Variation of initial Th-230/Th-232 and limits of high precision U-Th dating of shallow-water corals

One hundred eighty U-Th data, including 23 isochrons on 24 pristine modern and Holocene corals and 33 seawater samples, were analyzed using sector-field mass spectrometry to understand the variability of initial Th-230/Th-232 (Th-230/Th-232(0)). This dataset allows us to further assess the accuracy and precision of coral Th-230 dating method. By applying quality control, including careful sampling and subsampling protocols and the use of contamination-free storage and workbench spaces, the resulting low procedural blanks give an equivalent uncertainty in age of only +/-0.2-0.3 yr for 1-2 g of coral sample. Using site-specific Th-230/Th-232(0) values or isochron techniques, our study demonstrates that corals with an age less than 100 yrs can be Th-230-dated with precisions of 1 yr. Six living subtidal coral samples were collected from two continental shelf sites, Nanwan off southern Taiwan in the western Pacific and Son Tra off central Vietnam in the South China Sea; one coral core was drilled from an open-ocean site, Santo Island, Vanuatu, in the western tropical Pacific; and modern and fossil intertidal coral slabs, 17 in total, were cut from six sites around the islands of Simeulue, Lago, North Pagai and South Pagai of Sumatra in the eastern Indian Ocean. The results indicate that the main source of thorium is the dissolved phase of seawater, with variation of Th-230/Th-232(0) depending on local hydrology. With intense input of terrestrial material, low Th-230/Th-232(0) atomic ratios of 4.9 x 10(-6) and 3.2 x 10(-6) with a 10% variation are observed in Nanwan and Son Tra, respectively. At the Santo site, we find a value of 5.6 x 10(-6) at 4 horizons and one high value of 24 x 10(-6) in a sample from AD 1974.6 +/- 0.5, likely due to the upwelling of cold water during a La Nina event between AD 1973 and 1976. The natural dynamics of Th-230/Th-232(0) recorded in the intertidal corals at sites in the Sumatran islands are complicated so that this value varies significantly from 3.0 to 9.4 x 10(-6). Three of the 141 modern coral 230Th ages differ from their true ages by -23 to +4, indicating the presence of detrital material with anomalous 230Th/232Th values. Duplicate measurement of coeval subsamples is therefore recommended to verify the age accuracy. This improved high precision coral 230Th dating method raises the prospects of refining the age models for band-counted and tracer-tuned chronologies and of advancing coral paleoclimate research. (C) 2008 Elsevier Ltd. All rights reserved.