 230-240. Klinkhammer, G. P., Elderfield, H. & Hudson, A. (1983). Rare earth elements in seawater near hydrothermal vents. Nature, 305, 185-188. Konishi, Y. & Noda, Y. (2001). Precipitation Stripping of Rare-Earth Carbonate Powders from Rare- Earth Loaded Carboxylate Solution Using Carbon Diaxide and Water. Industrial & Engineering Chemistry Research, 40, 1793-1797. Koschinsky, A. & Halbach, P. (1995). Sequential leaching of marine ferromanganese precipitates: Genetic implications. Geochimica and Cosmochimica Acta, 59, 5113-5132. Koschinsky, A. & Hein, J. R. (2003). Uptake of elements from seawater by ferromanganese crusts: solid-phase associations and seawater speciation. Marine Geology, 198, 331-351. Kuhn, T., Bau, M., Blum, N. & Halbach, P. (1998). Origin of negative Ce anomalies in mixed hydrothermal-hydrogenetic Fe-Mn crusts from the Central Indian Ridge. Earth and Planetary Science Letters, 163, 207-220. Laveuf, C. & Cornu, S. (2009). A review on the potentiality of Rare Earth Elements to trace pedogenetic processes. Geoderma, 154, 1-12. Lentz, D. R. (1999). Carbonatite genesis: A reexamination of the role of intrusion-related pneumatolytic skarn processes in limestone melting. Geology, 27, 335-338. Linnen, R. L., Samson, I. M., Williams-Jones, A. E. & Chakhmouradian, A. R. (2014). 13.21 - Geochemistry of the Rare-Earth Element, Nb, Ta, Hf, and Zr Deposits - Holland, Heinrich D. In: Turekian, K. K. (ed.) Treatise on Geochemistry (Second Edition). Oxford: Elsevier. Lottermoser, B. G. (1990). Rare-earth element mineralisation within the Mr. Weld carbonatite laterite, Western Australia. Lithos, 24, 151-1677. Macleod, C. J. & Murton, B. J. (1995). On the sense of slip of the Southern Troodos transform fault zone, Cyprus. Geology, 23, 257-260.

195 Bibliography

Marchig, V. & Erzinger, J. (1986). Chemical composition of Pacific Sediments near 20°S: Changes with increasing distance from the East Pacific Rise. Initial Reports of DSDP, 92, 371-381. Marchig, V. & Gundlach, H. (1982). Iron-rich metalliferous sediments on the East Pacific Rise - prototype of undifferentiated metalliferous sediments in divergent plate boundaries. Earth and Planetary Science Letters, 58, 361-382. Martell, A. E. & Smith, R. M. (1977). Critical Stability Constants, Vol. 3, other Organic Ligands Plenum. Martell, A. E. & Smith, R. M. (1982). Critical Stability Constants, Vol. 5, First Supplement Plenum. 5. Martin, J. M., Hogdahl, O. & Philippot, J. C. (1976). Rare earth element supply to the oceans. Journal of Geophysical Research, 81, 3119-124. Mayanovic, R. A., Anderson, A. J., Basset, W. A. & Chou, I. M. (2008). REE-chloride complexing in hydrothermal fluids with new data on lanthanum chloroaqua complexes. Geochimica and Cosmochimica Acta, 72. Mayer, L. M. (1982). Aggregation of colloidal iron during estuarine mixing: kinetics, mechanism and seasonability. Geochimica and Cosmochimica Acta, 46, 2527-2535. Mc Arthur, J. M., Howarth, R. J. & Bailey, T. R. (2001). Strontium Isotope Stratigraphy: LOWESS Version 3: Best Fit to the Marine Sr‐Isotope Curve for 0–509 Ma and Accompanying Look‐ up Table for Deriving Numerical Age. The Journal of Geology, 109, 155-170. Mc Murtry, G. M., De Carlo, E. H. & Kim, K. H. (1991). Accumulation rates, chemical partitioning, and Q-mode factor analysis of metalliferous sediments from the North Fiji Basin. Marine Geology, 98, 271-295. Metz, S., Trefry, J. H. & Nelsen, T. A. (1988). History and geochemistry of a metalliferous sediment core from the Mid-Atlantic Ridge at 26°N. Geochimica and Cosmochimica Acta, 52, 2369- 2378. Michard, A. (1989). Rare earth element systematics in hydrothermal fluids. Geochimica and Cosmochimica Acta, 53, 745-750. Migdisov, A. A. & William-Jones, A. E. (2008). Speciation of REE(III) in chloride-, fluoride-, and sulphate-bearing solutions: solubility and spectroscopic studies. Geochimica and Cosmochimica Acta, 72. Miller, D. J., Vanko, D. A. & Paulick, H. (2006). Data report: petrology and geochemistry of fresh, recent dacite lavas at Pual Ridge, Papua New Guinea, from an active, felsic-hosted seafloor hydrothermal system. In: Barriga, F. J. a. S., Binns, R. A., Miller, D. J. & Herzig, P. M. (eds.) Proceedings of the Ocean Drilling Program, Scientific results. Ocean Drilling Program. Mills, R. A. & Elderfield, H. (1995). Hydrothermal activity and the geochemistry of metalliferous sediment. Seafloor Hydrothermal Systems: Physical, Chemical, Biological, and Geological Interactions, 392-407. Mills, R. A., Elderfield, H. & Thomson, J. (1993). A Dual Origin or the Hydrothermal Component in a Metalliferous sediment Core From the Mid-Atlantic Ridge. Journal of Geophysical Research, 98, 9671-9681. Mitra, A., Elderfield, H. & Greaves, M. J. (1994). Rare earth elements in submarine hydrothermal fluids and plumes from the Mid-Atlantic Ridge. Marine Chemistry, 217-235. Miyashiro, A. (1973). The Troodos Ophiolitic Complex was probably formed in an Island Arc. Earth and Planetary Science Letters, 19, 218-224. Moldoveanu, G. A. & Papangelakis, V. G. (2013). Recovery of rare earth elements adsorbed on clay minerals: II. Leaching with ammonium sulfate. Hydrometallurgy, 131–132, 158-166. Moore, W. S. & Vogt, P. R. (1976). Hydrothermal Manganese Crusts From Two Sites Near the Galapagos Spreading Axis. Earth and Planetary Science Letters, 29, 349-356. Moores, E. M. & Vine, F. J. (1971). The Troodos Massif, Cyprus and other ophiolites as oceanic crust: evaluation and implications. Phil. Trans. R. Soc. Lond., A268, 433-466. Morse, D. E. & Stevens, C. (1979). The Mineral Industry of Cyprus. Minerals Yearbook, 283-288. Mottl, M. J. & Mcconachy, T. F. (1990). Chemical processes in buoyant hydrothermal plumes on the East Pacific Rise near 21”N. Geochimica and Cosmochimica Acta, 54, 1911-1927.

