Slow oxygen diffusion rates in igneous zircons from metamorphic rocks

Empirical tests of oxygen exchange rate in zircon crystals from amphiboliteand granulite-facies metamorphic rocks of the Grenville Province demonstrate preservation of igneous dO through protracted igneous and metamorphic histories, forming the basis of quantitative estimates of diffusion rate. Granitic orthogneisses, which cooled slowly after granulite-facies metamorphism, show no consistent relationship between zircon size and dO, indicating slow oxygen diffusion. Detrital zircon crystals from granulite-facies quartzites are out of equilibrium with their host rocks, and no consistent correlation is seen between dO and grain size in high-precision analyses by laser fluorination of multiple grains, sieved for size. In a single sample, individual detrital zircon crystals preserve grain-to-grain variability in dO (determined by ion microprobe), ranging from 5.0 to 9.5‰. The inherited cores of some zircon crystals are up to 5.6‰ lower than igneous overgrowths, showing that gradients of 5.6‰ can be preserved over 50 mm even at magmatic conditions. All of these lines of evidence show that oxygen diffusion in zircon in these rocks was slow both during metamorphism and during slow cooling of 1–3 ∞/m.y. Calculations based on the measurements indicate that the oxygen diffusion rate in zircon (D) must be ≥ 10 cm/s at 600 ∞C to explain dO(zircon) values measured from Grenville quartzite and orthogneiss. This value is consistent with the experimentally determined value of D = 2 ¥ 10 cm/s for dry diffusion experiments extrapolated to 600 ∞C (Watson and Cherniak 1997). These results indicate that oxygen-isotope analysis of zircon may be used to see through granulite-facies metamorphism and anatexis, and to unravel crustal recycling processes in igneous rocks. PECK ET AL.: SLOW OXYGEN DIFFUSION IN ZIRCON 1004 studies should have exchanged completely with their host rocks, thus resetting dO(zircon). To understand this apparent discrepancy, and to constrain diffusion coefficients for zircon in nature, we examine the oxygen-isotope systematics of amphiboliteand granulite-facies lithologies that have cooled slowly after metamorphism. We use both high-precision laser fluorination of zircon separates and in situ ion-microprobe analysis to assess oxygen-isotope heterogeneity and zoning to constrain oxygen diffusion rates. ANALYTICAL TECHNIQUES Samples weighing 10 to 40 kg were crushed, and zircon crystals were separated using a shaking table, heavy liquids, and magnetic techniques. Polished grain mounts of zircon were imaged with back-scattered electrons (BSE) and cathodoluminenence (CL) using an SX-50 electron microprobe. Zircon splits were separated based on magnetism, hand-picked for purity, and soaked in cold HF to identify and dissolve metamict crystals. This screening was to identify samples that are obviously altered and may have experienced non-diffusional oxygen exchange. Note that if altered samples (e.g., metamict zircon with dampened CL) were analyzed inadvertently, it would cause the oxygen diffusion rate to be overestimated, and not change the conclusions of this study. Minerals were analyzed for dO by laser fluorination at the University of Wisconsin; data are given in Tables 1 and 2. This method analyzes 1–3 mg samples with high precision and accuracy (typically better than ±0.1‰). Oxygen was liberated from mineral separates by heating with a CO2 laser in the presence of BrF5 (see Valley et al. 1995). Forty-four aliquots of garnet standard (UWG-2) were measured on the 11 days of analysis. The overall raw dO of UWG-2 averaged 5.71 ± 0.12‰ (1 standard deviation