measurements al

advertisement
1
Supplementary Material Appendix A: Litlavíti Description and data.
2
Litlavíti is ~200 m across and ~60-70 m deep. Rubble fills the bottom of the
3
crater, but ~40 m of the crater wall exposes 10 distinct lava flows. The topmost flow
4
exposed in the Litlavíti crater can be traced back to the Stóravíti summit caldera. Using
5
technical climbing methods, a sequence of 13 lavas was collected from the vertical to
6
overhanging wall of the crater. This ~40 m vertical sequence samples the 10 flows
7
exposed along the crater wall. Except for a small 1-2 m section about halfway down the
8
wall of the crater, the flow contacts are fresh with no indication of weathered surfaces,
9
soil, or charcoal. The locations of the samples collected along the crater wall are shown in
10
Figure A1.
11
12
Figure A1. (a) Sample locations, flow contacts, and sample names collected from a
13
vertical cross section by technical methods in Litlavíti crater shown as a function of depth
14
measured from the crater rim. (b) Vertical profiles of Mg # and LaN/SmN, SmN/YbN, and
15
SrN/NdN from the Litlavíti lava section. Trace element normalization values are C1
16
chondrite: [La]=0.235 ppm, [Sm]=0.147 ppm, [Yb]=0.163 ppm. Over the 40 m vertical
17
sequence the Litlavíti lavas are nearly constant in composition, both for major and trace
18
elements (Table A1), except for LTLV-2, which has slightly lower MgO, Al2O3, and CaO
19
and has slightly higher concentrations of other major elements, as well as incompatible
20
trace elements, suggesting that this particular lava is slightly more differentiated. There
21
are no obvious systematic variations in trace elements with depth in the crater, which is in
22
sharp contrast to samples from the flanks of Stóravíti, the majority of which were
1
23
collected at distances of 5-25 km from the main caldera (Slater, 1996). The Litlavíti lavas
24
are generally more fractionated (i.e. lower Mg#s) and have less compositional variability
25
than the Stóravíti samples. Concentrations of FeO, incompatible trace elements, and
26
SmN/YbN and LaN/SmN are higher in Litlavíti compared to the rest of Stóravíti.
27
28
Table A1. Major and trace element data for Litlaviti and Krafla samples. Major and
29
trace element concentrations were measured at the GeoAnalytic Laboratory of
30
Washington State University (WSU) using XRF and ICP-MS, respectively. Detailed
31
descriptions of the XRF and ICP-MS techniques, including estimates of accuracy based
32
on measurement of international standards, are given in Johnson et al. (1999) and Knaack
33
et al. (1994), respectively. The precision (1σ) for major elements in basalts measured by
34
XRF is 0.11-0.33% for SiO2, Al2O3, TiO2, and P2O5 and 0.38-0.77% for other elements
35
(Johnson et al., 1999; Hart and Blusztajn, 2006). The precision (1σ) for trace elements in
36
basalts analyzed by ICP-MS is 0.77-3.2% for all elements except Th (9.5%) and U
37
(9.3%) (Knaack et al., 1994; Hart and Blusztajn, 2006). To establish the reproducibility
38
of the major and trace element measurements, repeated analyses were done of LTLV-
39
09B, the standard BHVO-1, and Hawaiian sample HK-10-KWWS-92 and the results
40
compared with accepted literature values (Gladney et al., 1988, and Sims et al., 1999,
41
respectively). These repeat analyses show that the reproducibility of major and trace
42
element data varies from element to element but in most cases varies <10%.
43
2
44
Supplementary Material Appendix B: Nd, Sr, Pb and Hf Isotopic Compositions-
45
Measurement Results and Analytical Details.
46
Sr, Nd and Pb isotopic compositions were measured at Woods Hole Oceanographic
47
Institution (WHOI) and hafnium isotopic compositions were measured at the Ecole
48
Normale Supérieure in Lyon, France. Measurement results are reported in Table B1. Our
49
analytical methods are detailed below.