196 Bibliography

Muiños, S. B., Hein, J. R., Frank, M., Monteiro, J. H., Gaspar, L., Conrad, T., Pereira, G. & Abrantes, F. (2013). Deep-sea Fe-Mn crusts from the northeast Atlantic Ocean: Composition and resource considerations. Marine Georesources and Geotechnology, 31, 40-70. Mukherjee, P. K. & Gupta, P. K. (2008). Arbitrary scaling in ISOCON method of geochemical mass balance: An evaluation of the graphical approach. Geochemical Journal, 42, 247-253. Murphy, E., Mcmurtry, G. M., Kim, K. H. & Decarlo, E. H. (1991). Geochemistry and geochronoly of a hydrothermal ferromanganese deposit from the north Fiji Basin. Marine Geology, 98, 297-312. Naden, J., Herrington, R. J., Jowitt, S. M., Mcevoy, F. M., Williamson, J. P. & Mnhemius, A. J. (2006). New methodologies for volcanic-hosted copper sulphide mineralization on Cyprus: A GIS-prospectivity analysis-based approach. British Geological Survey Internal Report CR/06/129. 241p. Newman, A. C. D. 1987. Chemistry of Clays and Clay Minerals, Mineralogical Society Monograph, No 6, 480p. Ohta, A., Ishii, S., Sakakibara, M., Mizuno, A. & Kawabe, I. (1999). Systematic correlation of the Ce anomaly with the Co/(Ni+Cu) ratio and Y fractionation from Ho in distinct types of Pacific deep-sea nodules. Geochemical Journal, 33, 399-417. Olivarez, A. M. & Owen, R. M. (1989). REE/Fe variations in hydrothermal sediments: implications for the REE content of seawater. Geochimica and Cosmochimica Acta, 53, 757-762. Olivarez, A. M. & Owen, R. M. (1991). The europium anomaly of seawater: implications for fluvial versus hydrothermal REE inputs to the oceans. Chemical Geology, 92, 317-328. Onuma, N., Higuchi, H., Wakita, H. & Nagasawa, H. (1968). Trace element partition between two pyroxenes and the host lava. Earth and Planetary Science Letters, 5, 47-51. Pantazis, T. M. (1966). The geology and mineral resources of the Pharmakas-Kalavasos area. Mem. Geol. Surv. Dep. Cyprus, 8, 190p. Pearce, J. A. (1975). Basalt geochemistry used to investigate past tectonic environments on Cyprus. Tectonophysics, 25, 41-67. Pearce, J. A., Lippard, S. J. & Roberts, S. (1984). Characteristics and tectonic significance of supra- subduction zone ophiolites. Geological Society, London, Special Publications, 16, 77-94. Pearce, J. A. & Robinson, P. T. (2010). The Troodos ophiolitic complex probably formed in a subduction initiation, slab edge setting. Gondwana Research, 18, 60-81. Piper, D. Z. (1973). Origin of metalliferous sediments from the East Pacific Rise. Earth and Planetary Science Letters, 19, 75-82. Piper, D. Z. (1974). Rare earth elements in ferromanganese nodules and other marine phases. Geochimica et Cosmochimica Acta, 38, 1007-1022. Piper, D. Z., Veeh, H. H., Bertrand, W. G. & Chase, R. L. (1975). An iron-rich deposit from the Northeast Pacific. Earth and Planetary Science Letters, 26, 114-120. Poulton, S. W. & Canfield, D. E. (2006). Co-diagenesis of iron and phosphorus in hydrothermal sediments from the southern East Pacific Rise: Implications for the evaluation of paleoseawater phosphate concentrations. Geochimica et Cosmochimica Acta, 70, 5883- 5898. Prichard, H. M. & Maliotis, G. (1998). Gold mineralization associated with low-temperature, off- axis, fluid activity in the Troodos ophiolite, Cyprus. Journal of the Geological society, 155, 223-231. Qu, Y. & Lian, B. (2013). Bioleaching of rare earth and radioactive elements from red mud using Penicillium tricolor RM-10. Bioresource Technology, 136, 16-23. Ravizza, G., Sherrell, R. M., Field, M. P. & Pickett, E. A. (1999). Geochemistry of the Margi umbers, Cyprus, and the Os isotope composition of Cretaceous seawater. Geology, 27, 971-974. Rhulin, D. E. & Owen, R. M. (1986a). Factors influencing the rare earth element composition of hydrothermal precipitates, East Pacific rise. In: Leinen, M. & Rea, D. K. (eds.) Initial reports of the Deep Sea Drilling Project. Washington D.C. Rhulin, D. E. & Owen, R. M. (1986b). The rare earth element geochemistry of hydrothermal sediments from the east Pacific rise: Examination of a seawater scavenging mechanism. Geochimica and Cosmochimica Acta, 50, 393-400.

