Tuesday, May 25, 2010

Mystery Behind normal Earthquake Faults Solved

Cristiano and the rocks arrived at Penn State in October to begin the lab-friction experiments. We wanted to see if the fault rocks were weak enough to explain why low-angle, normal faults are active in that region. I had done similar studies on fault rocks before, so we started with the standard approach, which is to crush the rock samples and construct layers of rock powder that we could shear (abrade).

But Cristiano was insistent that we also measure the properties of the intact rock, by shearing it in the orientation it existed in within the fault zone. Andre Niemeijer, who was then a post-doc in my lab, and Igor Faoro, an Italian graduate student, had been working to develop methods for cutting fragile samples, so we started testing ideas about how we could create a sample that was roughly 5 centimeters (cm) by 5 cm by 1.0 cm and which had the fault zone fabric parallel to the main sample faces.

Luckily, we had quite a bit of sample, because the first several attempts failed. Then, Igor had a revelation and set up a makeshift sculpting studio by duct-taping a Shopvac hose to the leveling jig that held the sample and using a Dremel tool to sculpt the blocks.

When we began to run experiments, we noticed a problem right away. The rock powders had typical coefficients of friction (~ 0.6), but the solid wafers of rock--that Andre and Igor were by now experts at sculpting--produced much lower values. We were perplexed because we had made the powders and wafers from exactly the same fault rock samples, so the material properties should have been identical.

It's not uncommon to find experiment-to-experiment variability in friction due to heterogeneities in the rocks, but differences this large were unheard of. Even though all the samples were labeled in the field and packed in labeled containers, I thought we must have mixed them up. So, we made more powders and wafers and started again.

The result was the same, and now we were perplexed. The wafers were not perfectly homogeneous, but there was nothing visible to explain such large differences in steady-state sliding friction. After we reproduced this curious result three times, on different pieces from the same fault zone unit, I decided to take the wafers, after shearing, and powder them. That way we'd be sure that the bulk chemistry was the same in both cases.

We took each of the wafers and crushed them, and then made layers with the powders. To our surprise, the new powders had friction values of ~ 0.6! It wasn't until we made thin sections and started to think about the thin- (less than 10 millionths of a meter) but-very-abundant seams of clays that we realized these fabric elements in the rock were acting in concert to produce a form of lubrication.

The role of fabric in rock deformation at high temperature had been well known for many years, but as a community, geophysicists working on faults in the brittle field had not considered that they could be so important as a possible mechanism for fault weakening.

No comments: