Crystallization thermometers for zircon and rutile

Abstract

Zircon and rutile are common accessory minerals whose essential structural constituents, Zr, Ti, and Si can replace one another to a limited extent. Here we present the combined results of high pressure–temperature experiments and analyses of natural zircons and rutile crystals that reveal systematic changes with temperature in the uptake of Ti in zircon and Zr in rutile. Detailed calibrations of the temperature dependencies are presented as two geothermometers—Ti content of zircon and Zr content of rutile—that may find wide application in crustal petrology. Synthetic zircons were crystallized in the presence of rutile at 1–2 GPa and 1,025–1,450°C from both silicate melts and hydrothermal solutions, and the resulting crystals were analyzed for Ti by electron microprobe (EMP). To augment and extend the experimental results, zircons hosted by five natural rocks of well-constrained but diverse origin (0.7–3 GPa; 580–1,070°C) were analyzed for Ti, in most cases by ion microprobe (IMP). The combined experimental and natural results define a log-linear dependence of equilibrium Ti content (expressed in ppm by weight) upon reciprocal temperature:

$$\log ({\text{Ti}}_{{{\text{zircon}}}}) = (6.01 \pm 0.03) - \frac{{5080 \pm 30}}{{T\;(\hbox{K})}}.$$

In a strategy similar to that used for zircon, rutile crystals were grown in the presence of zircon and quartz (or hydrous silicic melt) at 1–1.4 GPa and 675–1,450°C and analyzed for Zr by EMP. The experimental results were complemented by EMP analyses of rutile grains from six natural rocks of diverse origin spanning 0.35–3 GPa and 470–1,070°C. The concentration of Zr (ppm by weight) in the synthetic and natural rutiles also varies in log-linear fashion with T ⁻¹:

$$\log ({\text{Zr}}_{{{\text{rutile}}}}) = (7.36 \pm 0.10) - \frac{{4470 \pm 120}}{{T\;(\hbox{K})}}.$$

The zircon and rutile calibrations are consistent with one another across both the synthetic and natural samples, and are relatively insensitive to changes in pressure, particularly in the case of Ti in zircon. Applied to natural zircons and rutiles of unknown provenance and/or growth conditions, the thermometers have the potential to return temperatures with an estimated uncertainty of ±10 ° or better in the case of zircon and ±20° or better in the case of rutile over most of the temperature range of interest (∼400–1,000°C). Estimates of relative temperature or changes in temperature (e.g., from zoning profiles in a single mineral grain) made with these thermometers are subject to analytical uncertainty only, which can be better than ±5° depending on Ti or Zr concentration (i.e., temperature), and also upon the analytical instrument (e.g., IMP or EMP) and operating conditions.

Appendix

Natural rocks: sample descriptions and temperatures of zircon and rutile growth

Our knowledge of the growth temperatures of zircon and rutile crystals in the natural rocks is dependent upon temperature estimates obtained by other means, in most cases by other workers. We must assume, in addition, that the zircons and rutiles did in fact grow at these estimated temperatures. For the metamorphic rocks, we generally assume that the rutile and zircon crystals we examined grew simultaneously at peak conditions and that their compositions did not change during cooling (this assumption seems well founded for zircon, less so for rutile in some cases; see below and Diffusion issues: preview of new data for Zr in rutile and speculations on Ti in zircon of main text).

Bishop tuff (725°C)

The BT rhyolitic magma has been the target of several detailed investigations (e.g., Hildreth 1979; Anderson et al. 2000), so sample descriptions are not needed here. Moreover, because the BT magma was not saturated in rutile, the assigned zircon crystallization temperature (725°C) is addressed in the main text (Zircon). As noted, we ignore the apparent late heating event recorded by Ti in quartz phenocrysts (Wark et al. 2004), because zircon would likely dissolve slightly during this event, and because the capacity of a magma to crystallize zircon at this late stage is exceedingly small due to the low solubility (Watson and Harrison 1983).

Labait harzburgite (1,070°C)

The zircon and rutile in this harzburgite are found in diffuse metasomatic veins with phlogopite, orthopyroxene, olivine, chromite and sulfides (Rudnick et al. 1999; Lee and Rudnick 1999). The crystallization temperature of 1,070°C reported by Lee and Rudnick (1999) is based on the two-pyroxene thermometer of Brey and Köhler (1990) and the CaO-in-orthopyroxene thermometer, also of Brey and Köhler (1990). As noted in Zircon, the LB probably experienced a brief, late heating event immediately prior to eruption. The existence of such an event is implied by the high-Ca rims on orthopyroxene noted by Lee and Rudnick (1999); the zircons and rutiles we analyzed shown corroborating evidence. The zircons have locally high Ti abundances (up to ∼800 ppm) in patches near the grain margins, and the rutiles exhibit continuous, high-Ti rims suggesting diffusive uptake of Zr in response to increased T (see Fig. A1). The inferred late heating event does not appear to have affected the spatially dominant interiors of the zircon and rutile crystals, which are quite uniform in their Ti and Zr contents, respectively (see Tables 3, 5, Fig. 10). It is these broad interior regions that are assumed to have formed at 1,070°C. The LB is estimated to have equilibrated at P∼3.0 GPa on the basis of the low Al₂O₃ content of orthopyroxene (Lee and Rudnick 1999).