197 Bibliography

Richards, H. G. & Boyle, J. F. (1986). Origin, alteration and mineralization of inter-lava metalliferous sediments of the Troodos Ophiolite, Cyprus. In: Gallagher, M. J., Ixer, R. A., Neary, C. R. & Prichard, H. M. (eds.) Proceedings of the conference 'Metallogeny of basic and ultrabasic rocks'. Edimburgh, Scotland, from 9 to 12 April, 1985. Robertson, A. H. F. (1975a). Cyprus umbers: basalt-sediment relationships on a Mesozoic ocean ridge. Geological Society, 131, 511-531. Robertson, A. H. F. (1975b). Studies of the pre-miocene sedimentary cover of the Troodos Massif, Cyprus. Ph.D. Thesis, University of Leicester Robertson, A. H. F. (1976). Origins of ochres and umbers: Evidence from Skourioutissa, Troodos Massif, Cyprus. Institute of Mining Metallurgy Transactions, 85, 245-251. Robertson, A. H. F. (1977). The origin and diagenesis of cherts from Cyprus. Sedimentology, 24, 11- 30. Robertson, A. H. F. (2002). Overview of the genesis and emplacement of Mesozoic ophiolites in the Eastern Mediterranean Tethyan region. Lithos, 65, 1-67. Robertson, A. H. F. (2004). Development of concepts concerning the genesis and emplacement of Tethyan ophiolites in the Eastern Mediterranean and Oman regions. Earth Science Reviews, 66, 331-387. Robertson, A. H. F. & Boyle, J. F. (1983). Tectonic setting and origin of metalliferous sediments from the Mesozoic Tethys Ocean. In: Rona, P. A. (ed.) Hydrothermal Processes at Seafloor Spreading Centres. NATO Conference Series 1V, Marine Sciences. Robertson, A. H. F. & Degnan, P. (1998). Significance of mordern and ancient oceanic Mn-rich hydrothermal sediments, exemplified by Jurassic Mn-cherts from Southern Greece. Geological Society, London, Special Publications, 148, 217-240. Robertson, A. H. F. & Fleet, A. J. (1976). The origins of rare earths in metalliferous sediments of the Troodos massif, Cyprus. Earth and Planetary Science Letters, 28, 385-394. Robertson, A. H. F. & Fleet, A. J. (1986). Geochemistry and palaeo-oceanography of metalliferous and pelagic sediments from the Late Cretaceous Oman ophiolite. Marine and Petroleum Geology, 3, 315-337. Robertson, A. H. F. & Hudson, J. D. (1972). Cyprus umbers: chemical precipitates on a thetyan ocean ridge. Earth and Planetary Science Letters, 18, 93-101. Robertson, A. H. F. & Varnavas, S. P. (1993). The origin of hydrothermal metalliferous sediments associated with the Early Mesozoic Othris and Pindos ophiolites, mainland Greece. Sedimentary Geology, 83, 87-113. Robinson, P. T. & Malpas, J. (1990). The Troodos Ophiolite of Cyprus: new perspectives on its origin and emplacement. In: Malpas, J., Moores, E. M., Panayaiotou, A. & Xenophontos, C. (eds.) Ophiolites: Oceanic Crustal Analogues. Cyprus Geol. Surv. Dept., Nicosia, pp. 13–36. Robinson, P. T., Malpas, J. & Xenophontos, C. (2003). The Troodos Massif of Cyprus: Its role in the evolution of the ophiolite concept. Geological Society of America, Special Paper 373, 295- 307. Rode, H. (2015). Escondida, Enhancing our competitive advantage. BHP Billiton Investor Briefing, Escondida. BHP Billiton. Rona, P. A., Hannington, M. D. & Thompson, G. (1990). Evolution of hydrothermal mounds, TAG hydrothermal field, mid-Atlantic Ridge 26° N, 45° W. Geophys. Union, Trans.,, 71, 1650. Rona, P. A., Klinkhammer, G. P., Nelsen, T. A., Trefry, J. H. & Elderfield, H. (1986). Black smokers, massive sulphide and vent biota at the mid-Atlantic Ridge. Nature, 321, 33-37. Rozelle, P. L., Khadilkar, A. B., Pulati, N., Soundarrajan, N., Klima, M. S., Mosser, M. M., Miller, C. E. & Pisupati, S. V. (2016). A Study on removal of Rare Earth Elements from U.S. Coal Byproducts by Ion Exchange. Metallurgical and Materials Transactions, 12p. Ru'an, C., Jingming, X., Peijong, H. & Yongjun, Z. (1995). Recovering RE From Leach Liquor of Rare Earth Ore by Extraction. Transactions of NFsoc, 5, 5. Rudnicki, M. D. & Elderfield, H. (1993). A chemical model of the buoyant and neutrally buoyant plume above the TAG vent field, 26 degrees N, Mid-Atlantic Ridge. Geochimica et Cosmochimica Acta, 57, 2939-2957.

