Coffee bean porosity: a CT scan puts Mexican beans at 25.23%

SABy Sander · October 9, 2026 · 4 min read

An X-ray scan of roasted beans from four countries mapped the pore network inside them, and those pores decide how water gets through a brew.

A halved coffee bean on a scanner stage stops looking solid once light catches the cut face. The inside is a ragged honeycomb, much of it nothing at all. Coffee bean porosity is the share of a bean's volume that is gas rather than matter, and in roasted coffee beans it runs to a quarter or more. Two open access studies measured that structure rather than inferring it.

What roasting does to coffee bean pores

The pores are not in the green bean. They open during the roast, as heat and mass transfer drive pyrolytic reactions and thermal degradation. The core lags the surface while it happens. A 2025 modelling study in Frontiers in Food Science and Technology put the Biot number between 0.48 and 0.82 across roasting air of 200 to 280 degrees C, exothermic reactions starting above 150 degrees C.

The number, origin by origin

A 2019 study in Food Chemistry: X put shop-bought bags of roasted organic beans from Ethiopia, Colombia, Honduras and Mexico through a micro-CT scan. Percentage porosity, in ascending order: Honduras 23.5%, Mexico 25.23%, Colombia 30.42%, Ethiopia 31.43%. Mexican coffee beans sat near the bottom of that spread, nearer Honduras than Ethiopia.

Beans were scanned at 20 and 35 micrometre voxel resolution, 1600 to 3100 images per rotation, major cracks and the tegumentum void excluded, so the percentages describe the fine pore network, not the centre cut. Four bags off a shelf, one per country: the spread is real, but cannot establish an origin effect, and the paper never ranks cup quality by porosity.

Four roasted coffee beans drawn with rising pore density, standing for coffee bean porosity per origin

How uneven it is inside one bean

Features were heterogeneous within and between beans: two beans from one bag differ. Pore distribution was uniform, homogeneity of pore features across a bean's regions low, and Ethiopian beans showed the highest interconnectivity, individual pores fused into one larger cavity. Bean width differed between all four regions at p = 0.00, Honduras widest at 8.16 mm and Ethiopia narrowest at 7.19 mm, over 100 beans per region. The variability reaches the cup: coffee extraction averages over fragments that were never identical.

What the grinder does with the structure

Porosity changes how a bean breaks. The 2019 paper, citing work by Pittia and colleagues from 2001, notes that the porous microstructure affects mechanical properties such as brittleness, which govern grinding efficiency and the extraction yield of soluble solids. The open access paper carries a caveat: its discriminant analysis found no significant association between elemental levels and geographical origin.

What the burrs leave, once they have broken roasted coffee beans into fragments, is bimodal. A 2023 study in Scientific Reports puts the first peak at 30 to 40 micrometre, the fines, made of inner cell wall fragments. The fragments are not solid either: cell pockets of 30 to 60 micrometre sit inside them, the walls porous again at nanoscale. One grind size setting does not make one kind of particle. That study ground one medium roasted 100% arabica blend four ways, 6.7 g to a capsule, so its differences are grind, not coffee.

A coffee bed where water threads a crooked route between ground particles instead of running straight through

The water's route through the bed

The packed bed has a geometry of its own. In those 2023 simulations the bed porosity was 0.17 and the tortuosity 1.80, traced through 3706 labelled fluid particles: water covers about 1.80 times the straight line distance on its way out. Permeability came to 5.51 x 10-13 square metres and fell to 2.5 x 10-13 as the pressure head rose, with Reynolds numbers between 0.84 and 3.86. Coarser grinds raised the average permeability and its variance at once. Anything that narrows the particle distribution, including freezing the beans, a trick many specialty coffee drinkers try, works on that variance.

A limit belongs with those numbers. The resolution could not capture the pores inside the grains, so 0.17 is overwhelmingly the bed porosity of the cake: the gaps between particles, not the coffee bean pores counted one bean at a time. Comparing scans before and after extraction, the same authors found a decreasing porosity profile, bottom outlet to top inlet, always develops once water has been through.

Notice what the grinder gives back. One blend ground four ways produced measurably different beds, and coarser settings raised not only how fast water moved but how much that varied from spot to spot. When a brew runs unevenly, look first at the spread the grinder leaves behind, not the bean.