Fig. 10

Rim-to-rim electron microprobe scans across two rutile grains in the Labait harzburgite. The “background” level of ∼13,000 ppm is used in the Zr-in-rutile thermometer. Note the consistent enrichment in Zr near the grain margins, probably caused by pre-eruptive heating that caused “in-diffusion” of Zr from nearby zircons. The negative “spikes” in Zr content are exolved ilmenite lamellae. See text and Appendix for discussion

Adirondack migmatite (765°C)

Zircons and rutile crystals analyzed in this study were separated from two samples of a migmatitic metapelite outcrop in the southern Adirondack Highlands of New York State, USA (43° 14.630′N; 74° 19.732′W), designated ADK-01-6A and ADK-01-6B by Storm and Spear (2005). The major-phase assemblage is garnet + biotite + plagioclase + alkali-feldspar + quartz + sillimanite. Leucosome material dominates the outcrop, but dispersed restitic portions remain. Rutile grains from both leucosome and restite were analyzed for Zr (Table 5); all zircons came from the leucosome. Storm and Spear (2005) used garnet-biotite thermometry and GASP barometry to estimate peak P–T conditions of at least 790°C and 0.7–0.9 GPa. If biotite reacted out completely during progressive heating, the peak temperature may have been as high as 820°C. Storm and Spear (2005) placed the solidus at ∼740°C, which means that the system experienced a melt-present interval of at least ∼50°C. We assumed that the rutile crystals and the CL-dark rims on the zircons (Fig. 3) grew during cooling from a peak temperature of ∼790°C down to the solidus at ∼740°C; the 765°C midpoint was used for plotting purposes. The range in Ti content of the zircon rims may reflect real temperature variation over the growth interval, but the ion probe lacks the spatial resolution needed to systematically traverse the rims. Zoning of rutile with respect to Zr is not systematic, and the concentration differences among grains are larger than those within individual grains (see Table 5 and main text for discussion).

Stillup Tal aluminous schist (580°C)

This sample is from a shear zone through a granodiorite in which the rocks have been transformed into aluminous schist through extensive fluid–rock interaction. Our zircon + rutile separate was recovered from the biotite–phengite schist (zone III) described by Selverstone et al. (1991), who determined the P–T history of adjacent garnet-bearing rocks (zones I, II and IV) using garnet–biotite thermometry and garnet-plagioclase-muscovite-biotite barometry. The garnet–mica schists of zones II and IV recorded final equilibration conditions of 540–575°C and 0.7–0.8 GPa; zone I indicated slightly higher temperatures of 570–610°C. Lacking better constraints, we chose to plot our zircon and rutile information at 580°C.

Sifnos blueschist (470°C)

Metabasites and metapelites of Sifnos (Cycladic Islands), Greece preserve high-pressure assemblages from an Eocene (40 Ma) metamorphic event. The rutile grains analyzed for Zr in this study occur in both mafic and pelitic hosts, and are assumed to have grown at 470±30°C, an estimate based on garnet–omphacite geothermometry by Schliestedt (1986). The equilibration pressure is estimated to have been 1.5±0.3 GPa. See Spear et al. (2006) for additional discussion.

Vermont greenschist (485°C)

This sample is from the eastern Vermont metamorphic belt near Bellows Falls, Vermont (USA) and contains the major-phase assemblage garnet + biotite + chlorite + muscovite + plagioclase + quartz. Accessory rutile occurs in small (up to ∼300 μm), generally anhedral intergrowths with ilmenite. The P–T history of the rocks in this region has been determined using a variety of approaches, including thermobarometry, garnet zoning analysis, pseudomorph textures, and placement on petrogenetic grids, as summarized by Spear et al. (2002). The equilibration pressure of this sample is estimated to be 0.35±0.03 GPa.

Santa Catalina migmatite (660°C)

The rutiles and zircons analyzed are from the MM-RST suite of migmatites described by Sorensen and Grossman (1989). These migmatites occur as blocks in the upper part of the Catalina amphibolite unit, which is part of blueschist-bearing Cretaceous subduction complex. The migmatites represent the most altered lithologies of a probable metabasalt protolith, and contain (in addition to zircon and rutile) the assemblage garnet + hornblende + plagioclase + quartz + (minor) zoisite, clinozoisite, allanite, apatite ± muscovite ± (minor) titanite, kyanite and chlorite. The migmatites are believed to be the result of melting in the presence of low-salinity aqueous fluid (Sorensen and Barton 1987). Rutile and zircon crystals were sufficiently large and abundant to analyze in situ (i.e., without performing separations) by EMP and IMP, respectively, in thin sections generously provided by S. Sorensen. Sorensen and co-workers place the P–T conditions of migmatite formation at 0.8–1.1 GPa and 640–750°C, but were unable to narrow the temperature estimate any further. In this instance, we chose to narrow the range ourselves. We considered the zircons and rutiles to have crystallized near the lower end of this range (660°C) because this is much more consistent with the Zr contents of the rutiles. It is worth noting, also, that the water-saturated solidus of these rocks at 1 GPa is ∼650°C (Wyllie 1983; Johannes 1985). Our choice involves some circular reasoning—i.e., choosing a crystallization temperature for rutiles that are then included in the thermometer calibration—but this choice maximizes the consistency between the temperature range of Sorensen and co-workers and the temperatures indicated by the Zr content(s) of rutile and the Ti content(s) of zircon. As noted in the text, it is probable that rutile and zircon in most of the rocks studied co-crystallized over a range in temperature reflected by the ranges in Zr and Ti contents.