What I do at SNL: Part 4

People are always asking about what I actually do at my job at Sandia National Laboratories. I could claim that my work is too sensitive to discuss because it's related to national security or something, but (1) that's just not the case, (2) my research will be published in a public journal, and (3) that would stop me from talking about it (which isn't much fun). This series of posts will attempt to explain the basics of my work, but I can't promise it will be interesting or easily understood.

The Problem: Abnormal Grain Growth

Small grains are good for a variety of reasons, so grain growth is generally a bad thing once a material is actually being used; however, an even worse phenomenon (as the name might suggest) is abnormal grain growth, or AGG. Essentially, AGG has happened when a single grain (or just a few grains) grow very quickly. It is drastic enough that it falls under the category of "you know it when you see it." Here is a picture of a simulations where it has occured:

The blue grain obviously grew abnormally fast, and overtook the rest of the grains. This is bad not only because it eliminates the small grains, but it also cause irregularities in the structure. You don't want to have a grain structure that looks the same everywhere, except for in one spot where you have one huge grain. In fact, this is something that can happen when using aluminum in car body panels, and the abnormal grains can cause visible defects in the finish.

So, as I mentioned earlier, pinning is used by engineers in attempts to control grain structure. We all know things don't always work as intended, and that may be the case here. It turns out that Chris Roberts (a PhD recipient at CMU) did one isolated simulation in which a system grew abnormally because of pinning. This is a troubling finding because pinning is generally assumed to be a good way to prevent grain growth, especially AGG. Well, last summer my main finding was this: In a few occasions in microstructures that were already pinned (grain growth was very slow) AGG eventually happened if the simulation ran long enough. I verified Chris' findings, and was able to get a few limited statistics. In fact, the series of pictures above shows the first time I observed it. This is a fairly major finding (in the world of AGG) and will get my name on a peer-review journal publication.

This summer I'm working on improving statistics. I'm running simulations using a code package named SPPARKS on a supercomputer (named Thunderbird), and hope to see AGG at least 50 times. Last summer it happened in about 1/20,000 grains, so I just have to run lots ot trials, hoping to see a grain "explode." It's not good enough to know if it happens; I am trying to find out how often it occurs, how long before it occurs, what parameters prevent it, etc. Much of my work will be analyzing results trying to find patterns that can be used to predict AGG.

There you have it. This concludes the series, and I hope you enjoyed it. I get the feeling some of you were glad you didn't ask about my job now that you've seen these posts. It's pretty hard to explain for people who aren't familiar with materials science.