02 · Basic · Intermediate
Theory — Extraction, Variables, Water, Evaluation
What dissolves, and what decides how much of it does
Written by Chanho Hong
What this chapter covers
Extraction defined through soluble compounds, water and energy; why viscosity is a read-out of instantaneous concentration; what each of the seven brewing variables actually changes; and how water hardness and alkalinity divide the cup — with measured data throughout.
If roasting creates the potential of flavour, extraction is the work of translating that potential into the cup. This chapter covers what extraction is (theory), what we use to control it (variables), what we dissolve it into (water), and how we read the result (evaluation). It is the foundation of every technical chapter that follows.
Extraction Theory — Definition · What Actually Happens · Physics
Extraction is the phenomenon in which the soluble compounds of coffee are washed out into a liquid through chemical and physical reactions between coffee, water and physical energy. The three core elements: soluble compounds (what) · water (into where) · energy (by what).
What actually happens — the sequence of extraction
Coffee compounds leave in a set order. Starting with what comes out first at lower temperatures and shorter times: acidity → sweetness → bitterness and astringency. This sequence is the basis of all extraction control.
Every compound differs in solubility, molecular size and polarity. The acids are small and dissolve readily in water, so they come out first; the large molecules that carry bitterness (chlorogenic acid degradation products, melanoidins) demand more energy and time, so they come out later. This is why how much you extract changes the balance of flavour even from the same roasted coffee.
There are four levers for washing out more soluble compounds: solubility (temperature) · physical energy (agitation) · surface area (grinding) · time. Roasting (the intrinsic solubility of the compounds themselves) acts on top of these as a constant. Raise these levers and the extraction yield rises; lower them and it falls.
Physics — Brownian motion · hydrodynamics · viscosity
Extraction is ultimately an encounter between water molecules and flavour molecules. Three physical concepts govern that encounter:
- Brownian Motion — the phenomenon in which particles in a fluid collide and move at random. In extraction, water molecules constantly collide with the flavour molecules of the coffee and cause diffusion, and this movement dissolves flavour into the water. The more movement, the more extraction (moving from high energy to low energy).
- Hydrodynamics — the way water flows through the coffee bed. The flow must be uniform for every region to extract evenly (the opposite of channelling).
- Viscosity — the "stickiness" of a fluid. In espresso extraction the early flow is slow and highly viscous (concentrated), then becomes progressively faster and thinner. You can see with your own eyes the point after the ristretto phase where the flow accelerates sharply and turns watery.
What viscosity shows — a visual indicator of instantaneous concentration
Place viscosity alongside the quantity of soluble compounds being extracted and it becomes clear why it changes as it does. Viscosity reflects almost directly the concentration of dissolved solids in the water at that moment (plus emulsified lipids and suspended fines). In other words, viscosity = the concentration at this very moment.
Early in the extraction, the water meets coffee from which nothing has yet been removed. The concentration gradient between the interior of the coffee particle and the water is at its steepest, so the dissolved solids carried per unit of water are at a maximum and the flow is dark and thick. As extraction proceeds, the soluble compounds remaining in the puck are depleted, the gradient flattens, and the same volume of water carries fewer solids. The fall in viscosity is the depletion curve of the remaining soluble compounds.
The trap to watch for is confusing viscosity with total extracted mass. The early phase, where viscosity is highest, is the point at which little has yet been extracted in total; and in the later phase, where viscosity has fallen, the total continues to accumulate. So "blonding has arrived, therefore everything has been extracted" is a misreading — what enters from that moment on is water that is dilute but still adds to the cup's total yield and to the late-phase compounds (bitter and astringent). The cut-off point is decided not by viscosity but by target yield and taste.
The same principle operates in filter brewing. The liquid coming down from the first pour is dark, and the liquid from the final drawdown is noticeably pale. Put a glass server under the brewer and you can directly observe this concentration gradient as layers of differing colour mixing together. That said, viscosity is affected not only by dissolved solids but also by emulsified lipids and suspended fines, so use it as a directional indicator of concentration rather than a precise measurement. For an accurate value, measure with a refractometer.
The grinder as a variable
Grinding is the most powerful physical lever, because it sets surface area. The method of fracture changes the particle size distribution, and the particle size distribution in turn determines extraction uniformity.
| Method | Structure | Particle size distribution | Resulting tendency |
|---|---|---|---|
| Flat burr (planar) | Two discs facing each other | Relatively uniform; fast motor, slow grinding | Cleaner and more balanced, extraction↑ |
| Conical burr (conical) | Cone + ring | Relatively broad distribution; slow motor, fast grinding | More complex and aromatic, heavier body, extraction↓ |
| Roller mill (roller type) | Passes stage by stage through 2–3 pairs of intermeshing cylindrical rollers | Narrowest distribution, low proportion of fines | Stable flow rate and drawdown, repeatability↑ — mainly large-scale plant |
Roller Mill — why it is different
A Roller Mill is a method that fractures the beans by passing them between grooved cylindrical rollers turning against each other. Normally 2–3 pairs of rollers are arranged in series and the gap is narrowed stage by stage. The first pair (pre-breaker) breaks the beans to roughly the 1 mm level, and the rear pairs reduce them to the target particle size.
In a burr grinder the particles pass through a narrowing annular channel each by a different path and are repeatedly broken again. Because the number of fracture events differs from particle to particle, the distribution broadens and fines are generated. A roller mill is a single pass, in which the particle passes once through a fixed gap, and the two rollers are given different rotational speeds (e.g. 2:1) to apply shear — closer to being cut with scissors. Because the number of fracture events and the direction of the force are controlled, the result clusters narrowly. On top of this, three parameters — corrugation geometry, speed ratio and gap — let you design the shape of the distribution itself, so deliberate forms such as a bimodal distribution for espresso are also possible.
It is the standard for mass-production plant (capsules, pre-ground product), in the range of several hundred to 1,800 kg per hour. The reasons it has not come down to café scale are physical. ① A microscopic defect on the roller surface translates directly into particle size deviation, so centrifugal-casting-grade precision is required and replacement costs are very high. ② Both ends of a roller pair must be aligned simultaneously in three dimensions, which is far more demanding than with burrs. ③ At the fine gaps required for espresso, thermal expansion shifts the gap, so water cooling becomes necessary. ④ Because of the angle at which the rollers must bite the beans, the roller diameter must be at least around 200 mm, which limits miniaturisation.
"A roller mill produces fewer fines than a burr" is a claim that is industrially established and supported by equipment-manufacturer material, but no independent study directly comparing the particle size distribution of rollers and burrs in coffee has yet been published. The commonly cited "50–75% reduction in fines" figure is a comparison of roller mill vs hammer mill, not a comparison with burrs. Furthermore, a grinder that produces fewer fines must be set finer to give the same taste, so producing fewer fines at the same setting does not in itself mean there are fewer fines in the actual puck. The received wisdom that "coffee ground on a roller mill has a thin body" has no supporting evidence at any level. [to verify]
Heat Changes
Extraction temperature is not the machine water temperature alone. The actual total heat = machine water temperature + ground coffee temperature + ambient (environmental) temperature − heat lost during packing time. Once extraction begins, the temperature briefly recovers (heat recovery) at the start of contact and then gradually cools. This temperature curve interlocks with the acid → sugar → bitter extraction sequence seen above.
Measured data — grind, temperature and time change the acid profile
Evidence that brewing conditions genuinely change the chemistry of the cup. Below are measured values for the elution of the main organic acids (mg/L) by condition.
| Acid | Fine grind | Medium grind | Coarse grind |
|---|---|---|---|
| Lactic | 109.67 | 194.50 | 308.33 |
| Acetic | 242.67 | 225.67 | 209 |
| Citric | 325 | 461 | 440 |
| Malic | 119.33 | 137.00 | 163.67 |
| Chlorogenic (CGA) | 700 | 1064.67 | 1177 |
| Quinic | 435.33 | 495 | 510 |
| Acid | 70℃ | 94℃ | 100℃ |
|---|---|---|---|
| Lactic | 121 | 194.50 | 187.33 |
| Citric | 338.33 | 461 | 332 |
| Malic | 131 | 137 | 122 |
| Chlorogenic | 872.67 | 1064.67 | 1067.67 |
It is not simply that "everything increases the hotter and the finer you go". Citric acid, for example, peaks at 94℃ and then actually decreases at 100℃. In other words, each variable acts in a different direction on each particular acid. This non-linearity is the basis of the "Give & Take between variables" dealt with later (Chapter 08, Correlation).
① A sour, hollow cup (under-extraction). If a washed Kenyan feels sharply sour and hollow, it is a sign of under-extraction that has stopped at the acid stage. Grind slightly finer (surface area↑) or raise the water temperature (solubility↑) to pull it through into the sweetness and body range.
