— A student, moments ago. Bless. Let's fix this together. Tea steeps ground plant leaves in idle hot water. Pour-over drives heated water through a structured bed of precisely ground, carefully roasted coffee particles under the influence of gravity, pressure differentials, surface tension, and five simultaneously competing chemical extraction kinetics. They are not the same thing.
Every compound in a coffee particle dissolves into water according to its own rate equation. We model each one independently using a first-order differential equation — the same mathematics that governs radioactive decay, drug clearance from the bloodstream, and the cooling of a hot object:
Reading this aloud: the rate of change of dissolved concentration dC/dt is proportional to how much of the compound is still left to extract (C_max − C(t)). When you first pour water on fresh grounds, extraction is fast because the gradient is steep. As the cup fills with dissolved compounds, extraction slows — the driving force diminishes.
k_eff is where all the interesting physics lives. It is not a single constant — it is a product of four physical factors, each of which you control when you choose your method, grind, roast, and temperature.
Finer grinds expose exponentially more surface area. A 400 µm moka grind exposes ~3.4× the area of a 900 µm pour-over grind. This is the dominant variable for caffeine and CGA extraction speed.
Each compound has its own activation energy Ea. Diterpenes need Ea ≈ 55 kJ/mol — they barely extract below 85°C. Caffeine extracts across the full range (Ea ≈ 30 kJ/mol).
Modeled as 1 + P · 0.22. At 1.5 bar, moka gets a ~33% boost to keff. Pressure reduces boundary layer thickness and alters CO₂ partitioning in the extraction cell.
Roasting destroys compounds before brewing begins — this multiplier applies to C_max, not k_eff. CGAs lose 18–22% per roast unit. Melanoidins do the opposite — they form during roasting.
One factor sits outside the kinetics entirely. Diterpene filter gating is a physical barrier effect, not a rate constant problem: paper filters arrest cafestol and kahweol in their oil-droplet carriers regardless of kinetics. Pour-over through paper delivers ~5% of the diterpenes present in the grounds. Moka delivers close to 100%.
Coffee is not one thing. It is a mixture of at least 1,000 identified volatile and non-volatile compounds. Five compound classes dominate the pharmacological and sensory story:
| Compound | Moka yield | Pour-over yield | Roast sensitivity | Brain target |
|---|---|---|---|---|
| Caffeine | Minimal (–2%/unit) | All striatal regions (A2A/A1) | ||
| Chlorogenic acids | High (–18%/unit) | Ventral striatum (neuroprotection) | ||
| Trigonelline | Very high (–22%/unit) | Dorsomedial striatum (nAChR) | ||
| Diterpenes | Low (lipid-stable) | Dorsolateral striatum (PPAR-γ) | ||
| Melanoidins | Inverted (forms on roasting) | Gut–brain axis (prebiotic) |
* Paper filter blocks ~95% of diterpenes. Metal mesh filters restore yield.
Roasting is a series of overlapping pyrolytic and Maillard reactions from roughly 150°C to 240°C. What matters for extraction chemistry is which compounds are present in the bean when the water arrives.
Adjust the parameters below to see how each physical variable shifts the extraction curves and final compound yields. Use "Compare both" to see moka and pour-over side by side — then try pushing roast to dark and watch what happens to the diterpene lines.
Adjust the simulator as you read each insight — these are things that only become obvious when you watch the curves behave.
C_max in light roast. Stop a pour-over at 2 minutes and you get a caffeine-dominant cup with relatively few polyphenols — sharper, brighter, and less rounded. This is the under-extracted pour-over problem. It tastes bright not because pour-over is gentle, but because you interrupted the kinetics.
k_eff — the extraction rate — but it also drives acid hydrolysis of chlorogenic acids at the extraction interface. Moka reaches saturation faster, but C_max for CGAs is slightly lower than pour-over through the same beans at the same roast. You are trading polyphenol fidelity for speed and concentration.
k_eff for all compounds. But it also narrows the flow path in pour-over, lengthening contact time — which can cause over-extraction of bitter Maillard byproducts once most of the desirable compounds have already extracted. Grind size is not just about speed; it is about sequencing which compounds dominate at the end of the extraction window.
Tea steeps plant material in hot water. The sole variables are temperature and time. There is no grind geometry, no pressure differential, no Arrhenius activation energy landscape across five competing compound classes, no filter-mediated phase separation of lipophilic diterpenes, no roast degradation curve to account for before water touches the grounds.
Pour-over coffee is a precision extraction instrument operating five simultaneous first-order kinetic processes in a stratified particle bed, through a controlled gravity pressure head, at a temperature you chose for its Arrhenius consequences on specific molecular species, using a paper medium whose porosity gates which compound classes pass through.
It is not tea. Now go grind something.