Experimental Constraints on Lithium Exchange between Clinopyroxene, Olivine and Aqueous Fluid at High
Pressures and Temperatures
by
Natalie Caciagli Warman
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Department of Geology University of Toronto
© Copyright by Natalie Caciagli Warman 2010
ii
Experimental Constraints of Lithium Exchange between
Clinopyroxene, Olivine and Aqueous Fluid at High Pressures and
Temperatures
Natalie Caciagli Warman
Doctor of Philosophy
Department of Geology University of Toronto
2009
Abstract
Clinopyroxene, olivine, plagioclase and hydrous fluid lithium partition coefficients have been
measured between 800-1100oC at 1 GPa. Clinopyroxene-fluid partitioning is a function of
temperature (ln DLicpx/fluid = -7.3 (+0.5) + 7.0 (+0.7) * 1000/T) and appears to increase with
increasing pyroxene Al2O3 content. Olivine-fluid partitioning of lithium is a function of
temperature (ln DLiol/fluid = -6.0 (+2.0) + 6.5 (+2.0) * 1000/T) and appears to be sensitive to olivine
Mg/Fe content. Anorthite-fluid lithium partitioning is a function of feldspar composition, similar
to the partitioning of other cations in the feldspar-fluid system. Isotopic fractionation between
clinopyroxene and fluid, Licpx-fluid, has been measured between 900-1100oC and ranges from -
0.3 to -3.4 ‰ (±1.4 ‰).
Lithium diffusion has been measured in clinopyroxene at 800-1000oC and in olivine at 1000oC.
The lithium diffusion coefficient is independent of the diffusion gradient as values are the same
if the flux of lithium is into or out of the crystal and ranges from -15.19 ± 2.86 m2/s at 800oC to -
11.97 ± 0.86 m2/s at 1000oC. Lithium diffusion in olivine was found to be two orders of
magnitude slower than for clinopyroxene at similar conditions. Closure temperatures calculated
iii
for lithium diffusion in clinopyroxene range from ~400 to ~600oC. These results demonstrate
that lithium equilibration between fluids and minerals is instantaneous, on a geological
timescales.
The confirmation of instantaneous equilibration, combined with min-fluid partition coefficients
and values for Licpx-fluid, permits quantitative modeling of the evolution of lithium concentration
and isotopic composition in slab-derived fluids during transport to the arc melt source. Our
results indicate that fluids migrating by porous flow will rapidly exchange lithium with the
mantle, effectively buffering the fluid composition close to ambient mantle values, and rapidly
attenuating the slab lithium signature. Fluid transport mechanisms involving fracture flow are
required to maintain a slab-like lithium signature (both elemental and isotopic) from the slab to
the melt source of island arc basalts.
This study demonstrates that mineral-fluid equilibration is rapid, and as a result the lithium
content of minerals can only reliably represent chemical exchange in the very latest stages of the
sample’s history.
iv
Acknowledgments
This thesis dissertation marks the conclusion of work that began in 2001. There were many who
helped, supported, and cheered me on my way and I fear that to be able to acknowledge everyone
who assisted me would be an impossible task. I am certain that once this work has been
submitted I will realize that I have left out several important people.
First, I must thank Dr. James Brenan, my mentor and supervisor, who never gave up on me
despite everything. I am indebted to Dr. Lesley Rose Weston (my lab partner in crime) and Dr.
Boris Foursenko for all their time and invaluable mechanical, technical, and moral support. I am
grateful to my collaborators at Lawrence Livermore National Laboratory, Dr. Doug Phinney, Dr.
Ian Hutcheon and Dr. Rick Ryerson for access to and assistance with the SIMS. Thanks to Dr.
Bill McDonough and his lab at University of Maryland for the isotopic analyses. I also wish to
thank Dr. Grey Bebout at Lehigh University, for his patience and understanding when, to quote
Edison, “I found 10,000 ways that won’t work.” before I found one way that did work. I would
also like to thank Dr. Paul Tomascak, who pointed me in the direction that would bear the most
fruit and Dr. Jon Davidson for his unwavering encouragement for always having time for a ‘little
chat’.
Mainly I am indebted to my husband Tim, who never once let me give up on myself.
This work must be dedicated to my children, Olivia and Henry, whose existence shaped this
project more than any other thing.
v
Table of Contents
Acknowledgments.......................................................................................................................... iv
Table of Contents............................................................................................................................ v
List of Tables ................................................................................................................................. ix
List of Figures ................................................................................................................................. x
List of Appendices ....................................................................................................................... xiii
1 Introduction ................................................................................................................................ 1
1.1 Chemical Properties of Lithium.......................................................................................... 1
1.2 Sources and Concentrations of Lithium in the Earth .......................................................... 3
1.2.1 Subducted Materials (AOC and Seafloor Sediments) ............................................ 3
1.2.2 Eclogites.................................................................................................................. 4
1.2.3 Upper Mantle .......................................................................................................... 5
1.2.4 Convergent Margin Magmas .................................................................................. 6
1.3 The scale of lithium heterogenities ..................................................................................... 7
1.4 Previous experimental work ............................................................................................... 8
1.5 Focus of Thesis and Distribution of Work.......................................................................... 9
2 Lithium Partitioning and Isotopic Fractionation ...................................................................... 14
2.1 Introduction....................................................................................................................... 14
2.2 Methods............................................................................................................................. 16
2.3 Analytical Techniques ...................................................................................................... 18
2.3.1 Major Element Analyses....................................................................................... 18
2.3.2 Lithium Analyses .................................................................................................. 19
2.3.2.1 MC-ICPMS............................................................................................. 19
2.3.2.2 SIMS....................................................................................................... 19
vi
2.3.2.3 LA-ICPMS ............................................................................................. 20
2.4 Results............................................................................................................................... 21
2.4.1 Major Element Chemistry..................................................................................... 21
2.4.2 Mineral – Fluid Lithium Partitioning.................................................................... 21
2.4.2.1 Clinopyroxene ........................................................................................ 23
2.4.2.2 Olivine .................................................................................................... 24
2.4.2.3 Plagioclase .............................................................................................. 25
2.4.3 Olivine-Clinopyroxene Pair Experiments............................................................. 26
2.4.3.1 Experiments at variable fO2.................................................................... 27
2.4.3.2 Experiments with added REE................................................................. 27
2.4.4 Lithium Isotope Fractionation............................................................................... 28
2.4.4.1 Clinopyroxene- fluid .............................................................................. 28
2.4.4.2 Olivine-Clinopyroxene Lithium Isotope Fractionation .......................... 29
2.5 Discussion ......................................................................................................................... 29
2.5.1 Controls on Partitioning........................................................................................ 29
2.5.1.1 Clinopyroxene ........................................................................................ 29
2.5.1.2 Olivine .................................................................................................... 31
2.5.1.3 Plagioclase .............................................................................................. 31
2.5.1.4 Intermineral Partitioning......................................................................... 32
2.5.2 Controls on Isotopic Fractionation........................................................................ 33
2.5.3 Lithium Incorporation into the Mantle ................................................................. 34
2.5.4 The Mantle Wedge as a Chromatograph .............................................................. 35
2.5.5 Isotopic Evolution of Lithium-Bearing Fluids in the Mantle ............................... 39
2.5.5.1 Percolation and Rayleigh Distillation..................................................... 39
2.5.5.2 Generation of 6Li-rich fluids................................................................... 41
2.5.5.3 Generation of 6Li-rich zones in the mantle............................................. 42
vii
2.6 Conclusions....................................................................................................................... 44
3 Lithium Diffusion..................................................................................................................... 66
3.1 Introduction....................................................................................................................... 66
3.2 Experimental Methods ...................................................................................................... 68
3.3 Analytical Techniques ...................................................................................................... 69
3.3.1 Major Element Analyses....................................................................................... 69
3.3.2 Lithium Analyses .................................................................................................. 70
3.3.2.1 LA-ICPMS ............................................................................................. 70
3.3.2.2 Secondary Ion Mass Spectrometry (SIMS)............................................ 70
3.3.3 Data Reduction...................................................................................................... 71
3.4 Results............................................................................................................................... 72
3.4.1 Diffusion in Clinopyroxene .................................................................................. 72
3.4.2 fO2 Series Experiments ......................................................................................... 72
3.4.3 Diffusion in Olivine .............................................................................................. 73
3.4.4 Kinetic Fractionation of 7Li/6Li ............................................................................ 73
3.5 Discussion ......................................................................................................................... 74
3.5.1 Effect of fO2 on lithium diffusion in clinopyroxene ............................................. 74
3.5.2 Comparison with other lithium diffusion studies.................................................. 76
3.5.3 Comparison with diffusion of other cations in clinopyroxene.............................. 76
3.5.4 Geological Implications ........................................................................................ 77
3.5.4.1 Preservation of Lithium Signatures ........................................................ 77
3.5.4.2 Closure Temperature .............................................................................. 79
3.5.4.3 The Potential for Re-Equilibration of Lithium Composition ................. 79
3.5.4.4 Diffusion-Induced Isotopic Fractionation .............................................. 82
3.6 Conclusions....................................................................................................................... 84
4 Technique Development to Study Muscovite-Fluid Partitioning of Nitrogen....................... 104
viii
4.1 Introduction..................................................................................................................... 104
4.2 Theoretical Considerations ............................................................................................. 106
4.2.1 N-speciation and Isotopic Fractionation ............................................................. 106
4.2.2 Buffering pH ....................................................................................................... 108
4.3 Experimental Methodology ............................................................................................ 109
4.4 Analytical Methods......................................................................................................... 109
4.5 Results............................................................................................................................. 110
4.5.1 Nitrogen Partitioning .......................................................................................... 110
4.5.2 Nitrogen Isotopic Fractionation .......................................................................... 111
4.6 Discussion ....................................................................................................................... 112
4.6.1 Utility of NH4Cl as Nitrogen Source .................................................................. 112
4.6.2 Analytical Considerations................................................................................... 112
4.6.3 Experimental Considerations .............................................................................. 113
4.6.4 Isotopic Fractionation Experiments and Atmospheric Contamination............... 114
4.7 Suggestions for Future Work .......................................................................................... 114
5 Summary of Results and Conclusions.................................................................................... 125
References................................................................................................................................... 128
Appendix 1.................................................................................................................................. 140
6 Summary of Boron Work....................................................................................................... 140
ix
List of Tables
Table 2.1 Composition of Starting Material ................................................................................. 46
Table 2.2 Experimental Conditions .............................................................................................. 47
Table 2.3 Standards and Reference Material ................................................................................ 48
Table 2.4 Run Product Composition............................................................................................. 49
Table 2.5 Run Product Lithium Concentration............................................................................. 50
Table 2.6 Isotopic Composition of Starting Materials and Run Products .................................... 51
Table 3.1 Composition of Starting Material ................................................................................. 85
Table 3.2 Measurements of Standards .......................................................................................... 86
Table 3.3 Summary of Experiments ............................................................................................. 87
Table 4.1 Experiments and Results............................................................................................. 116
Table 4.2 Percentage of Nitrogen Contribution from Air........................................................... 116
x
List of Figures
Figure 1.1 Sources and Concentration of Lithium in the Earth .................................................... 11
Figure 1.2 Li/Y ratio and 7Li in Arc Lavas................................................................................. 12
Figure 1.3 Lithium Diffusion Coefficients ................................................................................... 13
Figure 2.1 Internal Reference Materials ....................................................................................... 52
Figure 2.2 Standards and Reference Material............................................................................... 53
Figure 2.3 Photomicrographs of Starting Material and Run Products.......................................... 54
Figure 2.4 Time Resolved Spectra................................................................................................ 55
Figure 2.5 lnDLi cpx/fluid vs 1000/T ........................................................................................... 56
Figure 2.6 lnDLi ol/fluid vs 1000/T.............................................................................................. 57
Figure 2.7 Anorthite/Fluid Lithium Partitioning .......................................................................... 58
Figure 2.8 Olivine/Clinopyroxene Lithium Partitioning .............................................................. 59
Figure 2.9 Lithium Partitioning From Mantle Xenoliths and Experimental Studies.................... 60
Figure 2.10 Mineral/Fluid Isotopic Fractionation of Lithium ...................................................... 61
Figure 2.11 Time for Li and B Transport to Top of Column........................................................ 62
Figure 2.12 Evolution of the Slab Derived Fluid by due to Rayleigh Distillation ....................... 63
Figure 2.13 Lithium Coordination and P-T Paths......................................................................... 64
Figure 2.14 Evolution of 7Li of Mantle Wedge due to Hydrofractures ...................................... 65
Figure 3.1 Li Elemental and Isotopic Gradients in San Carlos Opx............................................. 88
Figure 3.2 Effect of fO2 Anneal .................................................................................................... 89
xi
Figure 3.3 Zero time Experiment.................................................................................................. 90
Figure 3.4 Results for Experiment Kcpx-12 ................................................................................. 91
Figure 3.5 X-Ray Maps of Run Product ....................................................................................... 92
Figure 3.6 Time Series .................................................................................................................. 93
Figure 3.7 Measured Lithium Diffusion Coefficients................................................................... 94
Figure 3.8 fO2 Experiment Series ................................................................................................. 95
Figure 3.9 Lithium Diffusion Profile in San Carlos Olivine ........................................................ 96
Figure 3.10 Lithium Diffusion and Isotopic Fractionation in Kcpx-2.......................................... 97
Figure 3.11 Comparison of Lithium Diffusion Coefficients ........................................................ 98
Figure 3.12 Comparison of Diffusivities Measured in Pyroxene ................................................. 99
Figure 3.13 Retention of Lithium Composition.......................................................................... 100
Figure 3.14 Comparison of Closure Temperature of Li and Sr in Clinopyroxene ..................... 101
Figure 3.15 Lithium Isotopic Compositions of Kilauea Iki Lava Lake Rocks........................... 102
Figure 3.16 Li Isotopic Gradient in San Carlos Opx and Modeled Profile ................................ 103
Figure 4.1 Summary of N Concentration and Isotopic Composition ......................................... 117
Figure 4.2 Calculated N2-, and NH3-NH4+ Fractionation ........................................................... 118
Figure 4.3 Relationship of fH2, fN2, and fNH3............................................................................ 119
Figure 4.4 Scanning Electron Micrograph of Muscovite Texture .............................................. 120
Figure 4.5 Nitrogen contents of run products ............................................................................. 121
Figure 4.6 Nitrogen isotopic compositions of run products ....................................................... 122
xii
Figure 4.7 Isotopic shifts of run products ................................................................................... 123
Figure 4.8 Puncturing Device ..................................................................................................... 124
xiii
List of Appendices
6.1 11B notation ................................................................................................................... 140
6.2 Evidence of Boron Mobility from Arc Lavas ................................................................. 140
6.3 Evidence of Boron Mobility from Eclogites................................................................... 142
6.4 Summary of Experimental Methodology........................................................................ 143
6.5 Details of Boron Study.................................................................................................... 143
6.6 Boron Analyses............................................................................................................... 146
6.7 References for Boron Study............................................................................................ 148
1
1 Introduction
According to Davidson (1996), three fundamental questions remain unanswered in the study of
island arc magmagenesis:
“1. What is the (presubduction) composition of the mantle wedge source of arc magmas? 2. To what extent does it melt, and by what process? 3. What is the composition and amount of slab-derived component added to the wedge?”
Despite the significant strides that have been made in both our knowledge of earth processes and
our technical ability to analyze earth materials with greater precision and accuracy, these
questions remain unsatisfactorily answered today.
Low abundance, or trace elements, can provide essential information to address these issues. For
example, both convergent margin basalts and mid ocean ridge basalts (MORB) have similar
major element compositions; however, convergent margin basalts are differentiated by high
LILE/REE; (large ion lithophile element - Rb, K, Cs, Ba, Sr to rare earth element - actinides; La
through Lu) and high LILE/HFSE (high field strength element – Nb, Ta, Zr, Hf, Ti) signature
(Davidson, 1996). This pattern is interpreted to mean that arc magmas are products of the
overlying mantle wedge melt plus a LILE-rich fluid or melt originating from the subducted slab.
Boron, lithium and nitrogen have often been employed to identify the composition and amount
of the slab component in island arc magmas. These elements are relatively fluid mobile and
somewhat incompatible in mantle minerals and are easily released from the slab and
concentrated into the magmas at the arc front (Ishikawa and Tera, 1999; Leeman, 1996; Ishikawa
and Nakamura, 1994; Ishikawa et al., 2001; Moriguti and Nakamura, 1998).
1.1 Chemical Properties of Lithium
Lithium belongs to the Group 1 elements of the periodic table. Like the other alkali metals it is
characterized by low ionization energy and low electronegativity, and commonly forms
hydroxides, nitrides, carbonates and chlorides. When octahedrally coordinated it has an effective
ionic radius of 0.59 Å which is comparable to octahedrally coordinated Mg2+ (0.72 Å) and Fe2+
2
(0.78 Å), which allows it to substitute for these elements in olivine, pyroxenes, amphiboles and
clays (Brenan et al., 1998; Wenger and Armbruster, 1991). The oxygen co-ordination of lithium
can vary from 3 to 8; although lithium has a preference for tetrahedral co-ordination in melts and
fluids (Cormier et al., 1998; Majérus et al., 2003), it can be accommodated by octahedral co-
ordination, as is the case in many silicate minerals (Wenger and Armbruster, 1991). Lithium has
a single valence electron with a very low ionization potential, which makes it easily solvated.
Lithium has two stable isotopes, 6Li and 7Li, with a relative mass difference of ~16 % and
abundances of ~7.52 % and ~92.48 %, respectively. Enrichments in lithium isotopes are
described as either:
(1) 10001LiLi/
LiLi/Li
std76
smp76
6
or 10001
LiLi/
LiLi/Li
std67
smp67
7
where ‘smp’ refers to the sample and ‘std’ refers to the standard, typically the NBS lithium
carbonate L-SVEC (SRM #8545). The δ7Li notation is recommended by the International Union
for Pure and Applied Chemistry (Coplen et al., 1996) and will be used here.
Figure 1.1 shows the lithium abundance and isotopic composition of various geochemical
reservoirs. As with other isotopes, lithium isotopic fractionation between minerals and fluids
depends on the difference in the zero point potential energy (ZPE) between the phases of interest.
Heavier isotopes have lower vibrational frequencies, and therefore a lower ZPE than lighter
isotopes (Chacko et al., 2001). The molecule or phase that will undergo the greatest reduction in
ZPE with the substitution of the heavy isotope will become enriched in the heavier isotope
(Chacko et al., 2001). Ab initio calculations have demonstrated that during mineral-solution
reactions 6Li should be preferentially incorporated into octahedrally coordinated sites in the solid
(Yamaji et al., 2001). This appears to be the case in the formation of secondary minerals
produced during alteration of crustal rocks (Huh et al., 2001; Pistiner and Henderson, 2003;
Seyfried et al., 1998). During the formation of clays, 6Li is concentrated in the solid phase while
the resulting fluid becomes enriched in 7Li (Huh et al., 2001; Pistiner and Henderson, 2003).
Experimental measurements of lithium isotopic fractionation between spodumene and fluids also
confirm this behavior (Wunder et al., 2006). Interestingly, experiments measuring lithium
isotopic fraction between staurolite and fluids found that 6Li was preferentially enriched in the
3
fluids, and 7Li was enriched in the solids (Wunder et al., 2007). Considering that lithium is in
tetrahedral coordination in staurolite, this result also appears to confirm the ab initio calculations.
1.2 Sources and Concentrations of Lithium in the Earth
The various components of the convergent margin system have significant differences in lithium
concentrations and lithium isotopic compositions (see Figure 1.1). Inputs such as continental
crust (as pelagic clays) and altered oceanic crust tend to be enriched in lithium with respect to the
mantle, but the isotopic composition can vary considerably depending on the type and degree of
alteration or weathering. Mantle inputs tend to be more uniform in terms of lithium concentration
and isotopic composition; however, the MORB-source mantle can potentially contain both
elemental and isotopic heterogeneities. Very low δ7Li (-11 ‰ to +5 ‰) values are found in
samples of metamorphosed oceanic crust, possibly reflecting low temperature dehydration of the
slab during subduction (Zack et al., 2003). Conversely some pyroxenites from the Zabargad
peridotite, which is considered to be a fragment of exhumed mantle wedge, have high δ7Li
values (+8.4 ‰ to +11.8 ‰) suggesting that 7Li-rich fluids or melts derived from subducted
altered oceanic crust (AOC) may also be transferred to the mantle (Brooker et al., 2004). The
variability exhibited by the components of the arc magmagenesis system is not always reflected
in the output. Some island arc magmas display higher ratios of fluid mobile elements, such as
lithium, to relatively immobile elements, such as yttrium in the front-arc regions, which decrease
towards the back-arc. However, few arc lavas have δ7Li significantly greater than MORB, and
correlations between δ7Li and fluid enrichment are not always clear or consistent (Chan et al.,
2002; Tomascak et al., 2002; Tomascak, 2004; Leeman et al., 2004).
1.2.1 Subducted Materials (AOC and Seafloor Sediments)
Lithium in seafloor sediments ranges from 5 to 80 ppm with δ7Li ranging from -5 ‰ to +20 ‰
(Marschall et al., 2007 and references therein). In oceanic crust lithium can range from 5 to 6
ppm with a δ7Li of +1.5 ‰ to +5.6 ‰ in fresh mid-ocean ridge basalts (Moriguti and Nakamura,
1998; Tomascak et al., 2008; Chan et al., 1992) whereas altered oceanic crust may have >75 ppm
lithium with δ7Li of +14.2 ‰ (Moriguti and Nakamura, 1998; Tomascak et al., 2008; Chan et al.,
1992). Low temperature alteration of basalt results in concentration of lithium, and preferentially
4
6Li, into secondary minerals and enrichment of 7Li in seawater with a fractionation of up to +19
‰ with respect to the residual solid (Chan et al., 1992; Chan et al., 2002; Seyfried et al., 1998;
James et al., 2003). At temperatures greater than 350 oC, lithium is mobilized by saline fluids
and the extent of the isotopic fractionation decreases (Chan et al., 1992; Chan et al., 2002;
Seyfried et al., 1998; James et al., 2003).
The extent of solid-fluid fractionation has been inferred by various authors by measuring lithium
in sediment-derived pore fluids and serpentine diapirs (Chan et al., 1992; Chan et al., 2002; Chan
and Kastner, 2000; Benton et al., 2004). Pore fluids from the Costa Rican trench show a ~11 ‰
enrichment of 7Li with respect to the down-going sediments (Chan and Kastner, 2000). A larger
range of isotopic compositions, 7Li of -0.5 ‰ to +10 ‰ has been measured in so called
serpentinite diapers, which are super-hydrated mantle wedge extruded in the fore arc of the
Mariana trench (Benton et al., 2001; Benton et al., 2004). The variability in the lithium isotopic
composition of vent fluids (δ7Li ranging from +5 ‰ to +43.8 ‰) likely reflects differences in
temperature, reaction paths and fluid - rock ratios as well as source rock composition (Zhang et
al., 1998; Chan and Kastner, 2000; Foustoukos et al., 2004).
1.2.2 Eclogites
Alpine eclogites are thought to be exhumed remnants of subducted oceanic crust. The eclogites
at Trescolmen, Switzerland investigated by Zack et al. (2003) displayed extremely light 7Li
values ranging from -11 ‰ to +5 ‰. This study speculated that during subduction the slab was
progressively depleted in 7Li, via Rayleigh distillation during dehydration, and that the resulting 6Li enriched material was recycled into the mantle. A more comprehensive study of eclogites,
blueschists and other high pressure metamorphic rocks from classic European (Swiss Alps,
Münchberg, Aldalen and Greek Islands) and Asian localities (Qaidam, Dabieshan and Tianshan)
by Marschall et al. (2007) found an even larger range of lithium concentrations and isotopic
compositions. Lithium concentrations ranged from ~1 to ~50 ppm and 7Li values ranged from -
21.9 ‰ to > +6 ‰ (Marschall et al., 2007). More significantly, no correlation was found between
lithium content and 7Li within any single locality or within the full population. In fact, many of
the lightest samples had > 30 ppm lithium, contrary to what would be expected from a simple
Rayleigh-type distillation of the subducted slab. Marschall et al. (2007) concluded that many
exhumed eclogites have lithium compositions (both elemental and isotopic) that have been
5
influenced by influx of retrograde fluids and kinetically induced isotopic fractionation during
exhumation, and therefore primary subduction-induced fractionation has been modified by more
recent processes.
1.2.3 Upper Mantle
From analyses of pristine peridotite xenoliths, the upper mantle is estimated to contain 1.5 ppm
lithium with an average δ7Li of +4 ‰ (Jagoutz et al., 1979; Tomascak, 2004; Jeffcoate et al.,
2007). However, the fact that OIB and MORB can sometimes display a range of values has led
to the suggestion that variable amounts of recycled crustal material are sometimes present in the
mantle sources of these lavas (Tomascak et al., 2008). A study of MORB lavas from different
ridge systems found δ7Li ranging from +1.6 ‰ to +5.6 ‰. This represents a 5 ‰ heterogeneity
in the samples with no consistent correlation of δ7Li with major and trace elements or radiogenic
isotopes (Tomascak et al., 2008). Analysis of lithium in glass inclusions from Hawaii showed
δ7Li varying from –10.2 ‰ to +8.4 ‰ (Kobayashi et al., 2004). Similarly, low δ7Li (-3.3 ‰ to
+1.2 ‰) from glass inclusions in Iblean (Sicilian) Plateau tholeiites are thought to reflect melting
of an isotopically light region in the mantle (Gurenko and Schmincke, 2002). In rare cases, δ7Li
correlations with abundances of other trace elements or isotopes can be found. For example,
samples from the East Pacific Rise (EPR) show a weak correlation of increasing δ7Li with
increasing Cl/K, which was interpreted to reflect mixing or assimilation of recycled crustal
material (Tomascak et al., 2008). Other EPR samples show a positive correlation between δ7Li
and 143Nd/144Nd, (Elliot, 2004), again reflecting a possible recycled component. That reservoirs
with variable δ7Li exist in the mantle is also suggested by studies of peridotite massifs and
ultramafic xenoliths. Nishio et al. (2004) report the lithium isotopic composition of
clinopyroxene from xenoliths from Japan, SE Australia and eastern Russia. The δ7Li values from
NE Japan and SE Australia were high (+4 to +7 ‰), whereas Russian and SW Japan samples
were significantly lower (-17 to -3 ‰). In some of the samples from eastern Russia, δ7Li could
be positively correlated with 143Nd/144Nd but those correlations did not apply to the other sample
populations or even to all the eastern Russia samples. Ionov and Seitz (2008) also reported
lithium concentrations and isotopic data from xenoliths from the Kamchatka arc and the Vitim
(Siberia) volcanic field (an intra-plate continental volcanic setting). They found a relatively small
range of lithium concentrations, ~1 to ~2 ppm, and lithium isotopic composition of -3.6 to +6 ‰.
6
These studies demonstrate significant variations in the isotopic compositions of lithium in the
upper mantle, but suggest that such variation is fairly localized. Reports of significant correlation
between δ7Li and other elemental or isotopic tracer elements are rare within any given sample
locality and no global correlation has been found within a given tectonic setting.
1.2.4 Convergent Margin Magmas
Magmatism at convergent margins is commonly believed to be due to hydrous fluids from the
subducting slab fluxing the mantle wedge and lowering the mantle solidus. As mentioned
previously, the resulting magmas are very often characterized by higher ratios of fluid mobile
elements to relatively insoluble elements, such as the high field strength elements (HFSE). In
order to distinguish between crystal/melt fractionation effects and fluid involvement, fluid
mobile elements (such as lithium and boron) are measured against relatively insoluble elements
with similar solid/melt partitioning (i.e. B/Be, Li/Yb or Li/Y). The Izu arc in Japan is the
locality showing the clearest indication of a Li-bearing, slab-derived component. This is shown
by the correlated decrease in both Li/Y and δ7Li from the arc front lavas to those erupted in the
back arc region (Figure 1.2a, data from Moriguti and Nakamura, 1998). This is suggestive of
continuing mobilization of lithium into the arc source region by fluids derived from dehydration
reactions in the down going slab (Leeman, 1996; Ishikawa and Tera, 1999; Ishikawa and
Nakamura, 1994; Ishikawa et al., 2001). However, the trend displayed at Izu appears to be the
exception and not the rule (Figure 1.2b, data from Tomascak et al., 2002). Variations in slab age
and angle of subduction, which would influence the thermal regime, and therefore the extent of
dehydration, have been cited to explain these differences (Moriguti et al., 2004). Yet, a
difference in the Li/Y and 7Li behavior between the Kurile arc and the Izu arc, or even between
the Izu arc and the Japan arc where age of the subducted slab is similar, still persists, despite the
similarity of subduction regime (Morguti et al., 2004; Tomascak et al., 2002). Another
suggestion is that slab-derived fluids are significantly modified during transport through the
mantle wedge to the melt source, and that the lithium signal is attenuated by interaction with
mantle minerals (Tomascak et al., 2002). The extent of modification of the slab-derived
component during transport through, and interaction with, the overlying mantle wedge is
unknown, but it remains unanswered as to why more arc magmas do not display the same trends
as clearly as the Izu arc.
7
1.3 The scale of lithium heterogenities
Recent in situ micron scale analyses of lithium isotopes in geological samples have yielded
unexpected results. High temperature equilibrium is expected to impose minimal differences in
the lithium isotopic composition of individual minerals; however, samples from some magmatic
environments have revealed significant isotopic heterogeneity at the grain-scale. For example,
Rudnick and Ionov (2007) reported highly variable δ7Li in clinopyroxene and olivine grains in
peridotite xenoliths from eastern Russia. δ7Li values ranged from -0.8 to -14.6 ‰ for
clinopyroxene and -1.7 to +11.9 ‰ for corresponding olivine, and olivine/clinopyroxene
distribution coefficients varied from 0.2 to 1.0, which is lower than previously estimated for
equilibrium partitioning. Analyses of olivine and clinopyroxene pairs from a xenolith from the
Vitim volcanic field found δ7Li to range from -17 to -18 ‰ in the pyroxenes with a δ7Li of +6 ‰
in the corresponding olivine (Ionov and Seitz, 2008). Bulk measurements of olivine phenocrysts
in primitive magmas from a variety of localities found a relatively uniform δ7Li of +3.2 to +4.9
‰; however, measurements of clinopyroxene yielded highly variable δ7Li (+6.6 ‰ to -8.1 ‰;
Jeffcoate et al., 2007). Both the olivine and clinopyroxene phenocrysts from Solomon Island
lavas are zoned with respect to lithium and δ7Li (Parkinson et al., 2007). Rims of phenocrysts are
enriched in lithium compared to the cores and the δ7Li decreases from core to rim by as much as
20 ‰ in a W-shaped profile (Parkinson et al., 2007). This pattern was also observed by Jeffcoate
et al. (2007) who measured a 40‰ variation in a single orthropyroxene crystal from a San Carlos
xenolith. The extreme grain-scale variability exhibited by lithium and lithium isotopes is not
limited to terrestrial samples. The basaltic lunar meteorite NWA 479 examined by Barrat et al.
(2005) contains olivine and pyroxene phenocrysts that also display a wide range of δ7Li values
(+2.4 to +15.1 ‰ in olivine and -0.2 to 16.1 ‰ in pyroxene). Beck et al. (2004) examined
pyroxenes in the shergottite meteorite NWA 480 and found extreme zoning of δ7Li from -17 ‰
in the cores to +10 ‰ in the rims and an absence of lithium compositional variation.
The extreme fractionation of lithium isotopic values documented in these high temperature
samples suggests kinetic rather than equilibrium processes (Lundstrom et al., 2005; Beck, 2006;
Jeffcoate et al., 2007; Parkinson et al., 2007; Rudnick and Ionov, 2007; Marchall et al., 2007).
This kinetic effect has been experimentally demonstrated by Richter et al. (2003) for diffusion of
8
lithium in molten silicate where fractionation occurs due to slightly faster transport of 6Li than 7Li. The time scale for the development of diffusion-controlled isotopic fractionation is likely to
be quite short as documented by the rapid lithium exchange in natural samples. Berlo et al.
(2004) reported rapid mobilization of lithium in a study of plagioclase phenocrysts from the 1980
eruption of Mount St. Helens in Washington, USA. Plagioclase phenocrysts erupted prior to the
degassing event contained ~14 ppm lithium, whereas those erupted immediately after contained
~5 ppm. The implication is that the magma lost a significant amount of lithium in a seven-day
period, which was recorded in the lithium content of the plagioclase phenocrysts. Kent et al.
(2007) also found that the lithium contents of plagioclase phenocrysts from the Mount St. Helens
2004 dome lavas had increased due to the addition of a pre-eruptive lithium rich vapour phase.
Based on the lithium contents of plagioclase phenocrysts, melt inclusions, and plagioclase
encapsulated within gabbroic inclusions Kent et al. (2007) were able to estimate that the influx of
the lithium-rich volatile phase occurred within ~1 yr of the dome lava eruptions.
1.4 Previous experimental work
Previous experimental work has found lithium to be moderately incompatible in clinopyroxene
co-existing with either fluid or melt, and that partitioning is a function of clinopyroxene major
element composition, DLicpx/melt increasing with increasing Ca/Al ratio (Hart and Dunn, 1993;
Brenan et al., 1998a; Blundy et al., 1998; Blundy and Dalton, 2000; Hill et al., 2000; Bennett et
al., 2004). Lithium was also found to be moderately incompatible in olivine relative to melt
(Brenan et al., 1998a; Brenan et al., 1998b, Taura et al., 1998; Zanetti et al., 2004). To date,
lithium partitioning between olivine and fluids has not been measured. Coogan et al. (2005)
measured lithium partitioning between plagioclase and clinopyroxene and found DLiplag/cpx
increases with increasing temperature (900oC to 1200oC). In this case partitioning was not
investigated with respect to plagioclase or clinopyroxene major element chemistry.
Little work has been done to determine lithium isotope partitioning and diffusion at pressures
and temperatures corresponding to crustal and mantle processes. A study examining isotopic
fractionation between spodumene and hydrous fluids measured an enrichment of 7Li in the fluid
from +3.5 ‰ at 500oC to ~ +1.0 ‰ at 900oC and 2.0 GPa (Wunder et al., 2006). Isotopic
fractionation between fluids and mica (from 300oC to 500oC, 2.0 GPa) and staurolite (from
9
670oC and 880oC, 3.5 GPa) found fluids to be preferentially enriched in 7Li relative to the mica,
and staurolite to be slightly enriched in 7Li relative to the fluid (Wunder et al., 2007). These
studies are consistent with ab initio calculations where 6Li is preferentially fractionated in sites
with octahedral coordinations (e.g. mica, spodumene; Yamaji et al., 2001, Wunder et al. 2006)
and 7Li is preferentially fractionated in to sites with tetrahedral coordination (e.g. staurolite;
Wunder et al., 2007). Fractionation of lithium isotopes was also measured in the quartz-
muscovite-fluid system, from 400-500oC (Lynton et al., 2005). Lynton et al. (2005) found both
the quartz and the mica to be preferentially enriched in 7Li, with fractionation factors ranging
from +8 to +12 ‰ for quartz and +18 to +20 ‰ for mica. For reasons that are unclear, these
results are inconsistent with the subsequent studies of Wunder et al. (2006, 2007) or the results of
this study. Because mica has lithium in octahedral coordination the expectation is that the fluids
would be preferentially enriched in 7Li with respect to the solid.
Experimental studies of kinetic isotopic fractionation (Richter et al., 2003) found that 7Li could
be fractionated from 6Li by tens of per mil during diffusion between molten basalt and rhyolite or
when diffusing through fluids. Although estimates of lithium diffusion coefficients have been
made from natural samples, to date there have been few laboratory measurements of lithium
diffusion in rock forming minerals (Figure 1.3). Giletti and Shanahan (1997) measured the
diffusion rates of various alkali elements in plagioclase feldspar. They found that diffusion in
feldspars is a function of ionic radius and cation charge, and as a result of its small size, lithium
diffusion rates are very rapid. Coogan et al. (2005) measured the diffusion coefficient for 6Li in
clinopyroxene between 800oC and 1100oC by using SIMS analysis and found lithium to be
similarly rapid.
1.5 Focus of Thesis and Distribution of Work
Knowledge of lithium elemental partitioning and isotopic fractionation between fluids and
common rock forming minerals is essential to evaluate the variations seen in natural samples
adequately. Information on lithium diffusion in minerals can be used to more accurately assess
the time-scales of magmatic and hydrothermal processes and account for intermineral isotopic
differences.
10
This study examines the lithium, nitrogen and boron isotope fractionation that occurs in mineral-
fluid reactions during slab and mantle interaction. The existing experimental database is
insufficient to properly evaluate the degree of isotopic fractionation that occurs during fluid-
mineral partitioning of lithium, boron, and nitrogen. The results of this work provide the
essential input for modeling the behavior of lithium in the mantle.
Chapter 2 of this study examines the partitioning and isotopic fractionation that occurs between
clinopyroxene, olivine, plagioclase and aqueous fluids and the intermineral fractionation between
olivine and clinopyroxene. All experiments were conducted by N. Caciagli in the High Pressure
Laboratory at the University of Toronto. The elemental lithium was analyzed by laser ablation
inductively coupled plasma mass spectroscopy (LA-ICPMS) at the University of Toronto by N.
Caciagli. The multi collector inductively coupled plasma mass spectroscopy (MC-ICPMS)
analyses of the clinopyroxene starting material and bulk analyses of the run product
clinopyroxene were done at University of Maryland by W. F. McDonough. The lithium isotopic
composition of starting material anorthite and olivine, and in situ lithium isotopic compositions
were analyzed by secondary ionization mass spectroscopy (SIMS) at Lawrence Livermore
National Laboratory by N. Caciagli with the assistance of D. Phinney.
