Thursday, 21 February 2008

Wii Bowling: Serious Business

It's probably a good time to introduce my particular research area, "computational theoretical chemistry". As I discussed in the last long post, the ability of atoms to stick together comes from electrons' tendency to pair up, and atoms' need to fill up "shells". I'll elaborate on this in more detail, but the important point to take from this is that chemistry (the science of how matter comes together) is all about electrons. They're tiny, tiny little particles. The lightest nucleus, that of hydrogen, is two thousand times heavier than the electron it binds. That tiny mass means that the crazy rules of quantum mechanics dominate.

Now, the maths required to describe quantum mechanics is well known (Einstein was behind one of the seminal papers in at the start of the 20th century, and it got him the Nobel prize), and so we can use this maths to predict what a particular chemical system is like, but it's fairly torturous. It's possible to describe a hydrogen atom with a pen and paper, but as soon as you go above that (the simplest chemical entity, with just one electron to describe) you have to start making simplifications to even be able to solve the problems in principle. These approximations involve lots of repetitive methods. If you can remember trying to do long division, or find square roots by Newton's method, then you've done a similar sort of thing. We say that these methods of solving the problem are numerical rather than analytical.

Fortunately it wasn't too long before electronic computers (and theoreticians like John Pople who knew how to make them dance) came on the scene, so we could feed these long, boring problems into dumb but fast boxes and get the answers. This is computational, theoretical chemistry. These days there are lots of stupendously powerful computers to work with and handle all sorts of complicated problems. We work hand-in-hand with experimentalists, helping to figure out what could be going on in their experiments. In return, the experimentalists give us a way to test how accurate our predictions are.

As it happens, it's still very difficult to perform calculations on big systems. The way around this is to use simpler and simpler methods. Eventually you throw out quantum mechanics altogether and start describing molecules as little lumps (atoms) stuck together with springs (bonds). The springiness of the springs comes from experimental observations of similar bonds (a bond between a carbon atom and a hydrogen atom is always pretty much the same springiness, say). Then you solve the maths for that system, which is fortunately relatively simple. The theories behind weights and springs and so on are called "classical mechanics", and these were pretty well sussed by the end of the 19th century. Applying them to chemistry like this is called "molecular dynamics". It's got some limitations - you can't break the springs, usually, and some subtle effects can be missed - but it's still very powerful. The Folding@Home project uses this sort of method to study how proteins fold up, because they're absolutely huge.

There's no point in having a breath-takingly fast computer if you can't have a bit of fun with it, mind you. The Pittsburgh Supercomputing Centre decided to set up a simulation of a bunch of buckyballs - little spheres of carbon which are distant relations of the graphite in a pencil - and make a scientifically accurate game of microscopic Wii bowling. In fact they've hooked the remote into the spiffy molecular dynamics package NAMD and the spiffy visualisation tool VMD, to create something they've dubbed "WiiMD". Be sure to check out their YouTube videos. Many of these things are very difficult to look at by experiment. (You have to look at things sideways, and pick apart what's going on indirectly. I really do enjoy a good experimental method, mind you. I've kind of missed the detective work of decoding mysterious wiggly lines.)

Apologies for not updating sooner - I've been overrun in the lab (nothing like an upcoming presentation to convince you to get your work into order). I'll do a little video to discuss bonding in more detail, in particular how molecules interact with each other, and also a post about reactions, the changes which molecules go through. Then eventually some more quantum mechanics to put the stuff about valence bond theory into context. See you then!

No comments: