ggge20370-sup-0003-suppinfo01

advertisement
1
1. Analytical Methods
2
3
1.1. Hf, Pb, Sr and Nd chemical separation and mass spectrometry
4
New Hf-Pb-Sr-Nd isotopic measurements on glasses and whole rock powders are
5
reported in Table 2. The majority of analyses were carried out on >200 mg of fresh glass
6
chips (or whole rock powder), but only ~100 mg of glass available was available for a
7
few of the samples. Nineteen samples of glass chips and one whole rock powder (sample
8
4-1) were acid leached and dissolved at the Ecole Normale Supérieure in Lyon (ENS
9
Lyon) following the protocols of Blichert-Toft et al. [1997] and Blichert-Toft and
10
Albarède [2009]. The samples were leached in 2 ml of 6N HCl for 20 minutes at 120° C,
11
followed by 10 minutes of sonication. Samples were leached again at 120° C for an
12
additional 10 minutes, followed by 5 minutes of sonication and then 5 more minutes of
13
leaching at 120° C; the HCl was pipetted off and the samples were rinsed twice in milliQ
14
H2O.
15
Dissolution in concentrated HF and HNO3 and chemical separation of Hf by
16
anion- and cation-exchange columns followed the methods outlined by Blichert-Toft et
17
al. [1997]. An additional six samples were leached and dissolved at Boston University
18
(BU), and the leaching procedure followed the same protocol as that used at ENS Lyon.
19
Pb, Sr and Nd retained in the CaMg-fluoride precipitate (left after removal of Hf by HF
20
leaching at ENS Lyon) and Pb, Sr and Nd for samples leached at BU, were separated
21
from all 26 samples using cation-exchange chemistry at BU. A number of replicates also
22
were analyzed (Table 2), all on different aliquots of glass (from the same sample).
23
Replicates samples were picked separately, leached separately (in 6N HCl at 110° C for
24
20 minutes) dissolved separately, run on separate columns (on different days than the
25
original analyses) and analyzed during different analytical sessions on the mass
26
spectrometer. These replicate measurements provide conservative estimates of the long-
27
term reproducibility of isotopic analyses of unknown basaltic samples. Pb was purified
28
using Biorad AG1-X8 resin (100-200 μm), Sr was separated using Eichrom Sr resin (25-
29
50 μm), and Nd was purified using a two-step Nd separation method that first used
30
Eichrom TRU resin (100-150 μm) followed by Eichrom LN-Spec resin (50-100 μm).
31
Strontium and Nd were recovered from the clean wash of the Pb columns. This wash
32
fraction was split in two, one destined for the Sr separation protocol and the other
33
destined for the Nd separation protocol. In this manner, Hf, Pb, Nd, and Sr were all
34
separated from the same sample dissolutions, thereby minimizing glass consumption and
35
avoiding potential problems due to possible sample heterogeneity that could lead to
36
isotopic variations. The total procedural blanks for Hf, Pb, Sr and Nd were < 20 pg, < 30
37
pg, < 80 pg and < 35 pg, respectively, which are all negligible relative to the amount of
38
Sr, Nd and Pb analyzed in the rocks.
39
Hafnium isotopic compositions were determined by multi-collector inductively
40
coupled plasma mass spectrometry (MC-ICP-MS) on the Nu Plasma HR at ENS Lyon
41
during October and November of 2012. All Pb, Sr and Nd isotopic measurements were
42
performed on the Neptune MC-ICP-MS at Woods Hole Oceanographic Institution
43
(WHOI) at the end of 2012 and the first half of 2013, following recent replacement of the
44
Faraday collectors. The following outlines the mass spectrometry protocols used for the
45
analyses of Hf, Pb, Sr and Nd isotopic compositions of this study:
46
Hafnium. Hafnium isotopic measurements followed the protocol described by Blichert-
47
Toft et al. [1997]. The Hf standard JMC-475 was analyzed between every second to
48
fourth sample depending on machine stability. The unweighted mean 176Hf/177Hf ratios
49
of the JMC-475 standard obtained during each of three day-long sessions of collection of
50
the present Hf isotope data were 0.282166, 0.282170 and 0.282170. Since these values
51
are identical within error to the accepted value of 0.282163 ± 0.000009 [Blichert-Toft et
52
al., 1997] for JMC-475, no corrections were applied to the data collected on basalts. All
53
measured Hf isotopic ratios were corrected online for isobaric interferences of W and Ta
54
on mass 180 and Lu and Yb on mass 176 by monitoring the interference-free isotopes
55
183
56
fractionation by normalizing 179Hf/177Hf to 0.7325 using an exponential law. Lutetium
57
and Yb interferences on 176Hf were insignificant for all samples analyzed. The external
58
precision of 176Hf/177Hf based is estimated to be 10-30 ppm (2σ).
