Technical Comments

Comment on “Rapid cooling and cold storage in a silicic magma reservoir recorded in individual crystals”

See allHide authors and affiliations

Science  22 Dec 2017:
Vol. 358, Issue 6370, eaap8429
DOI: 10.1126/science.aap8429

Abstract

Rubin et al. (Reports, 16 June 2017, p. 1154) proposed that gradients in lithium abundance in zircons from a rhyolitic eruption in New Zealand reflected short-lived residence at magmatic temperatures interleaved with long-term “cold” (<650°C) storage. Important issues arise with the interpretation of these lithium gradients and consequent crystal thermal histories that raise concerns about the validity of this conclusion.

Rubin et al. (1) proposed that gradients in lithium concentration recorded in zircons from the 1314 ± 10 CE Kaharoa eruption (2) reflect diffusive modification of initially sharp step changes in lithium abundance. On the basis of diffusive modeling, they construed the gradients to reflect histories for the grains studied that could not have involved more than ~40 years above the rheological lockup temperature (750°C) or more than a total of ~150 to 1200 years above the solidus (650° to 700°C). Such a scenario was linked to prolonged “cold” (3) storage of the crystals and their host rocks, and as such was used to propose a short-lived history for the Kaharoa material as an eruptible magma. Such a scenario, although plausible in principle, is contradicted in this example by two key lines of evidence.

First, there is a mismatch between the natural scenarios modeled by Rubin et al. and the conditions under which the apparent diffusion characteristics for lithium in zircon used to undertake the modeling were obtained (4, 5). Rubin et al. report lithium abundances in their zircons typically of tens of parts per billion (ppb), yet the diffusion experiments on which the modeling is based (4, 5) involved abundances of hundreds of parts per million (ppm), three to four orders of magnitude higher than in the natural grains (1). The relevance of this discrepancy is that at the ppb level, movement of 1+ charged lithium in the zircon lattice will be heavily controlled by the requirement for charge balance, and will be locked especially to the position and/or movement of 3+ ions such as yttrium and the rare-earth elements (REE) (6, 7). Any elemental gradients in lithium at the ppb level are thus much more likely to be controlled by local charge balancing in the crystal and to bear no relationship to the diffusion rates used by Rubin et al. (1) that were measured (4) under lithium loadings of hundreds of ppm in a natural zircon (Mud Tank). This Mud Tank zircon used in the diffusion experiments (4, 5) contains only between 26 and 71 ppm total yttrium plus REE (8, 9), and ~35 ppb naturally occurring lithium (4), set against 100 to 600 ppm (4) or up to ~1000 ppm (5) lithium introduced into the crystals during diffusion experiments. In contrast, the content of yttrium alone in the Kaharoa zircons varies from ~40 to ~14,000 ppm [supplementary material of (1)], yet the peak lithium concentrations are at most ~350 ppb and generally <50 ppb (1).

We would contend that the distributions of lithium in the zircons described by Rubin et al. (1) are primarily controlled by local factors associated with charge balancing in the zircon lattice, consistent with the findings of (6, 7). Support for our alternative interpretation is given by the fact that the shapes (widths, slopes, and relative compositional contrasts) of peaks in the abundances of yttrium [figs. S1 and S2 of (1)] are similar to those of lithium in the same grain. Such a correspondence would, if the hypothesis of Rubin et al. were correct and yttrium and lithium had independent diffusion, imply that the diffusivity of lithium and yttrium in these grains was of a similar magnitude. Such a conclusion is wholly implausible with respect to the available data (48). That the positions of peaks in lithium do not always correspond with peaks in yttrium is not necessarily an issue, as charge balancing in zircon can span several different ionic species, including the REE, phosphorus, and hydrogen (with, for example, hydrogen abundances often being a factor of local yttrium, REE, and phosphorus contents) (10).

Second, there is a flaw in the logic used to match the inferred diffusion profiles in the zircons with the thermal history of the grains. Rubin et al. infer from textural information and U-Th disequilibrium model ages that the interior parts of their zircon crystals grew, incorporating a sharply defined step in the concentration of lithium. The time scales they then infer from diffusion modeling of the lithium profiles are then used to propose that the relevant parts of the crystal incorporating the lithium profile were held for all but a tiny fraction of their histories at temperatures at which they infer lithium diffusion to have been ineffective (e.g., a maximum of 1200 years at 650°C). The flaw in this logic is that between the crystal cores and the rims there has been growth of further, additional zircon, which demands [given the oscillatory and sector-zoned nature of the crystals under cathodoluminescence: figs. S1 and S2 of (1); (11, 12)] growth from a melt. The paradox is then that for the crystal to grow over the prolonged time scales indicated by the U-Th disequilibrium model ages given by Rubin et al. (1, 11) and others (12, 13), the zircon has to have been maintained in a melt (i.e., above the solidus) for much of that time [e.g., (12, 13)]. Of all the minerals present in rhyolites such as the Kaharoa (2), zircon is the one that is most mediated by diffusion to the growing crystal; the low supersaturations would require either substantial melt flow relative to the crystal or long enough time to diffuse to the nucleus for growth to occur. One cannot freeze a diffusive profile in an inner growth zone and at the same time immerse the crystal exterior in a melt to permit it to grow larger to the sizes reported (1, 1113).

In summary, we would contend that (i) the diffusion parameters used by Rubin et al. are inappropriate for the lithium concentrations reported by them in the Kaharoa zircons; (ii) the proposed lithium diffusion profiles represent near-equilibrium concentration profiles controlled by partitioning for charge balancing within the zircon structure; and (iii) the thermal histories proposed are wholly incompatible with evidence (1, 1113) for extended growth times of the zircons from melts. There is great potential in the use of lithium (both elemental and individually as 6Li and 7Li isotopes) as a geospeedometer in volcanic systems [e.g., (14, 15)], but diffusion modeling must first take account of compositional controls in the host minerals concerned [e.g., (6, 7, 14)] before time scales can be inferred with confidence.

References and Notes

Acknowledgments: We thank P. Wallace and M. Handler for discussions.
View Abstract

Navigate This Article