50
Sr, Nd and Pb isotopic compositions were measured at Woods Hole
51
Oceanographic Institution (WHOI) using a Thermo Finnigan Neptune multi-collector
52
inductively-coupled plasma mass spectrometer (MC-ICP-MS). The analyses were done
53
on powders leached for ~1 hour in hot 6.2 N HCl, then repeatedly rinsed with pure water.
54
This leaching is essential, particularly for Sr and Pb, as shown by a comparison of Nd, Sr
55
and Pb isotopic analyses for the Litlavíti samples on both leached and unleached samples
56
(Appendix Figure B1 and Table B2).
57
58
For the Sr and Nd separation chemistries, powders were dissolved in a concentrated
59
HF:HClO4 mixture, followed by conversion of fluorides to chlorides by drying down
60
three times with 6.2 N HCl. Strontium and Nd separation were done by conventional ion-
61
exchange chromatography using DOWEX 50 cation-exchange resin, followed by
62
HDEHP-coated Teflon powder (Taras and Hart, 1987). The total procedural blank for
63
Sr was <400 pg and that for Nd <100 pg.
64
3
65
Neodymium isotopic ratios were normalized for instrumental mass fractionation relative
66
to 146Nd/144Nd = 0.7219 using an exponential law. Strontium isotopic ratios were
67
normalized for instrumental mass fractionation relative to 86Sr/88Sr = 0.1194 also using an
68
exponential law. For Sr and Nd isotopic measurements, the internal precision was 5–
69
10 ppm (2σ). The external precision, after adjusting to 0.710240 and 0.511847 for the
70
NBS987 Sr and La Jolla Nd standards, respectively, was <30 ppm (2σ) for Sr and <30
71
ppm (2σ) for Nd (Jackson and Hart, 2006; Sims et al., 2008). εNd was calculated using
72
(143Nd/144Nd)CHUR(0)=0.512638 (Jacobsen, and Wasserburg, 1980).
73
74
Lead separation chemistry followed the HBr-HNO3 procedure of Abouchami et al. (1999)
75
using a single column pass. Lead isotope compositions were normalized for instrumental
76
mass bias relative to NBS/SRM 997 203Tl/205Tl = 0.41891. Additionally, NBS981 was
77
analyzed as a bracketing standard every few samples (White et al., 2000) and then
78
normalized against NBS981 using 206Pb/204Pb = 16.9356, 207Pb/204Pb = 15.4891, and
79
208
80
keep all WHOI data self-consistent. Internal precisions for 206Pb/204Pb, 207Pb/204Pb, and
81
208
Pb/204Pb were 15–60 ppm. Analyses of USGS standards AGV-1 gave
82
206
Pb/204Pb=18.9414, 207Pb/204Pb=15.6548, and 208Pb/204Pb=38.5615 and for BCR-1
83
206
Pb/204Pb=18.8215, 207Pb/204Pb=15.6356, and 208Pb/204Pb-38.7309, in good agreement
84
with Weis et al. (2006). External reproducibility for analyses of these standards ranges
85
from 75 ppm (2σ) for 207Pb/206Pb to 200 ppm (2σ) for 208Pb/204Pb. The total procedural Pb
86
blank was 120-150 pg. Further details can be found in Hart et al. (2004, open file reports
87
10 and 12) and Hart and Blusztajn (2006).
