Coffee Data Science

Measuring Fines Migration with a Laser

Another look at tiny particle movements

Robert McKeon Aloe
5 min readMar 14, 2023

Espresso extraction is a complicated process, and there is still so much to learn. Often, problems in extraction have been attributed to fines migrating towards the bottom of the puck during the shot.

However, the theory of fines migration seems to be light on evidence in favor, and I have done a far amount of experiments to show that fines don’t migrate as much as people thought. A major criticism of my experiments was the resolution of imaging techniques, which is why I used a laser this time!

Methods of Measurement

Previously, I have used imaging methods to examine fines migration. One of the challenges in imaging is that I only have two dimensions of the particle in question, and it is difficult to assume the shape of the article by it’s projection onto the camera.

I like imaging methods because one can examine the shape of the particles and possibly give a better understanding for how grinds differ not just in size but shape.

Imaging will get to around 50um for one pixel and potentially a little further to 30um using sub-pixel analysis. Some critics have agreed fine particles larger than 50um don’t migrate, but their concern was for even smaller particles. The definition of fines varies from 5 um to 200um in diameter, but most will agree less than 100 um in diameter is fine.

Laser particle measurement is more accurate and can measure much smaller particles using diffraction patterns. They can go down to 1 um and below in theory.

Design of Experiment

I used the same design of experiment as before which uses spent coffee to measure fines migration. The challenge with coffee particles is that they change after extraction. So how can you tell the difference between extraction and fine particle movement?

With spent coffee, you don’t have that problem, and the substance at hand is the same just lacking solubles and CO2. If anything, spent coffee is the worst case of fines migration because the flow will be unimpeded by solubles or degassing.

If fines migration doesn’t occur in spent coffee, I have major doubts it occurs in freshly ground coffee. My previous experiments showed fines migrated a little (<1% of the fines moved), and other experiments showed that coffee behaves similar to soil which means coffee needs a high volume of water to cause significant fines migration.

The puck of spent coffee was run through my Decent Espresso machine using a flat flow profile of 4 ml/s at 96C in 28 seconds. I put 20g in, pulled 63.65g liquid out with a TDS of 0.13%. I then sliced the puck in the top, middle, and bottom, and I dried these samples. The top was the top half of the puck, and the middle and bottom were both ~25% of the bottom of the puck (so ~50% total).

All images by author

Measurements

If fines migrates, one would expect the top slice to have the least amount of fines, the middle to have more, and the bottom to have the most. Some have claimed the fines come out into the puck, so I also measured what comes out in the cup. They weighed less than 0.01g because they didn’t register a weight on my Acaia Pyxis scale. Assuming it is the maximum of the scale error of 0.01g, then that would be 0.05% of all the grounds. This would be the worst case.

I sent these samples with coded labels to Christopher Hendon at the University of Oregon. He offered to run some samples on a Mastersizer, so I sent some without telling him what the samples were or what experiments I was running.

Here is the spent coffee and the three slices. There is no measurement lower than 15 um because nothing registered. However, considering that the amount of particles at 15um is 0.1%, I don’t expect many particles to be less than 15um.

Their patterns are very close, so let’s zoom-in on less than 100 um diameter particles.

This doesn’t follow the expected trend of the fines migration theory. We can look at this data using a cumulative bar chart for each bin measurement.

Alternatively, we can look at the cumulative distribution for these slices, and the bottom trends slightly lower. The middle doesn’t diverge until 50 um from the top slice.

We can look at the difference between cumulative distributions for these slices. Even if we assume the Top-Middle is the worse case of fines migration, it is 0.4% of the total grounds at 100 um particle diameter.

We can compare this to the original spent puck particles, and an interesting pattern does not emerge.

Lastly, I collected image based measurements. Those measurements found the bottom slice has more finer particles than the top.

All of these samples don’t show fines migration in a large percentage relative to how many fines there are.

In conducting these experiments, I have been very open to the idea that I could be wrong. I originally doubted fines migrating had a big impact on coffee because staccato espresso shots were able to work. So I ran some experiments, and the big brick wall was accuracy of measurement.

The remaining criticism is using spent coffee rather than fresh coffee. As the layers of the onion are pealed off, it stands to reason this theory should be reconsidered.

If you like, follow me on Twitter, YouTube, and Instagram where I post videos of espresso shots on different machines and espresso related stuff. You can also find me on LinkedIn. You can also follow me on Medium and Subscribe.

Further readings of mine:

My Book

My Links

Collection of Espresso Articles

A Collection of Work and School Stories

--

--

Robert McKeon Aloe
Robert McKeon Aloe

Written by Robert McKeon Aloe

I’m in love with my Wife, my Kids, Espresso, Data Science, tomatoes, cooking, engineering, talking, family, Paris, and Italy, not necessarily in that order.

No responses yet