reported, 1 s.e. uncertainty in the mean = 0.02). This is within the error of the long-term laboratory average for UWG-2 of 5.74 ± 0.15‰ (n = 1081, 1 s.e. = 0.005‰ ; Valley et al. 1995). The daily mean value of UWG-2 also averaged 5.71‰, and daily precision (1 s.d.) averaged 0.08 ± 0.03‰. Forty-eight zircon samples were analyzed in duplicate, and reproducibility averaged 0.07 ± 0.06‰. Some zircon splits were ground to a powder in a boron carbide mortar and pestle, which makes reaction while heating with the CO2 laser more reproducible (use of a powder reduces samples jumping from the holder). Zircon splits that have been analyzed both as powders and whole grains have identical average dO values, but grinding often improved reproducibility from ca. ±0.08‰ to ±0.06‰ (Peck 2000). Analyses were corrected to the VSMOW scale by the amount that daily UWG-2 values deviated from 5.80‰, its accepted value (Valley et al. 1995). This correction averaged 0.10‰ and was always less than 0.28‰. Oxygen-isotope ratios for single grains of zircon from quartzites 97ADK2 and 97ADK4 were measured by ion-microprobe/secondary ion mass spectrometry (SIMS) on a modified Cameca ims 4f ion microprobe at the University of Edinburgh (see Valley et al. 1998a). Precision and accuracy are ±1.0‰ (1 s.d.) for this method, based on counting statistics and duplicate analyses. Gold-coated samples were sputtered with a Cs beam defocused to a ~20 mm diameter spot. The resulting craters were ~3 mm deep. Secondary O and O ions were extracted with an energy offset of 350 ± 20eV. Sample analyses were standardized by bracketing analyses of a standard zircon of similar HfO2 content, KIM5 (Peck et al. 2001). 160 analyses of KIM-5 were made in the course of 192 sample analyses. Operating conditions, standardization, and data reduction are described in Peck et al. (2001) and raw data are given in Peck (2000). Table 3 summarizes ion-microprobe data. Core and rim analyses were made on some individual grains. Single zircon crystals mounted in superglue (cyanoacrylate) were ground and polished through the center of each crystal longitudinally parallel to the c-axis, and the interior was analyzed for oxygen-isotope ratio. Grains were then freed from grain-mounts with acetone, and pressed into indium with their crystal growth faces upward flush with the metal (see Fig. 2). This procedure allowed the outermost ~3 mm of the crystal (the depth of the ion sputtered pit) to be analyzed. EMPIRICAL TESTS OF OXYGEN DIFFUSION IN ZIRCON Two high-grade lithologies (orthogneiss and quartzite) were selected for empirical diffusion studies. In quartzite, nonmetamict detrital zircon and host rock were out of oxygen-isotope equilibrium before high-grade metamorphism. The approach of these zircon crystals toward isotopic equilibrium with their host rocks during metamorphism is rate limited by oxygen diffusion through the zircon crystal structure. We also examined zircon crystals from various metaigneous rocks, to assess oxygen diffusion in rocks from which zircon would most typically be analyzed. In orthogneiss, non-metamict zircon is assumed to be in initial magmatic equilibrium with its host rock, but to experience diffusive exchange during metamorphism and subsequent cooling. If oxygen diffusion rates in zircon are fast FIGURE 1. Arrhenius plot of experimentally determined and calculated oxygen diffusion rates in zircon. Hydrothermal “W&C (wet)” and anhydrous “W&C (dry)” experiments (Watson and Cherniak 1997). Dark line shows range of experimental temperatures, short dashes are extrapolation. “F&G (calc. wet)” (long dashes), “Z&F (calc. wet)”, and “Z&F (calc. dry)” (fine lines) are calculated