59
Lead. Lead isotopic ratios were corrected for instrumental mass bias by introducing Tl
60
(SRM 997) as an internal standard to sample Pb solutions prior to each run assuming a
61
205
62
correct for 204Hg interference, but this correction was small owing to low 202Hg/208Pb
63
ratios (typically < 1.0x10-5). Measured Pb-isotopic ratios of the samples were normalized
64
based on the offset between our average measured and the accepted SRM981 values from
65
Todt et al. [1996] (206Pb/204Pb =16.9356, 207Pb/204Pb =15.4891, 208Pb/204Pb =36.7006).
66
External reproducibility on runs of SRM981 at WHOI ranges from 17 ppm (2σ) for
67
207
68
Strontium. During each analytical session, intensities were measured on masses 82
69
through 88. Corrections for isobaric interferences of Rb on mass 87 and Kr on masses 84
W, 181Ta, 175Lu and 173Yb, respectively. The data were corrected for instrumental mass
Tl/203Tl ratio of 2.38709 and using an exponential law. Mass 202 was monitored to
Pb/206Pb to 117 ppm (2σ) for 208Pb/204Pb [Hart et al., 2004].
70
and 86 were made offline following the procedures outlined in Jackson and Hart [2006].
71
Runs with low intensities (i.e., <3 V on mass 88, with a 1011 Ω resistor) were discarded.
72
Strontium isotope ratios were corrected for instrumental mass bias relative to an 86Sr/88Sr
73
value of 0.1194 using an exponential law.
74
normalized by the offset between our average measured value of SRM987 during each
75
analytical session and the accepted 87Sr/86Sr of 0.710240 [Jackson and Hart, 2006]. The
76
external precision of the 87Sr/86Sr measurements is estimated to be 15–25 ppm (2σ) [Hart
77
and Blusztajn, 2006].
78
Neodymium. The data were corrected for instrumental mass fractionation relative to a
79
146
80
standards were run during each analytical session. The 143Nd/144Nd values for JNDi-1
81
were adapted to the La Jolla 143Nd/144Nd value using a using a ratio of 1.000503 [Tanaka
82
et al., 2000]. The La Jolla and La Jolla-renormalized-JNdi-1 143Nd/144Nd measurements
83
were averaged to give a final La Jolla average for each analytical session. Samples were
84
normalized based on the offset of this La Jolla average and the La Jolla 143Nd/144Nd value
85
of 0.511847 that we adopt here [White and Patchett, 1984]. The external precision of the
86
143
87
2006].
88
87
Sr/86Sr ratios for unknowns were then
Nd/144Nd value of 0.7219 using an exponential law. Both the La Jolla and JNdi-1
Nd/144Nd measurements is estimated to be 15–25 ppm (2σ). ) [Hart and Blusztajn,
The well-characterized international rock standards BCR-2 and AGV-2 were
89
dissolved, processed, and analyzed similarly to and along with unknowns (although rock
90
standards were not leached) during each batch of column chemistry and each analytical
91
session. Two BCR-2 standards were measured for Hf isotopes in this study
92
(Supplementary Table 1) and yielded 176Hf/177Hf of 0.282882±4 (2σ) and 0.282884±5
93
(2σ), which compare well with the value of 0.282884±7 (2σ) reported by Le Fevre and
94
Pin [2001] for BCR-2. Furthermore, the BCR-1 value of 0.282879±8 (2σ) reported by
95
Blichert-Toft [2001] and measured in the same laboratory (ENS Lyon) compares well
96
with the BCR-2 standards from this study. Results for Pb, Sr and Nd isotopic
97
measurements of the BCR-2 and AGV-2 rock standards (this study, Supplementary Table
98
1) likewise compare well with those of Weis et al. [2006] (Supplementary Fig. 1). In
99
particular, we note that the 87Sr/86Sr ratios for BCR-2 and AGV-2 measured on the
100
Neptune MC-ICP-MS in this study, compare well with the measurements of the same
101
rock standards determined on a Triton TIMS in Weis et al. [2006] (thermal ionization
102
mass spectrometer) (Supplementary Fig. 1 and Supplementary Table 1).
103
104
1.2. He isotopic analyses
105
Helium isotopes were measured on clean glass chips at WHOI using an automated
106
dual-collection, statically-operated helium isotope mass spectrometer. Measurements
107
were all made by crushing in vacuo, following the protocol of Kurz et al. [2004]. 4He gas
108
concentrations ranged from 9x10-10 to 5.5x10-6 cc STP/g, and 2σ internal precision of the
109
3
110
relative to atmospheric (R/Ra) using an atmospheric 3He/4He ratio of 1.384 × 10-6. New
111
3
He/4He measurements ranged from ±0.06 to ±0.52 Ra. 3He/4He ratios are reported
He/4He measurements on glasses reported in this study are shown in Table 2.
112
113
114
115
1.3. Major element analyses
All major element analyses, with the exception of samples 127-05, 127-11, and
126-18, were measured on glasses by electron microprobe at the Hawaii Institute of
116
Geophysics and the data are published previously [Sinton et al., 1985; Johnson and
117
Sinton, 1990; Sinton et al., 1993]. Whole rock powders were prepared for samples 127-
118
05 and 127-11 and measured by X-ray fluorescence on a ThermoARL XRF at
119
Washington State University (WSU) and the data are published in Jackson et al. [2010].