Chapter 3 of this study measures the diffusion coefficient of lithium in clinopyroxene and
olivine. All experiments were conducted by N. Caciagli in the High Pressure Laboratory at the
University of Toronto. The elemental lithium was analyzed by LA-ICPMS at the University of
Toronto by N. Caciagli and lithium isotopic analyses were done by SIMS at Lawrence
Livermore National Laboratory by N. Caciagli with the assistance of D. Phinney.
Chapter 4 of this study outlines the technique development for experimental measurements of
nitrogen partitioning and isotopic fractionation between fluids and muscovite. All experiments
were conducted by N. Caciagli in the High Pressure Laboratory at the University of Toronto. The
nitrogen elemental and isotopic analyses were done at Lehigh University with the assistance of
G. Bebout.
The exploratory work on techniques to measure boron partitioning and isotopic fractionation
between muscovite and fluid is summarized in Appendix 1.
11
0 0.01 0.1 1 10 100 1000
ave river water
sea water
est mean c.c.
marine seds
AOC
MORB
OIB
IAB
eclogites
xenoliths
est mean mantle
ppm
A
-30 -20 -10 0 10 20 30 40
river water
sea water
est mean c.c.
marine seds
AOC
MORB
OIB
IAB
eclogites
xenoliths
est mean mantle
7Li
B
Figure 1.1 Sources and Concentration of Lithium in the Earth
Lithium concentration (A) and isotopic composition (B) of various terrestrial reservoirs. River water: Huh et al. (1998);
seawater: Millot et al. (2004); estimated continental crust (c.c): Teng et al. (2004); marine sediments: Bouman et al.
(2004); AOC: Chan et al. (1992); MORB: Moriguti and Nakamura (1998); Tomascak et al. (2008); Nishio et al. (2002);
OIB: Kobayashi et al. (2004); IAB: Moriguti and Nakamura (1998); Tomascak et al. (2002); eclogites: Marschall et al.
(2007); xenoliths: Nishio et al. (2004); est mean mantle: Jagoutz et al., (1979) and Tomascak (2004).
12
0 0.2 0.4 0.6 0.8 1 1.2 1.41
2
3
4
5
6
7
8
Li/Y
7Li
A
0
1
2
3
4
5
6
7
8
0 0.2 0.4 0.6 0.8 1 1.2 1.4
7Li
Li/Y
B
Figure 1.2 Li/Y ratio and 7Li in Arc Lavas
A plot of 7Li as a function of Li/Y ratio in (a) Izu arc basalts and (b) basalts from other Sunda and Aleutian arcs.
Lavas from the Izu arc display a trend of increasing 7Li with increasing Li/Y ratio, and show an inverse relationship
with depth to the slab (Benioff zone) suggestive of progressively decreasing amounts of fluid being mobilized during
subduction. The trend of increasing 7Li with increasing Li/Y ratio is not consistently observed in other island arcs.
Izu arc data from Moriguti and Nakamura (1998), Sunda and Aleutian arc data from Tomascak et al. (2002).
13
-22.0
-20.0
-18.0
-16.0
-14.0
-12.0
-10.0
-8.0
-6.0
4 8 12 16 20
DLi (
m2 /s
)
10,000/T (K)
Si-crystal
albite & anorthite
cpx, Coogan et al. 2005
20040060080010001400T (oC)
Figure 1.3 Lithium Diffusion Coefficients
Plot of log DLi vs. 10,000/T (K) for lithium diffusion in geologically significant minerals. Lithium diffusion in a p-type Si-crystal data are from Pell (1960), feldspar data are from Giletti and Shanahan (1997), and cpx data are from Coogan et al. (2005)
14
2 Lithium Partitioning and Isotopic Fractionation
2.1 Introduction
Lithium and lithium isotopes are increasingly used as tracers of surface inputs to the mantle
during subduction. With a strong affinity for fluids, an incompatible nature during mantle
melting and a high relative mass difference (~16 %) between the two stable isotopes, (6Li and 7Li), lithium has the potential to serve as a robust indicator of fluid-rock interaction in a variety
of geological settings. Enrichments in lithium isotopes are described as either:
(2) 10001LiLi/
LiLi/Li
std76
smp76
6
or 10001
LiLi/
LiLi/Li
std67
smp67
7
where ‘smp’ refers to the sample and ‘std’ refers to the standard, typically the NIST lithium
carbonate L-SVEC. The δ7Li notation is recommended by the International Union for Pure and
Applied Chemistry (Coplen et al., 1996) and will be used here.
Several reservoirs of lithium, which are isotopically distinct from the mantle and each other, are
present within Earth. Seawater has 0.18 ppm lithium with δ7Li of +32 ‰ (James and Palmer,
2000) and the continental crust contains an average of 35 + 11 ppm lithium with a δ7Li that
ranges from –5 to +5 ‰ (Teng et al., 2004). Lithium in fresh mid-ocean ridge basalts (MORB)
can range from 5 to 6 ppm, with a δ7Li of +1.5 ‰ to +5.6 ‰ (Moriguti and Nakamura, 1998;
Tomascak et al., 2008; Chan et al., 1992). Altered oceanic crust (AOC) has a significantly
greater concentration of lithium, >75 ppm lithium, and is isotopically heavier than pristine
MORB with a δ7Li of up to +14.2 ‰ in the most altered oceanic crust (Moriguti and Nakamura,
1998; Tomascak et al., 2008; Chan et al., 1992). The mantle is estimated to contain 1.6 ppm
lithium with an average δ7Li of +4 ‰ (Jagoutz et al., 1979; Moriguti and Nakamura, 1998;
Tomascak, 2004; Tomascak et al., 2008; Teng et al., 2004). However, studies of mantle xenoliths
suggest that reservoirs with variable δ7Li exist in the mantle (Seitz et al., 2004; Nishio et al.,
2004; Brooker at al., 2004; Lundstrom et al., 2005). Very low δ7Li (-11 ‰ to +5 ‰) values are
found in orogenic eclogites which are thought to reflect low temperature dehydration of the slab
15
during subduction (Zack et al., 2003). Conversely some pyroxenites from the Zabargad
peridotite, which is considered to be a fragment of an exhumed mantle wedge, have high δ7Li
values (+8.4 ‰ to +11.8 ‰) suggesting that 7Li-rich fluids or melts derived from subducted
AOC are also transferred to the mantle (Brooker et al., 2004).
In an investigation of lithium isotopes from the Kilauea Iki lava lake, Tomascak et al. (1999)
demonstrated that neither partial melting nor low pressure differentiation results in significant (>
+2 ‰) variations in δ7Li. This has led to the interpretation that the variability evident in some
mantle-derived lavas is due to melting of a heterogeneous source. The Izu arc shows a trend with
the greatest lithium and 7Li enrichment occurring at the arc front where δ7Li in lavas ranges from
+7.6 ‰ to +1.1 ‰ (see Figure 1.2a) suggesting enrichment of the arc melt source by fluids
derived from the down going slab (Moriguti and Nakamura, 1998). A study of lithium in glass
inclusions from Hawaii showed the lithium isotopic composition to vary from –10.2 ‰ to +8.4
‰ (Kobayashi et al., 2004). Similarly, low δ7Li (-3.3 ‰ to +1.2 ‰) has been measured in glass
inclusions from the Iblean (Sicilian) Plateau tholeiites (Gurenko and Schmincke, 2002). A study
of δ7Li values in OIBs from Antarctica did not show any significant deviations from MORB
values (Ryan and Kyle, 2004); however, a study of MORB lavas from different ridge systems
found a 5 ‰ heterogeneity in the samples, and no significant correlation of δ7Li with major
elements, trace elements or radiogenic isotopes (a slight apparent correlation with Cl/K was
observed; Tomascak et al., 2008). The fact that both OIB and MORB can sometimes display a
range of values has prompted many researchers to suggest that variable amounts of recycled
material with modified δ7Li is transported into the mantle sources of these lavas (Tomascak et
al., 2008). But the origin and scale of these mantle heterogeneities are not well defined.
The extent of the modification of the down going slab by fluid-mineral exchange during
subduction remains ambiguous, as is the extent of modification of the slab derived fluid during
transport though the mantle to the arc source. Previous experimental work has found lithium to
be moderately incompatible in clinopyroxene relative to fluid or melt with DLicpx/fluid increasing
with increasing Ca/Al ratio (Hart and Dunn, 1993; Brenan et al., 1998a; Blundy et al., 1998;
Blundy and Dalton, 2000; Hill et al., 2000; Bennett et al., 2004). Lithium was also found to be
moderately incompatible in olivine relative to melt (Brenan et al., 1998a; Brenan et al., 1998b;
Taura et al., 1998; Zanetti et al., 2004) and fluid (Blundy and Dalton, 2000). Coogan et al. (2005)
measured lithium partitioning between plagioclase and clinopyroxene and found DLiplag/cpx
16
increasing with increasing temperature (900oC to 1200oC). In this case, partitioning was not
correlated with plagioclase or clinopyroxene major element composition.
Little experimental work has been done to examine lithium isotope fractionation at pressures and
temperatures relevant to crustal and mantle processes. Comparisons of δ7Li from metasomatized
and pristine peridotite xenoliths suggest that some olivine – clinopyroxene fractionation, (up to
3.5 ‰ enrichment in 7Li), may occur at mantle temperatures (950oC; Seitz et al., 2004). Fluid-
mineral partitioning will also fractionate lithium isotopes as documented in a single study which
measured an enrichment of 7Li in fluids relative to a Li-pyroxene (spodumene) by +3.5 ‰ at 500 oC to ~ +1.0 ‰ at 900oC and 2.0 GPa (Wunder et al., 2006).
To date, a systematic investigation of the degree of isotopic fractionation and the extent of
partitioning that occurs during mantle processes has been lacking. Both the isotopic
fractionation, Li, and the lithium partitioning between major mantle phases need to be known to
determine the extent to which a slab signal can propagate to the IAB source. This study presents
lithium partitioning and isotopic fractionation measurements between fluids and common rock
forming minerals. With this information, more accurate models can be developed to constrain the
origins of lithium anomalies in the mantle.
2.2 Methods
This study measured the partitioning of lithium between aqueous fluids and clinopyroxene,
olivine and plagioclase at pressures and temperatures corresponding to lower crustal and upper
mantle conditions (800oC to 1200oC; 1 GPa). Additional experiments were done to measure
olivine – clinopyroxene isotopic fractionation at similar conditions. Starting materials were
natural single crystals of: olivine (Fo82) from San Carlos, Arizona; plagioclase (bytownite) from
Crystal Bay, Minnesota; clinopyroxene (diopside) from Dekalb, New York; and plagioclase
(albite) from Mont St. Hilare, Quebec. Table 2.1 gives the composition of the starting materials.
In all cases, the mineral samples were first crushed to 1-3 mm grain size, after which grains free
of inclusions and alteration were hand picked and cleaned in dilute HNO3 and rinsed with ultra-
pure water in an ultrasonic cleaner. For the olivine and plagioclase experiments the minerals
were ground to a fine powder under ethanol with an additional SiO2 + Al2O3 (1:1) mixture (~3
17
wt. % of total) added to approximate natural mantle fluid compositions (Brenan et al., 1998b;
Holloway, 1971). Experiments containing olivine were not buffered for Fe loss to the noble
metal capsule; therefore, run product compositions are shifted to more magnesium rich
compositions (from Fo82 to Fo97 – 99). Experiments with bytownite as a starting mineral
composition were not buffered for Na2O loss to the fluid and as a result run product
compositions are shifted from bytownite (An75) to anorthite (An98-99). A few plagioclase
experiments contained additional albite to stabilize more sodic compositions of plagioclase. With
one exception all experiments with Dekalb diopside as starting material had ~3 wt. % SiO2 added
(no Al2O3) since clinopyroxene dissolution buffers the aluminum content of the fluid. To
minimize compositional zoning, a large fluid to solid ratio (4:1 by mass) was utilized for all
experiments. One clinopyroxene experiment was carried out with the addition of 3 wt. % albite
(+3 wt. % SiO2) to encourage compositional zoning with respect to the aluminum content of the
clinopyroxene.
A series of mineral pair experiments were run with clinopyroxene and olivine to constrain their
inter-mineral isotopic fractionation. These experiments used the same starting materials prepared
as above and mixed 80:20 cpx-olivine by mass.
Isotopically labeled solutions were made from ultra pure water with lithium added as either
Li2CO3 (LSVEC, SRM#8545) or some combination of LSVEC and a 6Li spike. For each
experiment a sample of either clinopyroxene, olivine or plagioclase powder (+/- SiO2, Al2O3 or
albite) and lithium bearing solution were added to a large volume Ni capsule with a Pt insert
(Ayers et al., 1992). To promote crystal dissolution and re-precipitation, the bottom of the
capsule was centered in the hotspot of the furnace. The temperature gradient over the length of
the capsule is less than 10oC (Ayers et al., 1992).
The experiments were conducted in an end-loaded piston–cylinder apparatus (Boyd and England,
1960) using a 1.9 cm bore pressure vessel, employing a cylindrical graphite heater and pressure
cells consisting of crushable MgO, Pyrex and NaCl. Samples were initially cold pressurized to
~0.5 GPa and then heated to 300oC to generate sufficient internal pressure to prevent capsule
deformation (Brenan et al. 1995). Temperature and pressure were then increased simultaneously
with the maximum pressure being achieved by the time the sample reached 600oC. Temperature
was monitored with W26% Re-W5% Re thermocouples uncorrected for the effect of pressure on
18
EMF. Experiments were run for 72 to 144 hours and quenched by cutting power to the sample
heater which resulted in temperatures dropping to < 300oC in 20 seconds. The capsules were then
recovered, punctured and dried. Fluid masses determined by weight loss were usually > 70 % of
the initial fluid mass; however, during the course of the experiment the capsule material became
work hardened; consequently, an undetermined amount of capsule material was sheared off
during puncturing. As a result, the fluid masses used in the mass balance calculations are the
initial fluid masses. In the case where a watertight seal was not maintained throughout an
experiment, a drop in pressure and a collapsed capsule would result.
One additional experiment was carried out in a cold seal vessel at 0.2 GPa and 800oC. In this
case the sample powder and lithium solution were loaded into a 5 mm O.D. Au capsule. The
capsule was then crimped, weighed, sealed by arc welding and reweighed to check for fluid loss.
The sample was loaded into a vertically mounted pressure vessel, first pressured to 0.2 GPa and
then externally heated. Temperatures were monitored with an internal type K thermocouple.
Experiments were quenched by removing the furnace and cooling the pressure vessel with
compressed air, which resulted in temperatures dropping to < 300oC in ~3 minutes. Table 2.2
provides a summary of experimental conditions.
2.3 Analytical Techniques
2.3.1 Major Element Analyses
Samples of starting material and splits of run products were mounted in epoxy, ground, polished
to 0.3 m and carbon coated. The major element compositions of the starting materials and run
product clinopyroxene, olivine and plagioclase were then obtained using the University of
Toronto’s Cameca SX50 Electron Probe X-ray Microanalyzer (EPMA). An accelerating voltage
of 15 kV and a focused 20 nA beam was used for all samples. The standards were albite for Na,
anorthite for Al, diopside for Ca, Mg, Si, basalt for Fe, and bustamite, (Mn,Ca)3Si3O9, for Mn.
X-ray intensities were converted to concentrations using modified ZAF or Phi-Rho-Z correction
schemes. The reported errors are the 1σ variations of (n) analyses.
19
2.3.2 Lithium Analyses
2.3.2.1 MC-ICPMS
Bulk lithium isotopic composition and element concentrations were determined for run product
clinopyroxene using a Multi-Collector Inductively Coupled Plasma Mass Spectrometry (MC-
ICPMS) at the University of Maryland. Samples were first rinsed in ultra-pure water to remove
any water-soluble Li-bearing residue and then analyzed following the method described in Teng
et al. (2004). The samples, 4-10 mg of run-product clinopyroxene, were digested in a mixture of
HF and HNO3 and dissolved in a 4M HCl solution. The lithium from the samples was then
separated from the dissolved matrix by thrice processing the solutions in cation exchange
columns. The solutions were then introduced to the Nu-Plasma MC-ICP-MS in a 2% HNO3
solution, and isotopic compositions were obtained by measurement of 7Li and 6Li simultaneously
on two high and low mass Faraday cups. Each sample analysis was bracketed by measurement of
the L-SVEC standard. The isotopic values are reported as δ7Li (equation 2.1) in which the
lithium isotopic standard is NIST L-SVEC Li2CO3. The 2σ precision of each analysis is ±1 ‰.
Table 2.3 lists measurements of standards and reference materials.
2.3.2.2 SIMS
In situ analyses of the lithium abundance and isotopic composition of the starting materials, run
product clinopyroxene, olivine, and plagioclase were obtained using the Cameca IMS 3f ion
microprobe at Lawrence Livermore National Laboratory. Secondary ions were generated by
bombardment with a 5-12 nA negatively charged 16O primary beam, accelerated through –12.5
kV and focused to ~20 μm. The positive secondary ions were accelerated through 4.5 kV. 6Li
and 7Li were measured with a mass resolving power of 1011, and no energy offset was applied.
The background, mass 5.8, 6Li and 7Li were counted on an electron multiplier for 2 s, 10 s and 2
s respectively over 120-400 counting cycles, depending on count rate. Figure 2.1 shows the
‘uncorrected’ δ7Li values measured by SIMS of the internal reference materials; Dekalb diopside
(measured in this study by MCICP-MS), San Carlos olivine (Magna et al., 2006), and
clinopyroxene from experiment NCDL6 (analysed by MC-ICPMS), plotted against the δ7Li
values measured by MC-ICPMS. The δ7Li values measured by MC-ICPMS of the internal
reference materials range from -3 ‰ to +10 ‰ whereas the corresponding ‘uncorrected’ δ7Li
values measured by SIMS range from +15 ‰ to +30 ‰. The discrepancy between values is due
20
to mass fractionations that are the result of both instrumental parameters and matrix effects
(Decitre et al., 2002, Tomascak, 2004). It is important to note that all the values plot on a single
line with a slope of ~1, suggesting that any matrix effect on the lithium instrumental isotopic
fractionation is still within the 2σ precision of the analysis (± 4‰). The 7Li/6Li ratios can be
corrected for this fractionation using the instrumental correction factor, ∆i (Decitre et al., 2002);
(3) ∆i = δ7LiSIMS – δ7LiMC-ICPMS
Because all of the fractionation measured is relative to the same starting material, the 7Li/6Li
ratio of Dekalb diopside was used as the internal reference material to determine ∆i. As a check
on the calibration of this correction factor, ∆i, was determined using San Carlos olivine, ∆i = -22
‰, and was found to be within 2σ (± 4 ‰) error of that for the Dekalb diopside, ∆i = -20 ‰.
However, the concentration of lithium in Dekalb diopside and the NCDL6 experiment are both
greater than that of San Carlos olivine (i.e. 8.6 ppm and 67 ppm vs. 2.5 ppm), therefore the
Dekalb diopside and NCDL6 were primarily used as internal reference materials.
2.3.2.3 LA-ICPMS
In situ analyses of lithium abundances in clinopyroxene, olivine, and plagioclase were also
determined by laser ablation inductively coupled plasma mass spectrometry (LA–ICP–MS) at
the University of Toronto. The system employs a frequency quintupled Nd:YAG laser operating
at 213 nm, coupled to a VG PQExcell quadrupole ICP-MS. The laser was operated at 10 Hz,
with He flushing the ablation cell to enhance sensitivity (Eggins et al. 1998), and produced spot
sizes ~50 μm in diameter and ~ 50 μm deep.
The torch position and lens settings were adjusted prior to each analytical session to optimize the
signal intensity while ablating NIST 610 with a spot size of approximately 75 μm and a laser
beam energy of less than 3 mJ, so that the sensitivity to 7Li was maximized. Data were collected
as time-resolved spectra with background levels determined by counting for 20 s prior to the 60 s
of sampling by laser ablation. Analyses were collected in blocks of >20, with the first and last
two spectra acquired on standard reference materials. Count rates were collected and exported as
CSV (comma-delimited values) files by “ThermoElectron Plasmalab” (TRA) software. All
subsequent data reduction was performed off-line using the GLITTER 5.3 software package,
supplied by Macquarie Research, Ltd. Ablation yields were corrected by referencing to the
21
known concentration of 43Ca or 55Mn as determined previously by electron microprobe analyses.
Lithium concentrations in clinopyroxene, olivine and anorthite were quantified using the “in
house” standard Kunlun diopside, which contains 42.6 ppm lithium. This standard was used
routinely because it generated a lower lithium background over the course of the analysis
compared to that produced by NIST 610. Kunlun diopside was, in turn, characterized using NIST
610 silicate glass, which contains 470.5 ppm lithium (Pearce et al., 1997). Table 2.3 compares
the accepted values for several standard and reference materials (fused into glass in sealed Pt
capsules at 1 GPa and 1200oC) which were also measured by LA-ICPMS using NIST 610 as a
standard. The precision for concentration measurements is better than ± 10 %. Figure 2.2 shows
lithium abundance of the cross-referenced samples plotted against lithium abundance determined
by LA-ICP-MS.
2.4 Results
2.4.1 Major Element Chemistry
Figure 2.3 shows the size and morphology of run product clinopyroxene, olivine and plagioclase.
The run products display considerable coarsening compared to starting materials, and well-
developed crystal faces. Crystals grew as aggregates of grains nucleating on the lid and/or upper
walls of the capsule. Table 2.4 lists the major element composition of the clinopyroxene, olivine
and plagioclase run products produced in this study. The minimum detection limits are less than
the lowest value cited for each element and the number in parentheses refers to 1σ of the
standard deviation for n analyses and reflects the degree of sample heterogeneity.
2.4.2 Mineral – Fluid Lithium Partitioning
Table 2.5 lists the lithium contents of experimentally produced crystals as analyzed by LA-
ICPMS as well as the calculated distribution coefficient (DLi) values for mineral/fluid
partitioning or where applicable the D-values for mineral/melt partitioning. Mineral/fluid
distribution coefficients (Dmin/fluid) were determined from final mass balance calculations:
(4) CLitotal = Cinitial
min Xinitial
min + Cinitialfluid X
initialfluid
(5) CLitotal = Cfinal
min Xfinal
min + Cfinalfluid X
finalfluid
22
Where Cmin is the lithium concentration of the mineral in ppm, Cfluid is the lithium concentration
of the fluid in ppm and Xmin is the mass fraction of the mineral, and Xfluid is the mass fraction of
the fluid. The initial concentration of fluid, Cinitialfluid, and the initial mass fraction of fluid,
Xinitialfluid, is assumed to be equal to the final concentration and mass fraction, Cfinal
fluid and
Xfinalfluid. Mineral solubility is also assumed to be negligible such that Xfinal
min is equal to
Xinitialmin.
(6) CLitotal/ C
finalmin = Xfinal
min + (Cfinalfluid X
finalfluid)/ C
finalmin
The Nernst distribution coefficient is defined as,
(7) Dmin/fluid = Cfinalmin/ C
finalfluid
Then,
(8) [CLitotal/ C
finalmin] – Xfinal
min = 1/ Dmin/fluid Xfinalfluid
(9) Dmin/fluid = Xfinalfluid/( [C
Litotal/ C
finalmin] – Xfinal
min)
The fluid-mineral ratio was such that the volume of the solution would serve as an infinite
reservoir and the lithium concentration would remain constant throughout the experiment.
Mineral/mineral distribution coefficients (Dmin/min) were calculated from:
(10) Dmin/min = CfinalminA/CfinalminB
where Cfinal is the lithium concentration of mineral A or mineral B in ppm.
The range of lithium content in the run products was 5 ppm to 7 ppm in the clinopyroxene, 13
ppm to 466 ppm in the olivines, and 20 ppm to 70 ppm in the plagioclase. Mineral – fluid
equilibrium was assessed in terms of run-product homogeneity. The concentration of lithium
within a single experiment typically varied by 14 % relative to the mean concentration from
grain to grain and in a few cases, lithium contents varied as much as 30 %. The 50 μm spot size
and 80-second ablation time often meant that the laser analysis consumed the whole grain. The
run product olivine grains typically had diameters < 75 m, the clinopyroxene > 75 m, and the
anorthite grains > 100 m. Figure 2.4 shows the time resolved spectra for 7Li and 43Ca, measured
by LA-ICPMS, for clinopyroxene produced in the lowest temperature partitioning experiment
23
(800oC). Time resolved spectra for all experiments, with a few exceptions thought to be the
result of fluid inclusions, display level spectra, which are interpreted as homogenous equilibrated
grains. The slight sloping of the spectra is due to signal decay as a result of the geometry of the
ablated pit and does not reflect sample heterogeneity as both the 7Li and 43Ca signals remain
consistent with respect to each other.
An additional test for equilibration between mineral samples and fluids was attempted by
measuring partition coefficients in reversal experiments where crystals previously equilibrated
with solution ‘A’ were re-equilibrated with solution ‘B’, containing a lower concentration of
lithium and a differing isotopic composition. These reversal experiments confirm isotopic
equilibrium; however, changes in mineral assemblages (zoisite in anorthite reversal, Mg-
hydroxides in olivine reversal, monticellite in olivine + clinopyroxene reversal, undetermined
phase in clinopyroxene reversal) make the mass balance calculations, used to determine the
distribution coefficients, impossible to resolve. A simple calculation using the diffusion rates of
lithium in clinopyroxene measured by Coogan et al. (2005) determines that at 800oC lithium
diffusion should penetrate to a distance of 150 μm in 72 hrs, which is greater than the radius of
the largest run product crystal (Figure 2.3). An experiment was attempted at 0.2 GPa and 800oC
to ascertain the effect of pressure on the partitioning behavior of lithium, with no significant
effect of decreasing pressure noted.
Lithium values were always shifted with respect to the starting material. Any change in the
lithium composition of the fluid due to uptake by the growing mineral is insignificant compared
to the total amount of lithium in the fluid even when the DLimin/fluid is greater than one. Generally,
the lithium mineral/fluid distribution coefficients decrease in the order: olivine (2.51 – 0.17),
plagioclase (0.32 – 0.090) and clinopyroxene (0.32 – 0.07).
2.4.2.1 Clinopyroxene
Run products produced in these experiments typically contained no phases other than
clinopyroxene. One experiment, NCDL3, consisted of approximately 50 % (by volume) olivine
crystals and 50 % clinopyroxene crystals, which was most likely a result of magnesium
contamination from the ceramic pressure cell during sample assembly or loading. Electron
microprobe traverses across individual clinopyroxene grains shows that the resulting crystals are
homogenous with respect to major elements. As shown in Table 2.4, these clinopyroxene grains
24
have much lower Al2O3 (~0.5 wt. % to ~0.2 wt. %) and FeO (~0.9 wt. % to ~0.4 wt. %) contents
than natural upper mantle clinopyroxene (~3 wt. % Al2O3, ~2 wt. % FeO; Lundstrom et al.,
2005). However, their MgO concentrations (~17 wt. % to ~21 wt. %) and Na2O concentrations
(~0.02 wt. % to ~0.30 wt. %) are similar to those in clinopyroxene from mantle xenoliths (~17
wt. % MgO and ~0.30 wt. % Na2O; Lundstrom et al., 2005). The FeO content of all the
experimentally produced clinopyroxene is low (< 1 wt. %) due to loss of Fe to the platinum
capsule. Additionally because the starting clinopyroxene material contained very low (below
detection limits) abundances of chromium and titanium these elements are absent in the run
products.
The range of DLi (0.07 to 0.613) for clinopyroxene/fluid measured in this study is similar to the
clinopyroxene/fluid DLi (0.08 to 0.25) measured by Brenan et al. (1998b) and the
clinopyroxene/silicate melt DLi (0.14 to 0.27) measured by Brenan et al. (1998a).
Lithium partitioning between clinopyroxene and fluid decreases from 0.32 to 0.09 with
increasing temperature from 800oC to 1100oC at 1 GPa. The temperature dependence of lithium
partitioning between clinopyroxene and hydrous fluids can be demonstrated on a plot of ln
DLicpx/fluid versus 1000/T, (Figure 2.5). A linear regression of the data yields the relationship:
(11) ln DLicpx/fluid = -7.3 (+0.5) + 7.0 (+0.7) * 1000/T (R2=0.98)
where T is temperature in Kelvins.
2.4.2.2 Olivine
Olivine partitioning experiments occasionally produced some oxide grains and in the reversal
experiment (NCOR), an unidentified magnesian phase. All are likely due to incongruent
dissolution of olivine. Electron microprobe traverses of individual crystals show the run product
olivines to be homogenous with respect to major element chemistry. Iron loss to the platinum
capsule resulted in considerably more magnesian (Fo# 98 to 99) olivines than those naturally
occurring in the mantle. Reversal experiments were attempted; however, re-equilibrating run
product material, from experiment NCOL2, caused Fe and water soluble elements (i.e. Na, Ca,
etc.) to become further depleted, such that the final solid composition was no longer in the
stability field of olivine. The same effect occurred with the plagioclase reversal run, NCAR,
which resulted in the crystallization of zoisite.
25
The range of DLi (0.17 to 0.57) measured in this study for olivine/fluid is similar to the
olivine/silicate melt DLi (0.13 to 0.35) measured by Brenan et al. (1998a) with the exception of
the experiment with added albite (OlAb10) which produced a value of 1.34. The temperature
dependence of lithium partitioning between olivine and hydrous fluids can be demonstrated on a
plot of ln DLiol/fluid versus 1000/T, (Figure 2.6). A linear regression of the data (excluding the
reversal, NCOLR) yields the relationship:
(12) ln DLiol/fluid = -6.0 (+2) + 6.5 (+2) * 1000/T (R2=0.82)
where T is temperature in Kelvins.
2.4.2.3 Plagioclase
The starting material for the anorthite experiments was bytownite (An76), but the experiments
that were not buffered for Na loss to the solution resulted in run product compositions very close
to end member anorthite (An96-99). The reversal run, NCAR, which consisted of re-equlibrating
material from NCA3, resulted in zoisite + unidentified Al-rich phase. At 1000oC run products
consisted of 30 % melt, 70 % anorthite crystals, at 900oC the amount of melt was negligible, and
runs at 800 oC were melt free. Electron microprobe traverses of these grains show that they are
homogenous with respect to major elements. In two of the experiments, AnAb10 and AnAb20,
the Na content of the fluid was buffered by addition of albite. This resulted in homogenous
anorthite crystals with only slightly more sodic compositions (see Table 2.4).
The range of DLi for anorthite – fluid partitioning measured in this study is 0.09 to 0.32. The
range of DSr and DBa for plagioclase with similar An content is 1 -3 and 0.1 to 0.2 respectively
(Blundy and Wood, 1991). Similar to the results from a study of Sr and Ba partitioning in
plagioclase (Bludy and Wood, 1991), the data show a linear relationship with a negative slope on
a plot of ln DLi versus XAn suggesting that lithium is more compatible in albite than in anorthite
(Figure 2.7). Linear regression of the six partitioning experiments yields the relationship, in
Jmol-1:
(13) RTlnDLi = 162,000 (+26,000)– 188,000 (+28,000) (XAn) (R2=0.96)
26
where R is the gas constant, T is temperature in Kelvins, and XAn is the anorthite content of the
plagioclase. Following Blundy and Wood (1991), RTlnDLi is used rather than lnDLi to minimize
the effect of temperature in the linear regression.
2.4.3 Olivine-Clinopyroxene Pair Experiments
A series of experiments were done with olivine + clinopyroxene + fluid to investigate inter-
mineral partitioning and isotopic fractionation. The run products from these experiments
typically consisted of coarse-grained intergrowths of olivine and clinopyroxene and in the case
of experiments 2m-lo and Yb-1, molybdenum oxide (from outer capsule material) and ytterbium
oxide, respectively. Due to the uncertainties in the mass balance of each phase after
equilibration, mineral-fluid D values have not been calculated for these runs. Experiment 2m-1 at
900oC contained only clinopyroxene; however, NCDL3 at 900oC stabilized both clinopyroxene
and olivine. Experiment 2m-hi, investigating partitioning at log fO2 of –5, stabilized only
enstatite. The reversal experiment, 2m-R, resulted in olivine + clinopyroxene + monticellite
(CaMgSiO4; see NCOR and NCAR above).
In all the experiments run at Ni-NiO, with the exception of the reversal experiment, 2m-R,
olivine preferentially incorporated lithium relative to clinopyroxene. The range of DLi for
olivine/clinopyroxene measured in this study is 1.2 to 6.7. Figure 2.8 shows how
olivine/clinopyroxene partitioning increases with increasing temperature between 800oC and
900oC. The lithium content of the run product olivine in the 800oC experiment, 2m-2, is 101 ± 59
ppm; this large standard deviation suggests that the run product olivine may be heterogeneous
with respect to lithium and may have formed lithium rich fluid inclusions. The reversal
experiment, which equilibrated a split of the 2m-2 experiment at 800oC with a solution of 96
ppm lithium, did result in a lower lithium content in the run products than in the starting
material. This experiment is complicated by the formation of monticellite, which contains 39
ppm lithium, more than either the forsteritic olivine or clinopyroxene (11 ppm and 26 ppm,
respectively). The constant Dol/cpxLi versus temperature further suggests that
olivine/clinopyroxene partitioning of lithium is independent of temperature. Also shown in
Figure 2.8 is the ratio of DLiolivine/fluid/ DLi
cpx/fluid calculated from single-phase experiments, which
at 1000oC is the same as that determined from the two-phase experiment.
27
2.4.3.1 Experiments at variable fO2
Three experiments were conducted to investigate the effect of oxygen fugacity on the lithium
partitioning behavior between olivine and clinopyroxene. Experiment, 2m-hi, which was run at
1000oC in a Pt + Re lined nickel capsule to generate an oxygen fugacity of log fO2 of –5,
stabilized enstatite. The enstatite had a lithium content of 5.28 ppm and resulted in a DLient/fluid of
0.02, which is much lower than the DLimin/fluid for either clinopyroxene (0.17) or olivine (0.48 at
Fo#98, or 0.17 at Fo#63) at the same temperature. Experiment 2m-lo was run at 1000oC in a Pt
lined Mo capsule to generate an oxygen fugacity of log fO2 of –15 and resulted in olivine with 30
ppm lithium and clinopyroxene with 42 ppm lithium which results in a Dol/cpxLi of 0.7. The
oxygen fugacity generated by the Ni lined Pt capsule at 1000oC is log fO2 = -10.3 and resulted in
olivine with 52 ppm lithium and clinopyroxene with 13 ppm lithium which results in a Dol/cpxLi of
4.0.
Higher oxygen fugacity results in higher abundances of Fe3+ relative to Fe2+. Incorporating a
higher proportion of Fe3+ into the olivine structure should generate more charge balancing
opportunities for the Li+ ion in the crystal structure; resulting in a coupled substitution where:
(14) Li1+ M1 + X3+
M2 (Mg2+, Fe2+)M1 + (Mg2+, Fe2+)M2
This is consistent with the results between the NNO run 2m-3, (higher fO2 and Dol/cpxLi = 4.0) and
the 2m-lo run (lower fO2 and Dol/cpxLi = 0.7).
2.4.3.2 Experiments with added REE
Experiment Yb-1, carried out at 1000oC with 0.25 mg of Yb2O3, was an attempt to determine the
effect of rare earth elements (REE) on the relative partitioning of lithium between olivine and
clinopyroxene. This experiment resulted in olivine with 175 ppm lithium and clinopyroxene with
17 ppm lithium, and resulted in a Dol/cpxLi of 10, which is significantly higher than the Dol/cpx
Li of
4.0 at 1000oC that results from no addition of REE. The increased partitioning of lithium into the
olivine is most likely a result of a coupled substitution with Yb3+, analogous to that in Equation
14 for Fe3+.
28
2.4.4 Lithium Isotope Fractionation
2.4.4.1 Clinopyroxene- fluid
Table 2.6 gives the isotopic ratios of the starting materials and run products from the isotopic
fractionation experiments as well as the calculated ∆7Licpx-fluid, where;
(15) ∆7Licpx-fluid = δ7Licpx (‰)- δ7Lifluid (‰)
With the exception of the reversal runs, all experiments had Dekalb diopside (δ7Li of +9.7 ‰) as
starting material. Two sets of solutions were used: two L-SVEC based solutions, (A) with 243
ppm lithium and δ7Li of 0 ‰ and (B) 96 ppm lithium and δ7Li -2.7 ‰ for the reversal and two 6Li doped solutions one (C) with 306 ppm lithium and δ7Li of -88.4 ‰ and (D) 180 ppm lithium
and δ7Li of –46.1 ‰ for the reversal. In all the experiments, the crystals were preferentially
enriched in 6Li with respect to the fluid. Duplicate experiments at 900oC and run times of 72 hrs
and 142 hrs had ∆7Licpx-fluid within + 2 ‰, which is within the precision of the analysis,
indicating that run times were sufficient for isotopic equilibrium. As a further test, a reversal
experiment, Ldi-12, was conducted in which a split of sample Ldi-10 with δ7Li of –90.9 ‰ was
reacted with solution B (δ7Li –46.1 ‰). The run product clinopyroxene from Ldi-12 had a δ7Li
of –49.5 ‰, an enrichment of 45 ‰ in the heavier isotope from its initial value, and resulted in a
∆7Licpx-fluid of –3.4 ‰, which is within the precision of the other experiments. The lithium
isotopic fractionation at high temperatures (T>900oC) is +2.5 ‰, which is just at the limit of the
analytical precision of this study. The range of ∆7Licpx-fluid measured in this study is from –0.3 ‰
to –3.4 ‰, and follows the trend of ∆7Licpx-fluid decreasing with increasing temperature. When the
run products were not rinsed prior to sample digestion and analysis, the data produced scattered
results, most likely due to the precipitation of lithium as the remainder of the solution was dried
down after sample recovery.