4
Pb/204Pb = 36.7006 (Todt et al., 1996). The Todt et al. (1996) normalization is used to
88
89
Hafnium isotopic compositions were measured at the Ecole Normale Supérieure in Lyon,
90
France, using a Nu Plasma HR MC-ICP-MS. Hafnium separation and isotope
91
measurement protocols are given in Blichert-Toft et al. (1997) and Blichert-Toft (2001)
92
with the only difference being the use here of a second-generation Nu Plasma instead of
93
the first-generation Plasma 54. Hafnium isotopic ratios were normalized for instrumental
94
mass fractionation relative to 179Hf/177Hf = 0.7325 using an exponential law. Accuracy
95
and external analytical uncertainties of < ±30 ppm for Hf isotopic measurements were
96
estimated from repeated runs of the JMC-475 Hf standard (Blichert-Toft et al., 1997),
97
which gave 0.28216±0.00001 (2σ) during the course of this study. Since this is identical
98
within error bars to the accepted value of 0.282163 ± 0.000009 (Blichert-Toft et al.,
99
1997) for JMC-475, no corrections were applied to the data. Internal run errors for all
100
samples are less than half or one third of the external reproducibility. The hafnium total
101
procedural blank was <20 pg. εHf values were calculated using (176Hf/177Hf)Chur(0) =
102
0.282772 (Blichert-Toft and Albarède, 1997).
103
104
The observation that the Litlavíti samples are compositionally similar places a potential
105
upper limit on the measure of external reproducibility of the data. Assuming that all of
106
the Litlavíti crater flows are uniform in their isotopic compositions, the reproducibility
107
(2
108
ppm for 87Sr/86Sr, ~700 ppm for 208Pb/204Pb and 207Pb/204Pb, and 1350 ppm for
109
206
5
) of the isotopic data is 22 ppm for 176Hf/177Hf, 42 ppm for 143Nd/144Nd, 40
Pb/204Pb (Table B2). Both leached and unleached Litlavíti samples were analyzed for
110
Nd, Sr and Pb isotopes. The effect of this leaching step was negligible for Nd, but
111
significant and variable for Sr and Pb isotopes (Figures B1 and B2, Table B2)
112
demonstrating the necessity of leaching for Sr and Pb isotopic analyses consistent with
113
previous studies (e.g. Hart et al., open file report 10; Nobre Silva et al., 2009, 2010).
114
115
Whether the much larger variability in Pb isotopes is due to analytical uncertainties or
116
greater sample variance in Pb isotopes will require further studies. As pointed out by Hart
117
et al. (2004), given the large improvement in precision provided by MC-ICP-MS and
118
double/triple spike methods, sample heterogeneity may now be the limiting influence in
119
Pb isotope studies; other local and global studies have shown that for a given set of lavas
120
there is often a larger variance in the isotopes of Pb than in those of Nd, Sr and Hf (e.g.
121
Sims et al., 2002; Sims and Hart, 2006; Waters et al., 2011). The fact that the
122
reproducibility of Pb isotopes in rock standards such as BCR-1 and AGV-1 is better than
123
observed within the Litlavíti sample suite itself suggests that the variability is due, in
124
large part, to real sample variance and not to mass spectrometric instabilities. It is also
125
possible that more aggressive leaching would bring the Pb isotopes into even tighter
126
agreement/uniformity (Figures B1 and B2, Table B2).
127
128
In this regard, another useful analytical comparison for examining intra-laboratory and
129
inter-laboratory reproducibility for Pb isotope measurements is the Mývatnseldar 1729
130
flow, which was analyzed twice in this study from two adjacent localities (Krafla 3 and
131
Krafla 4) and once (I646) by Thirlwall et al. (2004). In this comparison (Figure B3), we
6
132
again assume that all three measurements are from the same homogeneous flow. The
133
intra-laboratory difference for Krafla 3 and Krafla 4 is 91 ppm, 97 ppm, and 131ppm for
134
206
135
WHOI values with those of Thirlwall (2004), the difference is 324 ppm, 318 ppm, and
136
418 ppm for 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb, respectively. This difference is
137
similar in both direction and magnitude as the offset between the Todt et al. (1996) NBS
138
981 values used in this study and the more recent triple and double-spike NBS 981 values
139
measured by Galer and Abouchami, (1998), Thirlwall et al. (2002), and Baker et al.
140
(2004). The NBS 981 values in these later studies are all similar to each other (< 60 ppm
141
for 206Pb/204Pb, < 185 ppm for 207Pb/204Pb, and < 105 ppm for 208Pb/204Pb), but all
142
significantly higher (300-360 ppm for 206Pb/204Pb, 515-700 ppm for 207Pb/204Pb, and 638-
143
697 ppm for 208Pb/204Pb) than the values measured by Todt et al. (1996). See Baker et al.