diffusion rates (Fortier and Gilletti 1989; Zheng and Fu 1998). “K model” refers to “fictive” diffusion coefficients calculated for a cooling metabasite (see text; Kohn 1999). The circle is the anhydrous datum of Muehlenbachs and Kushiro (1974). The arrow shows the empirical estimate from this study, D < 10 cm/s at 600 ∞C. FIGURE 2. Cartoon of sample handling for single detrital zircons from Adirondack quartzites. Single grains were mounted in superglue and were ground through the center of each zircon longitudinally parallel to their C-axes. After ion-microprobe analysis, zircons were freed from grain mounts with acetone and were pressed into indium with their crystal growth faces upward and flush with the metal for additional ion microprobe analysis of the outermost 3 mm of the crystal. PECK ET AL.: SLOW OXYGEN DIFFUSION IN ZIRCON 1005 TABLE 1. Oxygen isotope ratios of zircon from granulite facies orthogneiss (Adirondack Mountains, New York) Sample Peak Mag. d18O d18O d18O d18O d18O d18O Meta. Zircon Zircon Zircon Zircon Zircon Zircon Temp >149mm 149–105mm 105–74mm 74–53mm <53mm Bulk LDT tonalite gneiss 675 ∞C M-1 7.29 ± 0.04 7.38 ± 0.04 7.48 ± 0.03 7.70 ± 0.10 7.84 ± 0.03 TOE metagranite 675 ∞C M0 8.33 ± 0.06 8.42 ± 0.03 8.25 ± 0.02 8.40 ± 0.05 8.28 ± 0.14 AM86-8 charnockite 750 ∞C M-2 8.19 ± 0.03 8.21 ± 0.07 8.24 ± 0.01 7.95 ± 0.11 8.30 98LB3 metasyenite 675 ∞C NM2 8.39 ± 0.08 8.11 ± 0.05 8.15 8.14 ± 0.00 8.27 ± 0.07 8.22 ± 0.01 AM86-6metagranite 725 ∞C M-2 7.74 ± 0.13 7.66 ± 0.06 7.70 ± 0.02 7.64 ± 0.02 7.75 ± 0.10 >149mm 149–74mm <74mm NOFO-1 metagranite 725 ∞C M0 8.26 ± 0.04 8.35 ± 0.02 8.33 ± 0.00 Notes: Analyses of d18O are reported in standard per mil (‰) notation relative to VSMOW. Analysis is discussed in text. Mag. indicates magnetic (M) or non-magnetic (NM) and angle of side tilt on the Frantz magnetic separator. Analyses of different size zircons separated in bulk by sieving (e.g., 149–105 mm, indicating zircon diameters) are indicated. All samples were powdered with a boron carbide mortar and pestle before analysis. Duplicate analyses are indicated by ±, with the average reproducibility shown (half the difference between analyses). Peak metamorphic temperatures are after Bohlen et al. (1985). Sample locations are given in Peck (2000). enough, zircon crystals of different sizes (separated by sieving) should show variable amounts of resetting reflecting the different closure temperatures of different crystal sizes. GRENVILLE QUARTZITES Quartzite samples were collected from granulite-facies localities in the Frontenac terrane (Ontario) and Adirondack Highlands (New York), and supplement the analyses of Frontenac terrane quartzites presented by Valley et al. (1994). Five samples were collected from the metasedimentary sequence in the Adirondack Highlands (ADK and SR sample numbers, Fig. 3). Samples 97ADK-2, 97ADK-3, and 97ADK-4 are from the southern Adirondack Highlands, and belong to the “Irving Pond Formation” as described by Wiener et al. (1984). Sample 92ADK7 was collected from the Swede Pond Quartzite (Eastern Adirondacks). Sample 99SR1 is an interlayered quartzite and calc-silicate lithology from the peak of Mount Pisgah (Central Adirondacks). This metasedimentary sequence is older than ~1.3 Ga (see Fig. 4; McLelland et al. 1988). In the Frontenac terrane, sample 97BM2 was collected from locality SL-1 of Sager-Kinsman and Parrish (1993). This sample contains striking heavy mineral layers that define relict cross-bedding. Sample 97WE10 was collected from locality CT-2 of Sager-Kinsman and Parrish (1993), and is a more typical massive quartzite. Most detrital zircon crystals in Frontenac terrane quartzites