Portal de Periódicos da CAPES

Reply to comment by I. Pineda‐Velasco, T. T. Nguyen, H. Kitagawa, and E. Nakamura on “Diverse magmatic effects of subducting a hot slab in SW Japan: Results from forward modeling”

2015; Wiley; Volume: 16; Issue: 9 Linguagem: Inglês

10.1002/2015gc005984

1525-2027

Jun‐Ichi Kimura, Takashi Miyazaki, Bogdan Stefanov Vaglarov, Satoru Haraguchi, Qing Chang, James B. Gill,

Geochemistry and Geologic Mapping

The Comment by Pineda-Velasco et al. [2015] examined Pb isotope data presented by Kimura et al. [2014] which had been obtained from lava samples from the Daisen and Aono regions of Japan. Pineda-Velasco et al. [2015] presented new Pb isotope data for Aono, together with published data for Daisen [Feineman et al., 2013] obtained by double spike-thermal ionization mass spectrometry (DS-TIMS) [Kuritani and Nakamura, 2003],this data differed from that presented by Kimura et al. [2014]. The authors' points are that (1) there is uncertainty in the analytical results of Kimura et al. [2014] due to the effect of mass fractionation, and therefore, (2) the interpretations of Kimura et al. [2014] based on the extent of crustal assimilation and the estimated Pb isotopic composition of the crustal component are erroneous. First, I (JIK), as the lead author of the paper, regret to report a serious flaw in the original paper of Kimura et al. [2014]. Kimura et al. [2014] reported that all the Pb isotopes were analyzed by thallium-spiked multicollector inductively coupled-plasma mass spectrometry (TS-MC-ICP-MS) at Shimane University [Kimura et al., 2006, 2010]. However, upon reviewing the original data in response to Pineda-Velaco et al.'s Comment, we found that the samples from the Aono, Daisen, and Kannabe regions were analyzed using conventional TIMS methods at the same institute [Kimura et al., 2003]. All other samples were analyzed using TS-MC-ICP-MS [Kimura and Nakano, 2004; Kimura et al., 2006]. Because of this difference, mass fractionations and thus analytical errors for the Daisen and Aono samples are larger than originally reported. I am grateful for the Comment of Pineda-Velasco et al. [2015] for finding this flaw. Because Kimura et al. [2014] analyzed Aono and Daisen samples by conventional TIMS, we agree that our original results for them are less precise than the high-precision DS-TIMS results presented by Pineda-Velasco et al. [2015] and Feineman et al. [2013], and that the conclusion that there is a "significant role of crustal contribution to Daisen and Aono lavas in terms of Pb isotopes" should be revisited. The best way is to do this would be to use same sample set. We agree that although not exactly the same, Pineda-Velasco et al.'s [2015] new Aono data are comparable to those in Kimura et al. [2014] based on their sampling locations. It is less clear for Daisen. Feineman et al.'s [2013] sample are from the Karasugasen and Misen regions where there is a potentially large Pb isotopic variation [Kimura et al., 2014]. Therefore, we have reanalyzed the sample powders from Karasugasen, Daisen, and Aono using TS-MC-ICP-MS. We present the analytical results and conclusions from the new data below. The detailed description of our single-column separation TS-MC-ICP-MS analytical method at JAMSTEC has been published [Hanyu et al., 2011; Miyazaki et al., 2009]. Pb isotopes were measured by MC-ICP-MS (Neptune, Thermo Fisher Scientific, Bremen, Germany) using a desolvating nebulizer (Aridus II, CETAC Technologies, Omaha, USA). Mass fractionation factors for Pb were corrected using Tl as an external standard with 203Tl/205Tl = 2.3889 [Thirlwall, 2002]. Residual instrumental fractionations between Tl and Pb isotopes (∼320 ppm/amu in Pb isotopes) were corrected by applying a standard bracketing method using SRM981 and assuming 206Pb/204Pb = 16.9416, 