Analytical techniques

10Be

Both the outer weathered surfaces and the interiors of five samples, collected from Solander Island by Harrington and Wood in 1957, were chosen for 10Be analysis. The outer surfaces were measured to examine the effect of surface weathering and the incorporation of meteoric 10Be (delivered in dryfall and by precipitation) into cracks and weathering products near the sample surface. The samples, including two full procedural blanks using only beryl-derived carrier (isotope dilution spike, carrier addition) and two additional process blanks using acid-etched sediment in addition to beryl-derived carrier, were prepared in a batch of 16 at the University of Vermont cosmogenic nuclide extraction laboratory following a modified version of the flux fusion method presented by Stone (1998). Approximately 0.5 g of finely milled material was mixed with KHF and NaSO4 along with ~300 ug of carrier. The mixture was fused in a platinum crucible for several minutes until the resulting melt was clear. After cooling, the crucible containing the solidified fusion cake was plunged into a teflon beaker containing Milli-Q water (18.2 Mohm). The Be was leached overnight, excess K removed by HClO3 precipitation, and BeOH precipitated as the hydroxide. The hydroxide was burned to BeO, mixed with an equimolar amount of Nb metal powder, and loaded into stainless steel cathodes for analysis at the Center for Accelerator Mass Spectrometry (AMS), Lawrence Livermore National Laboratory.

The spiked beryllium isotopic ratio (10Be/9Be after carrier addition) was measured in multiple interrogations (3 to 5) of each target until the beam current dropped 25% below its initial value at which point analysis ended. Three secondary standards were run repeatedly to verify linearity of the AMS. Results were normalized to Nishiizumi standard 3110 with an assumed 10Be/9Be ratio of 2.85 × 10-12 (Nishiizumi et al., 2007). Standard-corrected isotopic ratios for samples ranged from 4.3 to 34 × 10-15. A blank correction was made by subtracting the average of the two beryl-only process blanks run with the samples (m = 6.1+/–2.5 × 10-16) from each sample ratio. In addition to running 10 unknowns and 4 process blanks, two samples which had been analyzed previously, Raoul and Late, were analyzed (see George et al., 2005). During these earlier analyses, the nominal value of the AMS standards was about 10% higher (due to then accepted half life of 10Be, 1.5 My). Recent recalibration of the 10Be half-life (Nishiizumi et al., 2007) led to a reduction of 10% in the value of the standards, thus comparing previous and current results requires a correction (lowering) of previous values by 10.7% (equal to the difference in half lives between the current and past values, Table 1).

Oxygen isotopes

Oxygen fluorination data were collected at the Scottish Universities Environmental Research Centre (SUERC). Five trachyandesites-andesites, two porphyritic enclaves, a cumulate-textured nodule from Solander Island and two samples from Little Solander Island were chosen for analysis. The samples were disaggregated using a SELFRAG at the Geochemical Analysis Unit (GAU), Macquarie University. Amphibole and biotite phenocrysts were handpicked, ultrasonicated in acetone, rinsed in milli-Q water and oven dried prior to analysis. Oxygen fluorination data were collected using an offline purification line and CO2 laser system with gasses transferred, in sealed vials, to a VG SIRA 10 isotope ratio mass spectrometer (IRMS). Prior to gas collection, samples and standards were loaded into a 12 port holder, 8 samples and 4 standards, in an evacuated stainless steel chamber sealed with a kalrez® pre-fluorinated o-ring and BaF2 window. This chamber was prefluorinated with ClF3 combustion gas overnight, prior to any measurements, to condition the samples and to remove any moisture or contaminants from inside the chamber. Samples were combusted, individually, in an atmosphere of ClF3 using a CO2 laser, operated manually and step heated to ~2000oC. Laser combustion released O2 from the silicates as well as fluorine gasses and excess ClF3. Unwanted fluorine gasses and excess ClF3 were frozen into waste traps (using liquid N2), while the O2 passed through the system and was converted to CO2 when reacted with a platinum heated graphite rod. Resulting CO2 gas was trapped in a vial using liquid N2 and transferred to a VG SIRA 10 Mass Spectrometer. All of the samples were calibrated against a mix of in-house quartz and garnet standards (JJB8, SES and GP 147) which were themselves calibrated against international and reference standards (NBS 28, UWG2). Reproducibility of all standards analysed was <0.2‰ with an overall standard error, across all measurements, of 0.16‰ (Table 2).

Zircon trace elements, 206U/238Pb geochronology and 176Hf/177Hf isotopes

Zircon grains were collected by panning during the 2010 field season from modern day beach sand and colluvium eroded from volcanic rock exposures on Solander Island (sample locations are available on the on-line PETLAB database (http://pet.gns.cri.nz) of GNS Science). The zircon fractions were further concentrated using heavy liquids. The concentrated zircon separates were handpicked to maximize variation in zircon morphology, colour and size. The grains were mounted and polished for major and trace element analysis, U-Pb geochronology and Hf isotope analysis at the GAU, Macquarie University. Major element concentrations were determined using a Cameca SX-100 electron microprobe, equipped with five wavelength dispersive spectrometers. An accelerating voltage of 15 keV, a beam current of 20 nA and a beam diameter of 5 mm were used. Counting times were 10 seconds for peak and 5 seconds for background measurements on either side of the peak. Spectrometer calibration was achieved using the following standards: albite (Na), hematite (Fe), kyanite (Al), olivine (Mg), chromium (Cr), spessartine garnet (Mn), orthoclase (K), wollastonite (Ca, Si) and rutile (Ti).

