TRACE ELEMENT ZONING IN GARNET AS A TOOL FOR TECTONIC INTERPRETATION
Samples in this study are Proterozoic schist and gneiss from the southwest US that yield Lu-Hf garnet ages attributable to metamorphism during the Yavapai, Mazatzal, and Picuris orogenies. We measured qualitative Lu profiles of garnet in thin section by laser ablation.
Three samples from the Wet Mountains, Colorado, display varying Lu profile shapes. Sample 12W13 yields a Lu-Hf age of 1497 ± 4 Ma, and contains garnet with high Lu concentrations at both core and rim, which are separated by a trough. Sample OC14 yields a Lu-Hf age of 1601 ± 6 Ma. One garnet preserves a bell-shaped Lu profile, and a second garnet displays a flat core with a gradual Lu increase toward the rim. Sample 12W6 has a Lu-Hf age of 1476 ± 4 Ma. Crystals in this sample display a decrease in concentration from core to rim, with a high, narrow spike in Lu adjacent to the rim.
We use isochemical phase diagrams calculated for the samples to constrain reaction histories. Samples from the Wet Mountains contain minerals, textures, and inferred reaction histories indicative of garnet growth during melting. If a garnet is resorbed by melt shortly after growth, then the crystal may display complex Lu distributions without perturbing the Lu-Hf radiometric clock. This may be the case for sample 12W6, and its 1476 Ma age may date the partial melting of this sample. The Lu distribution of OC14 may indicate multiple garnet populations, or variable diffusion and resorption at the thin section scale. Sample 12W13 has Lu distributions and mineral textures that are consistent with extensive resorption, but with minimal diffusion, indicating initially rapid cooling rates.
We can test the resulting hypotheses using numerical models of growth, resorption, diffusion, and synthetic age calculations. These tools together lead to a more robust tectonic interpretation than would be possible from age or petrologic data alone. Hollister’s model of garnet growth is the essential concept that underlies these methods.