② A bitter, cloying cup (over-extraction). Conversely, if it is bitter, cloying and dry, that is over-extraction. Grind coarser, shorten the brewing time and lower the water temperature slightly, so as to stop before the bitter and astringent range.
③ Separating causes with data. Measuring TDS and extraction yield with a refractometer lets you cross-check sensation against numbers. If the yield is lower than the value you normally get and the cup is sour, it is under-extraction, so raise the variables. Conversely, if the cup is sour even though the yield is perfectly adequate, this is not a lack of extraction but the intrinsic acidity of the coffee or channelling, so start by checking distribution and pouring uniformity. What matters is not the absolute number but comparison against your own baseline for the same coffee and the same equipment. Once you know the sequence and the physics, "why does it taste like this" narrows from guesswork to diagnosis.
Brewing Variables — what determines what
The handles that control extraction are fixed in number. Memorise what each variable determines and you can diagnose a problem in the cup and steer it back in the direction you want. Below is the "variable → what it determines" mapping used in Chanho Hong's brewing courses.
| Variable | What it determines | ↑ Raise it | ↓ Lower it |
|---|---|---|---|
| Brew Ratio | The strength and flavour balance of the beverage | Longer ratio → weaker, watery, tea-like, bitter, dry and astringent | Shorter ratio → stronger, muddier, sour, acidity↑ |
| Grinding Size | Tactile quality (weight, texture, finish) and the tone of the acidity | Coarser → lighter and silkier, short and bright acidity | Finer → heavier and creamier, long, dry, astringency↑ |
| Brewing Time | Total extracted mass (extraction yield) | Longer → yield↑ (up to the limit) | Shorter → yield↓ |
| Bloom/Pre-infusion Time | The depth of flavour | Longer → depth↑, sweetness↑, complexity↓ | Shorter → depth↓, acidity↑, sweetness↓ |
| Bloom/Pre-infusion Rate (bloom water volume) | The complexity of the beverage | More → complexity↑, acidity↑, strength↓ | Less → sweeter and rounder, complexity↓ |
| Water Temperature | The tone of the acidity | Higher → brighter, malic↑, possibly slightly sharp | Lower → softer, citric↑, possibly slightly flat |
| Environmental Temp | Heat loss and stability during extraction | Warm → heat retained, extraction stable | Cold → bed cools, late-phase extraction falls off |
Variables in depth
Brewing Ratio — strength and balance
The ratio is coffee weight : beverage weight. It is the primary handle for setting strength, and it moves the overall nuance considerably. The filter standard is roughly 1:15–1:18. Lengthen the ratio and the cup becomes watery and tea-like with late-phase bitterness and astringency showing through; shorten it and it becomes dense and muddy with acidity in front.
Grinding Size — tactile quality and the tone of acidity
Grinding acts directly on extracted mass by way of surface area. The finer the grind, the greater the surface area → extraction↑ → heavier, dry, astringency↑; the coarser, the lower the extraction → lighter, bright acidity. "Particle size increases or decreases surface area, so more or less is extracted into the cup, and we perceive that difference as taste."
Water Temperature — the tone of acidity
Temperature changes which acids stand out. High temperature draws out malic acid (apple, green grape) brightly, but taken too far it turns sharp; low temperature moves towards citric acid (lemon, lime) and is softer, but taken too far it turns flat. (As summarised in the course material: lower temperature → phosphoric, lactic, tartaric, citric; higher temperature → malic, acetic.)
Blooming Time vs Rate — depth vs complexity
The two are easily confused. Bloom time sets the depth of flavour — the longer it is, the deeper and sweeter, but the less complex; the shorter it is, the more the acidity lives and the less the sweetness. Bloom water volume/rate sets complexity — pour a lot of water (quickly) and it becomes more complex with acidity↑ and strength↓; pour less and it becomes sweeter and rounder but complexity↓.
The standard course sequence: ① Brew Ratio (overall nuance) → ② Grind Size (detail) → ③ Water Temp → ④ Bloom Time → ⑤ Bloom Rate. Take hold of the big handles first and finish with the fine ones. Change only one thing at a time, or you cannot read cause and effect.
① Making it "sweet and round". Hold the ratio and reduce the bloom water volume (complexity↓, sweetness↑) while slightly lengthening the bloom time (depth↑, sweetness↑). Lowering the water temperature by 1–2℃ softens the tone. All three variables are then aligned in the direction of "sweet and round".
