I play hide and seek with atoms (and the stats help me win)

Atoms are small. Too small. If you were to compare a corn to a sugar cube, you could also compare a single grain of sugar to Antwerp’s cathedral, for example! Can you imagine a little? Nor am I at all.

Put a hundred to ten thousand atoms together, and you get a nanoparticle. One of the many applications of such nanoparticles is that you can make a very good catalyst with them. For example, it can help convert toxic carbon monoxide (CO) into carbon dioxide (CO.).₂) in your vehicle’s emissions, so that no toxic gases are released into our air. Whether it can help reduce carbon dioxide₂ into usable raw materials to combat the greenhouse effect. The success of this depends on the shape of the nanoparticles. The atoms will try to hide behind each other by forming different rows. In order to know what nanoparticles I am looking at, I have to know exactly how many and how many atoms there are, respectively. A few extra atoms in the right place could ensure the nanoparticles work even better!

That’s why I play hide and seek with atoms. I want to be able to find every atom! How can I do that? In my lab, which is really nothing more than a computer and notebook for jotting accounts. Since I started my studies in physics – and even before that to be honest – I’ve preferred more theoretical calculations. In the lab I was already able to successfully complete the experiments, but I liked the more abstract arithmetic exercises more. So I am a world without a white coat.

Looking inside the nano world

Whoever says playing hide and seek also wants to see where the participants in the game are. Atoms are just too small to be seen. It’s even smaller than our famous coronavirus. So we need to find a way to peek into this nanoworld of atoms.

This is where my teammates play. In the research group where I did my PhD (EMAT, University of Antwerp) there are also scholars who love the incantations – as well as the obstacles – associated with empirical research. First they make a sample of nanoparticles. They are not made from cornstarch, as that can take a very long time, rather they are made by making a large substance small. A bit like using a mortar and pestle to process herbs in your kitchen. They then place the nanoparticles on a stand that they insert into the microscope – carefully, preferably with a steady hand – to look at. Hence the exciting wait. What did we actually do?

The microscope – as you can see me beaming on the occasion in the photo at the top of this article – is illuminated not with light, but with electrons on nanoparticles. These electrons are a little crazy, because they are particles and waves at the same time. If you describe them as waves, it turns out that they have a wavelength much smaller than light, so you can also look at smaller things with them. So small, in fact, that you can see individual atoms with it! In order to be able to see those atoms, my colleagues must first adjust the microscope perfectly. Then they search for nanoparticles they can see clearly, preferably without being destroyed by the many electrons they shoot at. Hence the holder still had to be rotated precisely, so that we could accurately visualize the rows of atoms in the nanoparticles from above. From my lab, comfortably in an office chair (or sometimes from a sitting ball to train my muscles: there’s no superfluous luxury as a mother of two!) I sometimes forget how incredibly crazy it is in fact, that they still manage to get a nice picture of atoms in a particle nano. Good example of experimental science!

A declaration of love for statistics

I analyze such images, and I do it with my beloved statistic. Cheesy, I know. I am a big fan of statistics. This allows me to translate those beautiful pictures with atoms of a beautiful picture into useful information. To do this, we describe the shape of the images in mathematical formulas – or at least as accurately as possible.

I’m using statistics because there are a lot of uncertainties associated with my search for those atoms. The image can be compared to a terrain map, where the higher density in the image corresponds to a ‘higher’ portion of the nanoparticles. This means that there are more atoms hidden behind each other.

This is not a one-to-one relationship, so we describe the intensity with a Gaussian distribution. This typical bell-shaped curve is really the distribution any statistician should have. And perfectly appropriate in this case. From these Gaussen we can get one number for each row of atoms related to the number of atoms.

There they are!

Well then we are there, right? Just look at the size of each number, assign it to the number of atoms in the row, and you’re done. Unfortunately not, because there is always noise. Another thing that stats can come in handy! So we need to group these numbers for each row of atoms by the same type. And then we’re already there. However, if all the atoms sit quietly in the same place.

If we make a movie, we see that atoms sometimes dare to change their hiding places. So another element of the model, which is how often those atoms cheat. Now we’re here, really. We can now find any atom, even if it changes its hiding places while imaging the nanoparticles.

This way I can now see what nanoparticles my colleagues made and whether they are stable and keep the same shape over time. This allows other scientists to work on developing new materials that have exactly the right properties they are thinking of. Even the best catalysts, for example, that really work in tackling climate change. Product development at the atomic level.

In this way, my Ph.D. has contributed to creating a much more beautiful world, making me feel like a good scientist – without a white coat!