Simultaneous trace element analysis and U-Pb dating were determined using a New Wave 213 nm Nd:YAG laser system linked to an Agilent 7500cs ICPMS. The technique, along with analytical uncertainty, is described in Jackson et al. (2004). Analyses used a beam diameter of 40–80 μm, a 5 Hz repetition rate, and an energy output of 3.76 J/cm2. Synthetic glass standard NIST-610 was used for calibration of trace elements and GJ-1 zircon (GEMOC GJ-1, age 609 Ma; Jackson et al., 2004) was used for calibration of U-Pb isotopes. The data were reduced using GLITTER (GEMOC Laser ICPMS Total Trace Element Reduction) software. Hafnium was used as an internal standard for trace-element determinations, with HfO2 measured using a Cameca SX-100 electron microprobe. Near-concordant reference standards, 91500 (1064 Ma; Wiedenbeck et al., 1995; Griffin et al., 2006) and Mud Tank (732 Ma; Black & Gulson, 1978) were used to monitor reproducibility and instrument stability. Full trace element compositions and standards are given in the supplementary file. The low concentrations of 207Pb (80% of the grains fall below detection, n = 47) in the Solander zircons and associated low count rates and large analytical uncertainties in 207Pb/206Pb and 206U/238Pb preclude dating using conventional techniques. Therefore, the zircon grains were re-analyzed using a larger laser beam diameter of 100 μm with a 5 Hz repetition rate, to maximize ablated volume and count rate of daughter isotopes. The gem quality GJ zircon standard was used for calibration of U-Pb isotopes. Accuracy and precision of U-Pb isotopes were monitored by analyses of the Mud Tank zircon reference material (Black & Gulson, 1978). Individual grain ages and errors were determined using GLITTER v. 4.4 software (van Achterbergh et al., 2001). Only analyses with a homogeneous signal were used to infer age relationships (n = 22/38). Grains with heterogeneous signals contained relatively large spikes in measured Pb isotopes, interpreted to be regions of ablated zircon contaminated by common Pb (e.g. along subsurface cracks). These grains have spurious measured ages up to c. 4 Ma. Therefore, common-Pb contents were evaluated by monitoring the time-resolved signal, rather than using the algorithm described by Anderson (2002), where grains showing discordance, within 95% confidence limits, between 206Pb/238U, 207Pb/235U and 208Pb/232Th isotopic ratios are identified.

Hf-isotope analyses of zircon were obtained using a New Wave Research 213 nm Nd:YAG laser, attached to a Nu Plasma 005 multi-collector ICPMS, at GAU, Macquarie University. The complete analytical method with full analytical accuracy and precision is described in Griffin et al. (2000; 2004). Analyses used a beam diameter of 55 μm, a 5 Hz repetition rate and an energy output of 3.65 J/cm2. Reference standard Mud Tank yielded 176Hf/177Hf values of 0.282562 (n = 8) with 2.S.D. 0.000017. εHf values were calculated using the CHUR value of 0.282772 from Scherer et al. (2001).

Trace elements and Sr-Nd isotopes

The trace element concentrations for the DSDP site 279a basalt and sediment and five samples, collected from Solander Island by Harrington and Wood in 1957 were determined by inductively coupled plasma mass spectrometry (ICPMS), using established methods (Eggins et al., 1997) on an Agilent 7500cs Quadrupole ICPMS at the Geochemical Analysis Unit (GAU), Macquarie University. Fresh sample interiors were ultrasonicated 3 times in Milli-Q (18MΩ) water before being dried, crushed and powdered (<50 mm grainsize) in an agate mill. The samples and reference materials (BHVO-2, BIR-1 and BCR-2) were weighed (0.1 g) into clean 15 ml Savillex® Teflon beakers and digested in a 1:1 mixture of HF (Merck, suprapur grade) and distilled HNO3 (Ajax) at 160°C for 24 hours and dried down. This step was then repeated. A HF-HCl-HClO4 mixture was added to the digestions and fluxed for 2 days at 160°C. Samples were dried down at 200°C and fluxed in 6N HNO3 for 24 hours and dried down. Samples were then diluted to 10 ml in 2% HNO3 with trace HF. 1:1000 dilutions were spiked with a 15 µl aliquot of an internal standard solution of Li, As, Rh, In, Tm and Bi in 2% HNO3. Reference material BCR-2 was used as the calibration standard. Instrument sensitivity and drift were corrected using BCR-2 and the internal standard spike. The background was measured using the 2% HNO3 rinse solution. Standards and the full set of trace elements are given in the supplementary file.

Whole-rock DSDP site 279a basalt and sediment, along with BHVO-2 rock standard were digested in concentrated HF and HNO3 for Sr-Nd separation. Sr was isolated by a single pass through Teflon columns containing Biorad® AG50W-X8 (200–400 mesh) resin. The Nd fraction was purified using EIChrom® LN-spec resin (50–100 µm) following the method of Pin & Zalduegui (1997). Samples and standards for the Sr analysis were loaded onto single degassed rhenium filaments using 4 µl and 3 µl of TaCl5+HF+H3PO4+H2O (Birck, 1986) solution respectively. Samples and standards for Nd analysis were loaded onto double degassed rhenium filaments using 6 µl of 1NHCl:0.35NH3PO4 activator solution. Isotopic analyses of Sr and Nd were obtained using a Thermo Finnigan Triton thermal ionization mass spectrometer (TIMS) at the GAU, Macquarie University. The BHVO-2 standard yielded 87Sr/86Sr = 0.703477 ± 6 (n = 21) and 143Nd/144Nd = 0.512975 ± 6 (n = 25). SRM-987 Sr and JMC-321 Nd were analyzed to assess instrument performance during the time of analyses and gave long-term reproducibility values of 87Sr/86Sr = 0.710249 (2 S.D. = 0.000036; n = 18) and 143Nd/144Nd = 0.511115 (2 S.D. = 0.000044; n = 17), see Table 3 and the supplementary file.