② Making it "brighter and more complex". Conversely, if you want to go bright, increase the bloom water volume (complexity↑, acidity↑), raise the water temperature by 2–3℃ (malic tone↑) and grind slightly coarser. You concede a little sweetness, but the layers of aroma come alive.
③ One at a time (isolating variables). When the result is ambiguous, changing several variables at once makes the cause unreadable. Brew with only the grind moved one click, then change only the temperature, and this kind of isolated experiment builds your own data on what each variable actually does with this coffee. Keeping to the order of adjustment (ratio → grind → temperature → bloom) is the key.
Brewing Water — 98% of the coffee
About 98% of a cup is water. Water is not a simple solvent but an active ingredient that determines which flavours are drawn out and how much of them. With a different mineral composition, the same roasted coffee and the same recipe become an entirely different cup.
TDS (Total Dissolved Solids) is the total quantity of ions and solids dissolved in the water. Of these, what is decisive for extraction is the balance between the cations Ca²⁺ and Mg²⁺ (total hardness) and the anion HCO₃⁻ (alkalinity, buffer).
| Variable | Recommended range | Role |
|---|---|---|
| TDS (total dissolved solids) | 150 mg/L (75–250 acceptable) | Total mineral content |
| Calcium Hardness | 50–175 mg/L as CaCO₃ | The core of extracting power for flavour |
| Total Alkalinity | 40 mg/L as CaCO₃ (≤50) | Buffers acidity (pH stability) |
| pH | 6.5–7.5 | Neutral range (prevents corrosion and abnormal extraction) |
| Sodium | ≈10 mg/L | Salty taste in excess |
| Chlorine | 0 mg/L | Residual chlorine is an off-flavour |
Total Hardness · Temporary Hardness
The concentration of Ca²⁺ + Mg²⁺ dissolved in the water. These cations are strongly nucleophilic and bind to and transport flavour molecules, thereby strengthening extraction. Too little (distilled water) and extraction is impoverished and monotonous; too much and you get over-extraction and scale.
The concentration of HCO₃⁻ (bicarbonate). When hydrogen ions increase from the organic acids the coffee releases during extraction, HCO₃⁻ as a buffer neutralises them and returns the pH. Too high and the acidity is dulled (flat); too low and the acidity is excessive, with a risk of machine corrosion.
Magnesium · Calcium — why separate them at all
Ca²⁺ and Mg²⁺ are both divalent cations and both assist the extraction of flavour substances. Theoretically, Hendon et al. (2014, J. Agric. Food Chem.) modelled these cations as binding to the flavour compounds of coffee and thereby promoting extraction. In practice and across the industry, Mg is empirically associated with brighter, more complex acidity and Ca with rounder sweetness and body, and the Mg/Ca ratio of "custom water" is adjusted according to this tendency.
Some textbooks and courses (including older material) state that "Mg binds selectively to malic acid and Ca to citric acid", but the current literature does not support this specific binding. Subsequent experimental work (Nordkvist 2020 and others) suggests that, rather than selectively extracting particular acids, the cations may precipitate some acids (lactic, malic) or affect perception; and the claim that "Mg produces a different flavour" also has no rigorous experimental basis established as yet (Coffee ad Astra, 2018). The directional character of Mg and Ca is therefore most accurately understood as an empirical tendency.
"Brewing with purified (distilled) water gives the cleanest result" — false. With almost no Ca or Mg, flavour extraction is impoverished and the result is a monotonous, weak cup (a genuine misconception repeatedly pointed out in home-brewing and SCA education). What matters in water is not "cleanliness" but appropriate mineral content.
Alkalinity / Buffer
Alkalinity is the hidden variable that governs the perception of acidity in the cup. With the same roasted coffee, in water of high alkalinity the organic acids of the coffee are neutralised and the acidity is suppressed (rounded); in water of low alkalinity the acidity survives intact and springs forward. The reason speciality brewing generally prefers low alkalinity (≤40 mg/L as CaCO₃) is to preserve bright acidity. However, if it is excessively low, pH buffering capacity is weak and variation between cups grows, and water skewed towards acidity carries a risk of corroding the espresso machine.
① Same coffee, different café — a difference in alkalinity. If a washed Ethiopian brewed to the same recipe at Café A (hard, high-alkalinity tap water) and Café B (low-alkalinity filtered water) is "flat" at A and "vibrant" at B, the problem is not the coffee or the recipe but the alkalinity of the water. Switching A to remineralised RO water or a low-alkalinity filter brings the acidity back.