198 Bibliography

Ruttenberg, K. C. & Berner, R. A. (1993). Authigenic apatite formation and burial in sediments from non-upwelling, continental margin environments. Geochimica and Cosmochimica Acta, 57, 991-1007. Salman, H. M., El Hussaini, O. M., Abd-El Fatah, N. A. & Mahmoud, M. S. (2014). Extraction of yttrium from ferruginous sandstone, Southwestern Sinai, Egypt. The Journal of Applied Sciences Research, 1, 173-182. Savelli, C., Marani, M. & Gamberi, F. (1999). Geochemistry of metalliferous, hydrothermal deposits in the Aeolian arc (Tyrrhenian Sea). Journal of Volcanology and Geothermal Research, 88, 305-323. Schaller, T., Morford, J., Emerson, S. R. & Feely, R. A. (2000). Oxyanions in metalliferous sediments: tracers for paleoseawater metal concentrations? Geochimica et Cosmochimica Acta, 64, 2243-2254. Schijf, J. & Byrne, R. H. (2001). Stability constant for mono- and dioxalato-complexes of Y and REE, potentially important species in groundwaters and surface waters. Geochimica and Cosmochimica Acta, 65, 1037-1046. Schijf, J., Christenson, E. A. & Byrne, R. H. (2015). YREE scavenging in seawater: A new look at an old model. Marine Chemistry, 177, Part 3, 460-471. Schmidt, K., Garbe-Schönberg, C. D., Bau, M. & Koschinsky, A. (2010). Rare earth element distribution in >400°C hot hydrothermal fluids from 5°S, MAR: The role of anhydrite in controlling highly variable distribution patterns. Geochimica and Cosmochimica Acta, 74, 4058-4077. Schultz, H. D. (2006). Quantification of early diagenesis: Dissolved constituents in pore water and signals in the solid phase. In: Schultz, H. D. & Zabel, M. (eds.) Marine geochemistry. 2nd ed. . Heidelberg: Springer; 2006. p. 75–77. Severmann, S., Mills, R. A., Palmer, M. R. & Fallick, A. E. (2004). The origin of clay minerals in active and relict hydrothermal deposits. Geochimica et Cosmochimica Acta, 68, 73-88. Sherrell, R. M., Field, M. P. & Ravizza, G. (1999). Uptake and fractionation of rare earth elements on hydrothermal plume particles at 9°45′N, East Pacific Rise. Geochimica et Cosmochimica Acta, 63, 1709-1722. Sholkovitz, E. R. & Elderfield, H. (1988). The cycling of dissolved rare-earth elements in Chesapeake Bay. Global Biogeochemical Cycles. Sholkovitz, E. R., Elderfield, H. & Upstill-Goddard, R. C. (1986). Dissolved rare earth elements in Chesapeake Bay: riverine input and anoxia. EOS Trans. Am. Geophys. Union, 67, 1065- 1066. Sholkovitz, E. R., Landing, W. M. & Lewis, B. L. (1994). Ocean particle chemistry: The fractionation of rare earth elements between suspended particles and seawater. Geochimica and Cosmochimica Acta, 58, 1567-1579. Sinding-Larsen, R. & Wellmer, F. W. 2012. Non Renewable Ressource Issues: Geoscientific and Societal Challenges, Springer. Sinton, J. M., Ford, L. L., Chappell, B. & Mc Culloch, M. T. (2003). Magma genesis and mantle heterogeneity in the Manus Back-Arc Basin, Papua New Guinea. Journal of Petrology, 44, 159-195. Smith, P. A. & Cronan, D. S. (1983). The geochemistry of metalliferous sediments and waters associated with shallow submarine hydrothermal activity (Santorini, Aegean sea). Chemical Geology, 39, 241-262. Smith, R. M. & Martell, A. E. (1989). Critical Stability constants, Vol 6. Second Supplement Plenum. Sun, Z., Zhou, H., Yang, Q., Sun, Z., Bao, S. & Yao, H. (2011). Hydrothermal Fe–Si–Mn oxide deposits from the central and south Valu Fa Ridge, Lau Basin. Applied Geochemistry, 26, 1192-1204. Sverjensky, D. A. (1984). Europium redox equilibria in aqueous solution. Earth and Planetary Science Letters, 67, 70-78. Takeno, N. (2005). Atlas of Eh-pH diagrams, Intercomparison of thermodynamic databases. Geological Survey of Japan Open File Report No. 419, 287p.

