A matter of scale (ice spikes and silicon pillars)

One of the great things about science is how interlinked everything is once you start looking deeply enough. One example of this came a couple of months back when I was reviewing the electron micrographs from the latest experiment.

My work involves the growing of meso-scale (about a 100th of a millimeter) pillars on a silicon wafer by laser irradiation. A complex combination of melting, ablation (explosive boiling) and redeposition of material happens at the silicon surface, with the net result that pillars grow from the silicon towards the laser. The pillars spend most of their growing time bathed in a hot (>5,000 degree C) plasma, this tends to smooth out the tips and sides of the pillars by a combination of melting and sputtering.

In this particular experimental run, I’d been varying the conditions under which I grow silicon pillars. What I saw was most unexpected; nano-scale (around a billionth of a metre in size) spikes sticking out of the top of almost every pillar I’d grown. You can see in the pictures below, that the pillars are around a twenty microns long, and the tiny spikes stick out vertically from the tips.

Cross sectional view of silicon pillars showing nano-scale spikes, and a closeup of two pillars with spikes at their tips

After a bit of head scratching, it was realised that what we were seeing was just a much smaller scale version of the ice spike formation process.

If you’ve ever tried to make ice cubes with filtered or very pure water, you will probably have noticed that you sometimes get a spike or lump that grows upwards from the free surface of the ice. A very good example of one formed from a distilled water ice cube is shown below, it is about a hundred million times larger than the silicon spikes in the pictures above.

Ice spike growing from an ordinary ice cube. Image taken from the excellent ice spike site at Caltech

Rather amazingly, given the wildly different growing conditions, the process by which both ice spikes and silicon spikes grow is essentially the same. In both cases you start with a liquid – water or molten silicon which cools and begins to solidify on the outer edges until there is just a tiny space that remains unfrozen.

Both water and silicon expand as they freeze, liquid water is forced out of the center of the ice cube tray, up through the hole. If the conditions are correct, then water forced out of the hole will freeze into a hollow spike. The water freezes around the top of the spike, adding to its length. The spike continues to grow until all the water has frozen or the hollow in the spike freezes shut.

What happens with the silicon spikes is that the last pulse from the laser melts the tips of the pillars. These cool relatively slowly once the laser shuts off, the outside solidifying first and increasing the pressure on the liquid trapped inside the tips. The liquid silicon escapes though a hole or crack, forming a tube with frozen sides and a liquid centre. This grows like the ice spike until it freezes solid. In the case of the silicon spikes, the growth happens in a matter of microseconds, compared to the many minutes that ice spikes take to grow.

While you may have some trouble duplicating my work in your kitchen, you should be able to produce ice spikes quite easily. All the details are on the Caltech ice spike site