207Pb/204Pb = 15.5000, and 208Pb/204Pb = 36.7262 for it [Baker et al., 2004]. During the reanalyses, powder splits of AO-3 (Aono) and 7-1 (Daisen) samples were leached in 6 N HCl for 1 h at 100°C prior to analysis to compare with unleached samples. A two-stage column path was also tested using the unleached samples. All the results are given in Table 1 and Figure 1. Analytical results of Daisen, Aono, and GSJ igneous rock standard samples. TIMS: Kimura et al. [2003] for GSJ standards and Kimura et al. [2014] for Daisen and Aono samples; Plasma 54: Kimura and Nakano [2004]; Neptune: this study; DS-TIMS K03: Kuritani and Nakamura [2003]; DS-TIMS K06: Kuritani et al. [2006]; Aono (DS-TIMS): Pineda-Velasco et al. [2015]. SRM981 (inset): TIMS: Kimura et al. [2003], raw data; Plasma 54: Kimura and Nakano [2004], Tl corrected; Neptune: this study, Tl corrected; BW08: Barling and Weis [2008], Tl corrected. Red dotted lines connect results for the same samples. The magnitude of mass bias in TS-MC-ICP-MS is disputed and may be a function of instrumentation. A systematic bias of ∼2000 ppm was reported between Tl-spiked SRM981 and Tl-spiked silicate samples when using an IsoProbe MC-ICP-MS [Thirlwall, 2002]. A systematic bias of <830 ppm was reported between Tl-spiked SRM981 sample and a Tl-spiked Fe-Mn nodules when using an AXIOM MC-ICP-MS [Baker et al., 2004]. A mass bias of <200 ppm (Mg, Fe, Al) and one of up to 620 ppm (Ca) was found in the case of intentional doping with the major elements onto a SRM981 sample using a new generation double-focusing Nu Plasma MC-ICP-MS [Barling and Weis, 2008]. We present results for the analyses on JB-2 igneous rock standard samples from the Geological Survey of Japan (GSJ) carried out at JAMSTEC over the course of one year (n = 38 separate powder digests). These results show <189 ppm 2SD (two standard deviation) errors in all Pb isotope ratios and an accuracy of <129 ppm compared with the DS-MC-ICP-MS values of Baker et al. [2004] (Figures 1 and 2). This was accomplished with no systematic mass bias, as shown by the one-to-one correlation between the measured raw Tl isotope ratios in the SRM981 bracketing solution and the silicate sample solutions (Figure 2). Further evidence for no matrix bias was found between single and two-stage column separations for AO-3 and 7-1, (Table 1). Correlation between Tl isotope ratios measured by bracketing SRM981 standards and bracketed unknowns. No systematic mass bias was observed between SRM981, and matrix-bearing Pb separates from silicate rocks during analyses over the past 12 months. Data from Neptune at JAMSTEC. Averages and two standard deviations are shown in the right plot for results before and after correction for mass fractionation by SRM981 bracketing using values of by Baker et al. [2004]. Earlier, Kimura and Nakano [2004] compared the TS-MC-ICP-MS results obtained using a Plasma 54 at Shimane University with JB-3 results obtained by DS-TIMS [Kuritani and Nakamura, 2003; Kuritani et al., 2006], and JB-2 results obtained by DS-MC-ICP-MS [Baker et al., 2004]. These results showed an external reproducibility of <200 ppm. Their JB-2 value was also identical to that obtained by JAMSTEC, assuring comparable reproducibility in the spiking methods at JAMSEC, Shimane University, and Okayama University (Table 1 and Figure 1). The TS-MC-ICP-MS results obtained at Shimane are also comparable with those obtained by any other high-precision methods, such as Kimura et al. [2006] and Weis et al. [2011] for Hawaiian basalts, and Kimura et al. [2010] and Ishizuka et al. [2003] for N-Izu lavas. The conventional TIMS analyses used in Kimura et al. [2014] employed a constant percent correction per amu (0.092–0.099%) based on NIST SRM981 analyses. The results are therefore no worse than any other conventional TIMS