range from 1493 to 2580 Ma (37 single grain U-Pb ages), with two outliers yielding 1306 ± 16 Ma and 3185 ± 3 Ma (Sager-Kinsman and Parrish 1993). All but one quartzTABLE 2. Oxygen isotope ratios of zircon and other minerals from amphibolite and granulite facies Grenville quartzites (Ontario and New York) Sample Peak Mag. d18O d18O d18O d18O d18O d18O d18O Meta. Zircon Zircon Zircon Zircon Zircon Quartz Garnet Temp. >149 mm 149–105 mm 105–74 mm 74–53 mm <53 mm 92ADK7 725∞C NM2 9.10 ± 0.07p 9.35 ± 0.04p 9.68 ± 0.04p 10.14 ± 0.09p 14.75c 14.56c 12.48c 12.50c 14.91c 14.59c 14.38c 14.68c 97ADK2 675∞C NM2 9.00 ± 0.01 9.25 ± 0.01 9.51 ± 0.03 9.62 ± 0.08 13.38b 13.72b 11.44c 11.09c 13.02c 13.82c 11.13c 11.32c 13.59c 13.84c 11.24c 97ADK3 675∞C NM2 7.46 ± 0.21 7.81 ± 0.22 7.45±0.11 7.83 ± 0.12 12.90b 12.54b 12.56c 12.72c 12.65c 12.54c 97ADK4 675∞C NM2 8.85 ± 0.22 9.05 ± 0.09 9.04 ± 0.03 8.72 ± 0.08 12.29b 12.08b 10.42c 9.56c 11.81c 12.14c 9.81c 9.77c 12.08c 12.11c 99SR1 775∞C NM2 12.00p 14.62c 14.48c 97WE10 750∞C NM3 8.93 ±0.01 8.17 ± 0.16 8.98 ± 0.19 13.85b 13.45c 13.59c 13.57c 13.65c 97BM2 750∞C NM2 9.28 ± 0.05 8.72 ± 0.13 15.83b 15.81b 15.94c 15.64c 15.52c 15.73c SL2 750∞C 7.76 ± 0.03 7.38 Data from Valley et al. 1994 Rt81 Zircon <75 mm 9.15 16.71 16.20 CT1 Zircon 149–105 mm 9.60 ± 0.24 15.68 15.61 CT2 Zircon >105 mm 9.82 ± 0.13 16.12 SL1 Zircon >149 mm 7.91 ± 0.17 14.40 14.67 SL2 Zircon 149–105 mm 7.90 14.14 14.21 Notes: Analyses of d18O are reported in standard per mil (‰) notation relative to VSMOW. Analysis is discussed in text. Mag. indicates magnetic (M) or non-magnetic (NM) and angle of side tilt on the Frantz magnetic separator. Analyses of different size zircons separated in bulk by sieving (e.g., 149–105 mm, indicating zircon diameters) are indicated. ‘p’ indicates samples which were powdered with a boron carbide mortar and pestle before analysis. Duplicate analyses are indicated by ±, with the average reproducibility shown (half the difference between analyses). For quartz and garnet ‘b’ indicates a bulk mineral separate and ‘c’ indicates a single mineral fragment. Peak metamorphic temperatures are after Anovitz and Essene (1990) and Bohlen et al. (1985). Sample locations are given in Peck (2000). PECK ET AL.: SLOW OXYGEN DIFFUSION IN ZIRCON 1006 ite sample contains over 90% quartz, with minor to trace amounts of feldspars, micas, amphibole, pyroxene, garnet, oxides, and zircon. Sample 99SR1 is layered and contains abundant diopside and calcite in subequal proportions to quartz. ANALYSIS OF dO(ZIRCON) BY LASER FLUORINATION Different size splits of zircon (avg. radius ~20–90 mm) and quartz were analyzed from all quartzite samples, and garnet was analyzed from three samples (Table 2, Figs. 4 and 5). Zircon crystals from each quartzite sample are from a mixed population. Morphology and color are heterogeneous within each sample, and grains show variable amounts of rounding and abrasion. Note that these features are distinct from the broken grains often observed after separation of zircon crystals from igneous lithologies. We interpret the majority of zircon grains in quartzite to be detrital and from a mixture of sources, as is also illustrated by the range in ages in Frontenac zircons. The dO zircon from “pure” quartzites (>90% quartz) ranges from 7.4 to 9.6‰, and the average dO of quartz ranges from 12.1 to 15.7‰. There is a small variation in dO(quartz) observed in some samples when single pieces of quartz were analyzed (0.2 to 0.8‰, Table 2). This variability at the scale of 20 to 50 cm is expected given the large sample sizes crushed for analysis. In the three garnet-bearing quartzites, constant quartz-garnet