120
All major element data are summarized in Table 1.
121
122
123
1.4. Trace element analyses
Trace element concentrations were previously measured by X-ray fluorescence
124
for many of the samples examined in this study [Sinton et al., 1985; Johnson and Sinton,
125
1990; Sinton et al., 1993]. New trace element analyses were performed by ICP-MS at the
126
Washington State University GeoAnalytical Lab on a aliquots of the glass samples
127
examined by Sinton et al. [1985], Johnson and Sinton [1990], Sinton et al. [1993], and
128
Jackson et al. [2010] (Table 3). Measurements were made on glass chips (unleached, but
129
sonicated in milliQ H2O). For one sample (4-1), glass was not available and unleached
130
whole rock powder was prepared instead (sample 4-1 was crushed in W-carbide and,
131
hence, the Ta data are compromised). As a check on the reproducibility of the trace
132
element analyses, we included an analysis of a 200 mg aliquot of BHVO-2 powder with
133
the batch of samples analyzed in this study, and the data are reported in Table 3.
134
135
1.5. Comparisons of new isotopic data with previously published values
136
We note that 13 of the glass samples with new isotopic data reported here were
137
previous analyzed for Sr and Nd isotopes in the 1980’s at Lamont-Doherty Geological
138
Observatory (LDGO) [Sinton et al., 1993] and at Massachusetts Institute of Technology
139
(MIT) [Johnson and Sinton, 1990]. While all of the 143Nd/144Nd analyses agree within 60
140
ppm, two 143Nd/144Nd analyses show larger disagreement: sample 5-15, measured at
141
LDGO, is 107 ppm higher than our analysis, and sample 140-1A, measured at MIT, is
142
131 ppm lower. The 87Sr/86Sr agreement between this study and the two previous studies
143
is poorer. The 87Sr/86Sr analyses run 69 to 112 ppm higher than the MIT analyses. Five
144
of the seven 87Sr/86Sr analyses at LDGO agree with our analyses within 90 ppm, but
145
LDGO analyses of samples 5-14 and 6-54 ran 138 ppm and 439 ppm higher than the
146
present study. The LDGO glasses were not acid leached prior to analysis (they were
147
sonicated in distilled water), which may explain the elevated 87Sr/86Sr relative to the data
148
reported here.
149
However, we cannot easily explain the large offsets in 143Nd/144Nd between this
150
data set and the two MIT and LDGO measurements, as analyses at MIT and LDGO also
151
were made on glasses. Again, the lack of leaching may play an important role,
152
particularly if secondary Nd-rich material, such as ferromanganese oxyhydroxides or
153
phosphates, was present on the surface of the glass. We note that our measurements on
154
unknowns are reproducible and our results on basaltic standards are consistent with
155
measurements made in other modern laboratories (Supplementary Fig. 1 and
156
Supplementary Table 1). One of the two glasses that shows a large discrepancy in Nd
157
isotopic composition relative to the MIT analysis, sample 140-1A, was analyzed twice in
158
our laboratory on two different aliquots of glass chips prepared several months apart; the
159
aliquots were picked, leached, dissolved and processed through column chemistry
160
separately and run during different analytical sessions on the Neptune MC-ICP-MS, and
161
the 143Nd/144Nd measurements differ by only 50 ppm. Furthermore, multiple USGS rock
162
standards measurements of 143Nd/144Nd (and 87Sr/86Sr) made in this study show excellent
163
agreement with other labs (Supplementary Fig. 1).
164
As a final note, we point out that radiogenic isotopes were measured on whole
165
rock [Jackson et al., 2010] and glass (this study) aliquots for three samples from Wallis
166
and Futuna Islands, 127-05, 127-11 and 6-52. Where 87Sr/86Sr and 143Nd/144Nd were
167
measured on both glass and whole rock powder, there is excellent agreement. Lead
168
isotopic measurements of whole rocks and glasses show poorer agreement. One
169
explanation for this variance may be the variable affects that weathering has on the whole
170
rock, for which sample freshness is more difficult to ascertain than for glass samples,
171
which can be visually inspected for freshness. As demonstrated in Fig. 4, Futuna whole
172
rocks exhibit evidence of having been compromised by weathering, while glasses are
173
characterized by more pristine patterns. The whole rock and glass trace element analyses
174
of sample 6-52, which exhibit significant differences in Pb isotope compositions, also
175
show clear differences in spidergram shape. In particular, the Pb abundance in the glass
176
is nearly three times lower than in the whole rock, suggesting Pb mobilization.
177
Furthermore, it is not always apparent which portion of the rock is attacked by leaching
178
prior to sample dissolution and analysis, particularly for Sr and Pb isotopic compositions
179
in altered basalts [Nobre Silva et al., 2010]. Thus, we conclude that different weathering
180
patterns in the whole rocks may make whole rock Pb more susceptible to being
181
compromised relative to fresh glasses from the same sample.
182
183
Download