Figure 2.10 is a plot of the ∆7Licpx-fluid (‰) from this study, as well as the ∆7Lispodumene-fluid (‰)
from Wunder et al (2006), the ∆7Libasalt-fluid (‰) measured between basalt and seawater at 350oC
(Chan et al., 1993) and 2oC (Chan et al., 1992) versus 1000/T (K). Despite the fact that Wunder
et al. (2006) used both OH- and Cl-bearing fluids and this study was chlorine free, with the
lithium introduced as Li2CO3, all of the data plot on the same regression line empirically
29
determined by Wunder et al. (2006). There appears to be no difference in fractionation behavior
with pressure (seafloor to 2 GPa) or complexing anion.
It should be noted that measurements of lithium isotopic fractionation in the quartz-muscovite-
fluid system, from 400-500oC, found the quartz and the mica to be preferentially enriched in 7Li
(Lynton et al., 2005), which is inconsistent with this study.
2.4.4.2 Olivine-Clinopyroxene Lithium Isotope Fractionation
Table 2.6 gives the isotopic ratios of the starting materials and run products from the isotopic
fractionation experiments as well as the calculated ∆7Liol-cpx, where;
(16) ∆7Liol-cpx = δ7Liol (‰) - δ7Licpx (‰)
All experiments, with the exception of the reversal runs, had 82 wt. % Dekalb diopside (δ7Li of
+9.7 ‰) + 18 wt. % San Carlos olivine (δ7Li of +1.0 ‰) as starting material. Two L-SVEC
based solutions were used: (A) with 243 ppm lithium and δ7Li of 0 ‰ and (B) 96 ppm lithium
and δ7Li -2.7 ‰ for the reversal. In all the experiments, except the reversal run, the isotopic
composition of the olivine grains did not shift significantly; whereas, the isotopic composition of
the clinopyroxene became as much as 15 ‰ lighter (i.e. from δ7Li of +9.7 ‰ to ~-5 ‰). ∆7Liol-
cpx measured in this study is ~5 ±5 ‰, which is not resolvable with the precision of this study.
2.5 Discussion
2.5.1 Controls on Partitioning
2.5.1.1 Clinopyroxene
Clinopyroxene has two sites, M1 and M2; M1 has six-fold coordination with respect to oxygen
and M2 has eight-fold coordination. Given that the M1 site is slightly smaller than M2 (optimal
site radius (ro) of ~0.7 Å versus M2 ro of ~1.1 Å) and has lower defect energies for univalent
cations, it is likely that the primary site for lithium in the clinopyroxene structure is M1 where it
can exchange for Mg2+ (Purton et al., 1997). Such an exchange should be coupled by a trivalent
cation such as Al in jadeite (NaAlSi2O6) component or spodumene (LiAlSi2O6) or Fe3+ in a
aegirine-like molecule, NaFe3+Si2O6.
(17) Li1+M1 + X3+
M2 (Mg2+)M1 + (Ca2+) M2
30
Due to Fe loss to Pt capsule the total iron content of the run product clinopyroxene is low, less
than 2 wt. % and does not vary systematically with temperature. A Mössbauer study of natural
diopside crystals by De Grave (2003) found all to contain some component of Fe3+ in either M1
or M2 sites. Furthermore, increasing Fe3+ at the expense of Fe2+ also serves to increase the
Mg/(Mg+Fe2+) ratio (Luth and Canil, 1993) which should then increase the availability of sites
for lithium exchange. Experiments examining the effect of fO2 on lithium diffusion in
clinopyroxene appear to confirm this; lithium diffusion in clinopyroxene appears to increase with
decreasing oxygen fugacity (Caciagli, Chapter 3).
Previous work has shown lithium partitioning to increase slightly with increasing Al/Si ratio
(Brenan et al., 1998b). A compilation of the data collected in this study and other experimental
data displays a similar trend. In Figure 2.5, the partition coefficients determined from cpx/fluid
experiments in this study, and Brenan et al. (1998b), and the cpx/melt experiments of Hart and
Dunn (1993), Blundy (1998), Brenan et al. (1998a), and Blundy and Dalton (2000) are plotted as
a function of temperature and the data points are labeled with the wt. % Al2O3 content of the run
product clinopyroxene. It is important to note that the alumina contents of run product
clinopyroxene in this study are approximately constant. Generally, lithium-partitioning at a given
temperature increases with increasing Al2O3 content of the run product clinopyroxene, regardless
if the partitioning measured is min/melt or min/fluid. For example, the run product
clinopyroxene from the cpx/melt experiment of Blundy and Dalton (2000) at 1375oC (and 0.8
GPa) has a similarly low alumina content as the clinopyroxene from this study, and plots on the
regression line determined in this study. As the alumina content of the clinopyroxene increases,
lithium partitioning increases and the data fall above the regression line. Other elements, such as
Fe and Na, appear to influence the lithium partitioning as well. For example, the experiment of
Brenan et al. (1998b) labeled 1.5* was conducted at 900oC with 0.5 M aq NaCl, and has run
product clinopyroxene with 1.5 wt. % Al2O3, but resulted in a lower cpx/fluid partition
coefficient than measured in another 900oC experiment from that study containing 0.2 wt. %
Al2O3 and the 900oC experiment containing 0.3 wt. % Al2O3 from this study.
This relationship is consistent with Equation 17 where lithium substitution is coupled with Al3+.
The correlation that is observed here may exist because the total Al2O3 content is correlated with,
or somehow serving as an indicator of the amount of Al3+.
31
2.5.1.2 Olivine
The solution energies calculated by Purton et al. (1997) for forsterite demonstrate that the lowest
energy pairing is a 3+ cation in the M2 site coupled with Li+ in the M1 site. Previous studies have
shown lithium partitioning between olivine and silicate melt to be coupled with Al3+ (Suzuki and
Akaogi, 1995; Taura et al., 1998), suggesting the following mechanism.
(18) Li1+M1 + X3+
M2 (Mg2+, Fe2+)M1 + (Mg2+, Fe2+) M2
Comparison with other experimental data suggests that Fe may be affecting the partitioning of
lithium into olivine. In Figure 2.6, the partition coefficients measured from ol/fluid experiments
of this study, and the ol/melt experiments of Brenan et al. (1998a), Taura et al. (1998), Blundy
and Dalton (2000) and Zanetti et al. (2004) are plotted as a function of temperature and the data
points are labeled with the wt. % FeO in the run product olivine. The olivine-melt partitioning
experiments with low or no FeO plot on the same regression line as those determined in this
study. Experiments with run product olivine containing 8-10 wt. % FeO have higher lithium
partition coefficients than experiments with lower FeO contents in run product olivine, at a given
temperature.
Assuming that at a given fO2, and temperature the total Fe3+ content of olivine scales with total
Fe content, the trend of increasing lithium partitioning in olivine with increasing FeO, suggests
that lithium may be coupling with Fe3+ as substitution mechanism (see Equation 18). The
experiment of Zanetti et al. (2004), labeled 19*, further supports this suggestion. The Zanetti et
al. (2004) experiment was conducted at an fO2 equivalent to QFM-2, which is almost four orders
of magnitude more reducing than the experimental conditions of this study (NNO), and resulted
in a much lower lithium partition coefficient, despite its high (19 wt. %) FeO content. More
reducing experimental conditions in the Zanetti et al. (2004) experiment would result in lower
Fe3+ contents, and therefore less favorable conditions for Li+ substitution, than in the experiments
from this study, despite the fact that the olivine has a high FeO content.
2.5.1.3 Plagioclase
The primary control on lithium partitioning between plagioclase and hydrous fluids is the
composition of the feldspar (Figure 2.7). This is similar to the Sr and Ba partitioning behavior
observed between plagioclase and silicate melts or hydrothermal solutions (Blundy and Wood,
32
1991; Lagache and Dujon, 1987). In the case of Sr or Ba, both these cations are divalent and
based on size and charge balance considerations should be accepted more readily into the
anorthite structure in exchange for Ca2+ rather than Na+ in the albite structure. This apparent
discrepancy is explained by the highly elastic nature of the albite structure (Blundy and Wood,
1991). Albite has a lower bulk modulus and a lower shear modulus than anorthite, which results
in an increased “flexibility” of the albite crystal structure (Angel et al., 1988; Blundy and Wood,
1991). These results suggest that the albite crystal lattice would better accommodate Li+, despite
the fact that Na+ is larger than Ca2+, than the more rigid anorthite structure.
2.5.1.4 Intermineral Partitioning
Previous studies of the lithium content in mantle xenoliths have made a correlation between
lithium contents of olivine and clinopyroxene pairs and xenolith paragenesis, e.g. equilibrated,
metasomatised, etc., (Figure 2.9; Seitz and Woodland, 2000; Paquin and Altherr, 2002;
Woodland et al., 2002; Woodland et al., 2004). Specifically, olivine-clinopyroxene pairs that are
apparently equilibrated (are in chemical equilibration, have no major inhomogeneities or mineral
zoning; Seitz and Woodland, 2000) tend to fall on a linear ~1:1 trendline when lithium
abundance of the olivine is plotted against the lithium abundance of the clinopyroxene. The
xenoliths apparently metasomatised by silicate melt and hydrous fluids fall below the trend line
(depleted in olivine relative to clinopyroxene), and those altered by carbonatite melt fall on and
above the trend line (enriched in olivine relative to clinopyroxene). Figure 2.9 also includes
experimental data corresponding to similar metasomatic regimes. The higher concentrations of
lithium are due to experimental requirements and analytical detection limits. All the
experimental data fall in a relatively restricted range of Dol/cpxLi of ~1 or > 1. A silicate melt
equilibrated olivine-clinopyroxene pair, from Brenan et al. (1998b), plots slightly below the line
projected from the equilibrated mantle xenoliths; and carbonatite melt olivine-clinopyroxene
pairs, from Blundy and Dalton (2000), fall on the projected line, above the silicate melt
experiment. Olivine-clinopyroxene pairs equilibrated with hydrous fluids in this study fall above
the equilibrated mantle xenoliths trend, not below, where the hydrous fluid metasomatised
samples plot. Despite the diversity of experimental methods, the range of Dol/cpxLi exhibited in
natural samples is not reflected in experimental studies, as no experimental studies have shown
Dol/cpxLi
< 1.
33
One possible explanation is that the clinopyroxene compositions resulting from the hydrous fluid
experiments are very low in Al2O3, (0.2 wt. % to 0.3 wt. %); much lower than the Al2O3 content
of clinopyroxene found in mantle most xenoliths (2 – 6 wt. %, Seitz and Woodland, 2000).
Clinopyroxene-fluid partitioning experiments of Brenan et al. (1998b) have shown that lithium
partitioning increases with increasing Al content of the pyroxene. The high Dol/cpxLi values from
this study may be a reflection of the low Al2O3 content of the pyroxene. In the case of
carbonatite melt equilibrated olivine-clinopyroxene pairs from Blundy and Dalton (2000), their
experiments contain similar Al2O3 compositions as those found in carbonatite melt
metasomatised xenoliths, and the discrepancy between experimentally determined Dol/cpxLi,,
(values of ~1 or > 1) and those measured in mantle xenoliths (D < 1) still exists. Recent studies
of lithium and lithium isotopes in mantle xenoliths and other rocks have found that intermineral
partitioning and fractionation of lithium often will not correlate with expected equilibrium
values. These apparent disequilibrium signatures are attributed to remobilization of lithium with
differing rates of diffusion between olivine and clinopyroxene (Parkinson et al. 2007; Rudnick
and Ionov, 2007; Jeffcoate et al., 2007).
2.5.2 Controls on Isotopic Fractionation
Lithium isotopic fractionation between minerals and fluids depends on the difference in the zero
point potential energy (ZPE) between the phases of interest. 7Li is heavier and has a lower
vibrational frequency, and therefore a lower ZPE than 6Li (Chacko et al., 2001). The phase that
will undergo the greatest reduction in ZPE will preferentially take 7Li over 6Li (Chacko et al.,
2001). This has been demonstrated by Ab initio calculations, which have predicted that during
mineral-solution reactions 6Li, is preferentially incorporated into octahedrally coordinated sites
in the solid, and 7Li is preferentially incorporated into the dominantly tetrahedrally coordinated
sites in the fluid (Yamaji et al., 2001).
The coordination state and bonding environment of lithium in both spodumene, a clinopyroxene
with up to 7 % lithium content, and Ca-clinopyroxene, in this case diopside with 6 to 60 ppm
lithium, is octahedral. The consistent fractionation between spodumene and aqueous fluids at 2
GPa (Wunder at al., 2006); clinopyroxene and aqueous fluids at 1 GPa (measured in this study);
and altered seafloor basalts (Chan et al., 1992; Chan et al., 1993) suggests that the coordination
state of lithium in aqueous fluids does not change throughout this range of conditions.
34
2.5.3 Lithium Incorporation into the Mantle
Pristine mid-ocean ridge basalts (MORB) contain 5-6 ppm lithium, with an average δ7Li of +4
‰, and resemble the mantle with respect to lithium content and composition (Jagoutz et al.,
1979; Moriguti and Nakamura, 1998; Tomascak, 2004; Tomascak et al., 2008). The mantle
source that produces MORB is thought to also provide the source for IAB, after modification by
a slab-derived flux. Given that neither partial melting nor differentiation and crystallization will
cause significant fractionation of δ7Li (Tomascak et al., 1999), variations in lithium content and
isotopic composition in some arc lavas are believed to arise from the slab inputs to the melt
source regions. During prograde metamorphism the mineral assemblages in the subducted slab
become more anhydrous with increasing pressure and temperature. Fluids produced by
dehydration reactions in the subducting slab add fluid mobile elements to the overlying mantle
wedge; a signature that is believed to be reflected in the lavas derived from this re-hydrated
mantle. For example, Kamchatka arc lavas are most enriched in boron relative to Nb or Zr at the
arc front and the enrichment decreases to MORB values with increasing slab depth (Ishikawa et
al., 2001). This is suggestive of continuing mobilization of fluid mobile elements into the arc
source region by fluids derived from dehydration reactions in the down going slab (Leeman,
1996; Ishikawa and Tera, 1999; Ishikawa and Nakamura, 1994; Ishikawa et al., 2001). The Izu
arc is one of the few localities where a clear correlation between lithium content, boron content,
δ7Li, and distance to the arc front can be made. The δ7Li in the Izu lavas range from +7.6 ‰ in
the arc front, to +1.1 ‰ in the back arc; this is thought to reflect enrichment of the arc melt
source by fluids derived from the down going slab (Moriguti and Nakamura, 1998). Typically
the lithium isotopic composition of most arc lavas ranges from δ7Li approximately +5 ‰ to +1
‰ and shows a slight negative correlation with Li/Y ratio (Tomascak et al., 2002). It is not clear
why correlations between lithium content and δ7Li of arc lavas are not consistent among all arcs.
Variations in the extent of slab dehydration due to slab age, angle of subduction and overall
thermal regime have been suggested as possible factors (Moriguti et al., 2004). However,
differences in the Li/Y content and 7Li composition between the Kurile arc lavas and the Izu arc
lavas, or even the Japan arc lavas, persist despite similarities in the subduction regime and age of
the slab (Morguti et al., 2004; Tomascak et al., 2002). Another suggestion is that slab-derived
fluids are significantly modified during transport through the mantle wedge to the melt source,
and that the lithium signal is attenuated by interaction with mantle minerals (Tomascak et al.,
35
2002). Subtle differences in the mode of fluid transport through the mantle wedge could lead to
significant differences in the overall behavior of lithium in subduction zones.
2.5.4 The Mantle Wedge as a Chromatograph
The fluids derived from the dehydrating slab during subduction are potentially very lithium rich
depending on the nature of the subducted sediments, in some cases containing as much as 2000
ppm lithium or more (Chan et al., 2002). However, consistent and clear correlations of lithium
with other fluid mobile elements, such as boron, are rare. More commonly, a slab-like lithium
signal cannot be correlated with other indicators of fluid involvement such as B/Be ratios or
depth to slab. For example, some of the calc-alkaline lavas belonging to the Panamanian Old
Group lavas have high B/Be contents, suggesting high fluid input, and MORB-like δ7Li (+4.7 to
+5.6 ‰; Tomascak et al., 2000). Similar behavior is found in other Central American lavas
(Chan et al., 2002), as well as lavas from the Aluetian and Kurile arcs (Tomascak et al., 2002). It
has been suggested that the relatively compatible mineral-fluid partition coefficients for lithium,
the rapid diffusion of lithium into mantle minerals, and the high rock/fluid ratio experienced by
the fluids in the mantle wedge can provide a mechanism by which the lithium signal is decoupled
from other fluid mobile trace elements in slab-derived fluids (Tomascak et al., 2000; Tomascak,
2004; Wunder et al., 2006).
The diffusion coefficients measured in this study provide constraints as to the time required for
fluid-mineral equilibrium. This information, coupled with measurements of lithium partitioning
and isotopic fractionation between fluids and mantle minerals, allows for quantitative modeling
of the interaction between slab-derived fluids and the mantle wedge during fluid transport from
the slab to the arc melt source.
The effect of mantle wedge and fluid interaction can be evaluated following the method of
Navon and Stolper (1987), who modeled the distance traversed by various elements flowing
through an ideal mantle column of fixed porosity. This ideal column contains solid rock with an
interconnected fluid network along grain edges. Assuming that partition coefficients are
constant, the fluid fraction is uniform, and the densities and diffusivities of the solid and fluid do
not vary across the column length then:
36
(19) 0
z
CVX
t
C fff
f
where Cf is the trace element concentration in the fluid, Xf is the mass fraction of the trace
element in the fluid, Vf is the velocity of the fluid, t is time and, z is the distance traversed along
the column. Assuming solid-fluid equilibrium is maintained, the most incompatible trace
elements have fronts that travel farther than trace elements that are more compatible for any
given time. The transport velocity of a trace element relative to the transport velocity of the fluid,
(Vtr/Vfl), is equal to the mass fraction of the trace element in the fluid (Xf):
(20) Xf = f / (f +(1-)sD)
Where is the volume fraction of fluid in the column (assumed to be 0.03 by Navon and Stolper,
1987), f and s are the fluid density and mantle wedge density; (assumed to be 1 g/cm3 and 3
g/cm3 respectively) and, D is the bulk partition coefficient for the element of interest.
A bulk DLi of 0.42 was calculated, assuming 80 % olivine and 20 % clinopyroxene, using the DLi
of olivine-fluid and cpx-fluid measured in this study. The olivine-cpx boron data (from Brenan et
al., 1998a) and cpx-fluid boron (from Brenan et al., 1998b) were combined to estimate the bulk
DB for the same lherzolite assemblage and XfB was also calculated as above for comparison.
From Navon and Stolper (1987) the rate at which a point of constant concentration moves
through the column (Vtr) is:
(21) fftrCf
VXVt
z
So for a given column length, at the time that the fluid front reaches the melt source, the boron
front will be 91 % of the column length, and the lithium front will only be 2 % of the column
length. The maximum capacity of the column for each element can be determined by calculating
at what time the trace element front reaches the top of the column, relative to the time the fluid
reaches the top of the column.
(22) ff
ff
c XVL
VXL
t
t 1
37
where tc is the time for the fluid front to reach the top of the column, t is the time for the trace
element front to reach the top of the column, L is the column length, Vf is the fluid velocity, and
Xf is the mass fraction of the trace element in the fluid. For a given column length and a fluid
velocity, the boron front will reach the melt source at approximately the same time as the fluid
front (1.08tc); however, for the lithium front to reach the top of the column requires the column
to be filled ~42 times. This is a strikingly large volume of fluid, requiring a total of 1.26 cm3 of
fluid for every 1 cm3 of rock for a column with 3 % porosity, which is equivalent to ~40 wt. %
fluid. The highest estimates of fluid involved in arc magmatism from the literature is ~20 wt. %
(Ayers, 1998) and most estimates range from 1-5 wt. % (Stolper and Newman, 1994). Even if the
subducting slab has the capacity to generate such a large quantity of fluid, the time required to
deliver this amount of fluid to the melt zone needs to be considered.
Measurements of U-series disequilibria provide constraints on the timescales of fluid-mobile
element transport from the slab to the melt source (Elliott et al., 1997). Young lavas from
subduction zones often contain an excess of 238U relative to 230Th, or [238U]/[230Th] >1 (Elliott et
al., 1997). Unlike Th, U readily partitions into oxidized fluids, therefore a [238U]/[230Th] ratio >1
is believed to be the result of the addition of a slab derived fluid containing both 238U and 234U
(the parent of 230Th; half-life ( ~250 kyr), to the melt source within the last 30, 000 years
(Elliott et al., 1997). Fluid velocities predicted from U-series disequilibria range from 4 to 10
m/yr. Given these velocities, the time required for the lithium signal to reach the melt source can
be compared with that for the boron signal.
Assuming that partial melting of peridotite occurs at depths of 100 km below intra-oceanic arcs
(Plank et al., 2009) and a Benioff zone of ~125 km, then a maximum column length will be ~25
km. As shown in Figure 2.11, it would take the boron signal 100-5000 years to travel 25 km to
the melt source, whereas the lithium signal will need between 10,000 years and 200,000 years to
reach the melt source. If fluid velocities are 4-10 m/yr then, only the boron signal will reach the
melt source while 238U and 230Th still maintain measurable isotopic disequilibrium. Other studies
have made similar estimates of flux rates using 226Ra-230Th (Sigmarsson et al., 2002), the large
excess of 226Ra over 230Th displayed in many young lavas requires fractionation, presumably due
to fluid transport, to occur within 8 ka, or ~5half-lives (Sigmarsson et al., 2002). These time
constraints give rise to fluid velocities of 10-100 m/yr (Sigmarsson et al., 2002). Only at the
highest estimated fluid velocity, the lithium signal will reach the melt source within 10,000
38
years; this will satisfy the time constraints determined from measurements of U-series
disequilibria and Ra-Th disequilibria (~10,000 yr; Figure 2.11).
More recent studies have suggested that slab dehydration produces a zone of hydrated
lithosphere, mainly consisting of chlorite, which is down dragged by corner flow to depths where
melting may take place in the asthenosphere given the right subduction geometry (Grove et al.,
2009). In this scenario mantle melting occurs 50-100 km below intra-oceanic arcs depending on
the angle of slab-dip and the slab convergence rate. Assuming the Benioff zone is between 100
and 125 km, the minimum column length will be ~10 km. As shown in Figure 2.11, the boron
signal needs only 100-2000 years to travel 10 km, whereas the lithium signal will need 4000-
100,000 years to reach the melt source. Even if the column length is very short, in order for the
lithium signal to reach the melt source within the timescales constrained by U-series and Ra-Th
disequilibria, fluid velocities between 10 m/yr and 100 m/yr are necessary. It should be noted
that lithium partition coefficients for cpx/chlorite have been estimated from lithium
concentrations in the eclogites from Syros, Greece, and are ~1, therefore it is expected that
chlorite/fluid partitioning will be similar to clinopyroxene/fluid partitioning with respect to
lithium (Marschall et al., 2006).
This suggests that for the lithium signal to be correlated with the boron signal, as well as other
fluid mobile elements, extremely high fluid volumes and velocities are required. The rather
improbably high fluid volume and extremely rapid fluid velocity required to transport the lithium
signal from the slab to the melt source is consistent with the lack of a slab-derived lithium signal
in many arc volcanics. The lithium signal will not reach the melt source because it will
preferentially partition into the mantle wedge, relative to other fluid mobile elements (such as
boron).
Given that subducting slabs are unlikely to generate the large fluid volumes required to transport
a lithium signal to the melt source, the occurrence of a slab-like lithium signal in arc lavas
implies a mode of fluid transport other than percolation. The rapid fluid velocities required by
Ra-Th disequilibria have led to the suggestion of fluid transport by hydrofracture rather than
percolation (Davies, 1997). Because fluid transport would be limited to fractures on the scale of
300-1500 m long and 10-200 mm wide (Davies, 1997), the fluid volume needed to generate very
high fluid/rock ratios is more reasonable because the fluids only interact with a small volume of
39
rock. Since lithium partitions preferentially into the mantle relative to other fluid mobile
elements, minimal rock interaction would result in more lithium being transported to the melt
source.
2.5.5 Isotopic Evolution of Lithium-Bearing Fluids in the Mantle
2.5.5.1 Percolation and Rayleigh Distillation
Interaction of lithium bearing fluids with mantle minerals will result in changes to the isotopic
composition of both phases. If the shift in the isotopic composition of the fluid and the mantle
are known, then the amount of fluid: rock interaction can be estimated. The degree of
fractionation resulting from fluid-rock interaction can be modeled assuming a simple Rayleigh
distillation model:
(23) 3)1(377 10)10( fLiLi slabfluidfluid
Where 7Lifluid is the altered fluid, 7Lislabfluid is the initial composition of the slab-derived fluid,
and f is the fraction of the element in the fluid remaining after interaction with the mantle wedge.
In this case is calculated from the degree of cpx-fluid fractionation measured at 1100 oC in this
study and is defined as:
(24) = (7Limin + 1000)/(7Lifluid+1000)
The calculated here is 0.999, which is consistent with the calculated using data from the
study of Wunder et al. (2006). When is < 1, continued interaction of the fluid with mantle
minerals, i.e. distillation, will result in progressively heavier fluids. Figure 2.12 shows how the
isotopic composition of a fluid with an initial δ7Li of +9.7 ‰ (the slab input estimated by
Moriguti and Nakamura, 1998) would change during percolation through a mantle column. The
isotopic composition of the altered fluid increases and is heavier than the initial slab-derived
fluid and both the fore arc and back arc lavas of the Izu arc (δ7Li = +7.6 ‰ and +1.1 ‰,
respectively) for any amount of fluid/rock interaction.
It is important to note that the used in the above calculations is determined for cpx-fluid
fractionation at 1100oC. The temperature of the mantle will be lower near the subducted slab.
Depending on the age of the crust, the rate of subduction, and the degree of frictional shear
40
heating the temperature in the mantle above the slab may be as low as 700-800oC (Peacock,
1993). Because fractionation of stable isotopes tends to increase with decreasing temperature
(Urey, 1947) the fractionation occurring at the base of the mantle column, close to the top of the
slab, may be even larger. Additionally, the degree of fractionation between fluids and the mantle
may be greater than what has been calculated above since the value used in the above
calculation was determined from cpx-fluid fractionation experiments and assumes that the
fractionation of lithium between olivine and fluids is the same. For reference, curves for values
of 0.998 and 0.996 are plotted on Figure 2.12, showing the effects of greater fractionation factors
on the fluid composition. As the fluids percolate through the mantle wedge the fraction of
lithium in the fluid decreases, and the δ7Li of the fluid becomes progressively greater. Because
lithium is readily taken up by mantle minerals, the fraction of lithium remaining in the fluid
becomes very small, Xf → 0.2 (see above), and depending on , the isotopic composition
becomes extremely fractionated with δ7Li ranging from +15 ‰ to +35 ‰ (depending on ;
Figure 2.12).
Because 7Li preferentially fractionates into fluids, interaction of slab fluids with mantle minerals,
i.e. percolation, will generate heavier, more 7Li-rich fluids. A fluid with an initial δ7Li of +9.7 ‰
percolating through the mantle wedge could not generate the isotopic signature observed in the
Izu arc lavas. Either the initial slab fluid is lighter, or the isotopic signature is the result of mixing
the 7Li-rich altered fluid and a lighter mantle reservoir. Interestingly, to generate the isotopic
composition of the Izu fore arc lavas using the Xf as calculated above, requires an initial slab
input with δ7Li = +4 ‰; essentially a fluid with MORB-like δ7Li. For the Izu lavas of the back-
arc region, with δ7Li = +1.1 ‰, it is unlikely that any component of the slab derived fluid has
made its way to the melt source by percolation, as even the least altered slab-derived fluid is
isotopically heavier than the unaltered MORB source mantle. The implication here is that this
isotopically light lithium is not due to percolation of the fluid through the mantle, but must be the
signature of a component derived from an isotopically light reservoir, such as the residual slab or
oceanic sediments (δ7Li ~ -2 ‰; Moriguti and Nakamura, 1998). Another possibility is that this
signal is due to an entirely different mechanism of isotopic fractionation and transport, which is
discussed in the following section.
41
2.5.5.2 Generation of 6Li-rich fluids
Ab initio calculations have demonstrated that during mineral-solution reactions 6Li should be
preferentially incorporated into octahedrally coordinated sites in solid phases (Yamaji et al.,
2001). Recent work by Jahn and Wunder (2009) has examined how lithium speciation in hydrous
fluids affects isotopic fractionation. From Ab initio molecular dynamic (AIMD) calculations,
they have determined that during fluid-solid fractionation, 6Li will prefer sites with the higher
coordination. Lithium in pyroxene and olivine, the most abundant mantle minerals, is in
octahedral or six-fold coordination. When fluid densities are less than 1.0 g/cm3 coordination of
lithium in the fluid is mainly three-fold (Jahn and Wunder, 2009), and therefore 7Li will
preferentially partition into the fluid phase. As fluid density increases, the coordination of
lithium in the fluid also increases. When fluid density is greater than 1.2 g/cm3, the proportion of
5-fold and 6-fold coordinated lithium increases and the proportion of 3-fold and 4-fold
coordinated lithium decreases, and the overall average lithium coordination in the fluid is greater
than 4.5 (Jahn and Wunder, 2009). When lithium coordination in the fluid becomes greater than
lithium coordination in the mineral phase the sense of fractionation changes, and 6Li is predicted
to preferentially partition into the fluid phase. This change in sense of fractionation has been
observed during staurolite-fluid partitioning experiments at 3.5 GPa (Wunder at al., 2007).
Lithium is in tetrahedral coordination in staurolite and preferentially incorporates 7Li at 3.5 GPa.
Therefore, 6Li-rich fluids may be generated by mineral-fluid fractionation at high pressures.
Figure 2.13 is a plot of fluid density vs. temperature, with the average calculated Li-coordination
shown as degree of shading. Superimposed on this plot are the fluid densities calculated with the
CORK-EOS (Holland and Powell, 1991) using the Perple_X computer program (Connolly,
2005) for Franciscan and Alpine subduction zones (Ernst, 1988) as well as the most direct path
between the slab and a fore-arc volcano (Peacock, 1993). It is possible to generate 6Li-rich fluids
when mineral-fluid interaction occurs at depths greater than ~125 km. These results suggest that
as fluids percolate up through the mantle wedge, fluid density decreases and the average
coordination of lithium in the fluids will decrease; 6Li will once again preferentially fractionate
into the mineral phase and the fluids will become heavier. In order to preserve the 6Li-rich signal
generated at depth, fluids need to reach the melt source having undergone minimal interaction
with the mantle on their ascent path. Fluid transport by hydrofracture would satisfy this
42
requirement, as it results in high fluid-rock ratios, which would transport the 6Li-rich fluids to the
melt source quickly, and with least amount of interaction with the mantle wedge.
2.5.5.3 Generation of 6Li-rich zones in the mantle
Hyrdofracture of the mantle by slab-derived fluids is an appealing mechanism to transport
lithium through the mantle, as channelized flow through hydrofractures would satisfy the high
fluid: rock ratios and rapid fluid velocities required by mineral-fluid partitioning. This transport
mechanism would also minimize isotopic fractionation by limiting mineral-fluid interaction,
thereby effectively propagating a slab signal all the way to the melt source region. The isotopic
composition of the mantle wall rock of the fractures would also shift, generating a local
isotopically light region in the mantle wedge. The isotopic shift of the wall rock depends on the
extent of reaction between the mantle and the slab-derived fluids. If mineral-fluid exchange is
fast, then local mineral-fluid isotopic equilibrium will occur, and the isotopic shift will depend
on the fluid-rock ratio.
This process can be modeled after the approach of Abart (1995; after Taylor 1977) by calculating
the progress of the reaction, which is defined as the ratio between the observed isotopic shift
in the rock and the maximum attainable isotopic shift:
(25) RF
RRiLiFR
iLi
iLi
fLi
In this case, LifR is the final isotopic composition of the mantle, Li
iR is the initial composition
of the mantle, LifF is the composition of the metasomatizing fluid and R-F is the fractionation
between the mantle and the fluid. The value of will be between 0 (no equilibration) and 1
(complete equilibration). Where mineral-fluid exchange is rapid, as is the case for lithium
exchange, then the degree of equilibration depends on the lithium atom equivalent fluid-rock
ratio, N (Taylor, 1977):
(26) N = - ln(1 - )
The final isotopic composition of the mantle wall rock is estimated here given an initial mantle
composition of δ7Li = +4 ‰ (Moriguti and Nakamura, 1998; Tomascak et al., 2008; Nishio et
al., 2002). This model assumes complete equilibration between the mantle wall rock and fluid
43
will result in the mantle wall rock having a δ7Li‰ lower than the initial fluid composition
(cpx-fluid = -1 ‰ at 1100oC; this study, Wunder et al., 2005). Because the isotopic composition of
the initial slab fluid is not very well constrained, three different initial fluid compositions are
used in this illustration; δ7Li = +10 ‰ (estimate for the Izu arc fluids by Moriguti and
Nakamura; 1998), 0 ‰ and -10 ‰, the latter being arbitrary values reflecting generation of 6Li-
rich fluids at depth (Jahn and Wunder; 2009). Given a DLibulk = 0.42, N becomes (1/ DLi
bulk)*N in
weight units. Assuming the isotopic composition of the fluid does not change, which is the case
if the fluid/rock ratio is high, the isotopic composition of the vein/hydrofracture wall rock will
approach complete equilibrium with the fluid; equal to R-F of -1 ‰. Because the timescales of
lithium diffusion in olivine and clinopyroxene are rapid compared to the fluid transport times (2
m/1hr; see Ch.3, vs. 100 m/yr; Sigmarsson et al., 2002) the isotopic shift in the wall rock is a
function of the amount of fluid available.
Figure 2.14 shows the evolution of the final isotopic composition of the metasomatized mantle
wall rock with increasing fluid/rock ratio. The isotopic shift of the wall rock depends on the
extent of reaction between the mantle and the slab-derived fluids; if the fluid/rock ratio is high,
the isotopic composition of the vein/hydrofracture wall rock will approach complete equilibrium
with the fluid equal to R-F of -1 ‰. Also plotted are the δ7Li values of the Izu fore arc and back
arc lavas (which are typical of the range of δ7Li values found in many arc lavas; Tomascak et al.,
2002). The entire range of δ7Li values found in arc lavas can be achieved by metasomatizing the
mantle with fluids that have δ7Li between 0 ‰ and +10 ‰ and fluid/rock ratios >1.2. Both these
values are reasonable given the isotopic composition of subducted material and typical fluid/rock
ratios for hydrofractured zones. Values of δ7Li in seafloor sediments range from -5 ‰ to +20 ‰
(Marschall et al., 2007 and references therein) and altered oceanic crust has δ7Li of ~ +14 ‰
(Moriguti and Nakamura, 1998; Chan et al., 1992). Slab derived fluids with δ7Li between 0 ‰
and +10 ‰ could be achieved during dehydration of the slab, recalling that isotopic fractionation
in a cool slab at depths greater than ~125 km are likely to produce fluids that are isotopically
lighter than the solid (Jahn and Wunder, 2009). The fluid/rock ratios in vein systems can be very
high with typical values from calcite or quartz vein systems ranging from 70-100 cm3 of fluid per
100 cm3 of rock to as much as 1400 cm3 of fluid per 100 cm3 of rock (Spear, 1993).
44
2.6 Conclusions
Lithium is moderately incompatible in the mantle during mineral – fluid exchange reactions. The
DLi measured in this study ranges from 1.34 – 0.14 in olivine, to 0.32 – 0.09 in plagioclase and,
0.32 – 0.07 in clinopyroxene. Lithium partitioning between clinopyroxene and hydrous fluids is a
function of temperature, decreasing with increasing temperature from 800oC to 1100oC at 1 GPa
and appears to increase with increasing Al2O3 content of the pyroxene. Olivine-fluid partitioning
of lithium is not a function of temperature, but appears to be sensitive to Mg/Fe content, although
this needs to be investigated more systematically. Lithium partitioning in anorthite is a function
of feldspar composition, similar to the partitioning of other cations in the feldspar-fluid system.
Lithium partitioning between olivine and clinopyroxene appears to be independent of
temperature; however, preliminary experiments examining the effect of REE content and fO2
suggest that DLiol/cpx may be a function of crystal chemistry. Isotopic fractionation between
clinopyroxene and fluid has been measured as well as between olivine and clinopyroxene. The
isotopic fractionation between clinopyroxene and fluid at 900oC is ~ +1 ‰ (±2 ‰) and the
measured isotopic exchange between olivine and clinopyroxene is ~ +5 ‰ (±4 ‰). Isotopic
fractionation between clinopyroxene and fluids is a function of temperature and consistent with
what has been observed in the spodumene – fluid system. The fractionation between spodumene
and hydrous fluids results in an enrichment of 7Li in the fluid from +3.5 ‰ at 500oC to ~ +1.0 ‰
at 900oC and 2.0 GPa (Wunder et al., 2006).
Application of these data to models of fluid-rock interaction in the mantle wedge reveals that
lithium is a moderately incompatible element in the mantle during mineral-fluid exchange
reactions. Because lithium is not a conservative element, it cannot be used to deconvolve the
proportions of slab-derived fluid and altered and unaltered MORB-source involved in generating
arc lavas. However, constraining how lithium behaves in the mantle provides some insight into
the lithium and lithium isotopic trends, or lack thereof, observed in arc lavas. The absence of
high Li/Y ratios in arc lavas with high B/Be, or MORB-like δ7Li in lavas with high B/Be
contents (such as the lavas from the Sunda arc, Indonesia; Tomascak et al., 2002), can be
explained by partitioning lithium into mantle minerals as fluids percolate through the mantle
wedge. In these cases, transport through the mantle wedge completely removed the lithium
signal from the slab-derived fluid. Convergent margin lavas, such as the Izu forearc lavas, with
45
δ7Li values greater than the mantle values (δ7Li ~ +4 ‰) are likely the result of some component
of slab fluid-mantle interaction during percolation. It is important to note that very high fluid
fluxes are implied if a 7Li signal from slab-derived fluids is to reach the melt source by
percolation.
