Tuesday, April 06, 2010

Geochronology of the Cryogenian glaciations

A few weeks back Chris posted a lovely writeup of his work on the Oman cryogenian diamictites, and the snowball earth hypothesis. A recent paper on some Chinese diamictite ages has a nice summary of the chronology of this time period, so I thought I would throw it up here.

First, a short introduction. Between about 750 and 580 million years ago, a number of glacial sediments appear in the sedimentary record. Some of these appear to have been deposited fairly close to what was the equator at the time, and it was hypothesized that this may have been evidence for the planet having completely frozen over- a scenario known as the “Snowball Earth”. Subsequent stratigraphy identified three or four potential episodes of worldwide glaciation.

One of the ways of testing this theory is to see if all of the glacial strata assigned to a particular episode were actually deposited at the same time. Because there were no widespread diagnostic fossils this early in Earth’s history, biostratigraphy is not an option. One method used is to look at isotopic excursion in the sediments, on the assumption that they reflect a global signal, and not a local phenomenon. Another approach is to use radiometric dating to constrain the ages.

The problem with a radiometric approach is that there are relatively few sedimentary minerals that can be dated directly. So one generally looks for volcanic ash beds, crosscutting dykes, or other igneous rocks with an unambiguous relationship to the sediment in question. If these exist, then you can bracket the sediment in question with older-than / younger-than relationships.

There are several methods used to determine radiometric ages of neoproterozoic rocks. The most widespread is SHRIMP U/Pb geochronology. This method directly analyses the mineral zircon from igneous rocks of interest. Its advantage is that it is an in-situ method that only analyses a small portion of the grain, so it can be used on a wide variety of zircon crystals. The disadvantage is that achieving accuracy of better than 1% is difficult.

Another increasingly common method is CA-TIMS. In this method, zircon minerals are leached to remove damaged parts of the crystal, thin dissolved in acid and run on a traditional thermal ionization mass spectrometer. The advantage of this method is that it is much more accurate. The disadvantage is that it requires higher-quality zircons; inclusions and multiple growth episodes are more problematic.

A new method starting to be used is black shale Rhenium-Osmium dating, which directly dates the sedimentation of disseminated molybdenite in anoxic shales. This method should have similar accuracy to SHRIMP, but it is new enough that there could be unforeseen issues with its widespread application.

Neoproterozoic rocks have generally been metamorphosed since deposition, so K/Ar dating is usually not applicable.

In this paper, Xu et al. determine the age of three ash layers within a diamictite sequence using SHRIMP U-Pb geochronology. They then compare their results to those taken from allegedly correlated units from around the world, and those comparison charts (their figure 4) are shown below.

The following figures are laid out sideways, with older to the right. The lines represent the uncertainty in the measurement; the yellow arrow represents the relative age of the dated rock to the glacial event: younger, older, or synchronous.

As can be seen from the Gaskiers results, literature values do not contradict the existence of a short-lived, synchronous glaciation in the 580-590 Ma range. This study does not further refine this constraint, but the measurement is consistent with Xu et al’s uppermost unit being a Gaskiers correlate.

The Elatina/ Nantuo event is more commonly referred to as the Marinoan glaciation. However, the Marinoan rocks, in South Australia, have very poor age constrains, while the Nantuo is very well constrained.

So when testing for synchronicity, the Nantuo is a more sensible reference (in fact, it may be the golden spike). Once again, the Xu data does not shed much light on this event, but previous data does not contradict a single event in the 643-635 age range.

Finally, we get to the Sturtian. As is shown in this figure, ages for the glacial rocks below the Marinoan glaciation are all over the place.

Xu’s data is no exception; His data indicate a prolonged period of glaciation that precedes most Sturtian-aged rocks from other locales. As an examination of previous Sturtian work shows, a single world-wide Sturtian event is only allowed under the following conditions:
1. It is very long- with a minimum length of 40 million years.
2. It gradually thaws, with some areas deglaciating and precipitating carbonate caps tens of millions of years before other areas.

An open-access copy of the full paper can be found here.
XU, B., XIAO, S., ZOU, H., CHEN, Y., LI, Z., SONG, B., LIU, D., ZHOU, C., & YUAN, X. (2009). SHRIMP zircon U–Pb age constraints on Neoproterozoic Quruqtagh diamictites in NW China Precambrian Research, 168 (3-4), 247-258 DOI: 10.1016/j.precamres.2008.10.008


Anonymous said...

"Another increasingly common method is CA-TIMS. ___. The advantage of this method is that it is much more accurate."

Ack! Please don't make me get out the shotgun and toy ducks to explain the difference between accuracy and precision and why a precise date is not necessairily an accurate one....

Anonymous said...

Nice! I had missed this; thanks for blogging it up.

C W Magee said...

Go ahead, anonymous. CA-TIMS using a modern interlaboratory spike takes a lot of systematic errors out of the analysis. So that the accuracy is much greater, even at comparable count rates.

In geochronology, one usually stops collecting data at the point where systematic error dominates, because improving counting stats aren't helpful after that.

Unless you're doing laser ICP, of course, in which case you get heaps of counts instantaneously, and spend forever trying to figure out the systematic biases... ;)

Anonymous said...

...but a heterogeneous sample will still give you a mixed age regardless of how well it is measured... you just won't know it unless you characterise your sample first and then sub-sample. If that proecess is done routinely in CA-TIMS then I'll be impressed and stop shooting at toy ducks.

C W Magee said...

Traditionally, we use a well-characterized homogenous (in age) reference material to determine the degree of instrument fractionation, and apply that to the analytical error.

If you know a TIMS lab that wants to dedicate their days to confirming (or denying) in situ methodology, please give me their contact details...