144
(2004) for a compilation of double and triple spike TIMS and MC-ICP-MS
145
measurements of NBS 981. (Again we note that the Todt et al., 1996 value is used in this
146
study because this is the WHOI lab tradition and we feel it is important to maintain
147
consistency among all data produced from the same laboratory).
Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb, respectively. When comparing the average of the
148
149
With this difference in the NBS 981 normalization being noted, there is good agreement
150
between our data and the recent high-precision studies of Krafla and Theisturekyir of
151
Thirlwall et al. (2004), Baker et al. (2004), and Peate et al. (2010) (Figure B3). Our data
152
lie on an array between the glacial Gaesafjoll data (Peate et al., 2010; this study) and the
153
Arnahvammurhraun data (Baker et al., 2004; Peate et al., 2010); and as discussed above,
154
there is a reasonable match between our data and those of Thirwall et al. (2004) for the
7
155
1729 Mývatnseldar flow of Krafla, with much of the difference being an effect of the
156
different NBS981 values used for data normalization. However, as has been noted before
157
in other studies (Thirlwall et al., 2004; Baker et al., 2004; Peate et al. 2010) (Figure B3),
158
there is a significant off-set in the Stracke et al. (2003b) data from the ever evolving
159
Icelandic Pb isotope arrays, and this off-set is probably analytical.
160
161
That being said, for Nd, Sr and Hf isotopes there is a good agreement between our new
162
data and those of Stracke et al. (2003b). The Stracke et al. (2003b) study and this study
163
are the only two studies measuring Hf and Nd isotopes in samples from northern Iceland
164
(Figure B4). For these isotope systems these two data sets are well correlated within their
165
respective uncertainties and are consistent with the global basaltic array (Figure B4).
166
These two data sets are also complementary as the present study includes the
167
measurements of several glacial age lavas, while Stracke et al. (2003b) analyzed several
168
early postglacial lavas, in this way defining each end-member.
169
Table B1: 143Nd/144Nd, 87Sr/86Sr, 208Pb/204Pb, 207Pb/204Pb and 206Pb/204Pb of basaltic lava
170
samples from Krafla and Theistareykir.
171
Table B2: 143Nd/144Nd, 87Sr/86Sr, 208Pb/204Pb, 207Pb/204Pb and 206Pb/204Pb for leached and
172
unleached Litlavíti samples.
173
Figure B1: 143Nd/144Nd versus 87Sr/86Sr for leached and unleached Litlavíti samples.
174
Referential error bars represent 25 ppm uncertainties.
8
175
Figure B2: 208Pb/204Pb versus 206Pb/204Pb for leached and unleached Litlavíti samples.
176
Referential error bars represent 200 ppm uncertainties.
177
Figure B3. Comparison of A) 208Pb/204Pb and B) 207Pb/204Pb versus 206Pb/204Pb for Krafla
178
and Theisteraykyir samples measured by MC-ICPMS and internal Tl normalization and
179
external sample-standard bracketing using NBS 981 and MC-ICP-MS double spike data
180
for the same volcanic centers (Stracke et al., 2003b; Thirwall et al., 2004; Baker et al.,
181
2004; Peate, 2010; this study). The double spike data are shown with filled symbols,
182
while the Tl-doped data are shown with open symbols.
183
Figure B4: Hf versus Nd for glacial, early postglacial and late postglacial/recent lavas
184
show a comparison between the data of this study (filled symbols) and data from Stracke
185
et al. (2003b) (open symbols).
186
9
187
Supplementary Material Appendix C: Table C1. Compilation of ages, locations and
188
isotopic and major and trace element data for the entire suite of glacial (n =39), early
189
postglacial (n= 76) and late postglacial (n =16) lavas from Krafla and Theistareykir used
190
in this study.
191
192
10
Download