Particle size and fines

Size statistics#

A PSD is summarised by a few characteristic sizes. The textbook chapter on grinding uses x10, x50 and x90: the sizes below which 10%, 50% and 90% of the coffee lies; x50 is the median.³

Equivalent diameters compare each particle with a sphere of the same surface or the same volume.³ Averaged over the sample, they give the surface mean d[3,2] (Sauter mean diameter) and the volume mean d[4,3] (De Brouckere mean).^3,4,8 The surface mean is strongly pulled down by the fines, while the volume mean lies close to the median.³ In a bimodal grind the two can lie far apart: 27.34 µm and 457.84 µm for one drip-filter grind.⁴ A ratio of surface to volume gives the specific surface.³

Number versus volume (our own arithmetic). The volume of a sphere scales with the cube of its diameter. One 500 µm particle has the volume of 1,000 particles of 50 µm, but those 1,000 particles have 10 times its surface. A number-based distribution is therefore dominated by fines; a volume-based one by the coarse peak. A sample with about 25% of its volume below 100 µm consists, by count, mostly of fines.

Choice of basis. In the 2024 espresso study the median was volume-weighted and based on the projected particle area in images.² The 2023 microtomography study reported its laser-diffraction results as volume shares and specific surface.¹ Sieving gives mass fractions directly.³ Comparisons between studies need the same basis and method.

Measurement in practice#

In a 2023 espresso study, dry and wet laser diffraction of the same grinds gave highly similar distributions.⁸ Sieve openings and fines definitions do not always match: the 2024 espresso study defined fines as particles below 100 µm but prepared its fines fraction with a 120 µm sieve.²

Two-domain extraction models#

Models of extraction from a bed treat ground coffee as two domains. Easily accessible coffee sits in fines and on the broken surfaces of larger grains; less accessible coffee sits in the intact cells inside larger grains.⁴ Both are described by first-order rate equations.⁴ Because fines have a larger surface-to-volume ratio, they extract faster than coarse grains, so different particle sizes extract at different rates.⁴

Measurements support a particle-size effect, but its cause is less clear. In a 2017 espresso study, smaller particles gave more caffeine and trigonelline per gram of beverage, while tamping pressure had no detectable effect.⁷ That study introduced a reduced model with only physically meaningful parameters.⁷ A 2023 study from the same Munich group measured its grinds by laser diffraction and found that their Sauter mean diameters did not differ significantly between grind levels.⁸ Surface area alone therefore could not explain the effect of grind level in that study.⁸

The 2024 espresso study varied the share of fines directly. Coffee was ground at burr spacings of 160–250 µm; at the three coarsest settings, 1, 2 or 4 g of sieved fines replaced the same mass of coffee in a 20 g dose; shots ran at 9 bar to 40 g of beverage.² Added fines did not produce outlying combinations of extraction yield and time.² The authors conclude that fines change bed permeability, not the extraction mechanism, and that the surface-area effect on extraction efficiency seems marginal.² Fast shots were already efficient: shots of under 10 s and of 15 s reached 17–18% extraction yield.²

Permeability and channelling#

According to the brewing chapter, the coffee bed’s permeability decides how water flows; too low a permeability gives high pressure, low flow and long brew times.⁵ According to the grinding chapter, a wide or bimodal distribution packs to a lower porosity and a higher pressure loss than uniform particles.³ Fines lower permeability in both of the measuring studies opened here.^1,2 In the 2024 study, a regression on the full distributions linked particles up to 150 µm to longer extraction time, those above 250 µm to shorter time.²

Feedback in the bed. A 2023 model treats channelling as a positive feedback. Extraction dissolves coffee, which raises porosity and permeability, so paths that carry more water open further.¹⁰ Small differences in porosity are therefore amplified over the shot.¹⁰ In this model, the drop in yield at fine settings comes from one part of the bed being fully extracted, and the authors caution that average yield may be a poor guide to flavour at fine grinds.¹⁰ The 2020 espresso study had read the same drop as strong evidence of inhomogeneous flow at fine settings, with poor reproducibility and wasted coffee as results.⁹

In the conical geometry of a pour-over, flow is uneven for geometric reasons as well; for a fine drip grind in a cylindrical bed, a model predicted a variation of about 5 percentage points in extraction yield.⁴ The brew methods entry covers bed shape.

Fines migration and bed consolidation#

Fines migration is the idea that water carries fines downwards through the bed. If it happens, fines would collect above the filter and lower the porosity near the bottom, because the filter keeps them from leaving.¹ The 2023 microtomography study compared capsule beds before and after extraction.¹ After extraction, porosity always decreased from the outlet at the bottom to the inlet at the top: the bed was more open at the bottom, not less.¹ Its authors note that fines migration would push the profile the other way.¹

According to the brewing chapter, the bed also changes as it wets: particles swell and rearrange under flow and pressure, porosity falls and the pressure drop rises; the finer the distribution, the more the bed consolidates.⁵ No study opened for this entry measured fines migration directly in a pour-over or espresso bed.

Sifting: what the evidence shows#

The 2023 French press study separated one grind into six sieve fractions: below 124 µm, 125–249, 250–399, 400–599, 600–799 µm and above 800 µm.¹¹ Brews used 25 g of coffee in 250 g of water at 95 °C for 5 minutes.¹¹ The fine grind straight from the grinder gave the highest extraction, 21.4 ± 0.9%; the sieved 125–249 µm fraction gave the lowest.¹¹ The authors call this hard to explain and suggest that compounds may be distributed differently between particle sizes.¹¹ Homogeneous size classes appeared less able to extract the available solutes than bimodal grinds.¹¹

The textbook chapter on grinding describes the opposite expectation for espresso: a narrower distribution avoids the mix of over-extracted fines and under-extracted boulders.³ The French press study did not test espresso or pour-over, and the chapter reports no sensory measurement.^3,11

Reference values#

Open questions#

• Fines migration. Whether fines move far enough during a brew to change flow has not been measured directly in any study opened for this entry; the one imaging study found a porosity profile opposite to what migration would produce.¹
• Sifting and flavour. No peer-reviewed sensory test of sifted grinds for pour-over or espresso was found; the one French press study found sieved fractions extracted less.¹¹
• Surface area or permeability. Two espresso studies found that grind affects the cup mainly through flow, not surface area^2,8; how far this holds for low-pressure brews is not established.
• Grinder comparisons. Published distributions for named grinder types at matched settings are scarce in peer-reviewed work; the textbook chapter on grinding shows examples but no tabulated values.³

Key studies#