U-Series Dating of Carbonates by Mass Spectrometry With Examples of Speleothem Coral and Shell

Abstract

<p>Uranium series dating is a well established technique for dating carbonate deposits of the last 350,000 years. The most common carbonates which have been dated are cave calcites (speleothem) and corals. The technique relies on the buildup of ²³⁰Th over time by radioactive decay of ²³⁸U and ²³⁴U, in materials which were initially free of ²³⁰Th. The ratios of the activities of ²³⁰Th/²³⁴U and ²³⁴U²³⁸/U are entered into the standard equation for radioactive decay which is then solved for time. These ratios have traditionally been measured by counting the alpha particle emissions from each isotope. An alpha counting laboratory was set up in McMaster in the early 1970s and is still in operation today. The technique has been improved considerably but the precision of dating by this method is limited by the statistics of counting small numbers of emissions from the trace quantities of isotopes trapped in the carbonate crystal lattice. Typically U ratios can be measured to 1σ precision of 1% and Th ratios to 3%. This leads to an error in the estimated date of ~10%. Alpha counting requires a rather large sample, varying from ~10 g to ~40 g depending on the U content and age of the sample. It also requires extensive chemical preparation of the sample to isolate U and Th from the matrix and to completely separate U from Th.</p> <p>The development of high resolution and high abundance sensitivity thermal ionization mass spectrometry in the last decade has allowed for the measurement of heavy atom ratios to high precision with very small samples. The McMaster geology department acquired such a machine in the 1980s and it has been adopted for the measurement of ²³⁰Th/²³⁴U and ²³⁴U/²³⁸U ratios. The typical 1σ precision attainable for U ratio measurement is 0.03% and for Th ratios 0.15%, which leads to an error in the estimated age of only ~0.5%. This is more than an order of magnitude better than that attainable by alpha counting. The sample size required varies from ~0.3 g to ~6 g depending on the U content and age of the sample. Thus sampling can be at higher density or resolution than for alpha counting. The technique requires rigorous purification of U and Th from the matrix but separation of U from Th can be rough, so the chemical preparation is simpler than that for alpha counting and usually involves only anion exchange procedures. Because the samples are so small and the technique so sensitive, contamination must be carefully avoided: all operations take place in a very clean environment.</p> <p>Uranium and thorium ratios cannot be measured directly. Therefore, a tracer or "spike" of artificial isotopes of U and Th in a known ratio must be added. ²³⁶U and ²²⁹Th are used. The spike was made and its concentration and isotopic ratio calibrated against four standards of different origin. The accuracy of U ratio measurement was tested by measuring NBS U standards. The accuracy of Th ratio measurement cannot be directly tested because no Th isotopic standards are available. Instead it was tested by dating thc McMaster standard speleothem 76001.</p> <p>The main calcite sample to be dated was a piece of flowstone from -15 m in a flooded Bahamian cave. The many layers of calcite deposition separated by thin coatings of mud indicate that the cave was alternately air-filled (during calcite deposition) and water filled (cessation of calcite deposition) over the last 300,000 years. The drownings occur when sea level rises. The dates on the layers of calcite immediately below and above the mud-coated hiatuses give limits on the times when sea level must have been below -15 m elevation. The periods on non-deposition indicate when sea level was above -15 m. The dates on this sample allowed the construction of a sea level curve for this area of significantly higher precision than was formerly possible. The conclusion was reached that sea level events correlate with the marine foraminiferal isotopic signature which has been fitted to an orbitally-tuned timescale. This gives strong support for the orbital geometry theory (Milankovitch cycles) of climatic change.</p> <p>The second sample dated was a thermal water calcite crust from Wind Cave, South Dakota. High resolution dating of this 2 cm thick crust suggested that degassing is the principal control on sub-aqueous crust formation and revealed the behaviour of the aquifer over two glacials and one interglacial. This has been modelled in terms of aquifer recharge rate and mixing of waters of surface and deep groundwater origin.</p> <p>The third sample dated was a layered stalactite from Rat's Nest Cave, Alberta, which promised to yield a record of glacial advance and retreat in the Bow Valley, Rocky mountain foothills. Dating of the very thin layers simply revealed that calcite deposition correlates with interglacial periods and hiatuses correlate with glacials. A sample with thicker layers 3nd a greater number of layers is required for closer resolution of individual advance and retreat events.</p> <p>The rest of the study involved a small amount of test dating of other materials. Dating of calcite raft debris from Carlsbad Caverns gave some information on the rate cf drawdown of the water table. Dating of corals established that the technique works very well on this medium. Finally, dating of tiny samples of ostrich eggshells from the Sahara demonstrated that U-series dating may work on eggshells if they can be shown to have been closed to isotopic migration for most of their history. It also indicated that this now arid area was wetter during the last interglacial.</p> <p>Mass spectrometric dating of carbonates is now established at two world centres: Caltech and McMaster. In the future it may completely replace alpha counting for Th/U and U/U dating. However, alpha counting is still the best option for Pa/U dating and other techniques which require the measurement of trace amounts of isotopes with high activities.</p>