data [Dickin, 2005; Hamelin et al., 1985], namely within 3000 ppm (3 per mil) reproducibility [Kimura and Yoshida, 2006; Kimura et al., 2003]. Examples are shown in Figure 1 for JB-1 and JB-3 [Kimura et al., 2003]. The extent of mass bias is comparable to that of raw data for the SRM981 standards (Figure 1, inset). Some mass bias uncertainty remains in all conventional TIMS analyses. In contrast, standard rock data obtained by DS-TIMS and TS-MC-ICP-MS have an order of magnitude smaller mass bias (<300 ppm). Results by conventional TIMS can be more or less fractionated than by DS-TIMS or TS-MC-ICP-MS, but generally they are more fractionated (overcorrected) (Figure 1). Our new results obtained by TS-MC-ICP-MS for Daisen, Karasugasen, and Aono samples showed considerable overlap with those obtained by DS-TIMS in Feineman et al. [2013] and Pineda-Velasco et al. [2015]. Two-stage column separation or sample leaching made no difference (Table 1 and Figure 1). Therefore, the isotopic trends shown by the conventional TIMS in Kimura et al. [2014] were clearly analytical artifacts from mass bias (see red dotted lines in Figure 1). We thus conclude that the crustal assimilation proposed by Kimura et al. [2014] was erroneous in terms of Pb isotopes. This mistake may also apply to other published arguments for crustal contamination, or for addition of sediment to arc magma sources, when based only on conventional TIMS Pb isotopes. However, some crustal assimilation in the Karasugasen lava is evident from the chemical zoning of hornblende phenocrysts which shows deep (350–400 MPa) high-T (900–950°C) adakitic melt origin of the core versus shallow (120–200 MPa) low-T (750–820°C) sediment melt origin of the overgrowth mantle [Kimura et al., 2014, Figure 9] (additional P-T condition calculations by Ridolfi and Renzulli [2012] method based on the data in Auxiliary Material 3 in Kimura et al. [2014]). We are nevertheless grateful that the improved analytical precision in Pb isotope measurements corrects the misinterpretation of Pb isotopes in the original paper. The essential conclusion published in Kimura et al. [2014] concerned the sources and amounts of slab melts in the mantle wedge source as estimated by the ABS4 numerical mass balance model. Kimura et al. [2014] found a significant role of subducted sediments in the Daisen adakite (their Figure 17F), which agrees with and extends the conclusion of Feineman et al. [2013]. Although the Daisen, Aono, and Kannabe data in Kimura et al. [2014] were generated using conventional TIMS, this does not change the primary conclusion regarding the mantle source of the adakites. Our new high-precision Pb isotope compositions are almost the same as the model target compositions for both Aono and Daisen used by Kimura et al. [2014]. Those targets also are exactly the same as the data from Pineda-Velasco et al. [2015] for Aono, and <0.015 higher for 207Pb/204Pb and 208Pb/204Pb at 206Pb/204Pb = 18.30 for Daisen. Therefore, although the original Pb isotope argument for crustal contamination was wrong, the ABS4 modeling is unaffected. This is also true for the model mantle peridotite composition. The composition was estimated by averaging the Kannabe and Abu OIB basalts and weighted toward the Abu data, which were analyzed by TS-MC-ICP-MS [Kimura et al., 2014, Figure 8]. We are grateful for the Comment paper by Ivan Pineda-Velasco et al., which enabled us to correct the relevant discussion in Kimura et al. [2014]. We thank G-Cubed Editor Cin-Ty Lee for handling of this Reply to Comment. Data supporting Figure 1 are from this study and the literature, as noted in the figure caption. Data supporting Figure 2 are from this study together with a reference given in the figure. Data supporting Table 1 are from this study and references noted in the table.