Low δ7Li values (< MORB; δ7Li ~ +4 ‰) that correspond with high Li/Y ratios are likely
generated near the slab and transported to the melt source with a minimum amount of interaction
with the mantle wedge; here transport through hyrdofractures is a likely mechanism. The trend of
increasing δ7Li with decreasing Li/Y, which is observed in most arc lavas (Tomascak et al.,
2002), could be viewed as a spectrum between the two scenarios. Where low Li/Y values
correspond with high δ7Li, large fluid fluxes were most likely percolating through the mantle
wedge. Where high Li/Y values correspond with low δ7Li, the fluids were likely generated at
depth and transported through the mantle through hydrofractures, having minimal interaction
with the wedge. Intermediate values could be a result of some component of both these
mechanisms.
Transport of slab-derived fluids through hydrofractures in the mantle can also explain the lack of
clear and consistent correlations between lithium and other fluid mobile elements. Fluids
transported to the melt source through hydrofractures would be subject to differing degrees of
mantle interaction (variable fluid/rock ratios and transport velocities). Lithium is moderately
compatible in the mantle and diffuses rapidly; therefore, lithium contents and isotopic
compositions will be very sensitive to variations in the types of mineral-fluid interaction.
The lithium isotopic evolution of the mantle will also be affected by these processes, as it is such
an efficient sink for lithium. Dehydration reactions in the subducting slab at depths less than
~125 km, where fluid density is relatively low, and the predicted predominance of three-fold and
four-fold coordinated lithium in the fluid will generate 7Li-rich fluids and result in localized 6Li
enrichment of the mantle. Hydrous fluids generated deeper than ~125 km are predicted to contain
lithium in coordination states greater than four-fold, and therefore likely to be enriched in 6Li, at
least initially, giving rise to a zone of 7Li-rich mantle at depth. Xenoliths with δ7Li values
greater than MORB are uncommon, but have been found in blueschists from Syros (Greece),
eclogites from Dabishan (China), Cima di Gagnone and Trescolmen (Alps) and lherzolites from
Northern Japan and SE Austrailia (Nishio et al., 2004; Marschall et al., 2007).
46
Table 2.1 Composition of Starting Material
Dekalb Diopside San Carlos Olivine Crystal Bay Bytownite
Mt St. Hillaire Albite
SiO2 54.77 (0.80)1 40.95 (0.02) 49.16 (0.20) 67.88 (0.26)
Al2O3 0.66 (0.10) <0.01 32.67 (0.26) 19.61 (0.20)
FeO 0.85 (0.08) 9.31 (0.05) 0.50 (0.06) <0.03
MgO 17.31 (0.22) 49.19 (0.42) 0.13 (0.04) <0.02
CaO 25.17 (0.26) <0.02 15.09 (0.28) <0.02
Na2O 0.43 (0.08) <0.02 2.65 (0.12) 11.18 (0.46)
MnO 0.05 (0.06) 0.12 (0.02) <0.03 <0.03
NiO <0.03 0.39 (0.03) <0.03 <0.03
Total 99.29 100.86 100.27 98.75
n 11 3 8 4
Li ppm2 8.86 (0.60) 2.52 (0.46) 1.65 (0.08) <0.14
7Li (‰) +9.7 (1)3 +3.64 (0.15)4 1Numbers in parentheses represent 2 of the mean of n analyses 2Analyzed by LA-ICPMS, numbers in parentheses represent 2 of the mean of 5 analyses for diopside, 9 analyses for olivine and 2 analyses for feldspars 3Analyzed by MCICP-MS, numbers in parentheses represent 2 of the uncertainty on the measurement 4Magna et al., 2006
Table 2.2 Summary of Experimental Details
#3 0.2 800 74 Au DD 14.1 3.9 - - 48.3 A cpx + quench xtlsNCDL2-2 1 800 139 Pt DD 10.66 3.9 - - 57.09 A clear cpx + orange cpxNCDL4 1 1000 67 Pt DD 9.67 3.9 - - 49.46 A clear cpx + orange cpxNCDL5 1 1100 72 Pt DD 10.27 3.9 - - 36.02 A clear cpxNCDL6 1 900 72 Pt DD 6.65 3.9 - - 36.62 A clear cpx + orange cpxNCDLR 1 800 69 Pt NCDL1 9.20 - - - 21.35 B fine grained clear cpx + ? small xtls + quench DiAb10 1 800 66 Pt DD 6.82 - 9.5 - 42.01 A clear cpx + orange cpx
NCOL1 1 1000 72 Pt SCO 5.27 3.3 - 3.2 42.04 A clear olivine + pink/red/black oxidesNCOL2 1 900 68 Pt SCO 10.87 3.3 - 3.2 50.48 A clear olivine + pink/red/black oxidesNCOL3 1 1100 70 Pt SCO 7.83 3.3 - 3.2 37.21 A clear olivineLSCO8 1 1200 72 Pt SCO 12.63 3.3 - 3.2 57.92 C clear olivineNCOLR 1 800 67 Pt NCOL2 5.03 - - - 35.48 B clear olivine + ? Mg phase + pink/red/black oxidesOlAb5 1 900 66 Pt SCO 5.12 - 4.5 - 40.42 A clear olivine
NCA1 1 1000 77 Pt CBBy 11.31 3.1 - 2.6 45.28 A melt + large clear anorthiteNCA2 1 900 69 Pt CBBy 7.95 3.1 - 2.6 45.55 A clear anorthiteNCA3 1 800 70 Pt CBBy 18.36 3.1 - 2.6 46.20 A asicular green xtls (amphibole?) + clear anorthiteNCA5 1 800 48 Pt CBBy 9.34 3.1 - 2.6 28.69 A asicular green xtls (amphibole?) + clear anorthiteNCAR 1 800 72 Pt NCA3 5.92 - - - 34.15 B zoesite
AnAb10 1 800 68 Pt CBBy 5.49 - 12.6 - 46.36 A melt/quench + clear anorthiteAnAb20 1 800 55 Pt CBBy 7.68 - 19.3 - 40.93 A melt/quench + clear anorthite
2m-hi4 1 1000 48 Pt+Re DD + SCO3 6.60 - - - 25.45 A melt/quench + clear enstatite2m-lo5 1 1000 48 Pt+Mo DD + SCO3 5.73 - - - 29.66 A blue/grey cpx and ol + black oxides + CaMo oxideYb-16 1 1000 72 Pt DD + SCO3 4.80 - - - 30.19 A clear cpx and ol + black oxides
NCDL3 1 900 71 Pt DD 9.47 3.9 - - 48.03 A clear olivine + clear cpx + orange cpx 3
capsule solid (mg)+ wt% SiO2
+ wt% Abstarting
material1fluid (mg) run productsfluid2sample P (GPa) T (oC) t (hrs)
+ wt% Al2O3
2m-1 1 900 72 Pt DD + SCO3 5.33 - - - 40.34 A clear cpx + orange cpx, ol absent2m-2 1 800 68 Pt DD + SCO3 9.04 - - - 48.01 A clear olivine + clear cpx + orange cpx 2m-3 1 1000 72 Pt DD + SCO3 8.47 - - - 47.01 A clear olivine + clear cpx 2m-R 1 800 72 Pt 2m-2 5.57 - - - 39.63 B clear ol + cpx + monticellite + black oxides
LDi-107 1 900 142 Pt DD 16.3 3.9 - - 86.7 C clear cpx + orange cpxLDi-117 1 900 72 Pt DD 5.1 3.9 - - 52.4 C clear cpx + orange cpxLDi-127 1 900 68 Pt LDi-10 4.89 13 - - 66.35 D clear cpx + orange cpxLDi-15 1 900 75 Pt DD 6.86 3.9 - - 63.63 C clear cpxLDi-17 1 1000 20 Pt DD 9.25 3.9 - - 59.82 C clear cpx + black oxidesLDi-18 1 1100 70 Pt DD 9.66 3.9 - - 31.13 C clear cpx
1 DD: Dekalb Diopside, SCO: San Carlos Olivine, CBBy: Crystal Bay Bytownite2 Fluid compositions; A: 243ppm Li, B: 96ppm Li, C: 306ppm Li, D: 180ppm Li3 82wt% DD + 18wt% SCO 4 Capsule materials result in log f O2 of -55 Capsule materials result in log f O2 of -156 0.25 mg of Yb2O3 added7 Ti outer capsule
48
Table 2.3 Standards and Reference Material
Li (ppm)1 reference
International Standards
NBS 612 41.54 (2.87) Pearce et al. 1997
NBS 612 41.5 (2.2) this study2, LA-ICPMS
NBS 610 484.6 (21.7) Pearce et al. 1997
NBS 610 488.7 (39.6) this study3, LA-ICPMS
JG1a 79.5 (4.5) Imai et al. 1995
JG1a 92.4 (6.4) this study, LA-ICPMS
JB-2 7.78 (1.39) Imai et al. 1995
JB-2 7.9 (0.6) this study, LA-ICPMS
BCA1 13.3 Ryan and Langmuir, 1987
BCA1 12.3 (0.8) this study, LA-ICPMS
JGB-1 4.59 (.90) Imai et al. 1995
JGB-1 4.6 (0.4) this study, LA-ICPMS
In house Standards
San Carlos Olivine 2.52 (0.4) this study, LA-ICPMS
San Carlos Olivine 1.6 (0.08) Magna et al. 2006, MC-ICPMS
Dekalb Diopside 8.9 (0.6) this study, LA-ICPMS
Dekalb Diopside 7.8 (2) this study, MC-ICPMS
Kunlun Diopside 42.6 (3.) this study, LA-ICPMS
7Li(‰) reference
International Standards
IRMM016 +0.1 (1) this study, MC-ICPMS
IRMM016 -0.1 (1) Teng et al. 2004
IRMM016 +0.13 (1) Jeffcoate et a. 2004
In house Standards
UMD-1 54.8 (1) this study, MC-ICPMS
UMD-1 54.7 (1) Teng et al. 2004
San Carlos Olivine +3.64 (0.2) Magna et al. 2006, MC-ICPMS
San Carlos Olivine +1.1 (4) this study, SIMS corrected4
Dekalb Diopside +8.5 (4) this study, SIMS corrected4
Dekalb Diopside +9.7 (1) this study, MC-ICPMS
1) Numbers in parentheses represent 2 errors
2) Analysed using NIST 610 as standard unless otherwise noted
3) Analysed using NIST 612 as standard
4) Values are corrected for instrument mass fractionation (see text)
Table 2.4 Run Product Major Element Composition
Sample Total n ppm Yb Di% Jd% Tsc% Fo% An%
Single Phase ExpClinopyroxene
#3 55.87 (0.12) 0.59 (0.12) 0.82 (0.11) 0.05 (0.02) 17.94 (0.20) 25.06 (0.16) 0.37 (0.07) na 100.70 16 94 3 0NCDL2-2 54.72 (0.29) 0.22 (0.10) 0.57 (0.21) 0.06 (0.03) 18.26 (0.17) 25.26 (0.26) 0.02 (0.01) nd 99.11 16 97 0 1NCDL4 54.84 (0.30) 0.23 (0.06) 0.43 (0.18) 0.05 (0.03) 17.98 (0.16) 25.82 (0.13) 0.02 (0.02) nd 99.37 14 99 0 1NCDL5 55.02 (0.30) 0.21 (0.06) 0.05 (0.09) 0.03 (0.02) 18.36 (0.16) 25.77 (0.24) 0.02 (0.01) nd 99.46 11 100 0 0NCDL6 54.77 (0.23) 0.32 (0.11) 0.78 (0.27) 0.05 (0.01) 17.87 (0.18) 25.38 (0.12) nd nd 99.17 5 95 1 1NCDLR 54.99 (0.23) 0.47 (0.13) 0.76 (0.23) 0.05 (0.03) 17.48 (0.26) 25.41 (0.24) 0.27 (0.17) 0.11 (0.16) 99.54 14 96 2 0DiAb10 54.59 (0.43) 0.32 (0.09) 1.79 (0.44) 0.04 (0.03) 17.89 (0.27) 25.28 (0.11) nd 0.04 (0.03) 99.95 8 96 0 2
OlivineNCOL1 42.51 (0.22) 0.04 (0.01) 1.12 (0.10) 0.10 (0.04) 55.59 (0.18) 0.03 (0.01) nd 0.32 (0.05) 99.72 9 99NCOL2 42.45 (0.22) 0.04 (0.01) 2.07 (0.24) 0.13 (0.03) 54.74 (0.31) 0.04 (0.01) nd 0.33 (0.11) 99.80 9 98NCOL3 43.74 (0.27) 0.07 (0.01) 0.06 (0.02) 0.05 (0.03) 56.72 (0.20) 0.06 (0.01) nd 0.03 (0.03) 100.73 9 100LSCO8 42.51 (0.14) 0.09 (0.02) 1.32 (0.63) 0.16 (0.10) 56.61 (0.64) 0.04 (0.00) nd 0.31 (0.39) 101.04 7 99NCOR 41.81 (0.42) 0.03 (0.01) 0.12 (0.03) 0.09 (0.03) 53.55 (0.70) 0.01 (0.01) nd 4.53 (0.70) 100.14 13 100OlAb5 42.62 (0.05) 0.02 (0.01) 0.83 (0.11) 0.12 (0.02) 55.73 (0.10) 0.09 (0.01) nd 0.36 (0.02) 99.77 5 99
PlagioclaseNCA1 42.99 (0.14) 37.36 (0.15) 0.03 (0.03) nd nd 19.77 (0.14) 0.15 (0.03) na 100.30 22 98NCA2 43.04 (0.13) 37.19 (0.15) 0.22 (0.03) nd nd 19.78 (0.11) 0.15 (0.04) na 100.38 14 99NCA3 44.05 (0.41) 36.32 (0.31) 0.25 (0.03) nd nd 18.81 (0.29) 0.66 (0.17) na 100.09 23 94NCA5 44.57 (0.26) 35.66 (0.22) 0.06 (0.05) 0.03 (0.02) nd 18.23 (0.22) 0.97 (0.09) 0.03 (0.02) 99.55 8 91
NCAR - zoesite 39.53 (0.44) 34.18 (0.12) 0.24 (0.13) nd 0.55 (0.10) 24.38 (0.24) nd na 98.88 20AnAb10 43.09 (0.01) 36.66 (0.10) 0.14 (0.04) 0.02 (0.01) 0.01 (0.01) 19.23 (0.16) 0.36 (0.08) nd 99.52 4 97AnAb20 43.99 (0.26) 35.87 (0.24) 0.06 (0.06) nd 0.02 (0.01) 19.15 (0.18) 0.41 (0.03) nd 99.50 10 93
NCA1 melt 43.47 (0.99) 26.14 (0.36) 0.12 (0.03) nd 1.16 (0.06) 14.78 (0.54) 0.26 (0.05) na 85.93 10NCA2 melt 60.36 (2.59) 21.64 (1.46) 0.61 (0.09) nd 0.40 (0.10) 2.28 (0.48) 0.86 (0.31) na 86.15 4
Two Phase ExpNCDL3 olivine 43.97 (0.21) 0.03 (0.01) 0.17 (0.04) 0.11 (0.02) 54.95 (0.17) 0.96 (0.25) nd nd 100.19 6 100
NCDL3 cpx 54.20 (0.38) 0.33 (0.08) 0.65 (0.13) nd 17.91 (0.13) 25.93 (0.21) nd nd 99.02 9 100 2
CaO Na2O NiOSiO2 Al2O3 FeO MnO MgO
p ( ) ( ) ( ) ( ) ( )2m-hi opx 59.23 (0.13) 0.14 (0.06) 1.92 (0.09) 0.05 (0.02) 38.52 (0.23) 0.09 (0.01) nd 0.03 (0.04) 99.99 9
2m-lo olivine 42.56 (0.41) 0.04 (0.01) 0.12 (0.10) 0.14 (0.05) 57.05 (0.60) 0.11 (0.06) nd 0.01 (0.02) 100.03 9 1002m-lo cpx 56.04 (0.30) 0.53 (0.09) nd 0.07 (0.01) 21.06 (0.49) 22.71 (0.65) nd nd 100.41 8 85 1 1
Yb-1 olivine 42.60 (0.41) 0.05 (0.04) 0.25 (0.07) 0.16 (0.04) 55.47 (1.20) 1.11 (0.28) nd 0.04 (0.02) 99.67 13 2868 100Yb-1 cpx 55.73 (0.22) 0.20 (0.05) 0.13 (0.06) 0.03 (0.02) 18.69 (0.16) 25.84 (0.11) nd 0.16 (0.13) 100.78 12 1441 98 0 12m-1 cpx 54.40 (0.34) 0.52 (0.11) 1.30 (0.37) 0.05 (0.03) 17.87 (0.22) 25.46 (0.26) nd 0.03 (0.02) 99.63 6 97
2m-2 olivine 42.67 (0.30) 0.04 (0.01) 0.02 (0.01) 0.10 (0.06) 42.67 (0.30) 0.26 (0.18) nd nd 85.77 4 1002m-2 cpx 55.30 (0.34) 0.24 (0.13) 0.36 (0.49) 0.05 (0.02) 18.40 (0.33) 25.47 (0.16) nd 0.02 (0.02) 99.84 5 97 1 0
2m-3 olivine 42.80 (0.14) nd nd 0.04 (0.02) 55.96 (0.17) 0.95 (0.04) nd nd 99.75 9 1002m-3 cpx 55.35 (0.14) 0.23 (0.04) 0.21 (0.10) nd 18.50 (0.12) 25.59 (0.10) nd nd 99.87 10 98 1 0
2m-R monticillite 38.53 (0.10) 0.03 (0.02) 0.15 (0.02) 0.12 (0.04) 26.23 (0.25) 34.66 (0.27) nd 0.07 (0.04) 99.80 8 1002m-R cpx 54.80 (0.18) 0.29 (0.04) 0.93 (0.09) 0.02 (0.02) 18.04 (0.07) 25.67 (0.13) nd 0.02 (0.02) 99.76 9 98 0 2
2m-R olivine 42.40 (0.14) 0.04 0.02 0.35 (0.11) 0.12 (0.03) 55.55 (0.41) 0.91 (0.13) nd 0.16 (0.05) 99.52 4 100
50
Table 2.5 Run Product Lithium Concentration
sample Li ppm 21 n min/fluid 2 ol/cpx 2 min/melt 2
#3 13.66 3.12 2 0.07 0.01
NCDL2-2 97.87 32.12 2 0.32 0.14
NCDL4 33.20 13.96 3 0.14 0.06
NCDL5 26.64 12.04 3 0.11 0.05
NCDL6 61.96 32.29 2 0.27 0.14
NCDLR 6.67 1.66 3 0.07 0.02
DiAb10 49.92 0.61 2 0.21 0.03
NCOL1 101.36 10.18 3 0.47 0.03
NCOL2 124.3 13.32 4 0.57 0.06
NCOL3 85.6 35.00 2 0.38 0.16
LSCO8 49.20 12.22 3 0.17 0.04
NCOLR 13.37 1.84 3 0.14 0.02
OlAb5 278.90 22.66 3 1.34 0.02
NCA1 20.92 7.35 3 0.09 0.03
NCA2 20.89 2.26 2 0.09 0.02
NCA3 38.57 17.65 2 0.17 0.08
NCA5 70.21 33.57 3 0.32 0.15
NCAR* -zoisite 0.34 0.36 2
AnAb10 31.15 11.12 2 0.13 0.11
AnAb20 50.29 19.85 4 0.21 0.08
NCA1-melt 1119.93 223.99 1 0.019 0.008
NCA2-melt 1197.00 239.40 1 0.017 0.004
NCDL3 - olivine 66.27 8.10 3 0.30 0.02
NCDL3 - cpx 9.78 1.82 3 0.04 0.06 6.78 1.51
2m-1 cpx 13.56 2.83 2 0.06 0.01
2m-2 olivine 78.04 6.80 1
2m-2 cpx 50.22 4.03 2 1.55 0.18
2m-3 olivine 51.11 11.04 3
2m-3 cpx 12.79 7.07 2 4.00 2.37
2m-R olivine 31.04 2.14 1
2m-R cpx 25.84 5.16 1 1.20 0.25
2m-R monticillite 39.27 23.62 2
2m-hi enstatite 5.28 3.18 3 0.02 0.01
2m-lo olivine 29.21 8.90 2
2m-lo cpx 41.63 20.51 3 0.70 0.41
Yb-1 olivine 174.45 14.70 1
Yb-1 cpx 16.82 6.22 3 10.37 3.93
1) 2 refers to the standard deviation for n analyses and reflects the degree of sample heterogeneity
51
Table 2.6 Isotopic Composition of Starting Materials and Run Products
Starting Material Products Exp T (oC) fluid mineral mineral
7Li (‰) 7Li (‰) 7Li (‰) 2(‰) 7Licpx-fluid
(‰) 7Liol-cpx
(‰)
MC-ICPMS data (UMD)
cpx-fluid experiments using 6Li doped solution
LDi-10 900 -88.4 +9.7 -90.9 1 -2.5 ±1.4
LDi-11 900 -88.4 +9.7 -91.4 1 -3.0 ±1.4
LDi-12* 900 -46.1 -90.9 -49.5 1 -3.4 ±1.4
LDi-15 900 -88.4 +9.7 -89.1 1 -0.7 ±1.4
LDi-17 1000 -88.4 +9.7 -89.5 1 -1.1 ±1.4
LDi-18 1100 -88.4 +9.7 -88.7 1 -0.3 ±1.4
cpx-fluid experiments using LSVEC solution
NCDL-6 900 0 +9.7 -2.6 1 -2.6 ±1.4
SIMS data (LLNL)
cpx-fluid experiments using LSVEC solution
NCDL-2 800 0 +9.7 -2.1 4 -2.1 ± 4.1
NCDL-4 1000 0 +9.7 -1.1 4 -1.1 ± 4.1
NCDL-5 1100 0 +9.7 -2.9 4 -2.9 ± 4.1
NCDL-6 900 0 +9.7 -3.5 4 -3.5± 4.1
Diopside #3 800/0.2GPa 0 +9.7 -5.3 4 -5.3 ± 4.1
olivine-cpx experiments using LSVEC solution
clinopyroxene
2m-2 cpx 800 0 +9.7 -4.2 4
2m-3 cpx 1000 0 +9.7 -6.5 4
2m-R cpx* 800 -2.7 -4.2 -4.4 4
olivine
2m-2 ol 800 0 +1.0 1.5 4 +5.7 ± 5.6
2m-3 ol 1000 0 +1.0 -0.7 4 +5.9 ± 5.6
2m-R ol* 800 -2.7 +1.5 -7.6 4 -3.2 ± 5.6
2m-R mtc* 800 -2.7 +1.0 -14.8 4
* denotes reversal experiment
52
-8.0
-4.0
0.0
4.0
8.0
12
16
10 15 20 25 30 35 40
y = -16.287 + 0.80082x R2= 0.94585
7 Li m
easu
red
by
MC
-IC
PM
S (
o/o
o)
7Li 'uncorrected' measured by SIMS ( o/oo)
SCO
Dekalb
NCDL 6
Figure 2.1 Internal Reference Materials
Plot of δ7Li values measured by MC-ICPMS versus uncorrected δ7Li values measured by SIMS of the internal reference materials, Dekalb diopside and San Carlos olivine, as well as the run product from one experiment (NCDL6). The δ7Li values measured by MC-ICPMS for Dekalb diopside and NCDL6 are from this study and SCO is measured by MC-ICPMS from Magna et al. (2006). Error bars for ‘uncorrected’ SIMS δ7Li are 2, based on counting statistics. The error bars for δ7Li measured by MC-ICPMS are ±1 ‰ (2) or the published 2 errors (±0.3 ‰ ; Magna et al., 2006). Note that the discrepancy between values is due to instrument mass fractionation (Decitre et al. 2002). All values plot on a single line with a slope of ~1, suggesting the absence of any significant matrix effect on the lithium instrumental isotopic fractionation (see text).
53
1
10
100
1000
1 10 100 1000
Standard Reference Material
NBS 610NBS 612DeklabJG1aJB-2BCR1JGB-1SCO
Lith
ium
(pp
m)
Pub
lish
ed V
alue
s
Lithium (ppm)(LA-ICPMS, this study)
Figure 2.2 Standards and Reference Material
Lithium abundance of various standard reference materials (values taken from literature; see text) and internal reference material (SCO and Dekalb; measured by MC-ICPMS) plotted against lithium abundance determined in this study by LA-ICP-MS. NBS 610 was used as the standard reference material (SRM) for all analyses. The value for NBS 610 was determined using NBS 612 as the SRM. Error bars for lithium concentrations from this study are 2, based on standard deviation of replicate measurements (typically 5 or more analyses for each).
54
Figure 2.3 Photomicrographs of Starting Material and Run Products
(a) Starting material San Carlos olivine mounted in oil, (b) starting material Dekalb diopside mounted in oil, (c) run product olivine from NCOL3 mounted in epoxy, (d) run product diopside from NCDL 4 mounted in epoxy and (e) NCAL 2 mounted in epoxy. LA-ICPMS pits in run product crystals are 50m in diameter. All photomicrographs are taken in plane polarized light at 100x magnification.
55
100
101
102
103
104
105
106
107
0 10 20 30 40 50 60 70 80
2m-2, 800oC, 68 hrsco
unts
per
sec
ond
Time (seconds)
1.8 x105ppm 43Ca
50 ppm 7Li
laser off laser on
Figure 2.4 Time Resolved Spectra
Example of time resolved spectra for an individual clinopyroxene crystal from a cpx/fluid partitioning experiment at 800oC for 68 hrs. The first ~20 seconds of the analysis was done with the laser shutter in place for background measurements, followed by ~60 seconds of sample ablation. Note that the 7Li signal is consistent with respect to the 43Ca signal, which is an indication of homogeneity, confirming mineral-fluid equilibrium of both major and trace elements.
56
-6
-5
-4
-3
-2
-1
0
1
0.5 0.6 0.7 0.8 0.9 1
Hart & DunnBrenan 1998bBrenan 1998aBlundy & DaltonBlundy 1998this study
ln D
Li
1000/T
1200 1000 80014001600ToC
0.3
0.2
5
10
14
1.5*
0.5
8
3
1.3-1.6
Figure 2.5 lnDLi cpx/fluid vs 1000/T
A plot of ln DLicpx/fluid as a function of 1000/T for cpx/fluid partitioning measured in this study, demonstrating the temperature dependence of lithium partitioning between clinopyroxene and hydrous fluids. A linear regression of the data yields: ln DLicpx/fluid = -7.38 + 7.04 * 1000/T (R2 = 0.98) where T is temperature in Kelvins. Neither NCDL1, diopside #3 (unequilibrated samples) nor NCDLR (reversal) were used in the regression. Also shown are the data from the experiments of Hart and Dunn (1993), Brenan et al. (1998a, 1998b), Blundy and Dalton (2000). The data points are labeled with the wt. % Al2O3 content of the run product clinopyroxene and suggest that lithium may be coupled with Al3+ as a substitution mechanism in clinopyroxene. Error bars for the partition coefficients measured in this study are 2, based on the standard deviation of n analyses (see Table 2.5).
57
-3
-2.5
-2
-1.5
-1
-0.5
0
0.4 0.5 0.6 0.7 0.8 0.9
this studyBrenan 1998aTaura 1998Blundy & DaltonZanetti 2004
lnD
Li
1000/T (K)
1400 1200 1000 9001600
ToC
1800
1.3
19*
0.4na
98
10
Figure 2.6 lnDLi ol/fluid vs 1000/T
A plot of ln DLi ol/fluid as a function of 1000/T measured in this study, demonstrating the temperature dependence of
lithium partitioning between olivine and hydrous fluids. A weighted linear regression of the data (excluding the reversal, NCOLR) yields the relationship: ln DLi
ol/fluid = -5.93 + 6.46 * 1000/T (R2=0.82) where T is temperature in Kelvins. Also shown are the experiments of Brenan et al. (1998a), Taura et al. (1998), Blundy and Dalton (2000) and Zanetti et al. (2004). The data points are labeled with the wt. % FeO content of the run product olivine, and suggest that lithium couples with Fe3+ as an exchange mechanism in olivine. The experiment of Zanetti et al. (2004), 19*, was conducted at very reducing conditions and most likely contained less Fe3+ than the experiments in this study. Error bars for the partition coefficients measured in this study are 2, based on the standard deviation of n analyses (see Table 2.5).
58
-30000
-25000
-20000
-15000
-10000
-5000
0.9 0.92 0.94 0.96 0.98 1
RT
lnD
Li
Xan
Figure 2.7 Anorthite/Fluid Lithium Partitioning
Plot of RTln DLi as a function of the mole fraction of anorthite in plagioclase (XAn). Lithium partitioning between anorthite and fluid shows a linear relationship with a negative slope over the range of XAn from 0.91 to 0.99 indicating that lithium is more compatible in albite than in anorthite. Linear regression of the six partitioning experiments yields the relationship, in Jmol-1: RTlnDLi = 162,170 – 188,820(XAn) (R
2=0.96). The primary control on lithium partitioning between plagioclase and hydrous fluids is the composition of the feldspar (see text). 2 errors for the partition coefficients measured in this study are smaller than the symbol used, and based on the standard deviation of n analyses (see Table 2.5).
59
0.01
0.1
1
10
100
750 800 850 900 950 1000 1050 1100 1150
two phase exp
DiLi
ol/fluid/DLi
cpx/fluid
Li, p
pm, o
livin
e/cl
inop
yrox
ene
Temperature (oC)
reversal
Figure 2.8 Olivine/Clinopyroxene Lithium Partitioning
Dol/cpxLi as a function of temperature (oC) for two phase experiments (open symbols) and the ratio of DLi
olivine/fluid/ DLi
cpx/fluid calculated from single phase experiments (solid symbols). The constant Dol/cpxLi versus temperature from
800oC to 1000oC further suggests that olivine/clinopyroxene partitioning of lithium is independent of temperature. The partitioning of lithium between olivine and clinopyroxene calculated from single-phase fluid partitioning experiments is the same as those determined from the two-phase experiments. Error bars for the partition coefficients measured in this study are 2, based on the standard deviation of n analyses (see Table 2.5).
60
0.1 1 10 100 10000.1
1
10
100
1000
10000
silicate meltcarbonatite melt
equilibrated xenolithshydrous fluidsilicate meltcarbonatite melt
hydrous fluid
low fO2
high REEreversal
Li clinopyroxene (ppm)
Li o
livin
e (p
pm)
Experiments Natural Samples
Figure 2.9 Lithium Partitioning From Mantle Xenoliths and Experimental Studies
Lithium abundances in olivine and clinopyroxene from mantle xenoliths (Seitz and Woodland 2000, Paquin and Altherr 2002, Woodland et al. 2002, Woodland et al. 2004) and experimental data (this study, Brenan et al 1998b, Blundy and Dalton 2000). Equilibrated olivine-clinopyroxene pairs tend to fall on a linear ~ 1:1 trend, those metasomatised by silicate melts, and hydrous fluids fall below the trend line, and those altered by carbonatite melt fall on and above the trend. The experimental data have higher concentrations of lithium present due to experimental and analytical requirements nevertheless they appear to follow the same trends. A silicate melt equilibrated olivine-clinopyroxene pair from Brenan et al. (1998b) plots slightly below the line projected from the equilibrated mantle xenoliths, and carbonatite melt olivine-clinopyroxene pairs from Blundy and Dalton (2000) fall on the projected line, above the silicate melt experiment. Olivine-clinopyroxene pairs equilibrated with hydrous fluids in this study fall above the equilibrated mantle xenoliths trend, not below where the hydrous fluid metasomatised samples plot (see text).
61
-25
-20
-15
-10
-5
0
5
10
0.5 1 1.5 2 2.5 3 3.5 4
this study
Wunder et al 2006
reversal
Chan et al 1993Chan et al 1992
7 L
i cpx-
fluid (
o/o
o)
1000/T(K)
-15
-10
-5
0
5
10
0.6 0.8 1 1.2 1.4 1.6 1.8
Figure 2.10 Mineral/Fluid Isotopic Fractionation of Lithium
∆7Licpx-fluid (‰) as a function of 1000/T (K). Results are shown from this study, the spodumene-fluid experiments of Wunder et al. (2006), the basalt-seawater measurements of Chan et al. (1993; 350oC) and Chan et al. (1992; 2oC). Also shown is the regression line constrained by the experiments of Wunder et al. (2006). Error bars for this study are 2, based on counting statistics of the analyses.
62
10 10010
100
1000
10000
100000
1000000
z (km)
year
s
Vf = 100 m/yr
Vf = 10 m/yr
Vf = 4 m/yr
Lithium
Boron
max t from U-series
max t from Ra- Th
Figure 2.11 Time for Li and B Transport to Top of Column
Plot of the time (years) required for an element front to reach the top of a chromatographic column as a function of column height (km) for a fixed column length of 100 km. Time constraints given by U-series and Ra-Th disequilibria are also shown for reference (10,000 to 40,000 years). Because lithium is more compatible in mantle minerals than boron, a lithium signal transported by fluids percolating through the mantle will lag significantly behind the boron signal. The times of transport are calculated from life of the column with respect to each element, in other words; how many times the column can be re-used before its capacity to take up more lithium or boron is reached. The lithium signal will not reach the melt source because it will preferentially partition into the mantle wedge, relative to other fluid mobile elements (such as boron).
63
0
5
10
15
20
25
30
35
40
00.20.40.60.81
7 Li fl
uid
Fraction of Li in fluid remaining
0.996
0.999
0.998
Izu back arc
Izu fore arc
fraction of Li in fluid after equilibration with mantle
MORB
Figure 2.12 Evolution of the Slab Derived Fluid by due to Rayleigh Distillation
Plot of the evolution of δ7Li of a fluid percolating through the mantle as a function of fraction of lithium in the fluid remaining. The change in the isotopic composition of a slab-derived fluid with an initial δ7Li of +10 ‰ (as estimated for the Izu arc by Moriguti and Nakamura, 1998) during percolation through a mantle column. The isotopic composition of the altered fluid will increase and become heavier than the initial slab derived fluid with any amount of fluid:rock interaction. The mass fraction of Li remaining in the fluid after equilibration with the mantle column is also shown for reference. Because lithium is so readily taken up by the mantle wedge, only a small amount of lithium will remain in the fluid. This will result in extreme fractionation and lead to very 7Li-rich fluids at the top of the melt column. The isotopic composition of both the fore arc and back arc lavas of the Izu arc are shown for reference, as is the 7Li of MORB.
64
0 200 400 600 800 1000 1200 14000.8
0.9
1
1.1
1.2
1.3
1.4
T (oC)
fluid
de
nsi
ty (
g/c
m3)
Benioff Zone
(3.6 GPa, 700oC)
Franciscan Subduction
(0.9 GPa, 300oC)
Alpine Subduction
(1.1 GPa, 500oC)
Li[>4.5]
Li[<4.5]
Figure 2.13 Lithium Coordination and P-T Paths
A plot of fluid density as a function of temperature, with the average Li-coordination (from Jahn and Wunder, 2009) shown as degree of shading. Also shown are the fluid densities calculated with the CORK-EOS (Holland and Powell, 1991) using Perple_X program (Connolly, 2005) for Franciscan and Alpine subduction zone P-T paths (from Ernst, 1986) and an ascent path from the slab (Benioff zone) and a fore-arc volcano, ~125 km above the slab (calculated from the thermal model of Peacock, 1993). 6Li-rich fluids will be generated when mineral fluid interaction occurs at depths greater than 125 km, assuming a mantle-slab interface temperature of 700-800oC. As the fluids percolate up through the mantle wedge and fluid density decreases and the coordination of lithium in the fluids will decrease consequently, 6Li will once again preferentially fractionate into the mineral phase and the fluids will become heavier.
65
-15
-10
-5
0
5
10
15
0 0.5 1 1.5 2 2.5
Fin
al I
soto
pic
Co
mp
osi
tion
of
the
Wa
ll R
ock
Fluid/Rock
10 O/oo
0 O/oo
-10 O/oo
range of Izu arc lavas
range of most arc lavas
MORB
Figure 2.14 Evolution of 7Li of Mantle Wedge due to Hydrofractures
A plot of δ7Li of the altered wall rock as a function of fluid/rock ratio. The final isotopic shift of the wall rock depends on the extent of reaction between the mantle and the slab-derived fluids: if the fluid/rock ratio is high, the isotopic composition of the vein/hydrofracture wall rock will approach complete equilibrium with the fluid; equal to 1 ‰ less than the fluid (R-F of -1 ‰). The entire range of δ7Li values found in the Izu arc lavas (light grey shaded area) or other arc lavas (dark grey shaded are) can be achieved by metasomatising the mantle with fluids that have δ7Li between 0 and +10 ‰ and fluid/rock ratios >1.2. To generate arc lavas with δ7Li greater than MORB the isotopic composition of the metasomatising fluid must have δ7Li > +6 ‰. These values are reasonable given the isotopic composition of subducted material and typical fluid/rock ratios for hydrofractured zones.
66
3 Lithium Diffusion
3.1 Introduction
To date, there have been few studies on lithium diffusion in minerals, but observations of natural
samples and limited experimental work indicates that lithium diffusion may be extraordinarily
fast. For example, Berlo et al. (2004) reported rapid mobilization of lithium in plagioclase
phenocrysts from the 1980 eruption of Mount St. Helens in Washington, USA. Plagioclase
phenocrysts erupted prior to the degassing event contained ~14 ppm lithium, whereas those
erupted immediately after contained ~5 ppm. The implication is that the magma lost a significant
amount of lithium in a seven-day period, as recorded in the lithium content of the plagioclase
phenocrysts. Similarly, Kent et al. (2007) interpreted the lithium contents of plagioclase
phenocrysts from the Mount St. Helens 2004 dome lavas as having increased due to the addition
of pre-eruptive lithium rich vapour phase within one year of the dome lava eruptions.
