Thursday, October 05, 2006

A zircon that predates the universe 2

So, in the first episode, I decided to investigate the dread option two. It really doesn’t make sense. This is science, how could anything possibly go wrong? The whole point of SHRIMP is to make U/Pb geochronology simple enough for a trained monkey to perform. So what could the problem be?

Well, in theory, the zircon could contain common Pb, in addition to the radiogenic Pb that was measured. So, what happens to the results when I apply a common Pb correction? The date of 15.2 Ga drops to 15.1 Ga. And the error increases from 0.3 to 0.4. Next?

What else could happen? Well, the first rule of analysis is this: always make sure that the thing you’re measuring is actually what you think it is. For Th, there is no question. But what about Pb?

Pb is a trickier matter. Zirconium and Hafnium are so similar in chemistry that it took science 99 years to isolate Hf from Zr and identify it as a separate element. As a result, Zircon contains substantial Hf- generally 1-2%. Because SIMS creates numerous molecular species, it is worth investigating the possibility of a molecular interference, such as 174Hf16O2 on 206Pb, or 176Hf16O2 on 208Pb.

As it turns out, the whole reason for building SHRIMP back in the 70’s was to have an ion probe with enough mass resolution to separate these peaks. Hf and O both have higher binding energies, and therefore lower mass/amu, than does Pb. Specifically, 206Pb has a mass defect of -27mAMU, while 16O is -5.1 and 174Hf is -59.9. So the HfO2 peak should be 43.1 mAMU lighter. This is seen in figure 1, a mass scan taken on SHRIMP 1.

A few caveats about these figures. First of all, these are quick-and-dirty, low resolution mass scans, so SHRIMP-haters should not use these peak shapes as evidence that the machine is inherently imprecise. Secondly, these are all scans of the Temora standard, correctly centered on Pb, and not of the pre-universal grain. Thirdly, I know that the calibration of the X axis is off. Setting this is generally more trouble than it’s worth, as we measure peak positions relative to each other; the nominal “absolute” value is not all that important.

Anyway, as fig 1 shows, The HfO2 and Pb peaks are clearly resolvable. Furthermore, in this grain, the 206Pb peak is considerably larger than the HfO2 peak. This is not the case for the mass 208 peaks, as fig 2 shows.

There are two reasons for this. First, 176Hf is 33 times more abundant than 174Hf. Second, although the Th/U ratio of this zircon is greater than 0.3, Th has a much longer half-life, so relative to 206Pb, not much 208Pb has been produced in the 418 million year lifetime of this particular grain.

Thus, if the instrument was accidentally measuring HfO2 peaks instead of Pb peaks, one would expect to find very large Th ages, accompanied by modest 238U ages. That is exactly what was observed with the “pre-universal” grain. Unfortunately, the measured Th/HfO2 ratio is not very geologically useful.

As for how this happens, it is quite simple. The machine can be told to autocenter peaks, to allow it to correct for magnet instability. However, this centering assumes that the 206Pb peak will be larger than the HfO2 peak (such as is the case in fig1). If you have a young, or low U zircon, then the HfO2 peak may be larger. In such instances, the instrument will pick the Hf peak instead of the Pb peak, and the lab lemming who foolishly left centering on will get geologically meaningless results.

Anally observant readers will notice a discrepancy between my figures and my explanation. I won’t say what this problem is, but I will tell you this: The answer, in a single word, is:

1 comment:

C W Magee said...

First of all, I'd like to thank all the folks from, who have been pushing my traffic logs through the roof. I've noticed that despite this, nobody has guessed at the inconsistency, so I will offer a hint and a prize.

The prize is that the winner gets to pick a future blog topic for the lounge.

The hint is that discovering the inconsistency requires a quantitative, and not a qualitative reading of this post.