② Designing direction with custom water (competition, omakase). If you want to go "brighter" with the same Geisha, set up water with a higher proportion of Mg (e.g. Mg-led remineralisation); for "rounder and with more body", water with a higher proportion of Ca. But since this is an empirical tendency, always verify blind to confirm the difference is real.
③ Protecting the espresso machine vs taste. Lower the alkalinity too far and the cup is bright but the risk of machine corrosion and instability grows; raise it too far and scale builds up. One approach is a two-track operation: set shop espresso to the middle of the SCA recommended range for total hardness and alkalinity to compromise between taste and equipment life, and use a separate low-alkalinity custom water for the pour-over station alone.
Coffee Evaluation — how to read a roast
Evaluation is not a matter of preference but of gathering information. Read colour, aroma, information and sensory systematically with a cup (or a roast) in front of you and you can decide what to adjust and how. Here we learn in particular "how to read a roast through the flavour spectrum".
Flavour Spectrum — the axis on which a roast is read
The degree of roast moves the flavour of the cup along a single axis. One end (light roast) is acid, floral, fruit; the other end (dark roast) is bitterness, roasty, dark chocolate. Temperature (which molecules are driven off) and time (how long they are driven off for) set the position on this axis.
The four lenses of evaluation — colour · aroma · information · sensory
The colour of the roasted coffee and of the grounds is the primary indicator of degree of roast. Quantify it with an Agtron or colourimeter. Also look at the surface–core colour difference — if the core is lighter than the surface, that can be a sign of internal under-development. (Roasting approaches such as STLT/HTLT create this colour difference → Chapter 07)
Smell fragrance (dry grounds) and aroma (wet) separately. Read in the order intensity of aroma → complexity and clarity → descriptors. Aroma reflects roast, freshness and extraction all at once.
Contextual information such as origin, variety, processing, altitude, roast date and profile. It becomes the frame for interpreting the sensory data. Information explains "why does it taste like this".
The actual gustatory and tactile evaluation — acidity, sweetness, body, aftertaste, balance. The next chapter (Chapter 03, Sensory) refines this with cupping, brewing and espresso protocols.
Read in the order colour (roast hypothesis) → aroma (freshness, direction) → information (context) → sensory (verification), then adjust the roast or the extraction on that basis and evaluate again. This loop is the backbone of quality control (Chapter 10).
① Evaluating a newly arrived lot. Gauge the degree of roast from the colour of the beans (colour), check freshness and defects with the dry fragrance (aroma), confirm origin, processing and roast date (information), then verify by cupping (sensory). Pass it through the four lenses in order to decide "use it or not".
② Separating a problem into roasting vs extraction. If it is "core lighter than surface, with green and sour notes", insufficient development (roasting) is the likely cause, so do not struggle to cover it with extraction — work on the profile. Conversely, if colour, aroma and information are all normal and only the cup is bland, suspect the extraction and the water.
③ Shop QC routine. A quick spot-check of the day's coffee at opening, in the order colour (roast consistency) → aroma (freshness) → taste (reference shot), catches anomalies caused by a lot change or a shift in humidity before they go out to the customer. Evaluation is not preference but measurement in the service of reproducibility.
Questions this chapter answers
- How do I tell under-extraction from over-extraction?
- Soluble compounds broadly emerge in the order acid → sugar → bitterness and astringency. Stopping at the acid stage gives a sharp, sour, hollow cup (under-extraction); running all the way through gives a bitter, drying, astringent one (over-extraction). Where the balanced window sits differs by coffee, roast and brewer, so it is more accurate not to memorise a fixed yield figure.
- When the espresso stream goes pale, is extraction finished?
- No. Viscosity only reflects the concentration of dissolved solids at that instant. While viscosity falls, cumulative yield keeps climbing. What arrives after blonding is dilute but still adds to total yield and to the later compounds — bitterness and astringency. The cut-off should therefore be set by target yield and taste, not by viscosity.
- How should I choose water for coffee?
- Total hardness governs extracting power; alkalinity governs the buffering that neutralises acids. Too little hardness reads as flat, too much reads as muddy, and high alkalinity presses acidity down until the cup goes flat. If you want acidity to show, keeping alkalinity low is the key move.
Figures and regulations are verified against a source hierarchy: Tier 1 in-house knowledge base → Tier 2 SCA·WBC and peer-reviewed literature → Tier 3 web. Anything not adequately supported is marked [to verify] in the text.