fractionations [D(Qtz-Gt) = 2.2–2.3‰] are observed. Because D(Zrc-Gt)~0 at high temperatures (Valley et al. 1994, 2003), similar, systematic fractionations should be seen between quartz and zircon if they equilibrated during metamorphism. The results in Figure 5 are consistent with equilibrium between quartz and metamorphic garnet, but show a distinct lack of equilibration between quartz and detrital zircon. Only in 99SR1 [dO(zircon) = 12.00‰] is the fractionation between zircon and quartz consistent with metamorphic exchange, but this is likely to be a coincidence, as none of the other Adirondack samples have similar quartz-zircon fractionations. Values of dO(quartz) have a great deal of variability from sample-tosample in quartzite, probably because of mixing of different sedimentary and diagenetic sources of quartz. In contrast, values of dO(zircon) are more constrained, similar to dO values in igneous rocks (Peck et al. 2000, 2001; Valley 2003), and do not depend systematically on grain size (Fig. 6). ANALYSIS OF dO(ZIRCON) OF INDIVIDUAL CRYSTALS BY ION MICROPROBE Ion-microprobe analysis was undertaken in two quartzite samples to test for grain-to-grain oxygen-isotope variability among individual zircon crystals, and to test for diffusion profiles within single grains. Single zircon crystals (radius a88 mm) were handpicked from the largest size split (Table 1). In-situ ion-microprobe analysis of dO confirms the lack of equilibration during metamorphism. Variability is observed between the average dO values of the interiors of zircon grains from 97ADK2, FIGURE 3. Sample location map for quartzite and orthogneiss samples from the Adirondack Mountains, New York. The area of Precambrian outcrop is shown. Isotherms for regional metamorphism are from Bohlen et al. (1985) and Kitchen and Valley (1995). Quartzites: 1 = 99SR1, 2 = 92ADK7, 3 = 97ADK2, 4 = 97ADK3, 5 = 97ADK4; Orthogneiss: 6 = 98LB3, 7 = NOFO-1, 8 = AM86-6, 9 = AC85-6, 10 = AM86-8, 11 = TOE, 12 = LDT. CCMZ = Carthage-Colton Mylonite Zone, stippled pattern is Marcy anorthosite massif. TABLE 3. Summary of ion microprobe analyses of d18O in detrital zircon grains from Grenville quartzites 97ADK2 and 97ADK4 d18O d18O d18O D Crystal Grain n Inherited n Crystal n (CF–GI) Interior Core Face 97ADK2 1 6.6 ± 0.9 3 12.6 ± 0.5 3 6.1 2 5.9 ± 1.0 4 3 6.5 ± 0.0 2 11.3 ± 1.0 2 4.8 4 7.9 ± 0.0 2 5 8.0 1 6 5.7 ± 0.4 9 12.8 ± 0.5 6 7.2 7 7.5 ± 1.1 4 11.4 ± 0.9 4 4.0 8 7.8 ± 0.5 3 9.3 ± 0.7 3 1.5 9 8.1 ± 0.4 3 11.8 ± 1.1 3 3.7 10 7.4 ± 0.4 4 10.9 ± 0.3 2 3.5 11 5.1 ± 0.4 4 9.8 ± 0.7 7 4.6 12 4.9 ± 0.5 4 11.3 ± 0.8 4 6.4 13 9.7 ± 0.4 4 14 5.5 ± 0.6 5 15 9.0 ± 0.5 7 12.7 ± 0.5 4 3.8 16 5.5 ± 0.4 4 17 6.3 ± 0.6 4 10.5 ± 0.5 4 4.2 18 6.1 ± 1.0 4 11.1 ± 0.6 6 5.0 Average = 11.28 ± 0.24 48 97ADK4 1 9.4 ± 0.7 4 8.6 ± 1.2 2 10.7 ± 0.4 2 1.3 2 10.0 ± 0.1 3 7.1 ± 0.0 2 10.0 ± 0.5 4 0.0 3 9.7 ± 0.5 5 9.8 ± 0.5 3 0.1 4 11.2 ± 1.0 3 5 8.8 ± 0.7 4 10.1 ± 0.3 4 1.4 6 8.4 ± 1.3 3 6.5 ± 0.3 3 10.5 ± 0.3 4 2.1 7 11.3 ± 0.9 3 10.0 ± 2.0 2 –1.3 8 9.6 ± 0.6 9 10.0 ± 0.1 4 0.4 9 10.7 ± 0.5 3 5.1 ± 0.3 3 10.5 ± 0.9 4 –0.2 10 10.3 ± 1.1 3 11 9.8 ± 0.5 4 11.7 ± 1.0 2 1.9 Average = 10.30 ± 0.20 29 Notes: Analyses are reported in standard per mil (‰) notation relative to VSMOW. Full ion probe data and data reduction are detailed in Peck (2000). One sigma uncertainty in the mean of each oxygen isotope ratio (±) is shown, n = number of spot analyses. D(CF-GI) shows the fractionation between crystal faces and grain interiors, which is large and variable in 97ADK2 and small in 97ADK4. PECK ET AL.: SLOW OXYGEN DIFFUSION IN ZIRCON 1007 ranging from 5.0 to ~9.5‰ (Table 3, Fig. 7). This range is consistent with derivation