Recent high spatial resolution analyses of lithium isotopes have revealed significant isotopic
heterogeneity at the grain-scale that is suggestive of diffusive exchange. Both olivine and
clinopyroxene phenocrysts from Solomon Island lavas are zoned with respect to lithium and δ7Li
(Parkinson et al., 2007). The rims of the phenocrysts are enriched in lithium compared to the
cores, and the δ7Li decreases from core to rim by as much as 20 ‰ in a W-shaped profile
(Parkinson et al., 2007). A similar pattern was also observed by Jeffcoate et al. (2007) who
measured a 40 ‰ variation in a single orthropyroxene crystal from a San Carlos xenolith (Figure
3.1). The extreme grain-scale variability exhibited by lithium and lithium isotopes is not limited
to terrestrial samples. The basaltic lunar meteorite, NWA 479, examined by Barrat et al. (2005)
contains olivine and pyroxene phenocrysts that also display a wide range of δ7Li values (+2.4 to
+15.1 ‰ in olivine and -0.2 to +16.1 ‰ in pyroxene). Beck et al. (2004) examined pyroxenes in
the shergottite meteorite NWA 480, and found zoning of δ7Li from -17 ‰ in the cores to +10 ‰
in the rims but an absence of lithium compositional variation within the same crystals.
Bulk analyses of lithium isotopes in mantle xenoliths may also be reflect heterogeneity due to
kinetic effects operating on the grain scale. For example, Rudnick and Ionov (2007) reported
highly variable δ7Li in clinopyroxene and olivine grains in peridotite xenoliths from eastern
67
Russia. The δ7Li values ranged from -0.8 to -14.6 ‰ for clinopyroxene and -1.7 to +11.9 ‰ for
corresponding olivine, and olivine/clinopyroxene distribution coefficients varied from 0.2 to 1.0,
which is somewhat lower than previously estimated for equilibrium partitioning.
Diffusive fractionation of lithium isotopes would explain, in some cases, the apparent
disequilibrium that exists between olivine and clinopyroxene pairs of mantle xenoliths. If lithium
diffusion is significantly faster in one mineral phase compared the other, then one phase would
be more affected by introduction or loss of lithium during transport and cooling of the xenolith.
Bulk analyses of olivine and clinopyroxene pairs from a xenolith from the Vitim volcanic field
found δ7Li to range from -17 to -18 ‰ in the pyroxenes with a δ7Li of +6 ‰ in the
corresponding olivine (Ionov and Seitz, 2008). Bulk measurements of olivine phenocrysts in
primitive magmas from a variety of localities found a relatively uniform δ7Li of +3.2 to +4.9 ‰;
however, measurements of clinopyroxene yielded highly variable δ7Li (+6.6 ‰ to -8.1 ‰;
Jeffcoate et al., 2007).
The extreme fractionation of lithium isotopes documented in these studies indicates a kinetic
mechanism rather than an equilibrium process (Lundstrom et al., 2005; Beck, 2006; Jeffcoate et
al., 2007; Parkinson et al., 2007; Rudnick and Ionov, 2007; Marchall et al., 2007). This kinetic
effect has been experimentally demonstrated by Richter et al. (2003) for diffusion of lithium in
silicate melts. Richter et al. (2003) found that 7Li could be fractionated from 6Li by tens of per
mil during diffusion between molten basalt and rhyolite or when diffusing through hydrous
fluids (Richter et al., 2006).
Although estimates of lithium diffusion coefficients have been made from gradients measured in
natural samples (Parkinson et al., 2007), to date there have been few studies to determine the
diffusion coefficients or mechanisms of lithium diffusion in common rock forming minerals. Pell
(1960) measured lithium diffusion in a p-type silicon crystal (a semicounductor material with a
deficit of electrons, therefore allowing positively charged species, to move through the material)
and investigated the effect of diffusion on the 6Li/7Li ratio. Giletti and Shanahan (1997)
measured the diffusion coefficients of various alkali elements in plagioclase feldspars at high
temperature. Coogan et al. (2005) measured the diffusion coefficient for 6Li in clinopyroxene
between 800oC and 1100oC.
68
The current study presents new measurements of lithium diffusion in pyroxene and olivine
between 800oC and 1000oC, and provides the first demonstration of isotopic fractionation
induced by solid-state diffusion in a geological material. These data are critical to understanding
the origin of mineral zonation patterns and isotopic variations that have been documented in
natural samples. With this information, the times-scales of processes and events recorded by
mineral zonation patterns and isotopic variations can be extracted.
3.2 Experimental Methods
Lithium diffusion was measured parallel to the c-axis in clinopyroxene with experiments done at
atmospheric pressure, at temperatures between 800oC to 1000oC and controlled fO2. Starting
materials consisted of natural gem quality clinopyroxene (diopside) crystals from Dekalb, New
York and Kunlun, China. Table 3.1 gives the composition of the starting materials. The mineral
samples were oriented based on crystal habit and ~3 mm thick slabs were made by sectioning
perpendicular to the c-axis using a diamond saw. Slabs free of inclusions and alteration were
selected, cleaned in acetone and rinsed with ultra-pure water in an ultrasonic cleaner. The slabs
were then polished with diamond and alumina paste to 0.3 m. Slabs were sealed in silica tubes
with a solid state buffer (Ni-NiO, MnO-Mn3O4 or Fe3O4-Fe2O3) and ‘pre-annealed’ for 48 hrs at
the same fO2 and temperature conditions as subsequent diffusion experiments. This technique
was meant to homogenize lithium concentrations and equilibrate point defects in the crystals
prior to the diffusion experiments.
Examination of the annealed slabs revealed that the polished surface had roughened and that the
total concentration of lithium in the crystal had decreased uniformly, with the exception of the
outer 50 m of the crystal, which showed greater depletion than the remainder of the crystal
(Figure 3.2). The annealed slabs were then re-polished with diamond and alumina paste to 0.3
m to remove the depleted zone. The re-polished slabs were then cleaned in acetone and rinsed
with ultra-pure water in an ultrasonic cleaner. A single slab was then packed into a platinum
capsule with either a lithium-source, LiCl + 4 wt. % powdered Dekalb diopside for a diffusion-in
experiment, or a lithium-sink, NaCl + 4 wt. % powdered Dekalb diopside for diffusion-out
experiments. The platinum capsule was crimped shut, and loaded into a silica tube with a solid
state buffer. The silica tubes were gently heated in a water bath to ~100oC, and evacuated for a
minimum of 20 minutes, and sealed with a blowtorch. The ampoule containing the sample was
69
equilibrated in a box-type furnace for the duration of the experiment (3 minutes to 16 days) and
quenched by removal from the furnace and air-cooling.
The Pt capsule was recovered, the crystal slab extracted and rinsed with ultra-pure water in an
ultrasonic cleaner to remove the salt rind, mounted in epoxy, ground to half-thickness parallel to
the c-axis, and polished for analysis. This method allowed for measurement of two diffusion
profiles from each slab, on either edge of the slab towards the centre of the crystal. In one case
(sample Kcpx-900.72), the slab tilted slightly while the epoxy set and grinding of this sample
truncated one side, resulting in a single diffusion profile for this run. In some cases (samples
SCO-12, Kcpx-12, Kcpx-MH), the molten salt did not completely wet the slab, again resulting in
only a single profile measurement.
A ‘zero time’ experiment was carried out to assess the effects of the sample preparation, loading
procedure, and temperature run up on the lithium profile of the slabs. A slab that had been
previously analyzed by LA-ICPMS before and after annealing (sample from Figure 3.2) was
packed in NaCl, loaded, sealed into a silica tube as described above, and placed in a box-type
furnace at 1000oC. When the internal temperature stabilized at 1000oC, after 3 minutes, the
sample was removed from the furnace and extracted from the silica tube. This slab was prepared
as above, and lithium concentration was analyzed along the midpoint of the cross-section. No
measurable change in lithium was observed (Figure 3.3), indicating that the sample pre-treatment
and initial heating did not contribute to the diffusion profiles.
3.3 Analytical Techniques
3.3.1 Major Element Analyses
Major element compositions of the starting materials were obtained using the University of
Toronto’s Cameca SX50 Electron Probe X-ray Microanalyzer (EPMA). An accelerating voltage
of 15 kV and a focused 20 nA beam was used for all samples. Diopside, basalt, anorthite and
natural and synthetic oxides were used as standards. X-ray intensities were converted to
concentrations using ZAF and Phi-Rho-Z calculations. The reported errors are the 1 variations
of the reported number of analyses (n).
70
3.3.2 Lithium Analyses
3.3.2.1 LA-ICPMS
Elemental concentrations in the run-product clinopyroxene were measured by laser ablation
inductively coupled plasma mass spectrometry (LA–ICP–MS) at the University of Toronto,
using a frequency quintupled Nd:YAG laser operating at 213 nm, coupled to a VG PQExcell
quadrupole ICP-MS. The laser was operated at 10 Hz and 3 mJ, with He flushing the ablation
cell to enhance sensitivity (Eggins et al., 1998), and produced spot sizes ~25 m in diameter and
~ 25 m deep. At the start of each session, the quadrupole lens settings were adjusted to
maximize the signal on mass 7 during ablation of NIST 610. Data were collected as time-
resolved spectra with background levels determined by counting for 20 s prior to the 60 s of
sampling by laser ablation. Analyses were collected in blocks of 20, with the first and last two
spectra acquired on standard reference materials (SRM). SRMs employed include NIST 610
silicate glass, NIST 612 silicate glass, and “in house” standards of Kunlun diopside and Dekalb
diopside. Table 3.2 lists measurements of reference materials. Data reduction was performed off-
line using the GLITTER software package. Ablation yields were corrected by referencing to the
known concentration of 43Ca that was determined previously by electron microprobe analyses.
The precision for concentration measurements is better than ±10 %. The length of the diffusion
profile was determined by measuring the distance from the edge of the slab to the edge of the
spot using the digital measurement tool included with the laser operating software. The precision
of a repeated measurement is ±5 %.
3.3.2.2 Secondary Ion Mass Spectrometry (SIMS)
In situ analyses of the isotopic composition of the run product clinopyroxene for experiment
kcpx-2 were obtained using the Cameca IMS 3f ion microprobe at Lawrence Livermore National
Laboratory, Livermore, California. Secondary ions were generated by bombardment with a 5-12
nA negatively charged 16O primary beam, accelerated through –12.5 kV and focused to ~20 μm.
The positive secondary ions were accelerated through 4.5 kV. 6Li and 7Li were measured with a
mass resolving power of 1011, and no energy offset was applied. The background (mass 5.8), 6Li, and 7Li were counted on an electron multiplier for 2 s, 10 s, and 2 s respectively over 120-
400 counting cycles, depending on count rate. The isotopic composition of the sample is
expressed as per mil values relative to the core of the slab where,
71
(27) d7Li slab = [(7Li/6Li)slab - (7Li/6Li)core] /(
7Li/6Li)core x 1000
where ‘slab’ refers to the 7Li/6Li ratio from the salt-crystal interface and inwards toward the
core, and ‘core’ refers to the geometric centre of the slab, which is where the lowest 7Li/6Li were
recorded. Because the absolution value of 7Li is not needed here, this notation can used to
highlight the change in the 7Li/6Li of the salt-crystal interface (and inwards) with respect to the 7Li/6Li value of the core (which is assumed to be the original and unaltered 7Li/6Li value of the
slab). Using this notation also removes any ambiguity arising from the available standards not
matching matrix of the sample. The 2 precision of the 7Li/6Li measurements is based on
counting statistics and is approximately ±4 ‰.
3.3.3 Data Reduction
The experimental method was designed to provide a constant concentration of lithium at the
sample surface, a uniform initial concentration of lithium in the sample crystal with a fixed
diffusion boundary, and a semi-infinite diffusion medium. The solution of Fick’s Second Law
for a semi-infinite medium with a planar surface, and these boundary conditions is given by:
(28)
21,
2 Dt
xerfcCC otx
where Cx,t is the concentration of lithium at distance x (m) from the interface at time t (sec), Co is
the concentration of lithium in the crystal at x = 0, D is the diffusion coefficient (m2/s).
Following the method of Harrison and Watson (1983) the data are inverted through the error
function given:
(29)
21
,
)(21
Dt
x
C
Cinverf
o
tx
Fitting of the concentration data is accomplished by plotting x against the inverse error function
of (Cx / Co), and adjusting Co to force the intercept through the origin. This yields a straight line
with a slope of ½√(Dt), determined by least-squares regression. The modeled diffusion profile,
along with a typical concentration profile determined from LA-ICPMS, is shown in Figure 3.4a.
The resulting fit of the inverse error function is shown in Figure 3.4b.
72
3.4 Results
3.4.1 Diffusion in Clinopyroxene
Table 3.3 provides a summary of the experimental conditions and the calculated diffusion
coefficients. Run product crystal slabs emerged from the Pt tubing coated with recrystallized
NaCl or LiCl. The surfaces of the slabs developed a roughened texture, but the major element
chemistry of the crystal did not appear to be changed. Addition or loss of sodium or calcium
would be the most likely exchange; however, given the low diffusion rates of those cations
(Dimanov and Sautter, 2000; Brady and McCallister, 1983), and the short duration of the
experiments, very little exchange is expected. EDS X-ray map analysis of Si, Al, Ca, Na, Mg and
Cl for the highest temperature (1000oC) diffusion-out and diffusion-in experiments confirm this
The EDS X-ray map of kcpx-R, a diffusion in experiment, is shown in Figure 3.5.
Table 3.3 lists the diffusion coefficients calculated from the concentration profiles for each
experiment. Figure 3.6 shows the diffusion coefficients calculated from a series of experiments
at 1000oC and fO2 of NNO, with durations of 2 to 12 hours. The measured values are within error
of each other, demonstrating that the diffusion coefficients are independent of time. Lithium
diffusion coefficients are also independent of the diffusion gradient, as values are the same
whether the flux of lithium is into or out of the crystal. Diffusion coefficients are plotted as a
function of inverse absolute temperature in Figure 3.7, along with values measured by Coogan et
al. (2005). A least squares regression line can be fit to the data for experiments conducted at an
fO2 of NNO and temperature ranging from 800oC to 1000oC with the Arrhenius relationship:
(30) log DLicpx (m2/s) = 5.92 (±8.51) – 2.30 (±1.06)*10,000/T R2 = 0.93
From this, a pre-exponential factor of 8.31 x 105 m2/s and activation energy of 442 ±10 kJ mol-1
is determined.
3.4.2 fO2 Series Experiments
With the exception of Kcpx-MnOMn, and Kcpx-MH, all of the experiments were done at a log
fO2 of -10.3, buffered by Ni-NiO (calibration of O’Neill and Powceby, 1993a). Kcpx-MnOMn
was conducted at a log fO2 of -6.7 (calibration of O’Neill and Powceby, 1993b), and Kcpx-MH
73
was conducted at a log fO2 of -5.3, buffered by magnetite-hematite (calibration of Hemingway,
1990). The results of the experiments done at different fO2 are shown in Figure 3.8, along with
the 1000oC data of Coogan et al. (2005). In general, the diffusion coefficients are within error of
each other; however, there appears to be a slight but systematic trend of increasing values with
decreasing oxygen fugacity. A weighted least squares regression line can be fit to the data
resulting in the relationship:
(31) log DLicpx (m2/s) = -15.2 (±1.7) – 0.28 (±0.18) x log fO2 R2 = 0.75
3.4.3 Diffusion in Olivine
The low (2.5 ppm) lithium content of the San Carlos olivine starting material, relative to Kunlun
diopside (42 ppm) and Dekalb diopside (8 ppm), presented an analytical challenge in
determining diffusion coefficients by the diffusion-out method. Instead, a ‘Li-in’ method with a
Li-chloride source was used. Run durations were restricted as the high lithium vapour pressure
generated by the Li-source had a corrosive effect on the sealed silica tubes, resulting in
catastrophic failure of the ampoules after 12 hours. Nonetheless, a single measurement of lithium
diffusion into olivine was made that can be directly compared to values for clinopyroxene
(Figure 3.9). At 1000oC and fO2 of NNO the measured lithium diffusion into olivine is log D = -
14.1 (±0.12) m2/s, two orders of magnitude slower than the value for clinopyroxene at the same
conditions. The heating time of the olivine experiment was 12 hours and lithium concentration of
the crystal was unchanged beyond 100 m from the crystal-lithium source interface, whereas
given 12 hours at similar temperature and oxygen fugacity conditions, the concentration gradient
in a clinopyroxene continued to ~700 m from the crystal-source interface (Figure 3.4a).
3.4.4 Kinetic Fractionation of 7Li/6Li
An analysis of the 7Li/6Li ratio in experiment kcpx2 was carried out to investigate the
fractionation of lithium isotopes during diffusion. This sample was pre-annealed at an fO2 of
NNO at 1000oC for 48 hours, but unlike the other samples it was not re-polished to remove the
50 m zone of disturbed lithium (see above). Similar to other Li-out experiments the sample was
packed in NaCl, sealed in a Si-tube with NNO oxygen buffer and heated at 1000 oC for two
hours.
74
The diffusion coefficient for 7Li, calculated from the concentration profile acquired by LA-
ICPMS, is 5.28 x 10-13 m2/s, which is within the range of other 1000oC measurements from this
study. The measured lithium concentration profile is shown in Figure 3.10a, along with the
modeled diffusion profile, and represents lithium loss during the combined ‘pre-annealing’ step
and ‘Li-out’ diffusion experiment. The modeled diffusion profile which best fits the data is that
for 2 hours of diffusion time. This is consistent with the finding that the ‘pre-annealing’ step
served to homogenize the lithium concentrations in the slab.
The Co value that was determined from the fit of the 7Li data was then used, together with the
isotopic ratio that was measured by SIMS analysis, to generate a 6Li concentration profile. This
profile was used to calculate a diffusion coefficient of 5.44 x 10-13 m2/s. The modeled isotopic
gradient is shown in Figure 3.10b, along with the isotopic gradient measured by SIMS, and
reveals a +7 ‰ change between the NaCl-crystal interface and the centre of the slab.
Although the starting material for this experiment was not characterized for pre-existing isotopic
gradients, it is likely any gradients would have been eliminated in the subsequent annealing steps
as a result of the rapid lithium diffusion documented in this study. Rather, the main uncertainty
when interpreting this result is the timescale over which the gradient was produced. The
experiment was subject to two episodes of diffusive loss: ‘controlled’ loss into the NaCl lithium-
sink and an ‘uncontrolled’ loss during the ‘pre-annealing’ phase. Despite this shortcoming, there
are two notable aspects of the measured gradient. First, the isotopic gradient appears to penetrate
farther into the crystal than the chemical gradient, ~250 m versus ~100 m. This is consistent
with mass balance considerations as the bulk of the elemental profile will be dominated by the
more abundant 7Li, which is expected to diffuse more slowly than 6Li. The second and most
significant aspect of this profile is the large isotopic difference produced between the core and
the rim, ~ +7 ‰.
3.5 Discussion
3.5.1 Effect of fO2 on lithium diffusion in clinopyroxene
As described above, a negative correlation of lithium diffusion rates with fO2 is shown in Figure
3.8, with log DLicpx decreasing from -12.4 to -13.8 as oxygen fugacity increases from log fO2 -12
to log fO2 -5.5. Previous studies have demonstrated that the magnitude of the effect of fO2 on Pb
75
and Ca diffusion in clinopyroxene is about two log units in D-value over a range of ten log units
of fO2 (Cherniak, 2001; Dimanov et al., 1996). Given that the precision of this study is about 1
log unit (2), an effect of fO2 on lithium diffusion in clinopyroxene would just be discernable
from this data.
The negative dependence of lithium diffusion on fO2 is similar to that determined for Ca self-
diffusion in clinopyroxene by Dimanov et al. (1996), and in contrast to the positive fO2
dependence determined for Pb diffusion in clinopyroxene (Cherniak, 2001). Cherniak (2001)
interpreted the positive fO2 dependence of Pb diffusion in clinopyroxene as the result of the
oxidation of Fe2+ to Fe3+ creating point defects, or vacancies on the crystal lattice, which
accommodates the Pb cation as it diffuses through the crystal. According to Dimanov et al.
(1996), the negative fO2 dependence of Ca self-diffusion suggests an interstitial mechanism for
diffusion, where Ca moves through interstitial sites of the crystal lattice, and is not displacing
other cations in normal lattice sites. Tsai and Dieckmann (2002) describing the relationship
between oxygen content and point defects in olivines demonstrated how an increase of fO2 would
result in the oxidation of Fe2+ to Fe3+, thereby creating intrinsic point defects, or vacancies, in the
crystal lattice via the reaction:
(32) 6Fe2+o + 3O2 = 4Fe3+
o + 2Vo + 2FeO
where V is a vacancy in an octahedral site (o). Following the treatment of Ganguly et al. (2007),
interstitial diffusion of Li+ can be written as:
(33) Li+o + Vi ↔ Li+ i + Vo
where Li+ is the lithium cation in an octahedral site (o) or interstitial site (i), and V is either an
octahedral coordinated vacancy (o) or an interstitial vacancy (i). An increase in Vo would result
in an decrease of Li+i, and an increase in Li+
o, making interstitial diffusion less favorable.
The partitioning of lithium into olivine and clinopyroxene provides some insight into the lithium
transport process. Lithium incorporation into olivine or clinopyroxene appears to be a coupled
substitution with trivalent Fe or Al for Mg2+ or Fe2+ (see Chapter 2). Increasing the proportion of
trivalent Fe may result in a decrease in the degree of ‘misfit’ between the Li+ ion and a potential
site on the normal crystal lattice, thereby reducing the possibility of interstitial movement or
76
increasing the activation energy required for a jump. Given the small size of the lithium cation,
interstitial diffusion may be plausible. However, it should be noted that the effect of decreasing
diffusion with increasing fO2 is not large, and indicates that this interstitial mechanism is
probably only a minor component of the total diffusive flux.
3.5.2 Comparison with other lithium diffusion studies
Previous work on lithium diffusion in clinopyroxene by Coogan et al. (2005) employed a natural
diopside crystal as a Li-sink, and a powdered mixture of San Carlos olivine and 6Li enriched
Li2CO3 and Li2SiO3 as the Li-source. The diopside crystals and Li-source were then packed into
Al-crucibles and heated in a gas-mixing furnace using a CO and CO2 mix to control fO2. Unlike
this study, Coogan et al. (2005) did not pre-anneal the diopside slabs to pre-equilibrate point
defects under fO2 and temperature conditions of the diffusion experiments. Also, some of the
experiments of Coogan et al. (2005) were buffered at fO2s that differed from those used in this
study by as much as four log units (e.g. at 900oC). When equivalent fO2 conditions were
employed there is good agreement between the measurements of Coogan et al. (2005) and the
results presented here (Figure 3.7).
Figure 3.11 presents the lithium diffusion measurements for clinopyroxene and olivine from this
study, along with data from previous studies of lithium diffusion in other minerals. Lithium
diffusion in olivine is 2 orders of magnitude lower than in clinopyroxene at the same
temperature. Due to the different activation energies, the data for feldspars (Giletti and
Shanahan, 1997) overlap with the clinopyroxene data only at the lowest temperature
investigated. At high temperatures, lithium diffusivities in anorthite and albite are almost 4
orders of magnitude higher than in olivine or clinopyroxene. The measurements of Pell (1960) in
p-type Si-crystal are an order of magnitude higher than those in feldspars and two orders of
magnitude higher than those in clinopyroxene. Previous researchers have also proposed an
interstitial mechanism for lithium diffusion in both Si-crystal (Pell, 1960) and feldspars (Giletti
and Shanahan, 1997) consistent with the results of this study.
3.5.3 Comparison with diffusion of other cations in clinopyroxene
Figure 3.12 compares the lithium diffusivities measured in this study, and that of Coogan et al.
(2005), with experimentally determined diffusivities for other elements in clinopyroxene.
77
Lithium diffusivities measured in this study, and by Coogan et al. (2005), are more than five
orders of magnitude higher than Sr (Sneeringer et al., 1984), or Fe –Mn and Mg (Dimanov and
Sautter, 2001). Only hydrogen diffusion is more rapid (Woods, 2000). The large range of
diffusion coefficients measured in clinopyroxene most likely reflects the different mechanisms of
diffusion at work. Previous studies have suggested that the diffusion mechanism for many of the
larger, divalent cations, such as Pb, Sr, Fe, Mg, and Mn is point defect, or vacancy, controlled
(Dimanov and Wiedenbeck, 2006; Azough and Freer, 2000; Cherniak, 1998; -2001, Sneeringer
et al., 1984). In the case of REEs, U and Th an elastic diffusion model is suggested, where
movement is governed by the elastic strain of the crystal lattice, and diffusivities increase with
decreasing charge and radius (Van Orman et al., 2001). Some contribution from a point defect,
or vacancy, controlled mechanism may be generally applicable for clinopyroxene since diffusion
coefficients have been shown to increase in more Fe-rich pyroxenes (Cherniak, 2001; Woods,
2000) which have a greater ionic porosity (Cherniak, 2001 and references therein) or more‘free
space’ in their crystal lattices than Fe-poor pyroxene.
3.5.4 Geological Implications
3.5.4.1 Preservation of Lithium Signatures
Results of experimental measurements of lithium partitioning between mantle minerals and
hydrous fluids show that lithium is only slightly incompatible in olivine and clinopyroxene with
respect to hydrous fluids, and that the lithium content and isotopic signature of slab derived
fluids can be significantly modified during transport through, and interaction with the mantle
wedge (Caciagli, Chapter 2). Given the rapid diffusivities measured in this study, it is reasonable
to assume that mantle minerals in contact with lithium bearing fluids will quickly equilibrate.
This assumption can be quantitatively demonstrated following the treatment of Crank (1975).
The maximum time for centre preservation of lithium concentration in a spherical grain can be
calculated as a function of grain radius at a constant temperature. For diffusion in a sphere, the
centre will retain unaltered lithium concentrations for values of Dt/a2 ≤ 0.03, assuming that the
concentration of the sink/source remains constant at the sphere surface over the annealing
interval (Crank 1975). This relationship can be expressed as:
(34) t = 0.03 / (Da2)
78
where t is time, D is the diffusion coefficient, at a given temperature, and a is grain radius.
Figure 3.13 shows a plot of maximum annealing times at 1000oC for spherical olivine and
clinopyroxene grains of a given radius (in microns) and provides constraints on the time-scale of
processes which preserve lithium zonation. An essential assumption required for modeling the
effects of interaction between a fluid flux from a subducting slab and the overlying mantle
wedge is that fluid/solid equilibrium must be maintained. This calculation demonstrates that such
an assumption appears to be valid.
The rapid exchange portrayed in Figure 3.13 is also consistent with the evidence from mantle
xenolith studies that transient late stage events, such as entrainment and transport in a magma
followed by eruption and cooling, can perturb clinopyroxene and, to a lesser extent, olivine
compositions (Aulbach et al., 2008; Parkinson et al., 2007; Rudnick and Ionov, 2007). For
example, xenoliths from the Vitim volcanic field have pyroxenes with 2-5ppm lithium and 7Li
of -17 ‰ yet coexisting olivine has 1.2 ppm lithium and 7Li of +6.3 ‰ (Ionov and Seitz, 2008).
‘Normal’ mantle is estimated to contain 1.6 ppm lithium and an average 7Li of +4 to +5 ‰
(Jagoutz et al., 1979; Tomascak, 2004). Equilibrium partitioning of lithium between olivine and
clinopyroxene results in Dol/cpx ~1 (Chapter 2, this study) so the pyroxene and olivine should
contain similar lithium concentrations. The increased lithium content of the Vitim pyroxenes is
likely the result of an influx of lithium during entrainment and cooling; however, the lithium
content and isotopic composition of the olivine is relatively unperturbed (Ionov and Seitz, 2008).
This study has demonstrated that lithium diffusion is almost two orders of magnitude faster in
clinopyroxene than in olivine, therefore the lithium content and isotopic composition of
clinopyroxene will be modified to a greater extent than the coexisting olivine during short-lived
events. The elemental and isotopic disequilibrium evident in the Vitim xenoliths allows for
maximum transport times to be determined. Estimates of alkalic magma ascent rates range from
0.2 -0.5 m/s based on hydrogen diffusion studies (Peslier and Luhr, 2006), to 1.3-2.7 m/s based
on mineral dissolution studies (Brearley and Scarfe, 1987). The estimated depths of the Vitim
xenoliths are 40-50 km (Ionov et al., 1993), therefore a magma ascent rate of 0.5-1.0 m/s will
result in a transport time of ~12-25 hrs. From Figure 3.13, at 1000oC a clinopyroxene grain with
a radius of 1000 m will be re-equilibrated in ~29 hrs, whereas an olivine grain of same size will
preserve its original concentration for almost 1000 hrs. Assuming the temperature of the
entrainment magma was 1000oC, then the time for transport, eruption and cooling of Vitim
79
xenoliths was at least 29 hrs, to allow time to preturbe the lithium content of the clinopyroxene,
but less than 1000 hrs, to preserve the original olivine signature, consistent with an ascent rate of
~0.5 m/s.
3.5.4.2 Closure Temperature
Knowing the temperature for which mineral grains are ‘closed’ to lithium diffusion allows the
extent of re-equilibration during cooling to be estimated. A closure temperature for lithium
diffusion in clinopyroxene can be determined using the equation for mean closure temperature
given by Dodson (1973; eq. 23):
(35)
dtdTE
aDART
RT
E
a
oc
c
a22
ln
where Tc is the closure temperature, Ea is the activation energy, R is the gas constant, Do is the
pre-exponential factor, a is the characteristic dimension of the crystal (e.g. radius of a sphere or
cylinder or the half thickness of a plane), dT/dt is cooling rate, and A is a dimensionless
parameter relating to the weighted arithmetic mean of the closure temperature and geometry of
the grain (A = 55 for a sphere, 27 for a cylinder, and 8.7 for a plane sheet; Dodson, 1973). Figure
3.14 shows the calculated closure temperature of lithium as a function of characteristic radius for
a spherical grain for cooling rates of 1o, 10o and 100o/Myr. The closure temperatures calculated
for grains ranging from 1 to 1000 m in size is between 425oC and 550oC. This is approximately
100-200oC lower than the closure temperature for Sr in clinopyroxene for grains of equivalent
size (Figure 3.14).
3.5.4.3 The Potential for Re-Equilibration of Lithium Composition
The combination of extremely rapid lithium diffusion, and low closure temperatures results in a
significant potential for homogenization of elemental and isotopic differences between minerals.
In particular, the absence of any observed isotopic fractionation during crystallization and melt
differentiation needs to be re-assessed. Kilauea Iki lava lake, Hawaii formed in 1959 due to a
single eruption of picritic tholeiite magma; a crust formed within a few weeks and the whole lake
crystallized as a closed system afterwards. The resulting rocks span a wide compositional range,
from olivine-cumulates, and olivine tholeiites to ferrodiabase, and silicic veins (Helz et al.,
1989). The isotopic composition of a suite of samples from Kilauea Iki lava lake are plotted as a
80
function of MgO content in Figure 3.15, and show no significant (> ±2 ‰) fractionation between
the more primitive, MgO-rich rocks and the more evolved, MgO-poor rocks (Tomascak et al.,
1999). However, lithium isotope fractionation is observed in granitic systems during
crystallization at 500-600oC (Tomascak, 2004). Because fractionation factors for stable isotopes
are temperature dependant (Urey, 1947), it has been concluded that crystallization of more mafic
magmas occurs at sufficiently high temperatures, 1200-1000oC, such that appreciable (>1 ‰)
equilibrium fractionation of lithium isotopes does not occur (Tomascak et al., 1999).
Recent experiments have found lithium isotopic fractionation between clinopyroxene and fluids
to persist to high temperatures (Chapter 2, Wunder et al., 2006). Furthermore, the case can be
made that the degree of lithium isotopic fractionation between minerals and fluids should be
comparable to the degree of fractionation that occurs between minerals and melts. Lithium
isotopic fractionation between minerals and fluids depends on the difference in the zero point
potential energy (ZPE) between the phases of interest. 7Li is heavier and has a lower vibrational
frequency, and therefore a lower ZPE than 6Li (Chacko et al., 2001). The phase that will undergo
the greatest reduction in ZPE will preferentially take 7Li over 6Li (Chacko et al., 2001). This has
been demonstrated by Ab initio calculations, which have predicted that during mineral-solution
reactions 6Li, is preferentially incorporated into octahedrally coordinated sites in the solid, and 7Li is preferentially incorporated into the dominantly tetrahedrally coordinated sites in the fluid
(Yamaji et al., 2001). Experimental measurements of lithium isotopic fractionation between
clinopyroxene and fluids (Chapter 2) and spodumene and fluids (Wunder et al., 2006) also
confirm this behavior. Given that silicate melts also have a tetrahedral structure, with lithium in
tetrahedrally coordinated sites (Cormier et al., 1998; Majérus et al., 2003) then a degree of
isotopic fractionation, similar to that which occurs between minerals and fluids, is expected
between minerals and silicate melts. This study found the isotopic fractionation between
clinopyroxene and fluids at 1000oC to be approximately -1‰, which is in agreement with the
empirically determined regression of Wunder et al. (2006), which predicts min-fluid
fractionation of -0.65 ‰ to -1.1 ‰ at 1200oC to 1000oC. Although this fractionation is small,
the overall magnitude of this effect can be significant, resulting in gradients of up to 6-7 ‰
across the Kilauea differentiation sequence, (Chapter 2, Figure 2.12) if closed system Rayleigh
distillation continues to completion.
81
Surprisingly, significant (> ±2 ‰) lithium isotopic fractionation is not evident in the Kilauea Iki
lava lake rocks. Magmatic differentiation of Kilauea Iki lava lake took place between 1200-1100 oC, with the last veins forming at ~1000oC (Helz et al., 1989). The eruption occurred in 1959,
and samples from the upper portion of the lake were sub-solidus by mid-1969, which gives a
maximum cooling rate of 25 o/yr for the rocks in the upper zone. Because closure temperatures
are dependant on cooling rate (see Equation 35), the closure temperature for grains ranging from
1-1000 m would increase to 600-800oC respectively. Even considering the rapid cooling rate
and increased closure temperatures, the timescales required for re-equilibration of lithium
isotopes by diffusion (weeks) are much less than the timescales required for cooling and
crystallization (years). Thus, it seems plausible that the overall homogeneity of lithium isotopes
between samples from the Kilauea Iki lava lake implies that isotopic re-equilibration occurred
beyond the grain-scale with an external reservoir.
In support of this notion is the existence of a convecting geothermal system within the upper
portion of the lava lake (Hardee et al., 1981). The geothermal system did not extend to the rocks
of the lower part of the lake; however, interaction with a volatile phase is still possible in this
zone via the fractures and pores in these rocks (Hardee et al., 1981). A convecting volatile phase
such as the one that developed in the upper zone of the lava lake, or volatiles moving through
fractures and pores in the rocks of the lower zone, can efficiently transport lithium. Evidence for
lithium transport in a volatile phase is provided by several studies that report changes in the
lithium content of phenocrysts phases from other volcanic conduit systems. For example,
plagioclase phenocrysts erupted prior to the 1980 Plinian eruption of Mout St. Helens contained
~14 ppm lithium, whereas those erupted seven days later contained ~5 ppm. Similarly, Kent et
al. (2007) found that the lithium contents of plagioclase phenocrysts from the Mount St. Helens
2004 dome lavas had increased due to the addition of pre-eruptive lithium rich vapor phase
within one year of the dome lava eruptions. The implication is that the Mount St. Helens magmas
can gain or lose a significant amount of lithium over very short timescales, as recorded in the
lithium content of the plagioclase phenocrysts. Likewise, a study of lithium the Tin Mountain
pegmatite and host rocks found gradients in both lithium concentration and isotopic composition
to persist much farther (>30 m) into the country rock than other alkali elements such as Rb and
Cs, which were limited to <2 m from the contact (Teng et al., 2006). These findings demonstrate
that even in a crystalline rock, when a volatile phase is present, lithium will be extremely mobile;
82
the estimated lithium diffusivity in the host rock amphibolite interacting with an interconnected
fluid or volatile network along grain edges in the Tin Mountain pegmatite is approximately 2 x
10-16 m2/s at 350oC (Teng et al., 2006). Given the rapid diffusivity of lithium and the efficiency
of lithium transport in a volatile phase, a setting such as the Kilauea Iki lava lake would not be
expected to preserve signatures of isotopic fractionation that might arise from crystallization and
magma differentiation. Isotopic fractionation due to crystallization would be most reliably
investigated by direct experimental measurements.
The potential for lithium re-equilibration during emplacement and cooling of rocks exists in
many different settings. The lithium contents of rocks exhumed at convergent margins,
especially eclogites thought to represent remnants of subducted oceanic crust, have been the
focus of much study with the aim to understand the potential for lithium recycling between the
crust and mantle (Zack et al., 2003; Marschall et al., 2007). The P-T path of exhumed eclogites
must be taken into account when examining the lithium concentrations of these rocks. A
Franciscan type P-T path, where the retrograde and prograde paths are similar, requires that the
rocks be cooled as they are exhumed. In this setting exhumation, and consequently, cooling, is
relatively slow (Ernst, 1988). It is unlikely that these rocks will preserve the lithium signatures
that developed at peak metamorphic conditions, unless the peak metamorphic temperatures are <
500oC, which is the case for blueschist facies rocks. Alpine metamorphic paths are characterized
by nearly isothermal decompression due to very rapid exhumation (Ernst, 1988). In this case, the
lithium signatures displayed in exhumed Alpine eclogites may be representative of those
achieved at peak metamorphic conditions, assuming interaction with retrograde fluids is at T <
500oC. Given the relatively low closure temperature, any post emplacement heating or secondary
metamorphic events must be carefully considered when interpreting lithium concentrations.