199 Bibliography

Taylor, R. N. & Nesbitt, R. W. (1988). Light rare-earth enrichment of supra subduction-zone mantle: Evidence from the Troodos ophiolite, Cyprus. Geology, 16, 448-451. Taylor, S. R. & Mclennan, S. M. 1985. The continental Crust: Its composition and Evolution, Blackwell Scientific, 312 p, Boston, Mass. Thompson, G., Humphris, S. E., Schroeder, B., Sulanowska, M. & Rona, P. A. (1988). Active vents and massive sulfides at 26°N (TAG) and 23°N (Snakepit) on the mid-Atlantic Ridge. Canadian Mineralogist, 26, 697-712. Tian, J., Chi, R.-A. & Yin, J.-Q. (2010). Leaching process of rare earths from weathered crust elution- deposited rare earth ore. Transactions of Nonferrous Metals Society of China, 20, 892- 896. Toyoda, K., Nakamura, Y. & Masuda, A. (1990). Rare earth elements of Pacific pelagic sediments. Geochimica and Cosmochimica Acta, 54, 1093-1103. Tsakanika, L. V., Ochsenkühn-Petropoulou, M. T. & Mendrinos, L. N. (2004). Investigation of the separation of scandium and rare earth elements from red mud by use of reversed-phase HPLC. Analytical and Bioanalytical Chemistry, 379, 796-802. U.S.G.S (2013). Mineral Commodity Summaries. Rare Earth Reston, Virginia, U.S. Geological Survey. U.S.G.S (2014). Mineral Commodity Summaries. Rare Earth. Reston, Virginia, U.S. Geological Survey. U.S.G.S (2015). Mineral Commodity Summaries. Rare Earth. Reston, Virginia, U.S. Geological Survey. U.S.G.S (2016). Mineral Commodity Summaries. Rare Earth. Reston, Virginia, U.S. Geological Survey. Ujaczki, E., Zimmermann, Y. S., Feigl, V. & Lenz, M. (2015). Recovery of rare earth elements from Hungarian red mud with combined acid leaching and liquid-liquid extraction. Bauxite residue valorisation and best practices. Leuven 5-7 October 2015. Upstill-Goddard, R. C., Findlay, J. S., Elderfield, H. & Sholkovitz, E. R. (1986). Rare earth elements (REE) in rivers and estuaries. EOS Trans. Am. Geophys. Union, 67. Vahidi, E., Navarro, J. & Zhao, F. (2016). An initial life cycle assessment of rare earth oxides production from ion-adsorption clays. Resources, Conservation and Recycling, 113, 1-11. Van Der Zee, C., Slomp, C. P. & Van Raaphorst, W. (2002). Authigenic P formation and reactive P burial in sediments of the Nazaré canyon on the Iberian Margin (NE Atlantic). Marine Geology, 185, 379-392. Varentsov, I. M., Drits, V. A. & Gorshkov, A. I. (1991). Rare earth element indicators of Mn-Fe oxyhydroxide crust formation on Krylov Seamount, Eastern Atlantic. Marine Geology, 96, 71-84. Varga, R. J. & Moores, E. M. (1985). Spreading structure of the Troodos ophiolite, Cyprus. Geology, 13, 846-850. Varshal, G. N., Senyavin, M. M. & Yartseva, R. D. (1975). Forms of calcium and REE in river waters. In: Tugarinov, A. E. (ed.) Recent contributions to Geochemistry and Analytical Chemistry. Wiley. Von Damm, K. L., Bray, A., Buttermore, L. G. & Oostings, S. (1998). The geochemical controls on vent fluids from the Lucky Strike vent field, Mid-Atlantic Ridge. Earth and Planetary Science Letters, 160, 521-536. Wegorzewski, A. V. & Kuhn, T. (2014). The influence of suboxic diagenesis on the formation of manganese nodules in the Clarion Clipperton nodule belt of the Pacific Ocean. Marine Geology. Weng, Z., Jowitt, S. M., Mudd, G. M. & Haque, N. (2015). A Detailed Assessment of Global Rare Earth Element Resources: Opportunities and Challenges. Economic Geology, 110, 1925- 1952. Wilson, R. a. M. (1959). The geology of the Xeros-Troodos area, Cyprus. Cyprus Geological Survey Departement, Mem. 1, 73 pp.

200 Bibliography

Xiong, Y. (2011). Organic species of lanthanum in natural environments: Implications to mobility of rare earth elements in low temperature environments. Applied Geochemistry, 26, 1130- 1137. Yang, K.-F., Fan, H.-R., Santosh, M., Hu, F.-F. & Wang, K.-Y. (2011). Mesoproterozoic carbonatitic magmatism in the Bayan Obo deposit, Inner Mongolia, North China: Constraints for the mechanism of super accumulation of rare earth elements. Ore Geology Review, 40, 122- 131. Yang, X. J., Lin, A., Li, X. L., Wu, Y., Zhou, W. & Chen, Z. (2013). China's ion-adsorption rare earth resources, mining consequences and preservation. Environmental Development, 8, 131- 136. Yuan, Z., Bai, G., Wu, C., Zhang, Z. & Ye, X. (1992). Geological features and genesis of the Bayan Obo REE ore deposit, Inner Mongolia, China. Applied Geochemistry, 7, 429-442. Zhang, J. & Nozaki, Y. (1996). Rare earth elements and yttrium in seawater: ICP-MS determinations in the East Caroline, Coral Sea, and South Fiji of the western South Pacific Ocean. Geochimica and Cosmochimica Acta, 60, 4631-4644. Zinin, D. S. & Bushuev, N. N. (2014). Production and physicochemical study of oxalate and oxide REE concentrates. Russian Journal of Applied Chemistry, 87, 1611-1618.

201