Cabot Phillips

This is a sample can.

So this machine is specifically designed to measure the structure of whatever you put in that can.

And that can be anything you want.

I mean, it could be table salt.

But if you put a material in there and put it in the neutron beam, the data you get out of the instrument will tell you exactly with atomic resolution where all the atoms are inside.

So what they do is they take that data and then they shove it into a machine learning or AI and the AI looks at it and says, OK, now for the next day you need to do this because it's getting too strained in this place.

So it's steering the experiment as it goes.

So this is one of the big things.

And what's interesting, you can measure real engineering materials.

So that's a cylinder head assembly for a car.

That's an actual operating gas engine.

You can put that whole thing in the neutron beam and you can run the engine and you can synchronise the engine with the neutron beam pulse and you can look at what's happening in the combustion chamber.

What's the interest is to make materials stronger, lighter and cheaper.

With the AI, I imagine also you can model out a million different variables.

You have the possibility of having so much data that you can train the AI model on that it will become very accurate in its predictions.

I mean, they can train a chat GPT size model on Frontier very, very quickly.

And that's what the research community are doing.

This is what's coming out of the detectors.

These peaks are what we would call diffraction peaks.

The position in that direction and the height, that's the data which is telling you where the atoms are.