3.5.4.4 Diffusion-Induced Isotopic Fractionation
Richter et al. (2003) showed that slightly faster transport of 6Li than 7Li results in diffusive
fractionation in silicate melts. Diffusive fractionation has been suggested as the mechanism for
generating the 20-40 ‰ gradients observed in some phenocrysts and xenoliths. This study has
documented isotopic fractionation due to diffusion in a solid medium. A 7 ‰ gradient was
generated 300 m into the crystal during the course of the 2-hour diffusion experiment (Figure
3.10b). Although preliminary, this result is a demonstration of the magnitude of the isotopic
83
gradient that can be produced by diffusion in a very short time-scale. This fractionation occurs
because lighter isotopes diffuse faster than heavy ones; the relative diffusivity between isotopes
is given by:
(36) D1/D2 = (m2/m1)
where D1/D2 is the ratio of the diffusivity of isotopes of mass m1 and m2, and the exponent is
empirically determined (Richter et al., 1999; Richter et al., 2003; Richter et al., 2009). A of ½
is characteristic of self-diffusion in a gas, for diffusion in liquids or melts ranges from 0.025 to
0.215, depending on the isotopes (Ge isotopes have < 0.025; Ca isotopes, of 0.05 to 0.1; Li
isotopes, ~0.215; Richter et al., 1999; Richter et al., 2003).
A -value was estimated from the calculated diffusion coefficients of 7Li and 6Li. A diffusion
coefficient for 7Li was calculated from the measured concentration profile following the method
described in section 3.3.3. The modeled diffusion profile for 7Li was then used, together with the
isotopic ratio that was measured by SIMS analysis, to generate a 6Li concentration profile. This
profile was used to calculate a diffusion coefficient of 5.44 x 10-13 m2/s. A value of 0.2 was
determined from the ratio of diffusivities calculated from the concentration profile and isotopic
profile of experiment kcpx-2. This is similar to the of 0.215 determined for lithium isotopes
diffusing in melt by Richter et al. (1999).
This study has shown that isotopic fractionation can occur due to diffusion of lithium in a
mineral. Diffusive isotopic fractionation has been proposed as the means to produce the extreme
isotopic compositions and complex isotopic profiles observed in xenoliths and phenocrysts
(Parkinson et al., 2007; Jeffcoate et al., 2007). In Figure 3.16 one half of the W-shaped profile
observed in an orthropyroxene crystal from a San Carlos xenolith (Figure 3.1; Jeffcoate et al.,
2007) is modeled using the diffusivities and the (0.2) determined from this study. The lithium
isotopic gradient is calculated for 10 hrs, 1000oC and with a surface concentration/initial mineral
concentration (Cs/Co) of 3.25. Because the lithium concentration at the surface of the grain is
higher than the interior of the grain, the faster diffusion of 6Li into the grain results in a 7Li
profile with a ‘trough’ or one-half of a W-profile. The calculated model is a very good fit for the
orthopyroxene data and demonstrates that lithium diffusion in a solid is a viable mechanism for
producing extreme and complex isotopic gradients.
84
3.6 Conclusions
Measured lithium diffusion in a natural clinopyroxene at 800-1000oC ranges from 6.5 x 10-16 to 1
x 10-12 m2/s and is 3.5 x 10-15 m2/s in San Carlos olivine at 1000oC. The lithium diffusion
coefficient is independent of the diffusion gradient as values are the same if the flux of lithium is
into or out of the crystal. These rapid diffusivities can be used to determine the timescales for
retention of lithium signatures in mantle minerals, which can in turn be used constrain transport
times for xenoliths.
Closure temperatures have also been calculated for clinopyroxene and have been found to be
low, ranging from 400-600oC, depending on cooling rates and grain size. The low closure
temperatures and rapid diffusivities indicate that there is great potential for lithium re-
equilibration during emplacement and cooling of rocks in many geological scenarios.
Furthermore, this implies that the absence of discernable isotopic gradients in high temperature
differentiation sequences is not necessarily evidence for a lack of isotopic fractionation during
crystallization. Direct experimental measurements would be the most reliable way to determine
the degree of isotopic fractionation during crystallization; however, to date this information is
lacking.
Lithium and lithium isotopes are frequently used as tracers of surface inputs into the mantle;
however, great care must be taken when interpreting the lithium contents of minerals. Given the
rapid diffusion of lithium and its low closure temperature, lithium contents in minerals can be
significantly modified due to diffusion over very short timescales. The results of this study show
that lithium is most suitable when interpreting short duration events, such as volcanic eruptions
or degassing events (Berlo et al. 2004, Kent et al 2007) or cooling rates (Coogan et al., 2005).
Furthermore, due to the high mass ratio (1.166) between 7Li and 6Li, lithium isotopes are subject
to diffusive fractionation, even during diffusion in solids, and the isotopic composition of lithium
in minerals can be modified over very short timescales.
85
Table 3.1 Composition of Starting Material
Kunlun Diopside Dekalb Diopside San Carlos
Olivine
SiO2 55.39 (0.23)1 54.77 (0.8) 40.95 (0.02)
Al2O3 0.90 (0.04) 0.66 (0.10) <0.01
FeO 0.74 (0.05) 0.85 (0.08) 9.31 (0.05)
MgO 17.83 (0.13) 17.31 (0.22) 49.19 (0.42)
CaO 24.82 (0.1) 25.17 (0.26) <0.02
Na2O 0.59 (0.03) 0.43 (0.08) <0.02
MnO 0.07 (0.03) 0.05 (0.06) 0.12 (0.02)
NiO nd nd 0.39 (0.03)
Total 100.34 99.29 100.86
n 25 11 13
Li ppm2 42.6 (2.5) 8.9 (0.6) 2.5 (0.5)
7Li (‰) +13.0 (1.0)3 +9.7 (1.0) +3.64 (0.15)4 1Numbers in parentheses represent 2 of the mean of n analyses 2Analyzed by LA-ICPMS, numbers in parentheses represent 2 of the mean of 10 analyses 3Analyzed by MCICP-MS, numbers in parentheses represent 2 of the uncertainity on the measurement 4Magna et al., 2006
86
Table 3.2 Measurements of Standards
LA-ICP-MS1 Li (ppm) reference
International Standards
NBS 612 41.5 41.54 (2.87) Pearce et al. 1997
NBS 610 488.7 484.6 (21.7) Pearce et al. 1997
JG1a 92.4 79.5 (4.5) Imai et al. 1995
JB-2 7.9 7.78 (1.39) Imai et al. 1995
BCA1 12.3 13.3 Ryan and Langmuir, 1987
JGB-1 4.6 4.59 (.90) Imai et al. 1995
In house Standards
Kunlun Diopside 42.6 this study
1) 2 errors for LA-ICPMS abundance analyses are better than 10% based on counting statistics
87
Table 3.3 Summary of Experiments
Experiment T
(oC) log fO2 t (min) sample
Li source/sink
1D (m2/s) Co log D (m2/s)
zerotime 1000 -10.3 5 kunlun NaCl nd
kcpx3 1000 -10.3 180 kunlun NaCl 4.06E-13 (0.73)2 4.25 (0.42)2 -12.39 (1.12)2
kcpx6 1000 -10.3 360 kunlun NaCl 1.06E-12 (0.17) 2.83 (0.51) -11.97 (0.93)
kcpx12 1000 -10.3 720 kunlun NaCl 1.01E-12 (0.11) 13.08 (2.35) -11.99 (0.65)
kcpx2 1000 -10.3 120 kunlun NaCl 5.28E-13 (0.47) 6.28 (0.31) -12.27 (0.76)
kcpxMH 1000 -5.3 345 kunlun NaCl 1.71E-14 (0.21) 4.19 (0.29) -13.77 (0.86)
kcpxMnOMn 1000 -6.7 364 kunlun NaCl 1.10E-13 (0.21) 6.82 (0.55) -12.96 (1.24)
kcpx900-72 900 -10.3 4485 kunlun NaCl 4.47E-14 (0.71) 3.27 (0.29) -13.35 (1.05)
kcpx900-36 900 -10.3 2150 kunlun NaCl 6.23E-15 (0.88) 2.06 (0.21) -14.21 (1.)
kcpx800-16d 800 -10.3 23100 kunlun NaCl 6.47E-16 (2.43) 6.56 (0.52) -15.19 (2.86)
kcpxR 1000 -10.3 362 kunlun LiCl 1.18E-13 (0.17) 1.77 (0.12) -12.93 (0.95)
sco12 1000 -10.3 735 sco LiCl 3.44E-15 (0.91) 2.64 (0.32) -14.46 (1.92)
1D shown is the average of Ds calculated from two traverses, with the exception of SCO12, Kcpx-12, Kcpx-MH, and Kcpx-900.72, see text for details 2number in parenthesis represents 2 error
88
-50
-40
-30
-20
-10
0
0
0.5
1
1.5
2
2.5
3
0 500 1000 1500 2000 2500
7 Li
Li pp
m
microns
Figure 3.1 Li Elemental and Isotopic Gradients in San Carlos Opx
7Li (‰) and lithium concentration (ppm) plotted against distance from crystal edge (m) for cross sections of a orthopyroxene xenolith from San Carlos, Arizona. The isotopic gradient shows a 35 ‰ decrease from rim to core, in a W-shaped profile. Data from Jeffcoate et al. (2007).
89
0
10
20
30
40
50
60
70
0 200 400 600 800 1000
DekalbKunlun
Li (
ppm
)
X(m)
starting concentration
Figure 3.2 Effect of fO2 Anneal
Lithium concentration (ppm) plotted against distance from the salt-crystal interface (m) for cross sections of a Dekalb diopside slab (circles) and a Kunlun diopside slab (squares) after the fO2 anneal. The total concentration of lithium in the crystal is uniformly decreased with respect to pre-treated material (solid lines) with the exception of the outer 50 m of the crystal, which is slightly more depleted. The annealed slabs were then re-polished with diamond and aluminum paste to 0.3 m to remove the depleted zone. Error bars are 2, based on analytical uncertainty of lithium concentration.
90
10
20
30
40
50
60
0 50 100 150 200 250
lith
ium
(p
pm)
x (m)
Figure 3.3 Zero time Experiment
Lithium concentration (ppm) plotted against distance from the salt-crystal interface (m) for a cross section of the ‘zero time’ experiment. No measurable change in lithium was observed, indicating that the sample preparation and sample loading had no effect on the lithium concentration. Error bars are 2, based on analytical uncertainty of lithium concentration.
91
-14.0
-12.0
-10.0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
0 200 400 600 800 1000
bac
kgro
un
d c
orr
ecte
d li
thiu
m (
pp
m)
microns
0.0
0.2
0.4
0.6
0.8
1.0
1.2
0 100 200 300 400
kcpx12
inv
erf
(C
/Co
)
microns
Figure 3.4 Results for Experiment Kcpx-12
(a) Background corrected lithium concentration measured by LA-ICPMS (circles), plotted against distance from salt-crystal interface for a cross section of Kcpx12. Also plotted is the model diffusion profile (curve) calculated from the concentration gradient. (b) Plot of distance from the salt-crystal interface, x (m), versus the inverse of the error function of (Cx / Co) and the resulting fit of least-squares regression. Error bars are 2, based on analytical uncertainty of lithium concentration.
a)
b)
92
Figure 3.5 X-Ray Maps of Run Product
SEI and EDS x-ray maps of Kcpx-R (1000oC, 6 hour heating in LiCl). No detectable gradients were noted in major element concentration (Si, Al, Ca, Na, Mg) or Cl.
SEI
93
-18.0
-16.0
-14.0
-12.0
-10.0
-8.0
0 5 10 15
diffusion outdiffusion in
log
D (
m2 /s
)
t (hrs)
Figure 3.6 Time Series
Diffusion coefficients measured from a series of diffusion-out experiments (solid circles) at 1000oC, fO2 of NNO and times ranging from 2 to 12 hours. The measured diffusion coefficients are within error of each other (2 based on analytical uncertainty of lithium concentration) and are independent of experiment duration. Also within error of the other experiments is Kcpx-R, a diffusion-in experiment (open circle), which demonstrates that the lithium diffusion coefficient is independent of the diffusion gradient as D-values are the same if the flux of lithium is into or out of the crystal. Error bars are 2, based on analytical uncertainty of lithium concentration.
94
-22.0
-20.0
-18.0
-16.0
-14.0
-12.0
-10.0
-8.0
7 7.5 8 8.5 9 9.5
this studyCoogan et al., 2005olivine
log
D (
D in
m2 /s
)
10000/T(K)
1100oC 1000oC 900oC 800oC
Figure 3.7 Measured Lithium Diffusion Coefficients
Log DLi values measured in this study (solid circles), and Coogan et al. 2005 (open circles), plotted against 10,000/T (K). The diffusion coefficient for lithium in clinopyroxene is temperature dependent from 800 to 1000 oC with an Arrhenius relationship of log DLi
cpx = 5.92 (±8.51) – 192 (±10)/RT (R2=0.87) Also shown is the datumn from the olivine experiment. Error bars are 2, based on analytical uncertainty of lithium concentration. Error bars are 2, based on analytical uncertainty of lithium concentration.
95
-18.0
-17.0
-16.0
-15.0
-14.0
-13.0
-12.0
-11.0
-10.0
-16 -14 -12 -10 -8 -6 -4
Coogan et al., 2005
log
D (
D in
m2 /s
)
log fO2
Figure 3.8 fO2 Experiment Series
Log DLicpx measured in this study (solid circles), and Coogan et al. 2005 (open circles), plotted as a function of log fO2.
As log fO2 increases from -12 to -5.3 (Kcpx-MH; log fO2 buffered by magnetite-hematite) log DLicpx decreases from -
12.4 (from the study of Coogan et al., 2005) to -13.8 (Kcpx-MH). A weighted least squares regression line yields the equation: log DLi
cpx = -15.2(±1.7) – 0.28 (±0.18) x log fO2 (R2=0.75). The least squares regression line does not
include the data of Coogan et al., (2005). Error bars are 2, based on analytical uncertainty of lithium concentration. Error bars are 2, based on analytical uncertainty of lithium concentration.
96
0
1
2
3
4
0 50 100 150 200
bac
kgro
un
d c
orr
ecte
d li
thiu
m (
pp
m)
microns
Figure 3.9 Lithium Diffusion Profile in San Carlos Olivine
The background corrected lithium measured by LA-ICPMS (circles) plotted against distance from the salt-crystal interface and model diffusion profile (curve) from a cross section of the 12-hour SCO12 experiment. At 1000oC and fO2 of NNO the measured lithium diffusion into olivine is log D = -14.1 (±0.12) m2/s, which is two orders of magnitude slower than the measured lithium diffusion into clinopyroxene at the same conditions, see text. Error bars are 2, based on analytical uncertainty of lithium concentration.
97
-12
-10
-8
-6
-4
-2
0
2
4
0 100 200 300 400 500 600
Kcpx2, Li out
1000oC, 2 hoursLi corrmodel
back
gro
und
cor
rect
ed
Li (
ppm
)
x (m)
-5.0
0.0
5.0
10
15
0 100 200 300 400 500 600 700
Kcpx2, Li out
1000oC, 2 hours
(7 Li/6 Li
- 7 Li
/6 Li c
ore
)/7 Li
/6 Li c
ore
* 1
000
m from rim to core
Figure 3.10 Lithium Diffusion and Isotopic Fractionation in Kcpx-2
(a) The background-corrected lithium (ppm) measured by LA-ICPMS (circles) plotted against distance from the salt-crystal interface and model diffusion profile (curve) from a cross section of Kcpx-2. (b) A plot of the7Li/6Li ratio (normalized by 7Li/6Li ratio of the core of the slab) measured by SIMS plotted against distance from the salt-crystal interface (m) and a model isotopic gradient (dashed curve). The solid line is the assumed core value (the original and unaltered 7Li/6Li ratio of the slab). Error bars are 2, in (a) based on analytical uncertainty of lithium concentration and in (b) based on counting statistics from SIMS analysis.
a)
b)
98
-22.0
-20.0
-18.0
-16.0
-14.0
-12.0
-10.0
-8.0
-6.0
4 8 12 16 20
log
DLi (
m2 /s
)
10,000/T (K)
20040060080010001400T (oC)
Si-crystal
albite & anorthite
cpx, this study
cpx, Coogan
ol, this study
Figure 3.11 Comparison of Lithium Diffusion Coefficients
A plot of log DLi as a function of10,000/T (K) for lithium diffusion in geologically significant minerals. Lithium diffusion, as calculated from the Do and Ea of the following studies: in a p-type Si-crystal (short-dash line) from Pell (1960), feldspar data (anorthite and albite; dashed-dotted line) from Giletti and Shanahan (1997), clinopyroxene (long-dash line) from Coogan et al. (2005) and this study (solid line). Also shown is the measurement of lithium diffusion in olivine (solid square) from this study.
99
-30.0
-25.0
-20.0
-15.0
-10.0
5 6 7 8 9 10 11
log
D (
m2 /s
)
10,000/T (K)
HLi, Coogan et al
Li, this study
Fe, Mn-Mg
Sr
Ca-Mg
Ca
Fe
Pb
Si
YbDyNdCeLu
U
Th
O
1400 1200 1000 800
T(oC)
Figure 3.12 Comparison of Diffusivities Measured in Pyroxene
A plot of log DLi vs. 10,000/T (K) for the lithium diffusivities measured in this study (solid circles) and that of Coogan et al. (2005, open circles) with experimentally determined diffusivities for other elements in clinopyroxene as calculated from the Do and Ea of those studies. Hydrogen data (H) are from Woods (2000), strontium (Sr) from Sneeringer et al. (1984), Fe, Mn-Mg interdiffusion from Dimanov and Sautter (2000), Ca-Mg interdiffusion from Brady and McCallister (1983), lead (Pb) from Cherniak (2001), Ca self diffusion (Ca) from Dimanov et al. (1996), iron (Fe) from Azough and Freer (2000), oxygen (O) from Ryerson and McKeegan (1994), uranium (U) and thorium (Th) from Van Orman (1998), REE (Yb, Dy, Nd, Ce, Lu) from Van Orman (2001) and silicon (Si) from Béjina and Jaoul (1996).
100
-3
-2
-1
0
1
2
3
4
1 10 100 1000
log
tim
e (
ho
urs
)
grain size (microns)
olivine
clinopyroxene
100 hrs
10 hrs
1 hr
10 min
30 sec
10 sec
1000 hrs
1 year
Figure 3.13 Retention of Lithium Composition
Centre retention time (hours) at 1000oC as a function of grain size (m) for spherical grains from 1 to 1000 m. Curves are calculated for Dta2 = 0.03 and represent the maximum time an olivine or clinopyroxene grain can remain at 1000 oC and retain its original core concentration, unaltered by diffusion. See text for details.
101
400
500
600
700
800
900
1 10 100 1000
T (
oC
)
characteristic radius (microns)
100o/Myr
10o/Myr
1o/Myr
dT/dt = 100o/MyrLiSr
Figure 3.14 Comparison of Closure Temperature of Li and Sr in Clinopyroxene
A plot of calculated closure temperatures (oC) versus characteristic grain size (m) for a spherical grain with cooling rates of 1o, 10o and 100 o/Myr. The solid curves are values calculated for lithium using the diffusion parameters measured in this study. The dashed curve is for Sr calculated using the diffusion parameters of Sneeringer et al. (1994). Note the large difference in the closure temperature for the two elements. Whereas the value for Sr (and other more highly charged cations) is more than 1000 oC higher for a given grain size and cooling rate.
102
0
1
2
3
4
5
6
7
8
0 5 10 15 20 25 30
7 Li
MgO (wt%)
Figure 3.15 Lithium Isotopic Compositions of Kilauea Iki Lava Lake Rocks
A plot of measured lithium isotopic composition of Kilauea Iki samples as a function of MgO content, which represents degree of magmatic differentiation. There is no significant fractionation observed in these samples; however, the expected isotopic fractionation due to Rayleigh distillation in a closed system is approximately 6-7 ‰, assuming ~0.999 (consistent with predicted from min-fluid fractionation at 1100oC). Kilauea Iki lava lake data from Tomascak et al. (1999). Error bars are based on the external reproducibility (±1.1 ‰)
103
-50
-40
-30
-20
-10
0
0 200 400 600 800 1000 1200
7 Li
microns from crystal edge
Figure 3.16 Li Isotopic Gradient in San Carlos Opx and Modeled Profile
7Li (‰) and lithium concentration (ppm) plotted against distance from crystal edge (m) for cross sections of a orthopyroxene xenolith from San Carlos, Arizona (closed symbols) and the modeled isotopic gradient (curve). The right half of the W-shaped profile observed in an orthropyroxene crystal from a San Carlos xenolith is modeled using the diffusivities and the (determined from this study. The lithium isotopic gradient shown here is calculated for 10 hrs, 1000oC and with a surface concentration/initial mineral concentration (Cs/Co) of 3.25.
104
4 Technique Development to Study Muscovite-Fluid Partitioning of Nitrogen
The following represents work that was undertaken in collaboration with Gray Bebout at Lehigh
University, Bethlehem, Pennsylvania. The results of this study demonstrated the feasibility of the
technique and provided the basis for a successful NSF application awarded to Bebout et al. in
2007.
4.1 Introduction
Nitrogen isotopes are often employed to unravel the mechanisms involved in a wide variety of
systems such as: N-cycling at convergent margins and recycling of surface material to the mantle
(Sadofsky and Bebout, 2004; Bebout and Fogel, 1992), the origins of fluids involved in orogenic
Au-deposits (Jia and Kerrich, 1999), the origin of the Earth’s atmosphere and hydrosphere and
their chemical evolution (Jia and Kerrich, 2004; Pinti et al., 2001; Sephton et al., 2002) and the
production and migration of hydrocarbons due to metamorphism of organic matter (Williams et
al., 1995).
Nitrogen has two stable isotopes, 14N and 15N, whose abundances are ~99.64 % and ~0.36 %,
respectively. Enrichments in nitrogen isotopes are described as:
(37)
10001NN/
NN/N
std1415
smp1415
15
where ‘smp’ refers to the sample and ‘std’ refers to the standard, typically atmospheric N2.
Nitrogen is a crucial nutrient in biological systems and its global cycle is thought of as primarily
a biological one. The present day nitrogen cycle is a result of the rise of oxygenic photosynthesis
and aerobic respiration (Falkowski and Godfrey, 2008). Biological cycling of nitrogen involves
the reduction of N2 to NH4+, for the synthesis of protein, as it is transferred from the atmosphere
and hydrosphere to the biosphere (Falkowski and Godfrey, 2008). A large portion (7.5 x 1020 g)
of the Earth’s nitrogen content is also stored in sedimentary rocks (Holloway and Dalhgren, 2002
and references therein). The transfer of nitrogen from surface reservoirs to the mantle and then
105
back to the atmosphere is an important component of the global cycle (Berner, 2006). Burial of
organic matter and NH4+ in sedimentary rocks and the subsequent weathering of these rocks will
cycle some nitrogen from the biosphere to the atmosphere (Holloway and Dahlgren, 2002).
Subduction of sedimentary rocks and transfer of nitrogen to the mantle from these rocks will also
result in the cycling of nitrogen back to the atmosphere via volcanic and metamorphic degassing
(Berner, 2006; Bebout and Fogel, 1992).
Nitrogen in the atmosphere occurs primarily as N2 with a 15N = 0 ‰ (Faure, 1986). Nitrogen in
the hydrosphere and in soils can occur as nitrate (NO3-), nitrite (NO2
-) ammonium (NH4+),
ammonia (NH3), oxides (NO, NO2 and N2O) and amino acids. Pelagic sediments are enriched in
nitrogen, ranging from 100 ppm N to as much as 2000 ppm N and have 15N ranging from +2 ‰
to +10 ‰ (Holloway and Dalhgren, 2002; Tolstikin, 1998), relative to mid-ocean ridge basalts
(MORB) which have 0.1 to 0.3 ppm N and 15N -5‰ (Marty and Humbert, 1997; Busigny et al.,
2005). Therefore, nitrogen isotopes can serve as an ideal tracer of surface-mantle interaction, as
well as a tracer of organic and sedimentary sources of fluids and melts in the mantle (Figure 4.1;
Hallam and Eugster, 1976; Sadofsky and Bebout, 2000; Bebout, 1997).
In silicate rocks, nitrogen occurs primarily as ammonium (NH4+), which replaces K+ (Hallam and
Eugster, 1976); however, 10-20 % of the ammonium in igneous rocks is extractable with weak
KCl solutions, suggesting that some of the ammonium may occur as soluble salts or introduced
by biological activity along grain boundaries (Faure, 1986). The primary mineral hosts for
nitrogen in igneous and metamorphic rocks are sheet silicates, and especially white micas
(Bebout et al., 1999; Bebout, 1997; Sadofsky and Bebout, 2000).
Various researchers have documented a correlation between nitrogen content, nitrogen isotopic
composition, and metamorphic grade of sedimentary rocks. Häendel et al. (1986) found that the
nitrogen content of both regional-metamorphic and contact-metamorphic rocks from Erzgebirge,
Germany, decreased from ~500 to ~20 ppm towards the contact zone, whereas the δ15N of the
rocks increased from ~ +5 ‰ up to +15 ‰ towards the contact. This correlation of nitrogen,
δ15N, and metamorphic grade was also reported in many other settings including: the contact
aureole of the Skiddaw Granite, Lake District, England (Bebout et al., 1999), the contact aureole
of the Cooma metamorphic complex, southeastern Australia (Jia, 2006), and the Catalina schist
subduction-zone metamorphic complex, California, USA (Bebout and Fogel, 1992). The
106
increasing δ15N and decreasing nitrogen content with increasing temperature (and therefore,
increasing metamorphic grade) has been interpreted as progressive release of nitrogen enriched
in 14N, leaving the residual rocks enriched in 15N (Häendel et al., 1986; Bebout et al., 1999; Jia,
2006).
To date, a systematic investigation of the degree of isotopic fractionation and the extent of
partitioning that occurs during dehydration reactions has been lacking. Both the isotopic
fractionation, 15Nsolid-fluid, and the nitrogen partitioning between mica and fluids needs to be
known if nitrogen contents are to be used to determine the extent of slab dehydration . This study
presents a technique to measure nitrogen partitioning and isotopic fractionation between fluids
and mica. With this information, more accurate models can be developed to constrain the extent
of slab dehydration.
4.2 Theoretical Considerations
4.2.1 N-speciation and Isotopic Fractionation
In crustal metamorphic and magmatic settings the most likely nitrogen species are N2 (Bebout,
1997) or NH3 (Equation 4, see below). Based on analyses of mineral separates from
metamorphosed rocks, many workers have postulated that NH4+ - N2 or NH4
+ - NH3 exchange
during devolatilization increases the δ15N in the residual metamorphic rocks (Bebout, 1997;
Sadofsky and Bebout, 2000; Bebout et al, 1999). That is:
(38) 14NH4+
musc + 15N14N,aq = 15NH4+
musc + 14N2,aq
(39) 14NH4+
musc + 15NH3,aq = 15NH4+
musc + 14NH3,aq
Using MINDO/3, a semi-empirical molecular orbital calculation method, Hanschmann (1981)
predicted a large difference in the isotopic fractionation between N-species in fluids (as N2 or
NH3) and NH4+ in solids (Figure 4.2). All previous studies have based their calculations on the
work of Hanschmann (1981), who calculated fractionation factors between N2, NH3, and NH4+
molecules. At any given temperature, the fractionation that occurs during NH4+ - NH3 is larger
than would occur at that temperature during NH4+ - N2 exchange. For example, at 500 oC NH4
+ -
NH3 exchange results in a 7 ‰ fractionation, whereas NH4+ - N2 results in only a 3 ‰
fractionation. To date, direct experimental measurements of isotopic fractionation during NH4+ -
NH3 or NH4+ - N2 exchange are lacking.
107
Therefore, knowledge and control of the fN2 and fNH3 of the experiments is necessary to
determine N speciation and understand the isotopic fractionation. Estimation of N- species is
possible given knowledge of the activity product ratios (K) for the speciation of water and N2 in
the experiment:
(40) H2 + ½O2 H2O K1 = fH2O / (fH2 * fO2 ½)
(41) ½ N2 + 3/2H2 NH3 K2 = fNH3 / (fN2 ½ * fH2
3/2)
(42) PT = Pgas = (fH2O/H2O) + (fH2/H2) + (fO2/O2) + (fN2/N2) + (fNH3/NH3)
fNH3 can be calculated from K2 (from JANF tables) if fH2 or fN2 are known.
One possible fN2-buffer is the Cr-CrN buffer (CCN) used by Hallam and Eugster (1976);
(43) CrN Cr + ½N2 K4 = fN2 ½ log fN2 = -2GoCrN/2.303RT
Assuming the fH2 external to capsule (the intrinsic fH2 of the reaction vessel), is equal to the
internal fH2, then both fH2 and fN2 are known. The CCN buffer can be used together with the
Inconel pressure vessels (which intrinsically generate an fO2 of approximately one log unit below
Ni-NiO; Matthews et al., 2003). The two buffers, NNO-1 and CCN, result in an fH2 such that fN2
<< fNH3 (Figure 4.3). In this manner, control of fH2 can function to buffer fN2 and fNH3. The
appropriate fH2 buffers needed such that fN2 >> fNH3 have yet to be determined; however, there
are a variety of solid materials (e.g. Cu metal, graphite) that can be used as spacers inside
pressure-vessels during experiments that can be employed to buffer redox conditions inside the
pressure vessel (Matthews et al., 2003).
Many studies have also found that the degree of fractionation is consistent with isotopic
exchange shifting from N2-dominated release to NH3-dominated release (Häendel et al., 1986;
Bebout and Fogel, 1992; Jia, 2006). For example, the data for greenschist and amphibolite (300-
600 oC) facies rocks of the Cooma metasedimentary complex have isotopic compositions that are
shifted only +1 ‰ from lowest grade metasediments, which suggest N2-NH4+ dominates the
isotopic exchange. Whereas the upper amphibolite facies (> 600oC) rocks are shifted almost +7
‰, which implies NH3-NH4+ fractionation (Jia, 2006). However, this process is not well
understood (Häendel et al., 1986; Bebout and Fogel, 1992; Jia, 2006). To date there is very little
108
information on the degree of isotopic fractionation that may occur during fluid-mineral
partitioning of nitrogen, and no experimental determinations have been made. These data are
essential for developing quantitative models of N-isotope systematics and N exchange during
fluid-mineral partitioning and constrain devolatilization processes.
4.2.2 Buffering pH
The experimental method employed needs to ensure mica stability. Previous experimental work
on micas has used KCl solutions to buffer pH and stabilize mica (Lynton et al., 2005):
(44) 1.5 k-spar + H+ 0.5 mus + 3 qtz + K+
However, this reaction generates potassium feldspar (k-spar), which also takes up NH4+. This
introduces some ambiguity to the mass balance assumptions necessary for quantitative
determination of partitioning and fractionation; therefore, keeping the proportion of k-spar to a
minimum is necessary. NH4Cl is ideal because it serves both as a nitrogen source and potentially
provides a way to buffer the pH, which is necessary for subsequent fluid speciation calculations.
(45) NH4+ ↔ NH3 + H+
By adjusting the concentration of the NH4Cl solution, the pH of the system can be controlled.
For example, a 1 M NH4Cl solution has a pH of 4.6, and a 0.1 M solution has a pH of 5.1. As
Equation 45 shows, a fixed pH would also constrain the nitrogen speciation.
Another method to stabilize mica is to use mineral mixtures with a very high ratio of finely
powdered muscovite to quartz and k-spar (50:1:1; Table 4.1). The powdered mica would
promote fluid-mineral exchange, probably by some combination of dissolution/re-precipitation
and diffusive exchange. The dissolution of some of the mica would also serve to buffer the pH of
the solution. To determine the contribution of the k-spar to the nitrogen content of the solid
residue, experiments using variable ratios of k-spar to quartz + muscovite can be conducted.
Another method to assess the contribution from potassium feldspar would use larger mica
fragments. These larger fragments would undergo diffusive exchange with the fluid, and then be
mechanically separated from the residual solid after the experiment.
109
4.3 Experimental Methodology
Experiments to measure N isotopic fractionation involved equilibrating mica, of known N-
isotopic composition, with a solution of known N-isotopic composition using the cold-seal
pressure vessels at the University of Toronto's High Pressure Lab. All experiments utilized a
natural muscovite from Ontario, Canada (40 ppm N, 15N = +6.2 ‰), and synthetic quartz and
orthoclase that were ground and screened to 200-300 mesh, as a starting material. Isotopically
characterized NH4Cl (15N = -4.1 ‰) was combined with ultra-pure water to make a 1M
solution. Table 4.1 lists the details of the preliminary experiments and the initial results.
Experiments using mineral mixtures with a very high ratio of muscovite to quartz and k-spar
(50:1:1; Table 4.1), were conducted. For these experiments, N-02, N-07, N-08, and N-09 the
muscovite was ground and screened to 100-200 mesh.
Another strategy to ensure mica stability and limit interference from potassium feldspar used
larger mica fragments. These fragments, which would undergo diffusive exchange with the fluid,
could be mechanically separated from the residual solid after the experiment. Experiments N-04
and N-05 utilized 3 x 10 mm flakes of muscovite.
The sample mixtures and the 1M NH4Cl solution were loaded into Pt-capsules (5 mm OD x 1.5
cm height) and sealed. The whole assembly was weighed, placed in a drying oven for several
hours, and weighed again to determine if the welding had resulted in an airtight seal. The
capsules were loaded into Inconel pressure vessels, pressurized to 15,000 psi (~1 kbar), and
heated while the pressure was adjusted accordingly to maintain 15000 psi. Experiments were
quenched from maximum temperature to room temperature in 4 min; this was accomplished by
cutting power to the heaters and alternately misting the pressure vessels with water and air-
cooling them with pressurized air. The capsules were recovered; the samples extracted and rinsed
with ultra pure water to remove NH4Cl precipitate and quench products.
4.4 Analytical Methods
All samples were analyzed at Lehigh University on a Finnigan MAT 252 isotope ratio mass
spectrometer, using the new Gas Bench II metal-high vacuum extraction line (Bebout et al.,
2007). Powdered samples were loaded, along with varying amounts of the Cu/CuOx reagent, into
110
6 mm (o.d.) quartz tubes that were previously combusted in atmosphere at 550°C for 2 h to
remove organic and other contaminants. Samples were evacuated overnight, heated with heating
tape to 100°C, and then sealed under vacuum. The tubes were then heated to 950–1050°C in a
programmable furnace, held at these peak temperatures for at least 3 h, then cooled slowly,
particularly through the 700–500°C temperature range, to ensure proper speciation of gases as
H2O, CO2, and N2 (Bebout and Fogel, 1992). After loading the sealed tube onto the tube cracker,
and at least 1 h of evacuation, the tube was cracked and the released gas expanded into a series of
traps to remove any condensable gases, presumably mostly H2O and CO2 but also some Ar and
other minor contaminants. After passing through the last trap, the gas was expanded and isotopic
analyses were undertaken using the GBII system. Nitrogen concentrations are obtained by
measurement of voltage on the m/z 28 peak for calibrated volumes in the mass spectrometer
(either a variable-volume/bellows or micro-volume inlet); voltage vs. moles N2 is calibrated by
extractions and analyses of non-silicate standards with known N contents. A 1 of 0.14 ‰ and
0.12 nmol for analyses of 12-225 nmol of N2 is reported by Bebout et al. (2007) for this system.
4.5 Results
4.5.1 Nitrogen Partitioning
The addition of K was found to promote potassium feldspar growths (k-spar) as minute
intergrowths of ms + k-spar on the surface of the mica, which could not be mechanically
separated. The potassium feldspar forms clusters of idiomorphic, monoclinic crystals 2-5 m in
length near the edges of the muscovite sheet, as well as where the surface of the muscovite has
formed ‘steppes’ either due to dissolution or re-precipitation (Figure 4.4).
Another method employed to ensure mica stability and limit interference from potassium
feldspar used larger mica fragments. These fragments diffusively exchanged K+ and NH4+ with
the fluid, and were mechanically separated from the residual solid after the experiment.
However, experiments that used this technique (N-04, N-05; Table 4.1) produced unsatisfactory
results. The mica fragments were found to have lower nitrogen contents than experiments
utilizing finely powdered mica. The samples that utilized muscovite pieces as a starting material
contained ~250 ppm N, whereas the experiments that had muscovite powder as a starting
material contained ≥1000 ppm N (Figure 4.5).
111
The experiments using mineral mixtures with a very high ratio of muscovite to quartz and k-spar
(50:1:1; Table 4.1), resulted in some loss of sample (presumably dissolved in solution during the
experiment); however, they resulted in a remarkably reproducible nitrogen content (1250 ±182
ppm). This suggests that using finely powdered mica promotes fluid-mineral exchange, probably
by some combination of dissolution/re-precipitation and diffusive exchange.
4.5.2 Nitrogen Isotopic Fractionation
Experiments, in which muscovite (δ15N = +6.2 ‰) was equilibrated with a 1 M NH4Cl solution
(δ15N = -4.1 ‰) resulted in solid residues that showed an overall decrease in δ15N (δ15N = +0.35
‰, +1 ‰, +1.5 ‰; Table 4.1, Figure 4.6). Consistent with previous work, which has shown that 15N is preferentially concentrated in the solid relative to the fluid, the final solid in the
experiments has a more positive δ15N than the NH4Cl of the fluid. The apparent mica-fluid
fractionation is ~4.5 ‰ to 5.6 ‰, assuming that the fluid composition is constant and equal to
the NH4Cl. The work of Hanschmann (1981) predicts a fractionation of 2.5 ‰ between N2,aq-
NH4+ and ~8 ‰ between NH3,aq-NH4
+ at 500oC.