from igneous rocks. It is interesting to note that the lowest observed dO is ~5‰, a value in equilibrium with the mantle and consistent with derivation from a juvenile magma. For example, zircons from juvenile plutonic rocks of the 2.7–3.0 Ga Superior Province have dO = 5.6 ± 0.5‰ (King et al. 1998), whereas high dO(zircon) (>7‰) is consistent with derivation from magmas with some supracrustal component (e.g., Peck et al. 2000, Valley 2003). Metamorphic zircon grown in high dO metasedimentary rocks (e.g., ≥9‰) would also have high dO values. Comparison of interior and rim dO(zircon) values In-situ analyses of the outer 3 mm of crystal faces of zircon crystals in 97ADK2 gave consistently higher dO values than grain interiors (Fig. 7, Table 3). The average of all spot analyses of crystal faces in this sample is 11.3 ± 1.7‰ (n = 48 spots; 1 s.e. = 0.2‰). This result is consistent with oxygen diffusion profiles across the outermost few micrometers of the crystal. The dO values of the crystal faces are the same (within uncertainty) as dO(garnet) from the same rock [average dO(garnet) = 11.24 ± 0.14‰, n = 5, 1 s.e. = 0.06‰], indicative of approach to oxygen-isotope equilibrium between zircon faces, garnet, and quartz during metamorphism. We interpret the crystal face value of dO as the result of diffusion into the igneous growth face and not growth of new zircon during metamorphism of the quartzite. Detrital rounding and abrasions are observed on most grains, and no metamorphic rims are visible by BSE or CL. To further evaluate the FIGURE 4. (A) Summary of temperature-time evolution of the Adirondack Highlands, after Bohlen et al. (1985); Mezger et al. (1991); Chiarenzelli and McLelland (1993); McLelland et al. (1996); Spear and Markussen (1997); and McLelland et al. (2001). Estimates of peak temperature of metamorphic events are shown, and durations of orogenic deformation are given as horizontal bars. Note that temperatures are well constrained for the Ottawan, but only generally known for the Elzevirian orogeny. Ages of minerals that formed above their closure temperatures (Tc) are plotted vs. T. M = Monazite, S = Sphene, H = Hornblende, R = Rutile; B = Biotite. Minerals that formed below their individual Tcs (metamorphic zircon and garnet) are not plotted against temperature, but likely formed ca. 600–700 ∞C (see McLelland et al. 2001). See text for geochronology of Adirondack Highlands samples (minimum ages shown by arrows). All Adirondack Highlands samples examined in this study experienced granulite-facies metamorphism during the Ottawan orogeny followed by slow cooling. (B) Summary of 4A showing the integrated thermal history of the Adirondack Highlands. Note that the thermal pulse of 1150 Ma anorthosite-suite magmatism may lead into Ottawan heating (shown by striped pattern), or temperature may drop to low values before rising again during orogeny. The thermal paths of modeling approaches are shown [Fast Grain Boundary (FGB) cooling and isothermal models], and are conservative relative to the integrated thermal history of the rocks modeled. FIGURE 5. Values of dO(quartz) versus dO(detrital zircon) from Grenville quartzites. Data are from Table 2 and Valley et al. (1994). Analyses of zircons of different sizes (Fig. 6) from the same sample are connected by a vertical line, and garnet analyses (boxes) are indicated. Isopleths of temperature are the equilibrium DO(Qtz-Zrc)· DO(Qtz-Grt) (Valley et al. 1994, 2003). Values of dO(zircon) are clearly not in equilibrium with host quartz for the peak of metamorphism at 675 to 775 ∞C. In contrast, metamorphic garnet consistently shows high-temperature metamorphic fractionations (see