An average mica-fluid = +4.8 (+0.6) ‰ is calculated from this preliminary data (Figure 4.7). The
magnitude of 15Nmica-fluid measured in this study falls between the NH4+-N2 and NH4
+-NH3
curves determined by Haschmann (1981). Because the speciation of N was not controlled in this
study, it is uncertain if the intermediate 15Nmica-fluid measured here is a result of isotopic fraction
between some mix of N2 and NH3 species or isotopic fractionation due to quench and sample
retrieval procedures. This result is very encouraging as it is consistent with the shift observed in
the δ15N of mica samples from the Catalina Schist, where the residual solids are enriched in 15N
with respect to the fluids released by dehydration (Bebout, 1997). This is also consistent with the
behaviour of N-isotopes in metamorphic rocks in the contact aureole of the Skiddaw Granite,
Lake District, England (Bebout et al., 1999), the contact aureole of the Cooma metamorphic
complex, southeastern Australia (Jia, 2006), and the regional-metamorphic and contact-
metamorphic rocks from Erzgebirge, Germany (Häendel et al., 1986). The magnitude of the
isotopic shifts measured in these rocks range from ~1 ‰ to 7 ‰, which is consistent with
fractionations occurring between the NH4+-N2 and NH4
+-NH3 curves (as determined by
Hanschmann, 1981). It is interesting to note that natural samples have calculated 15Nmica-fluid
values that also fall between the two curves as well, this phenomenon has been noted by several
112
researchers and has been attributed to a shift in the speciation of the fluids; however, this process
is not well understood (Häendel et al., 1986; Bebout and Fogel, 1992; Jia, 2006).
4.6 Discussion
4.6.1 Utility of NH4Cl as Nitrogen Source
The use of NH4Cl seemed ideal because it would serve both as a nitrogen source and provide a
way to buffer the pH, which is necessary for subsequent fluid speciation calculations. However,
a nitrogen source that does not cause precipitation of N-rich solids would be preferable because
it would allow for ‘on-line’ (see below) analysis of vapor, liquid and solid components of the
experiment. Silver azide, AgN3, has been employed in previous studies as a N-source (Keppler,
1989), and has the benefit of producing a large amount of N, and a relatively inert residue (Ag
metal). However, at unconfined conditions, such as in an improperly sealed capsule, AgN3
releases N explosively making it rather hazardous in the laboratory, both for laboratory personnel
and equipment. Anovitz et al. (1998) have evaluated a variety of solid-N sources for
experiments. They concluded that Cu3N is an ideal N source; the release of N is much slower
than AgN3 and it is inexpensive and commercially available. They also examined CrN and
concluded that it did not dissociate readily enough at metamorphic temperatures to generate a
sufficient amount of nitrogen; however, the benefit of using CrN is that the CrN-N buffer fixes
the relative fugacities of N2 and NH3 in the fluid (see section 4.2.2). It is important to note that
using CrN, AgN3 or Cu3N as a nitrogen source would require some strategy to buffer pH and
stabilize muscovite. Because a satisfactory alternative to using NH4Cl as a pH buffer has not yet
been determined, a 1 M NH4Cl solution was used as the nitrogen source for these experiments.
Considering the difficulties that were encountered with the NH4Cl residue during analyses (see
below) the experimental method will ultimately need to make use of one of the other nitrogen
sources discussed.
4.6.2 Analytical Considerations
Originally, it was planned to pierce the capsule on-line, step-heat, and measure the isotopic
composition of the gas, liquid and solid components as they were released to determine the
fractionation by measuring each phase. This method would forgo the rinsing, powdering and
combusting of the sample, thereby reducing change for contamination and having the ability to
113
analyze all run-products, not just the solid. To this end a device was constructed that would
enable the capsules to be punctured and heated on the extraction line (Figure 4.8); however,
difficulties were encountered with the discovery of a NH4Cl residue that formed on the inner
surface of the puncturing tool and inside the capsule. The use of NH4Cl as a N-source and a pH
buffer, while desirable from an experimental point of view, resulted in complications during the
analysis. A natural consequence of taking a solution from high pressure and temperature to room
temperature and pressure is that the solubility of various components of the solution (such as
dissolved gases and salts, as well as hydrous silica and alumina complexes) decreases
dramatically. This change in solubility is accommodated by the exsolution of the dissolved gases
and the precipitation of salts and amorphous silica and alumina complexes – described as quench
products. Typically, the run products of interest are the solid phases and these are recovered by
puncturing the capsule, evaporating off the solution (which results in further precipitation of the
remaining solute load) and rinsing away the evaporite and quench products.
In puncturing and extracting the run products ‘on-line’, all the run products (vapor phase, liquid
phase and solid phase), are of interest. However, it is necessary to ascertain what proportion of
the total N component this NH4Cl residue represents. That is, what proportion of this solid was
dissolved in the solution at the conditions of the experiment and how much precipitated during
quenching? Was some of the nitrogen dissolved in the vapor phase at the conditions of the
experiment? It is uncertain whether the NH4Cl residue was the result of precipitation during
quenching of the experiment or evaporation due to heating during analysis, (it is most likely a
combination of both). This also complicated the N analysis of vapor phase, as there is likely
isotopic fractionation occurring between the NH4Cl dissolved in the solution and the coexisting
liquid/vapor phase during the exsolution of the gas phase at quench. Due to the ambiguities
regarding the nature of the vapour and liquid phases, partitioning was determined from the
composition of the solid phase (with any residual NH4Cl rinsed away).
4.6.3 Experimental Considerations
The mica fragments were found to have lower nitrogen contents than experiments utilizing finely
powdered mica. The samples that utilized muscovite pieces as a starting material contained ~250
ppm N, whereas the experiments that had muscovite powder as a starting material contained ≥
1000 ppm N. The discrepancy between the results from muscovite powder experiments and the
114
muscovite flake experiments may be a result of nitrogen exchanging faster in the smaller,
powdered muscovite grains. This suggests that equilibration had not been reached in the larger
grain sizes by 6 days at 550oC, which is consistent with the sluggish exchange kinetics
documented in past studies (Lynton et al., 2005; Bos et al., 1988). Another possibility is that the
orthoclase is contributing nitrogen to the analysis. The ambiguity of these results highlights the
importance of quantifying the amount of nitrogen incorporated into orthoclase and its effect on
the mass balance of the experiment. Even if complete separation of orthoclase and muscovite is
not possible, a control on the amount of orthoclase would at least allow an evaluation of the
contribution of this mineral to the overall nitrogen content of the run product.
4.6.4 Isotopic Fractionation Experiments and Atmospheric Contamination
A complicating factor in calculating mica-fluid fractionation by mass balance is the presence of
trapped atmospheric nitrogen (δ15N = 0 ‰) in the capsule. This amount of added nitrogen would
be variable and difficult to control. In the above experiments, trapped nitrogen would result in a
positive shift of the overall composition of the fluid. The magnitude of the fractionation
calculated neglecting the additional nitrogen-source would represent a maximum value. It is
possible to calculate the percentage of nitrogen contribution from trapped air in the capsule using
estimated capsule volumes and the known atmospheric nitrogen concentration. Estimates of the
contribution from atmospheric nitrogen to the mass balance are provided in Table 4.2. The above
calculations also show that an increase in the concentration of the NH4Cl solution of up to 3 M
would produce enough nitrogen to buffer the solution against the atmospheric contribution, and
provide an ‘infinite reservoir’ of constant N-isotopic composition. For the mass of solution used
and the estimated capsule volume, the composition of the solution is estimated to have shifted a
maximum of +0.2 ‰ from the value for the pure NH4Cl solution; this would result in a
maximum mica-fluid of +5.4 (+0.6) ‰.
4.7 Suggestions for Future Work
This study has produced promising new results for nitrogen partitioning and isotopic
fractionation between aqueous fluids and muscovite. This work also highlights the challenges in
making these measurements and their interpretations. From an experimental design point of
view, minimizing the contribution from orthoclase to the nitrogen content during analysis,
115
ensuring mica stability and equilibrium between fluids and mica is established, could be
circumvented by growing mica in a nitrogen-rich environment. This technique has been
successfully utilized by Pöter et al. (2004) and has not been investigated here. From an analytical
point of view, the choice of a N-source needs to be carefully considered so as not to add
ambiguity during the analysis. A solid source such as Cu3N (Anovitz et al., 1998) is attractive
because the residue (Cu) is relatively inert and would not result in further reactions/fractionation
during quench or analysis.
The experiments conducted in this study have demonstrated that measurable fractionation occurs
between fluids and micas equilibrated at metamorphic conditions. The magnitude of
fractionation is consistent with estimates from analysis of metasedimentary rocks (Häendel et al.,
1986; Bebout and Fogel, 1992; Bebout, 1997; Jia, 2006) and theoretical estimates based on N2-,
NH3-NH4+ calculations (Hanschmann, 1981). The above techniques can be utilized to better
constrain the N-isotopic fractionation that occurs during fluid-rock interaction. With better
constraints, the relationship between N and δ15N might be used to estimate metamorphic grade
for rocks with no indicator mineralogy, or provide insight to the degree of fluid-rock interaction
or retrograde metamorphism. This data would allow the δ15N content of metasedimentary
convergent margin rocks to be used to better quantify the composition of material re-cycled deep
into the mantle, pinpoint the origins of fluids involved in orogenic Au-deposits or even further
our understanding of the evolution of the Earth’s atmosphere and oceans.
116
Table 4.1 Experiments and Results
Final
Name T (oC) t (days)
Composition (wt. %)
Mica (mg)
sol'n (mg)
prep ppm N 15N M-F
N-02 550 5 1 mus pdr : 1 qtz : 4 orth 46 50 rinsed 900
N-04 550 6 1 mus pcs : 1 qtz : 3 orth 72 rinsed - qtz/orth removed 270
N-05 650 6 1 mus pcs : 1 qtz : 4 orth 70 rinsed - qtz/orth removed 230
N-07 500 6 50 mus pdr : 1 qtz : 1 orth 47 60 30.1 mg recovered/rinsed 1302 +0.35 4.5
N-08 500 16 50 mus pdr : 1 qtz : 1 orth 51 57 25.7 mg recovered/rinsed 1407 +1 5.1
N-09 500 10 50 mus pdr : 1 qtz : 1 orth 61 58 31.5 mg recovered/rinsed 1052 +1.5 5.6
Table 4.2 Percentage of Nitrogen Contribution from Air
% atmospheric N2 in capsule
solution concentration
composition of atmospheric N2 and
1MNH4Cl solution mix in capsule
Solution added to capsule (mg)
1 M 2 M 3 M 15N
50 11 % 6 % 4 % -3.9
75 7 % 4 % 3 % -4.0
100 6 % 3 % 2 % -4.0
117
0.1 1 10 100 1000 10000
Marine Seds
AOC
MORB
Seawater
Lakes, Rivers
N content (ppm)
a)
-10 -5 0 5 10 15 20 25 30
Marine Seds
AOC
MORB
Seawater
Lakes, Rivers
15N
b)
Figure 4.1 Summary of N Concentration and Isotopic Composition
The nitrogen content (ppm) a) and isotopic composition b) of terrestrial reservoirs. Data for seawater, lakes, rivers from Faure (1986), nitrogen content of marine sediments from Holloway and Dahlgren (2002), data for MORB from Busigny et al. (2005), and isotopic composition of marine sediments from Tolstikin (1998).
118
2 4 6 8 10 12300
400
500
600
700
800
15N
T (
oC
)
NH4
+-NH3
NH4
+-N2
Figure 4.2 Calculated N2-, and NH3-NH4+ Fractionation
Plot of the isotopic fractionation as calculated from quantum theory that occurs during NH4+ - N2 and NH4
+- NH3 exchange as a function of temperature. Data from Hanschmann (1981) is based on isotopic fractionation predicted to occur between N2, NH3 and NH4
+ molecules.
119
(i)
(ii)
Figure 4.3 Relationship of fH2, fN2, and fNH3
(i) Plot of resulting log fN2/fNH3 versus temperature when using CCN buffer Inconel pressure vessel to buffer fH2
(ii) (ii) Plot of fH2 versus temperature showing lines of fNH3/ fN2 ratios
120
Figure 4.4 Scanning Electron Micrograph of Muscovite Texture
Scanning electron micrograph of Ontario Muscovite sample heated to 540oC and 1.5 kbar with 1 M KCl solution for seven days. K-feldspar forms clumps of 2-5 m long monoclinic crystals on the surface of the mica sheet.
Mus
Mus + k-spar
121
0
200
400
600
800
1000
1200
1400
1600
4 6 8 10 12 14 16 18
500oC powder
550oC powder
550oC piece
650oC piece
Nit
rog
en (
pp
m)
time (days)
Figure 4.5 Nitrogen contents of run products
Plot of measured nitrogen content (ppm) of run product micas versus experiment duration (days). Error bars are 2 and represent the error of the analysis (10 %). The experiments using powdered muscovite (circles) contained the highest nitrogen contents suggesting that experiments using larger pieces of mica (squares) have not equilibrated with respect to nitrogen content of the solution. The experiments using powdered mica as a starting material have a constant N content in experiments conducted for times of 6 to 16 days, suggesting that these samples have reached equilibrium values.
122
-5
-2.5
0
2.5
5
7.5
10
4 6 8 10 12 14 16 18
15N
(o
/oo
)
time (days)
NH4Cl
ont mus
Figure 4.6 Nitrogen isotopic compositions of run products
Plot of the measured nitrogen isotopic composition of experiments (15N in ‰) versus experiment duration (in days). Also shown for reference is the isotopic composition of the starting material, Ontario Muscovite and the NH4Cl used for the solution. Consistent with the N content measurements, isotopic equilibrium appears to have been reached, as the isotopic composition does not change with experiment duration from 6 to 16 days. 2 errors (approximately symbol size) are ±0.2 ‰ and represent the error of the analysis.
123
2 4 6 8 10 12300
400
500
600
700
800
15N
T (
oC
)
N2-NH
4
+
NH3-NH
4
+
Figure 4.7 Isotopic shifts of run products
Plot of temperature versus 15N between NH4+, N2 and NH3 molecules. Curves constructed with data from
Hanschmann (1981). Open circles are measured 15N from 500oC experiments. 2 errors (approximately symbol size) are ±0.6 ‰ and represent the error of the analysis propagated through the calculation of 15N (15Nmin-
15Nfluid).
124
Figure 4.8 Puncturing Device
Puncturing device constructed at the University of Toronto Physics Machine shop based on plans generously provided by Ethan Baxter. The sample is placed in the sample well; the device is sealed and then attached to the nitrogen extraction line. The sample is punctured by turning the screw, which lowers the puncturing tool, then the sample is heated (by means of an electrical heating coil or heating tape) to extract the nitrogen.
125
5 Summary of Results and Conclusions
This study has found lithium to be moderately incompatible in the mantle during mineral – fluid
exchange reactions. The measured DLi ranges from 1.34 – 0.17 in olivine, to 0.32 – 0.09 in
plagioclase and, 0.32 – 0.04 in clinopyroxene. Lithium partitioning between clinopyroxene and
hydrous fluids is a function of temperature, decreasing with increasing temperature from 800oC
to 1100oC at 1 GPa, and appears to be a function of clinopyroxene Al2O3 content. Lithium
partitioning between olivine and fluid is not strongly a function of temperature, but appears to be
sensitive to FeO content. Lithium partitioning in anorthite is a function of feldspar composition,
similar to the partitioning of other cations in the feldspar-fluid system. Lithium partitioning
between olivine and clinopyroxene is independent of temperature; however, preliminary
experiments examining the effect of REE content and fO2 suggest that DLiol/cpx may be a function
of crystal chemistry.
Additionally, isotopic fractionation between clinopyroxene + fluid and olivine + clinopyroxene
has been measured. The isotopic fractionation between clinopyroxene and fluid at 900oC is ~ +1
‰ (±2 ‰) and the measured isotopic exchange between olivine and clinopyroxene is ~ +5 ‰
(±4 ‰). Isotopic fractionation between clinopyroxene and fluids is a function of temperature and
consistent with what has been observed in the spodumene – fluid system. The fractionation
between spodumene and hydrous fluids results in an enrichment of 7Li in the fluid from 3.5 ‰ at
500oC to ~1.0 ‰ at 900oC and 2.0 GPa (Wunder et al., 2006).
Given the incompatibility in mantle minerals, and the slight fractionation that occurs during
clinopyroxene-olivine-fluid interaction, the lithium content and isotopic signature of slab-derived
fluids can be significantly modified during transport through and interaction with the mantle
wedge. Because diffusion is so rapid, complete equilibration of the fluid with the mantle wedge
can be assumed if fluid transport from the slab to the melt source occurs by percolation;
therefore, characteristics of the lithium signal, such as the Li/Y ratio and the isotopic
composition, can provide some insight to the mechanism of transport. For example, the absence
of high Li/Y ratios in arc lavas with high B/Be, or MORB-like δ7Li in lavas with high B/Be
contents (such as the lavas from the Sunda arc, Indonesia; Tomascak et al., 2002), can be
126
explained by partitioning lithium into mantle minerals as fluids percolate through the mantle
wedge. In these cases, transport through the mantle wedge completely removes the lithium signal
from the slab-derived fluid. In cases where island arc lavas have δ7Li values greater than the
mantle values (δ7Li ~ +4 ‰), such as the Izu fore arc lavas, significant slab fluid-mantle
interaction has likely occurred, as would be the case if transportation was by percolation. It is
important to note that very high fluid fluxes are implied if a lithium signal from slab-derived
fluids is to reach the melt source by percolation. When low δ7Li values (< MORB; δ7Li ~ +4 ‰)
correspond with high Li/Y ratios the fluids transported to the melt source with a minimum
amount of interaction with the mantle wedge, and transport through hyrdofractures is a likely
mechanism.
The trend of increasing δ7Li with decreasing Li/Y, which is observed in most arc lavas
(Tomascak et al., 2002), could be viewed as a spectrum between the two scenarios. Where low
Li/Y values correspond with high δ7Li, large fluid fluxes were most likely percolating through
the mantle wedge. Where high Li/Y values correspond with low δ7Li, the fluids were likely
generated at depth and transported through the mantle through hydrofractures, having minimal
interaction with the wedge. Intermediate values could be a result of some component of both
these mechanisms.
Transport of slab-derived fluids through hydrofractures in the mantle can also explain the lack of
clear and consistent correlations between lithium and other fluid mobile elements. Fluids
transported to the melt source through hydrofractures would be subject to differing degrees of
mantle interaction (variable fluid/rock ratios and transport velocities). Lithium is moderately
compatible in the mantle and diffuses rapidly; therefore, lithium contents and isotopic
compositions will be very sensitive to variations in mineral-fluid interaction.
The diffusion coefficient of lithium in clinopyroxene measured in this study is temperature
dependent, increasing from -15.19 ± 2.86 m2/s at 800oC to -11.97 ± 0.86 m2/s at 1000oC. These
diffusion coefficients are consistent with those determined by Coogan et al. (2005) for 6Li in
clinopyroxene. Lithium diffusion is independent of the diffusion gradient; values are the same if
the flux of lithium is into the crystal or out of the crystal. The lithium diffusion coefficient in
clinopyroxene has a slight negative dependence with fO2, which suggests a component of
interstitial diffusion. A single measurement of lithium diffusion in olivine was made at 1000oC
127
and an fO2 of NNO, the measured lithium diffusion into olivine is log D = -14.1 (m2/s), two
orders of magnitude slower than the measured lithium diffusion into clinopyroxene at similar
conditions.
More significantly, isotopic fractionation of lithium isotopes can occur as a result of diffusion in
a mineral. This means that even the isotopic composition of lithium grains can be modified over
very short timescales. If a component of interstitial diffusion exists, as this study suggests, then
the isotopic diffusivity ratio, D6/D7 may increase with increasing temperature.
Closure temperatures calculated for lithium diffusion in clinopyroxene range from ~400 to
~500oC. Considering the rapid cooling rate and increased closure temperatures, the timescales
required for re-equilibration of lithium isotopes (weeks) are much less than the timescales
required for cooling and crystallization of magmatic rocks (kyr-Ma). These results suggest that
isotopic fractionation during crystallization and magmatic differentiation is unlikely to be
preserved due to rapid diffusion and re-homogenization of lithium isotopic compositions. The P-
T history of the samples must be evaluated before lithium signatures are interpreted. It is likely
that the lithium content of minerals can only reliably represent chemical exchange in the very
latest stages of the sample’s history, or if there is no inter-granular reservoir for lithium
exchange.
This study has developed new and promising techniques to measure isotopic fractionation of
nitrogen between muscovite and fluids. These experiments have demonstrated that measurable
fractionation occurs between fluids and micas equilibrated at metamorphic conditions. These
results are very encouraging as they are consistent with the shift observed in the δ15N of
metasedimentary rocks (Häendel et al., 1986; Bebout and Fogel, 1992; Bebout, 1997; Jia, 2006)
and theoretical estimates based on N2-, NH3-NH4+ calculations (Hanschmann, 1981). The above
techniques can be utilized to better constrain the N-isotopic fractionation that occurs during
fluid-rock interaction. This data would allow the δ15N content of metasedimentary convergent
margin rocks to be used to better quantify dehydration processes during subduction, pinpoint the
origins of fluids involved in orogenic Au-deposits or even further our understanding of the
evolution of the Earth’s atmosphere and oceans.
128
References
Abart, R. (1995). Phase equilibrium and stable isotope constrains in the formation of metasomatic garnet-vesuvianite veins (SW Adamello, N Italy). Contributions to Mineralogy and Petrology, 122: 116-133.
Angel, R. J., R. M. Hazen, T. C. McCormick, C. T. Prewitt, and J. R. Smyth. (1988). Comparative compressibility of end-member feldspars. Physics and Chemistry of Minerals, 15 (4): 313-318.
Anovitz, L. M., J. G. Blencoe, D. B. Joyce, and J. Horita. (1998). Precise measurement of the activity/composition relations of H2O-N2 and H2O-CO2 fluids at 500oC, 500 bars. Geochimica et Cosmochimica Acta, 62 (5): 815-829.
Aulbach, S., R. Rudnick, and W. F. McDonough. (2008.) Li-Sr-Nd isotope signatures of the plume and cratonic lithospheric mantle beneath the margin of the rifted Tanzanian craton (Labait). Contributions to Mineralogy and Petrology, 155 (1): 79-92.
Ayers, J. C. (1998). Trace element modeling of aqueous fluid; peridotite interaction in the mantle wedge of subduction zones. Contributions to Mineralogy and Petrology, 132 (4): 390-404.
Ayers, J.C., J. B. Brenan, E. B. Watson, D. A. Wark, and W. G. Minarik. (1992). A new capsule technique for hydrothermal experiments using the piston-cylinder apparatus. American Mineralogist, 77 (9-10): 1080-1086.
Azough, F., and R. Freer. (2000). Iron diffusion in single-crystal diopside. Physics and Chemistry of Minerals, 27 (10): 732-740.
Barrat, J. A., M. Chaussidon, M. Bohn, Ph. Gillet, C. Göpel and M. Lesourd. (2005). Lithium behavior during cooling of a dry basalt; an ion microprobe study of the lunar meteorite northwest Africa 479 (NWA 479). Geochimica et Cosmochimica Acta, 69 (23): 5597-5609.
Bebout, G. E. (1997). Nitrogen isotope tracers of high-temperature fluid-rock interactions; case study of the Catalina Schist, California. Earth and Planetary Science Letters, 151 (1-2): 77-90.
Bebout, Gray E., and M. L. Fogel. (1992). Nitrogen-isotope compositions of metasedimentary rocks in the Catalina schist, California; implications for metamorphic devolatilization history. Geochimica et Cosmochimica Acta, 56 (7): 2839-2849.
Bebout, G. E., J. G. Ryan, W. P. Leeman, and A. E. Bebout . (1999). Fractionation of trace elements by subduction-zone metamorphism; effect of convergent-margin thermal evolution. Earth and Planetary Science Letters, 171 (1): 63-81.
Bebout, G. E., B. D. Idelman, L. Li, and A. Hilkert. (2007). Isotope-ratio-monitoring gas chromatography methods for high-precision isotopic analysis of nanomole quantities of silicate nitrogen. Chemical Geology, 240 (1-2): 1-20.
Beck, P., J. A. Barrat, M. Chaussidon, Ph. Gillet, and M. Bohn. (2004). Li isotopic variations in single pyroxenes from the northwest Africa 480 shergottite (NWA 480); a record of degassing of Martian magmas? Geochimica et Cosmochimica Acta, 68 (13): 2925-2933.
129
Béjina, F., and O. Jaoul. (1996). Silicon self-diffusion in quartz and diopside measured by nuclear micro-analysis methods. Physics of the Earth and Planetary Interiors, 97: 145-162.
Bennett, S. L., J. Blundy and T. Elliott. (2004). The effect of sodium and titanium on crystal-melt partitioning of trace elements. Geochimica et Cosmochimica Acta, 68 (10): 2335-2347.
Benton, L.D., J. G. Ryan, and I. P. Savov. (2004). Lithium abundance and isotope systematics of forearc serpentinites, conical seamount, Mariana forearc; insights into the mechanics of slab-mantle exchange during subduction. Geochemistry, Geophysics, Geosystems – G3, 5 (8).
Benton, L. D., J. G. Ryan, and F. Tera. (2001). Boron isotope systematics of slab fluids as inferred from a serpentine seamount, Mariana forearc. Earth and Planetary Science Letters, 187 (3-4): 273-282.
Berlo, K., J. Blundy, S. Turner, K. Cashman, C. Hawesworth, and S. Black. (2004). Geochemical precursors to volcanic activity at Mount St. Helens, USA. Science, 306 (5699): 1167-1169.
Blundy, J. D., and J. Dalton. (2000). Experimental comparison of trace element partitioning between clinopyroxene and melt in carbonate and silicate systems, and implications for mantle metasomatism. Contributions to Mineralogy and Petrology, 139 (3): 356-371.
Blundy, J. D., and B. J. Wood. (1991). Crystal-chemical controls on the partitioning of Sr and Ba between plagioclase feldspar, silicate melts, and hydrothermal solutions. Geochimica et Cosmochimica Acta, 55 (1): 193-209.
Blundy, J. D., J.A. C. Robinson, and B. J. Wood. (1998). Heavy REE are compatible in clinopyroxene on the spinel lherzolite solidus. Earth and Planetary Science Letters, 160 (3-4): 493-504.
Bos, A., W. Duit, Ad. M. J. van der Eerden, J. Jansen, H. Ben. (1988). Nitrogen storage in biotite – An experimental study of the ammonium and potassium partitioning between 1M-phlogopite and vapor at 2 kb. Geochimica et Cosmochimica Acta, 52 (5): 1275-1283.
Boyd, F. R., and J.L. England. (1960). Apparatus for phase-equilibrium measurements at pressures up to 50 kilobars and temperatures up to 1750oC. Journal of Geophysical Research, 65 (2): 741-748.
Brady, J. B., and R. H. McCallister. (1983). Diffusion data for clinopyroxenes from homogenization and self-diffusion experiments. American Mineralogist, 68 (1-2): 95-105.
Brearley, M. and C. M. Scarfe. (1986). Dissolution rates of upper mantle minerals in an alkali basalt melt at high pressure: an experimental study and implications for ultramafic xenolith survival. Journal of Petrology, 27 (5): 1157-1182.
Brenan, J.M., E. Neroda, C. C. Lundstrom, H. F. Shaw, F. J. Ryerson, and D. L. Phinney. (1998a). Behaviour of boron, beryllium, and lithium during melting and crystallization constraints from mineral-melt partitioning experiments. Geochimica et Cosmochimica Acta, 62 (12): 2129-2141.
130
Brenan, J.M., F. J. Ryerson, and H. F. Shaw. (1998b). The role of aqueous fluids in the slab-to-mantle transfer of boron, beryllium, and lithium during subduction experiments and models. Geochimica et Cosmochimica Acta, 62 (19-20): 3337-3347.
Brooker, R. A., R. H. James, and J. D. Blundy. (2004). Trace elements and Li isotope systematics in Zabargad peridotites; evidence of ancient subduction processes in the Red Sea mantle; lithium isotope geochemistry. Chemical Geology, 212 (1-2): 179-204.
Busigny, V., C. Laverne, and M. Bonifacie. (2005). Nitrogen content and isotopic composition of oceanic crust at super fast spreading ridge: A profile in altered basalts from ODP Site 1256, Leg 206. Geochemistry, Geophysics, Geosystems G3, 6 (12).
Chacko, T., D. R. Cole and J. Horita. (2001). Equilibrium oxygen, hydrogen and carbon isotope fractionation factors applicable to geologic systems. Reviews in Mineralogy and Geochemistry, 43 (1): 1-81.
Chan, L. H., J.M. Edmond, G. Thompson, and K. Gillis. (1992). Lithium isotopic composition of submarine basalts; implications for the lithium cycle in the oceans. Earth and Planetary Science Letters, 108 (1-3): 151-160.
Chan, L. H., J.M. Edmond, and G. Thompson. (1993). A lithium isotope study of hot springs and metabasalts from mid-ocean ridge hydrothermal systems; special section on Atlantic hydrothermal activity. Journal of Geophysical Research, 98 (B6): 9653-9659.
Chan, L. H., and M. Kastner. (2000). Lithium isotopic compositions of pore fluids and sediments in the Costa Rica subduction zone; implications for fluid processes and sediment contribution to the arc volcanoes. Earth and Planetary Science Letters, 183 (1-2): 275-290.
Chan, L. H., W. P. Leeman, and C. F. You. (2002). Lithium isotopic composition of Central American volcanic arc lavas; implications for modification of subarc mantle by slab-derived fluids; correction. Chemical Geology, 182 (2-4): 293-300.
Cherniak, D. J. 2001. Pb diffusion in Cr-diopside, augite, and enstatite, and consideration of the dependence of cation diffusion in pyroxene on oxygen fugacity. Chemical Geology, 177 (3-4): 381-397.
Cherniak, D. J. 1998. Pb diffusion in clinopyroxene. Chemical Geology, 150 (1-2): 105-117.
Connolly, J. A. D. (2005). Computation of phase equilibria by linear programming: A tool for geodynamic modeling and its application to subduction zone decarbonation. Earth and Planetary Science Letters, 236: 524-541.
Coogan, L. A., S. A. Kasemann, and S. Chakraborty. (2005). Rates of hydrothermal cooling of new oceanic upper crust derived from lithium-geospeedometry. Earth and Planetary Science Letters, 240 (2): 415-424.
Coplen, T. B. (1996). Atomic weights of the elements 1995. Pure and Applied Chemistry, 68 (12): 2339-2359.
Cormier, L. P. H. Gaskell, G. Calas, J. Zhao, and A. K. Soper. (1998). Environment around Li in the LiAlSiO4 ionic conductor glass: A neutron-scattering and reverse Monte Carlo study. Physical Review B, 57 (14): R8067-R8070.
Crank, J., (1975). The Mathematics of Diffusion. 2nd edn. Oxford: Oxford University Press.
131
Davidson, J. P. (1996). Deciphering mantle and crustal signatures in subduction zone magmatism. Subduction top to bottom, Geophysical Monograph 96: 251-262.
De Grave, E. and S. G. Eeckhout. (2003). 57Fe moessbauer-effect studies of Ca-rich, Fe-bearing clinopyroxenes; part III, diopside. American Mineralogist, 88 (7): 1145-1152.
Decitre, S., E. Deloule, L. Reisberg, R. James, P. Agrinier, and C. Mével. (2002). Behaviour of Li and its isotopes during serpentinization of oceanic peridoites. Geochemistry, Geophysics, Geosystems - G3, 3 (1)
Dimanov, A., O. Jaoul, and V. Sautter. (1996). Calcium self-diffusion in natural diopside single crystals. Geochimica et Cosmochimica Acta, 60 (21): 4095-4106.
Dimanov, A., and V. Sautter. (2000). "Average" interdiffusion of (Fe,Mn)-Mg in natural diopside. European Journal of Mineralogy, 12 (4): 749-760.
Dimanov, A., and M. Wiedenbeck. (2006). (Fe, Mn)-Mg interdiffusion in natural diopside; effect of pO2. European Journal of Mineralogy, 18 (6): 705-718.
Dodson, M. H. (1973). Closure temperature in cooling geochronological and petrological systems. Contributions to Mineralogy and Petrology, 40 (3): 259-274.
Eggins, S. M., R. L. Rudnick, and W. F. McDonough. (1998). The composition of peridotites and their minerals; a laser-ablation ICP-MS study. Earth and Planetary Science Letters, 154 (1-4): 53-71.
Elliott T., T. Plank, A. Zindler, W. White, and B. Bourdon. (1997). Element transport from slab to volcanic front at the Mariana Arc. Journal of Geophysical Research, 102 (B7): 24,991-15,019.
Elliott T., A. Jeffcoate, and C. Bouman. (2004). The terrestrial Li isotope cycle: light-weight constraints on mantle convection Earth and Planetary Science Letters, 220: 233-245
Ernst, W. G. (1988). Tectonic history of subduction zones inferred from retrograde blueschist P-T paths. Geology, 16 (12): 1081-1084.
Faure, G. (1986). Principles of isotope geology. 2nd edn. New York: John Wiley & Sons.
Falkowski, P. G., and L. V. Godfrey (2008). Electrons, life and the evolution of the Earth’s oxygen cycle. Philosophical Transactions of the Royal Society B, 363 (1504): 2705-2716.
Foustoukos, D. I., R. H. James, M. E. Berndt, and W. E. Seyfried. (2004). Lithium isotopic systematics of hydrothermal vent fluids at the main Endeavour field, northern Juan de Fuca ridge. Chemical Geology, 212 (1-2): 17-26.
Ganguly, J., M. Ito and X. Zhang. (2007). Cr diffusion in orthopyroxene; experimental determination, 53Mn-53Cr thermochronology, and planetary applications. Geochimica et Cosmochimica Acta, 71 (15): 3915-3925.
Giletti, B. J., and T. M. Shanahan. (1997). Alkali diffusion in plagioclase feldspar. Chemical Geology, 139 (1-4): 3-20.
Grove, T. L. C. B. Till, E. Lev, N. Chatterjee, and E. Médard. (2009). Kinematic variables and water transport control the formation and location of arc volcanoes. Nature, 459: 694-697.
132
Gurenko, A. A., and H. U.Schmincke. (2002). Orthopyroxene-bearing tholeiites of the Iblean plateau (Sicily); constraints on magma origin and evolution from glass inclusions in olivine and orthopyroxene; melt inclusions at the millennium; toward a deeper understanding of magmatic processes. Chemical Geology, 183 (1-4): 305-331.
Häendel, D., K. K. Mühle, H. M. Nitzsche, G. Stiehl, and U. Wand. (1986). Isotopic variations of the fixed nitrogen in metamorphic rocks. Geochimica et Cosmochimica Acta, 50: 749–758.
Halama, R., W. F. McDonough, R. L. Rudnick, and K. Bell. (2008). Tracking the lithium isotopic evolution of the mantle using carbonatites. Earth and Planetary Science Letters, 265 (3-4): 726-42.
Hallam, M. and H. P. Eugster (1976). Ammonium silicate stability relations. Contributions to Mineralogy and Petrology, 57(3): 227-244.
Hanschmann, G. (1981). Berechnung von isotopieeffekten auf quantenchemischer grundlage am beispiel stickstoffhaltiger molecule. ZfI-Mitteilungen Akademie der Wissenschaften der DDR : Berichte des Zentralinstituts für Isotopen- und Strahlenforschung, 41: 19–31.
Hardee, H. C. J. C. Dunn, J. D. Ward, and R. D. Hills. (1981). Probing the melt zone of Kilauea Iki Lava Lake, Kilauea Volcano, Hawaii. Geophysical Research Letters, 8 (12): 1211-1214.
Harrison, T. M. and E. B. Watson. (1983). Kinetics of zircon dissolution and zirconium diffusion in granitic melts of variable water content. Contributions to Mineralogy and Petrology, 84 (1): 66-72.
Hart, S. R., and T. Dunn. (1993). Experimental cpx/melt partitioning of 24 trace elements. Contributions to Mineralogy and Petrology, 113 (1): 1-8.
Hemingway, B. S. (1990). Thermodynamic properties for bunsenite, NiO, magnetite, Fe3O4, and hematite, Fe2O3, with comments on selected oxygen buffer reactions. American Mineralogist 75, (7-8): 781-790.
Hill, E., B. J. Wood, and J. D. Blundy. (2000). The effect of Ca-tschermaks component on trace element partitioning between clinopyroxene and silicate melt; element partitioning in geochemistry and petrology. Lithos, 53 (3-4): 203-215.
Holland, T. and R. Powell. (1991). A Compensated-Redlich-Kwong (CORK) equation for volumes and fugacities of CO2 and H2O in the range 1 bar to 50 kbar and 100-1600oC. Contributions to Mineralogy and Petrology, 109: 265-273.
Holloway, J. R. (1971). Composition of fluid phase solutes in a basalt-H2O-CO2 system. Geological Society of America Bulletin, 82 (1): 233-237.
Holloway, J.M., and Dahlgren, R.A., (2002). Nitrogen in rock: Occurrences and biogeochemical implications. Global Biogeochemical Cycles, 16: 1118.
Huh, Y., L. H. Chan, and J. M . Edmund. (2001). Lithium isotopes as a probe of weathering processes; Orinoco river. Earth and Planetary Science Letters, 194 (1-2) 189-199.
Huh, Y., L. H. Chan, L. Zhang, and J. M. Edmund. (1998). Lithium and its isotopes in major world rivers; implications for weathering and the oceanic budget. Geochimica et Cosmochimica Acta, 62 (12): 2039-2051.
133
Imai, N., S. Terashima, S. Itoh and A. Ando. (1995). 1994 Compilation of analytical data for minor and trace elements in seventeen GJG geochemical reference samples, “Igneous Rock Series”. Geostandards Newsletter, 19 (2): 135-213.
Ionov, D. A., I.V. Ashchepkov, H. G. Stosch and G. Witt. (1993). Garnet peridotite xenoliths from the Vitim volcanic field, Baikal region: the nature of the garnet-spinel peridotite transition zone in the continental mantle. Journal of Petrology, 34 (6): 1141-1175.
Ionov, D. A., and H. M. Seitz. (2008). Lithium abundances and isotopic compositions in mantle xenoliths from subduction and intra-plate settings; mantle sources vs. eruption histories. Earth and Planetary Science Letters, 266 (3-4): 316-331.
Ishikawa, T., and E. Nakamura. (1994). Origin of the slab component in arc lavas from across-arc variation of B and Pb isotopes. Nature, 370 (6486): 205-208.
Ishikawa, T., and F. Tera. (1999). Two isotopically distinct fluid components involved in the Mariana arc; evidence from Nb/B ratios and B, Sr, Nd, and Pb isotope systematics. Geology, 27 (1): 83-86.
Ishikawa, T., F. Tera, and T. Nakazawa. (2001). Boron isotope and trace element systematics of the three volcanic zones in the Kamchatka arc. Geochimica et Cosmochimica Acta, 65 (24): 4523-4537.
Jahn, S. and B. Wunder. (2009). Lithium speciation in aqueous fluid at high P and T studied by ab initio molecular dynamics and consequences for Li-isotope fractionation between minerals and fluids. Geochimica et Cosmochimica Acta, 73 (18): 5428-5434.
Jagoutz, E., H. Palme, H. Baddenhausen, K. Blum, M. Cendales, G. Dreibus, B. Spettel, V. Lornez and H. Wänke. (1979). The abundances of major, minor and trace elements in the earth's mantle as derived from primitive ultramafic nodules; early solar system and lunar regolith. Proceedings of the Lunar and Planetary Science Conference, 10 (2): 2031-2050.
James, R. H., D. E. Allen, and W. E. Seyfried. (2003). An experimental study of alteration of oceanic crust and terrigenous sediments at moderate temperatures (51 to 350oC); insights as to chemical processes in near-shore ridge-flank hydrothermal systems. Geochimica et Cosmochimica Acta, 67 (4): 681-691.
James, R. H., and M. R. Palmer. (2000). The lithium isotope composition of international rock standards. Chemical Geology, 166 (3-4): 319-326.
Jeffcoate, A. B., T. Elliott, S.A. Kasemann, D. Ionov, K. Cooper, R. Brooker. (2007). Li isotope fractionation in peridotites and mafic melts. Geochimica et Cosmochimica Acta, 71 (1): 202-218.
Jia, Y., and R. Kerrich. (1999). Nitrogen isotope systematics of mesothermal lode gold deposits: metamorphic, granitic, meteoric water, or mantle origin? Geology, 27: 1051-1054.
Jia, Y., and R. Kerrich. (2004). Secular evolution of 15N in crustal fluids; implications for the origin of earth's early atmosphere and hydrosphere; proceedings of the eleventh international symposium on water-rock interaction. Proceedings - International Symposium on Water-Rock Interaction, 11: 945-949.
Jia, Y. (2006). Nitrogen isotope fractionations during progressive metamorphism: A case study from the Paleozoic Cooma metasedimentary complex, southeastern Australia. Geochimica et Cosmochimica Acta, 70: 5201-5214.
134
Kent, A. J., J. Blundy, K. V. Cashman, K. M. Cooper, C. Donnelly, J. S.Pallister, M. Reagan, M. C. Rowe, and C. R. Thornber. (2007). Vapor transfer prior to the October 2004 eruption of Mount St. Helens, Washington. Geology, 35 (3): 231-4.
Keppler, H., (1989). A new method for the generation of N2-containing fluids in high-pressure experiments. European Journal of Mineralogy, 1: 135-137.
Kobayashi, K., R. Tanaka, T. Moriguti, K. Shimizu, and E. Nakamura. (2004). Lithium, boron, and lead isotope systematics of glass inclusions in olivines from Hawaiian lavas; evidence for recycled components in the Hawaiian plume; lithium isotope geochemistry. Chemical Geology, 212 (1-2): 143-161.
Lagache, M., and S. C. Dujon. (1987). Distribution of strontium between plagioclases and 1 molar aqueous chloride solutions at 600oC, 1.5 kbar and 750oC, 2 kbar. Bulletin De Mineralogie, 110 (5): 551-561.
Leeman, W. P. (1996). Boron and other fluid-mobile elements in volcanic arc lavas; implications for subduction processes. Subduction top to bottom, Geophysical Monograph 96: 269-276.
Leeman, W. P., S. Tonarini, L. H. Chan, and L. E. Borg. (2004). Boron and lithium isotopic variations in a hot subduction zone; the southern Washington Cascades; lithium isotope geochemistry. Chemical Geology, 212 (1-2): 101-124.
Lundstrom, C. C., M. Chaussidon, A. T. Hsui, P. Kelemen, and M. Zimmerman. (2005). Observations of Li isotopic variations in the Trinity ophiolite; evidence for isotopic fractionation by diffusion during mantle melting. Geochimica et Cosmochimica Acta, 69 (3): 735-751.
Luth, R.W., and D. Canil. (1993). Ferric iron in mantle-derived pyroxenes and a new oxybarometer for the mantle. Contributions to Mineralogy and Petrology, 113 (2): 236-248.
Lynton, S. J., R. J. Walker, and P. A. Candela. (2005) Lithium isotopes in the system Qz-Ms-fluid: An experimental study. Geochimica et Cosmochimica Acta, 69 (13): 3337-3347.
Magna, T., U. H. Wiechert, and A. N. Halliday. (2004). Low-blank separation and isotope ratio measurement of small samples of lithium using multiple-collector ICPMS. International Journal of Mass Spectrometry, 239 (1): 67–76.
Magna, T., U. Wiechert, and A. N. Halliday. (2006). New constraints on the lithium isotopic compositions of the Moon and terrestrial planets. Earth and Planetary Science Letters, 243 (3-4): 336-353.
Majérus, O., L. Cormier, G. Calas and B. Beuneu. (2003). Structural modifications between lithium-diborate glasses and melts: implications for transport properties and melt fragility. Journal of Physical Chemistry B, 107: 13044-13050.
Marschall, H. R., R. Altherr, T. Ludwig, A. Kalt, K. Gméling, and Z. Kasztovszky. (2006). Partitioning and budget of Li, Be, and B in high-pressure metamorphic rocks. Geochimica et Cosmochimica Acta, 70: 4750-4769.
Marschall, H. R., P. A. E. Pogge von Strandmann, H. M. Seitz, T. Elliott and Y. Niu. (2007). The lithium isotopic composition of orogenic eclogites and deep subducted slabs. Earth and Planetary Science Letters, 262 (3-4): 563-580.
135
Marty, B. and F. Humbert (1997). Nitrogen and argon isotopes in oceanic basalts. Earth and Planetary Science Letters, 152 (1-4): 101-112.
Matthews, W., R. L. Linnen, and Q. Guo. (2003). A filler-rod technique for controlling redox conditions in cold-seal pressure vessels. American Mineralogist, 88 (4): 701-707.
Millot, R., C. Guerrot, and N. Vigier. (2004). Accurate and high-precision measurement of lithium isotopes in two reference materials by MC-ICP-MS. Geostandards and Geoanalytical Research, 28 (1): 153-159.
Moriguti, T., and E. Nakamura. (1998). Across-arc variation of Li isotopes in lavas and implications for crust/mantle recycling at subduction zones. Earth and Planetary Science Letters, 163 (1-4): 167-174.
Moriguti, T., T. Shibata, and E. Nakamura. (2004). Lithium, boron and lead isotope and trace element systematics of Quaternary basaltic volcanic rocks in northeastern Japan: mineralogical controls on slab-derived fluid composition. Chemical Geology, 212: 81-100.
Navon, O., and E. Stolper. (1987). Geochemical consequences of melt percolation; the upper mantle as a chromatographic column. Journal of Geology, 95 (3): 285-307.
Nishio, Y., S. Nakai, J. Yamamoto, H. Sumino, T. Matsumoto, V. Prikhod’ko, and S. Arai. (2004). Lithium isotopic systematics of the mantle-derived ultramafic xenoliths; implications for EM1 origin. Earth and Planetary Science Letters, 217 (3-4): 245-261.
O'Neill, H. St C., and M. I. Pownceby. (1993). Thermodynamic data from redox reactions at high temperatures; I, an experimental and theoretical assessment of the electrochemical method using stabilized zirconia electrolytes, with revised values for the Fe- "FeO", Co-CoO, Ni-NiO and Cu-Cu2O oxygen buffers, and new data for the W-WO2 buffer. Contributions to Mineralogy and Petrology, 114 (3): 296-314.
O'Neill, H. St C., and M. I. Pownceby. (1993). Thermodynamic data from redox reactions at high temperatures; II, the MnO-Mn3O4 oxygen buffer, and implications for the thermodynamic properties of MnO and Mn3O4. Contributions to Mineralogy and Petrology, 114 (3): 315-320.
Paquin, J., and R. Altherr. (2002). Subduction-related lithium metasomatism during exhumation of the Alpe Arami ultrahigh-pressure garnet peridotite (Central Alps, Switzerland). Contributions to Mineralogy and Petrology, 143 (5): 623-640.
Parkinson, I. J., S. J. Hammond, R. H. James, and N. W. Rogers. (2007). High-temperature lithium isotope fractionation; insights from lithium isotope diffusion in magmatic systems. Earth and Planetary Science Letters, 257 (3-4): 609-621.
Peacock, S. M. (1993). The importance of blueschist → eclogite dehydration reactions in subducting oceanic crust. Geological Society of America Bulletin, 105 (5): 684-694.
Pearce, N. J. G., et al., (1997). A compilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials. Geostandards Newsletter, 21(1):115-144
Pell, E.M., (1960). Diffusion of Li in Si at high T and the isotope effect. Physical Review, 119 (3): 1014-1021
136
Peslier, A. H., and J. F. Luhr. (2006). Hydrogen loss from olivines in mantle xenoliths from Simcoe (USA) and Mexico: Mafic alkalic magma ascent rates and water budget of the sub continental lithosphere. Earth and Planetary Science Letters, 242: 302-319.
Pinti, D., K. Hashizume, and J. Matsuda. (2001). Nitrogen and argon signatures in 3.8 to 2.8 Ga metasediments; clues on the chemical state of the Archean ocean and the deep biosphere. Geochimica et Cosmochimica Acta, 65 (14): 2301-2315.
Pistiner, J. S., and G. M. Henderson. (2003). Lithium-isotope fractionation during continental weathering processes. Earth and Planetary Science Letters, 214 (1-2): 327-339.
Plank, T., L. B. Cooper, and C. E. Manning. (2009). Emerging geothermometers for estimating slab surface temperatures. Nature Geoscience Progress Article, NGEO0614. Retrieved from www.nature.com/naturegeoscience
Pöter, B., M. Gottschalk, and W. Heinrich. (2004). Experimental determination of the ammonium partitioning among muscovite, K-feldspar, and aqueous chloride solutions. Lithos, 74: 67-90.
Purton, J. A., N. L.Allan, and J. D. Blundy. (1997). Calculated solution energies of heterovalent cations in forsterite and diopside; implications for trace element partitioning. Geochimica et Cosmochimica Acta, 61 (18): 3927-3936.
Richter, F. M., A. M. Davis, D. J. DePaolo, and E. B. Watson. (2003). Isotope fractionation by chemical diffusion between molten basalt and rhyolite. Geochimica et Cosmochimica Acta, 67 (20): 3905-3923.
Richter, F. M., N. Dauphaus, and F. Z. Teng. (2009). Non-traditional fractionation of non-traditional isotopes: Evaporation, chemical diffusion and Soret diffusion. Chemical Geology, 258: 92-103.
Rudnick, R. L., and D. A. Ionov. (2007). Lithium elemental and isotopic disequilibrium in minerals from peridotite xenoliths from Far East Russia: Product of recent melt/fluid-rock reaction. Earth and Planetary Science Letters, 256 (1-2): 278-293.
Ryan, J. G., and P. R. Kyle. (2004). Lithium abundance and lithium isotope variations in mantle sources; insights from intraplate volcanic rocks from Ross island and Marie Byrd Land (Antarctica) and other oceanic islands; lithium isotope geochemistry. Chemical Geology, 212 (1-2): 125-142.
Ryan, J. G., and C. H. Langmuir. (1987). The systematics of lithium abundances in young volcanic rocks. Geochimica et Cosmochimica Acta, 51 (6): 1727-1741.
Ryerson, F. J., and K. D. McKeegan. (1994). Determination of oxygen self-diffuion in åkermanite, anorthite, diopside, and spinel: Implications for oxygen isotopic anomalies and the thermal histories of Ca-Al-rich inclusions. Geochimica et Cosmochimica Acta, 58 (17): 3713-3734.
Sadofsky, S. J., and G. E. Bebout (2000). Ammonium partitioning and nitrogen-isotope fractionation among coexisting micas during high-temperature fluid-rock interactions; examples from the New England Appalachians. Geochimica et Cosmochimica Acta, 64 (16): 2835-2849.
137
Sadofsky, S. J., and G. E. Bebout (2004). Nitrogen geochemistry of subducting sediments; new results from the Izu-Bonin-Mariana margin and insights regarding global nitrogen subduction. Geochemistry, Geophysics, Geosystems – G3, 5.
Seitz, H. M., G. P. Brey, Y. Lahaye, S. Durali, and S. Weyer. (2004). Lithium isotopic signatures of peridotite xenoliths and isotopic fractionation at high temperature between olivine and pyroxenes; lithium isotope geochemistry. Chemical Geology, 212, (1-2): 163-177.
Seitz, H. M., and A. B. Woodland. (2000). The distribution of lithium in peridotitic and pyroxenitic mantle lithologies; an indicator of magmatic and metasomatic processes. Chemical Geology, 166 (1-2): 47-64.
Sephton, M. A., K. Amor, I. A. Franchi, P. B. Wignall, R. Newton and J. P. Zonneveld. (2009). Carbon and nitrogen isotopic disturbances and an end-Norian (Late Triassic) extinction event. Geology, 30: 1119-1122.
Seyfried, W. E., X. Chen, and L. H. Chan. (1998). Trace element mobility and lithium exchange during hydrothermal alteration of seafloor weathered basalt; an experimental study at 350oC, 500 bars. Geochimica et Cosmochimica Acta, 62 (6): 949-960.
Shannon, R. D., and C. T. Prewitt (1969). Effective ionic radii in oxides and fluorides. Acta Crystallographica Section B, 25 (5): 925-946.
Sneeringer, M., S. R. Hart, and N. Shimizu. (1984). Strontium and samarium diffusion in diopside. Geochimica et Cosmochimica Acta, 48 (8): 1589-1608.
Stolper, E., and S. Newman. (1994). The role of water in the petrogenesis of Mariana Trough magmas. Earth and Planetary Science Letters, 121 (3-4): 293-325.
Suzuki, T., and M. Akaogi. (1995). Element partitioning between olivine and silicate melt under high pressure. Physics and Chemistry of Minerals, 22 (7): 411-418.
Taura, H., H. Yurimoto, K. Kurita, and S. Sueno. (1998). Pressure dependence on partition coefficients for trace elements between olivine and the coexisting melts. Physics and Chemistry of Minerals, 25 (7): 469-484.
Taylor, H. P. Jr. (1977). Water/rock interactions and the origin of H2O in granitic batholiths. Journal of the Geological Society of London, 133 (6): 509-558.
Teng, F. Z., W. F. McDonough, R. L. Rudnick, C. Dalpé, P. B. Tomascak, B. W. Chappell and S. Gao. (2004). Lithium isotopic composition and concentration of the upper continental crust. Geochimica et Cosmochimica Acta, 68 (20): 4167-4178.
Teng, F. Z., W. F. McDonough, R. L. Rudnick, and R. J. Walker. (2006). Diffusion-driven extreme lithium isotopic fractionation in country rocks of the Tin Mountain pegmatite. Earth and Planetary Science Letters, 243 (3-4): 701-710.
Tomascak, P. B. (2004). Developments in the understanding and application of lithium isotopes in the earth and planetary sciences; geochemistry of non-traditional stable isotopes. Reviews in Mineralogy and Geochemistry, 55: 153-195.
Tomascak, P. B., C. H. Langmuir, P. J. le Roux, and S. B. Shirey. (2008). Lithium isotopes in global mid-ocean ridge basalts. Geochimica et Cosmochimica Acta, 72 (6): 1626-1637.
138
Tomascak, P. B., E. Widom, L. D. Benton, S. L. Goldstein, and J. G. Ryan. (2002). The control of lithium budgets in island arcs. Earth and Planetary Science Letters, 196 (3-4): 227-238.
Tomascak, P. B., J. G. Ryan, and M. Defant. (2000). Lithium isotopic evidence for light element decoupling in the Panama subarc mantle. Geology, 28 (6): 507-510.
Tomascak, P. B., F. Tera, R. T. Helz, and R. J. Walker. (1999). The absence of lithium isotope fractionation during basalt differentiation: New measurements by multi-collector ICP-MS. Geochimica et Cosmochimica Acta, 63: 901-910.
Tsai, T. L., and R. Dieckmann. (2002). Variation of the oxygen content and point defects in olivines, (FexMg1-x)2SiO4, 0.2 ≤ x ≤ 1.0. Physics and Chemistry of Minerals, 29: 680-694.
Ulmer, P. (2001). Partial melting in the mantle wedge – the role of H2O in the genesis of mantle-derived ‘arc-related’ magmas. Physics of the Earth and Planetary Interiors, 127 (1-4): 215-232.
Urey, H. C. (1947). The thermodynamic properties of isotopic substances. Journal of the Chemical Society, 1947: 562-581.
Van Orman, J. A., T. L .Grove, and N. Shimizu. (2001). Rare earth element diffusion in diopside; influence of temperature, pressure, and ionic radius, and an elastic model for diffusion in silicates. Contributions to Mineralogy and Petrology, 141 (6): 687-703.
Van Orman, J. A., T. L .Grove, and N. Shimizu. (1998). Uranium and thorium diffusion in diopside. Earth and Planetary Science Letters, 160 (3-4): 505-519.
Wenger, M., and T. Armbruster. (1991) Crystal chemistry of lithium; oxygen coordination and bonding. European Journal of Mineralogy, 3 (2): 387-99.
Williams, L.B., R. E. Ferrell Jr., I. Hutcheon and A. J. Bakel, M. M. Walsh and H. R. Krouse. (1995). Nitrogen isotope geochemistry of organic matter and minerals during diagenesis and hydrocarbon migration. Geochimica et Cosmochimica Acta, 54: 765–779.
Woodland, A. B., H. M. Seitz, and G. M. Yaxley. (2004). Varying behaviour of Li in metasomatised spinel peridotite xenoliths from western Victoria, Australia; trace element fingerprinting; laboratory studies and petrogenetic processes. Lithos, 75 (1-2): 55-66.
Woodland, A. B., H. M.Seitz, and R. Altherr, H. Marschall, B. Olker, and T. Ludwig. (2002). Li abundances in eclogite minerals; a clue to crustal or mantle origin? Contributions to Mineralogy and Petrology, 143 (5): 587-601.
Woods, S., S. Mackwell, and D. Dyar. (2000). Hydrogen in diopside; diffusion profiles. American Mineralogist, 85 (3-4): 480-487.
Wunder, B. A. Meixner, R. L. Romer, and W. Heinrich. (2006). Temperature-dependent isotopic fractionation of lithium between clinopyroxene and high pressure hydrous fluids. Contributions to Mineralogy and Petrology, 151 (1): 112-120.
Wunder, B. A. Meixner, R. L. Romer, A. Feenstra, G. Schettler, W. Heinrich. (2007). Lithium isotope fractionation between Li-bearing staurolite, Li-mica and aqueous fluids: An experimental study. Chemical Geology, 238 (3-4): 277-290.
Yamaji, K., Y. Makita, H. Watanabe, A. Sonoda, H., Kanoh, T. Hirotsu, and K. Ooi. (2001) Theoretical estimation of lithium isotopic reduced partition function ratio for lithium ions in aqueous solution. Journal of Physical Chemistry A, 105 (3): 602-613
139
Zack, T., P. B. Tomascak, R. L. Rudnick, C. F. Dalpe, and W. F. McDonough. (2003). Extremely light Li in orogenic eclogites; the role of isotope fractionation during dehydration in subducted oceanic crust. Earth and Planetary Science Letters, 208 (3-4): 279-290.
Zanetti, A., M. Tiepolo, R. Oberti, and R. Vannucci. (2004). Trace-element partitioning in olivine; modelling of a complete data set from a synthetic hydrous basanite melt; trace element fingerprinting; laboratory studies and petrogenetic processes. Lithos 75, (1-2): 39-54.
Zhang, L., L. H. Chan, and J. M. Gieskes. (1998). Lithium isotope geochemistry of pore waters from ocean drilling program sites 918 and 919, Irminger Basin. Geochimica et Cosmochimica Acta, 62 (14): 2437-2450.
140
Appendix 1
6 Summary of Boron Work
Boron is commonly used as a tracer of mineral-fluid reactions; however, to date there is very
little information on the degree of isotopic fractionation that may occur during fluid-mineral
partitioning. Without this information, the mixing models developed by many workers to
describe such systems as mantle metasomatism or arc magmatism are qualitative at best.
6.1 11B notation
Boron has two stable isotopes, 11B and 10B, whose abundances are ~80 % and ~20 %,
respectively. Most commonly, enrichments in boron isotopes are described as:
(1)
10001BB/
BB/B
std1011
smp1011
11
where smp refers to the sample and std refers to the standard, typically NBS SRM 951 for silicate
materials.
6.2 Evidence of Boron Mobility from Arc Lavas
Boron and boron isotopes are often employed to unravel the mechanisms involved during the
slab to mantle transfer of material in subduction zones (Ishikawa et al., 2001; Bebout et al., 1999;
Sano et al., 2001; Ishikawa and Tera, 1999; Benton et al., 2001). Due to boron’s affinity for
phyllosilicates, both altered oceanic crust (AOC) and pelagic sediments are enriched in these
elements relative to both mid-ocean ridge basalts (MORB) and oceanic island basalts (OIB)
(Table A1; Leeman, 1996).
141
Table A1: Summary of B concentration and isotopic composition
B (ppm) 11B (‰)
Marine Sediments 15-160a -6.6 to +4.8a
AOC 3-63b +1 to +10b
OIB < 2c -15 to +1c
MORB < 0.1d -6 to -1d
Seawater 24e +39e
aIshikawa and Nakamura, 1993, bIshikawa and Nakamura, 1992 cPeacock and Hervig ,1999; dIshikawa and Tera, 1999; eVengosh et al.,
1995;
Because pelagic sediments and AOC are isotopically distinct from MORB and OIB, it is thought
that this slab input is reflected in the boron isotopic composition of arc lavas, which often differs
from the mantle (Ishikawa and Nakamura, 1992; --, 1993; Smith et al., 1995).
The extent to which these elements are lost by either dehydration or metasomatism to the
overlying mantle wedge or are incorporated into the mantle is not well known. Fluid and
sediment collected in fore arc environments provide evidence for mobilization and isotopic
fractionation of boron as a result of dehydration reactions in the subducting slab (Benton et al.,
2001). Clasts and muds from a serpentinite seamount in the Mariana fore arc were found to have
higher concentrations of boron and are enriched in 11B compared to seafloor sediments of the
area.
Several island arcs display higher ratios of boron to relatively immobile elements (such as Zr and
Nb in the front-arc regions) which systematically decrease towards the back-arc regions. For
example the Kamchatka arc lavas have the greatest enrichments in boron to Nb or Zr at the arc
142
front; however, these enrichments decrease to MORB values with increasing slab depth
(Ishikawa et al., 2001; Wunder et al., 2005).
Figure A1 11B as a function of B/Nb in Kamchatka Arc Lavas
Plot of 11B versus B/Nb in arc lavas. Because the solid/melt partition coefficients for B and Nb are indistinguishable from one another, this trend cannot be the result of igneous processes. Rather this is suggestive of continuing mobilization of boron into the arc source region by B and 11B enriched fluids derived from dehydration reactions in the down going slab (data from Ishikawa et al., 2001).
Arc lavas display boron isotopic compositions that differ from mantle and MORB (Sadofsky and
Bebout, 2000). In both the Kamchatka arc and the Mariana arc, δ11B is most enriched at the arc
front where the highest B/Nb or B/Zr ratios occur (Ishikawa et al., 2001; Ishikawa and Tera,
1999). The δ11B in Kamtchatka lavas ranges from -4 to +6 ‰, and in Mariana lavas δ11B ranges
from +2.9 to +6.2 ‰ (Ishikawa et al., 2001; Ishikawa and Tera, 1999). In many cases researchers
point to the 11B enrichments in arc lavas as tracers of input from the subducted slab, either as
AOC or sediments, and propose various mixing models for these volcanic arcs; however, the
extent of boron isotope fractionation between fluid and residual solid are unknown and
quantitative modeling of this process is not possible.
6.3 Evidence of Boron Mobility from Eclogites
Only a few studies have examined the effects of progressive dehydration on isotopic
fractionation of boron isotopes. The δ11B values obtained from subduction zone metamorphic
rocks range from –11 to –3 ‰, which is generally lower than the δ11B values for seafloor
sediments and AOC (Peacock and Hervig, 1999).
143
Figure A2 Evidence of Boron Mobility from Eclogites
Plot of 11B in eclogites versus estimated peak metamorphic temperature. Progressive dehydration results in lighter boron isotopic fraction in metamorphic rocks (from Peacock and Hervig, 1999)
These light stable isotopes are fractionated between phases, namely the exsolved fluids and the
host minerals, and unless the extent of this shift is known these isotopes cannot be fully utilized
as tracers.
6.4 Summary of Experimental Methodology
Experiments were planned to equilibrate muscovite with a fluid of known isotopic composition
at high pressures and temperatures using the cold-seal pressure vessels and piston-cylinder
apparatus at the University of Toronto's High Pressure Lab. Natural muscovite crystals from an
unknown locality in Ontario, Canada were to be used for most, if not all, of the boron
experiments.
6.5 Details of Boron Study
In silicate minerals boron is typically bonded to O; however, in silicate minerals the boron occurs
in tetrahedral coordination and is known to substitute for Al3+ or Si4+ (Schreyer et al., 2000). In
fluids, the isotopic enrichment depends on coordination of the species. Tetrahedrally coordinated
boron is 10B enriched, and trigonally coordinated boron is 11B enriched; therefore, the isotopic
fractionation depends almost entirely on the relative partitioning of B(OH)4- and B(OH)3
o
(Palmer, 1992), which in turn will depend on the speciation of boron in aqueous fluids. The boric
acid – borate equilibria can be written as:
(1) H OHB(OH)B(OH) 2-4
o3
144
And the equilibrium constant K3 is expressed as:
(2) o3B(OH)
H4B(OH)3 a
aaK
Where a is the activity of the subscripted species.
The equilibrium constant (K) for reaction (2) can be derived from previous work of Mesmer et
al. (1972) calculated with SUPCRT92 (Helgeson et al., 1978; Johnson et al., 1992; Shock et al.,
1989). The Helgeson-Kirkham-Flowers equation of state for aqueous species limits the
applicability of SUPCRT92 to ≤5 kbar, and so extrapolation is necessary. It has been shown that
the logarithm of the equilibrium constant for many mineral hydrolysis reactions is linear with
logH 2O at constant T (Eugster and Baumgartner, 1987; Mesmer et al., 1988; Anderson et al.,
1991; Manning, 1994).
Figure A05 log K3 vs log density of water at 500°C, 2 - 4.5 kbar,
Calculated at 0.5 kbar increments using SUPCRT92. Error bars correspond to propagated uncertainties in thermodynamic data from Shock et al. (1989). These values can be reasonably fit by a straight line. This implies that a limiting slope method can be used to extrapolate K3 to P>5 kbar.
Using extrapolated values of K3, pK3 values for up to 30kbar have been calculated from:
(3) o3B(OH)4B(OH) logloglog 3 aa pHK
Using these values, the boric acid – borate equilibria was calculated for 2 – 20 kbar and 400-800 oC and are shown in Figure A06. At low P and T these values are consistent with those
145
determined by Mesmer et al. (1972). Also shown for reference are neutral pH, and the pH that
would be buffered by the assemblage muscovite + potassium feldspar + quartz + KCl. In the case
of a dehydrating slab at the blueschist – eclogite transition, pH would be buffered at ~ 6
(Manning, 1998). As Figure A6 shows, a fluid with pH = 6 would have B(OH)4- as the dominant
species at 20 kbar but end up with B(OH)3o by 5 kbar. The role that boric acid – borate equilibria
during dehydration reactions plays in the isotopic fractionation has not been previously
addressed.
Palmer et al. (1992) conducted a study on boron-isotope fractionation with synthetic tourmaline,
and proposed that B(OH)3o initially is adsorbed onto the mineral surface. This initial adsorption
is suggested to control the isotopic fractionation because the boron symmetry changes to
‘psuedo-tetrahedral’, becomes enriched in 10B, and is incorporated into the structure without
further fractionation. Adsorption of dissolved B(OH)4- is not favoured because it involves the
breaking of a B-O bond. In their study, Palmer et al. (1992) concluded isotopic fractionation
between tourmaline and aqueous fluids decreased with increasing pressure. Palmer et al. (1992)
further state that B(OH)3o was the only B-species in the experiments because the formation of
trigonally coordinated species is favored at high pressures. This is contrary to recent studies that
have shown an increase in polymerization of hydrated species with increasing pressure (Zotov
and Keppler, 2002) and with predictions based on thermodynamic data.
146
400
450
500
550
600
650
700
750
800
0 2 4 6 8 10 12 14
2 kbar
Te
mpe
ratu
re (
o C)
pH
B(OH)3 B(OH)
4
-
mu
s+
ksp
ar+
qtz
+K
Cl
neu
tral
pH
blu
esc
his
t -
ecl
og
ite
400
450
500
550
600
650
700
750
800
0 2 4 6 8 10 12 14
5 kbar
Te
mpe
ratu
re (
o C)
pH
B(OH)3
B(OH)4
-
neu
tral
pH
blu
esc
his
t -
ecl
og
ite
mu
s+
ks
pa
r+q
tz+
KC
l
400
450
500
550
600
650
700
750
800
0 2 4 6 8 10 12 14
10kbar
Te
mpe
ratu
re (
o C)
pH
B(OH)3 B(OH)
4
-
neu
tral
pH
mu
s+
ksp
ar+q
tz+
KC
l
blu
esc
his
t -
ecl
og
ite
400
450
500
550
600
650
700
750
800
0 2 4 6 8 10 12 14
20kbar
Te
mpe
ratu
re (
o C)
pH
B(OH)3
B(OH)4
-
neu
tral
pH
eclo
git
e -
blu
esch
ist
mu
s+
ksp
ar+
qtz
+K
Cl
Figure A6 calculated boric acid – borate equilibria for 2 – 20 kbar and 400-800 oC
Also shown for reference is neutral pH and pH as would be buffered by the assemblage muscovite + potassium feldspar + quartz + KCl. Note the effect of pH on speciation of boron in fluids, a fluid with pH = 6 would start off with B(OH)4
- as the dominant species at 20 kbar but end up with B(OH)3o by 5 kbar.
6.6 Boron Analyses
Initially it was planned to analyze the boron isotope composition of the run product mica using
the CAMECA ims1270 high-resolution secondary ion mass spectrometer at the Department of
147
Terrestrial Magmatism under the supervision of Erik Hauri. The main concern going into the
study was to produce a large enough mica grain suitable for in situ analysis. Reconnaissance
analyses at UCLA revealed that there are significant challenges to boron analyses in micas. High
amounts of sample charging occurred during analyses (resulting in an apparent 20 ‰
fractionation from one side of the mount to the other, reproducible when sample mount was
rotated 90o). Switching to mono-collector mode appeared to resolve the initial charging issue;
however, this resulted in very long analyses. Compounded by the fact that the ion yield for micas
is very low so collection times are, by necessity, long; the amount of charging on the sample
increased. Increasing the concentration of boron in the micas would improve the counting
statistics and facilitate the analyses somewhat. One possibility would be to begin with boron rich
mica and attempt to measure diffusion out.
148
6.7 References for Boron Study Anderson, G. M., S. Castet, et al. (1991). "The density model for estimation of thermodynamic
parameters of reactions at high temperatures and pressures." Geochimica et Cosmochimica Acta 55(7): 1769-1779.
Benton, L. D., et al. (2001). "Boron isotope systematics of slab fluids as inferred from a serpentine seamount, Mariana Forearc." Earth and Planetary Science Letters 187(3-4): 273-282.
Eugster, H. P. and L. Baumgartner (1987). "Mineral solubilities and speciation in supercritical metamorphic fluids
Faure, G. (1986). "[Monograph] Principles of isotope geology."
Helgeson, H. C., J. M. Delany, et al. (1978). "[Monograph] Summary and critique of the thermodynamic properties of rock-forming minerals."
Ishikawa, T. and E. Nakamura (1992). "Boron isotope geochemistry of the oceanic crust from DSDP/ODP Hole 504B." Geochimica et Cosmochimica Acta 56(4): 1633-1639.
Ishikawa, T. and E. Nakamura (1993). "Boron isotope systematics of marine sediments." Earth and Planetary Science Letters 117(3-4): 567-580.
Ishikawa, T. and F. Tera (1999). "Two isotopically distinct fluid components involved in the Mariana Arc; evidence from Nb/B ratios and B, Sr, Nd, and Pb isotope systematics." Geology 27(1): 83-86.
Ishikawa, T., et al. (2001). "Boron isotope and trace element systematics of the three volcanic zones in the Kamchatka Arc." Geochimica et Cosmochimica Acta 65(24): 4523-4537.
Johnson, J. W., E. H. Oelkers, et al. (1992). "SUPCRT92: A software package for calculating standard molal thermodynamic properties of minerals, gases, aqueous species, and reactions from 1 to 5000bar and 0 to 1000oC." Comp. Geosci 18: 899-947.
Leeman, W. P. (1996). "Boron and other fluid-mobile elements in volcanic arc lavas; implications for subduction processes
Manning, C. E. (1994). "The solubility of quartz in H2O in the lower crust and upper mantle." Geochimica et Cosmochimica acta 58: 4831-4839.
Manning, C. E. (1998). "Fluid composition at the blueschist-eclogite transition in the model system Na2O-MgO-Al2O3-SiO2-H2O-HCl." Swiss bulletin of Mineralogy and Petrology 78(2): 225-242.
Manning, C. E. and S. L. Boettcher (1994). "Rapid-quench hydrothermal experiments at mantle pressures and temperatures." American Mineralogist 79(11-12): 1153-1158.
Mesmer, R. E., C. F. Baes, et al. (1972). "Acidity Measurements at Elevated-Temperatures .6. Boric-Acid Equilibria." Inorganic Chemistry 11(3): 537-&.
Mesmer, R. E., W. L. Marshall, et al. (1988). "Thermodynamics of Aqueous Association and Ionization Reactions at High-Temperatures and Pressures." Journal of Solution Chemistry 17(8): 699-718.
149
Palmer, M. R., D. London, et al. (1992). Experimental determination of fractionation of 11B/10B between tourmaline and aqueous vapor; a temperature- and pressure-dependent isotopic system. Frontiers in isotope geosciences. R. S. Harmon and R. W. Hinton. Amsterdam, Elsevier. 101: 123-129.
Paquin, J. and R. Altherr (2001). "New constraints on the P-T evolution of the Alpe Arami garnet peridotite body (Central Alps, Switzerland)." Journal of Petrology 42(6): 1119-1140.
Peacock, S. M. and R. L. Hervig (1999). Boron isotopic composition of subduction-zone metamorphic rocks. Interactions between slab and sub-arc mantle; dehydration, melting and element transport in subduction zones. D. S. Draper, A. D. Brandon and H. Becker. Amsterdam, Elsevier. 160: 281-290.
Sano, T., T. Hasenaka, et al. (2001). "Boron contents of Japan Trench sediments and Iwate basaltic lavas, Northeast Japan arc; estimation of sediment-derived fluid contribution in mantle wedge." Earth and Planetary Science Letters 186(2): 187-198.
Schreyer, W., U. Wodara, et al. (2000). "Synthetic tourmaline (olenite) with excess boron replacing silicon in the tetrahedral site; I, Synthesis conditions, chemical and spectroscopic evidence." European Journal of Mineralogy 12(3): 529-541.
Shock, E. L., Helgeson, H.C., and Sverjensky, D.A. (1989). " Calculation of the thermodynamic and transport properties of aqueous species at high pressures and temperatures: Standard partial molal properties of inorganic neutral species." Geochimica Cosmochimica Acta 53: 2157-2183.
Smith, H. J., A. J. Spivack, et al. (1995). The boron isotopic composition of altered oceanic crust. The mantle-ocean connection. H. Staudigel, F. Albarede, D. Hilton and T. Elliott. Amsterdam, Elsevier. 126: 119-135.
Tolstikhin, I. N. and B. Marty (1998). The evolution of terrestrial volatiles; a view from helium, neon, argon and nitrogen isotope modelling. The degassing of the Earth [modified]. M. R. Carroll, S. C. Kohn and B. J. Wood. Amsterdam, Elsevier. 147: 27-52.
Tomascak, P. B., et al. (2000). "Lithium isotope evidence for light element decoupling in the Panama subarc mantle." Geology 28(6): 507-510.
Vengosh, A., , et al. (1995). "Chemical and boron isotope compositions of non-marine brines from the Qaidam Basin, Qinghai, China." Chemical Geology 120(1-2): 135-154.
Wunder, B., et al. (2005). “The geochemical cycle of boron; constraints from boron isotope partitioning experiments between mica and fluid”. Lithos 84 (3-4): 206-16.
Zotov, N. and H. Keppler (2002). "Silica speciation in aqueous fluids at high pressures and high temperatures." Chemical Geology 184(1-2): 71-82.