Maillard Reaction: What Browning Actually Is

Browning Is a Condensation, and It Starts With Almost Nothing

Browning is not burning. Nothing combusts and nothing is oxidised to CO2; the Maillard reaction is a condensation — an amine joins a sugar and one water is expelled. That single water is the opening move; everything downstream is the wreckage of it.

It is slow to start because both partners are nearly unavailable. A reducing sugar reacts only through its open-chain form, and in water sugars sit overwhelmingly as cyclic hemiacetals: only ~0.002% of D-glucose in water is the reactive open-chain aldehyde (fructose ~0.7%). That is roughly 350 times more free carbonyl from fructose, which is why honey and invert syrup brown faster than glucose — and why sucrose, not a reducing sugar at all, must hydrolyse first.

The amine is throttled the same way. The nucleophile is a lysine side chain's epsilon-amino group, and it attacks only as the neutral -NH2; the protonated ammonium has no lone pair to give. With a lysine epsilon-ammonium pKa ~10.5 — only the neutral -NH2 attacks, so higher pH browns faster — the free-base fraction follows f = 1/(1 + 10^(pKa − pH)). At pH 6.0 that is 1/(1 + 10⁴·⁵) = 3.2 × 10⁻⁵, about 0.003% of the lysine; at pH 8.0 it is 1/(1 + 10²·⁵) = 3.2 × 10⁻³, about 0.32% — a clean hundredfold for two pH units. Multiply the two scarcities: at pH 6 only about 6 in 10 billion glucose–lysine encounters involve a reactive pair. Heat is what makes an improbable collision happen often enough to see. Alkali is the shortcut: baking soda at pH ~8 darkens onions in minutes, pretzels get a lye dip, Dutch-process cocoa is alkalised almost black.

Carbinolamine, Schiff Base, Amadori Product

Step one is ordinary carbonyl chemistry: the neutral amine lone pair attacks C1 of the open-chain aldehyde, and the tetrahedral adduct is a carbinolamine (hemiaminal), nitrogen and OH on the same carbon.

Step two is the condensation proper: acid-catalysed loss of that OH as water gives an iminium, which deprotonates to the Schiff base. One H2O leaves per Schiff base formed. The sugar Schiff base ring-closes, nitrogen now at the anomeric position — an N-glycosylamine. Everything to this point is reversible: dilute or cool it and it comes apart again.

Step three is the point of no return: the acid-catalysed Amadori rearrangement, described by Mario Amadori in 1925. The ring opens to the protonated Schiff base, a proton leaves C2 to give a 1,2-enaminol, and that tautomerises to a stable ketone. The product is a 1-amino-1-deoxy-2-ketose — from glucose and a lysine residue, epsilon-fructosyl-lysine. The sugar has isomerised from aldose to ketose with the amine welded to C1, and it will not dissociate back.

Ketoses take the mirror-image route: fructose plus an amine gives a Schiff base that rearranges the other way — the Heyns rearrangement — to a 2-amino-2-deoxyaldose. Same destination, a stable amino-sugar primed to fall apart.

The Amadori Branch Point and the Alpha-Dicarbonyls

The Amadori compound is where one clean mechanism ends and a combinatorial cascade begins: it can enolise two ways, and pH picks the winner.

1,2-enolisation dominates in acid. The amine is eliminated and successive dehydrations give 3-deoxyglucosone and then HMF (5-hydroxymethylfurfural) from hexoses, or furfural from pentoses. Plenty of colour, little aroma, and the nitrogen is handed back rather than built in.

2,3-enolisation takes over near neutral and above. It yields 1-deoxyglucosone, which fragments by retro-aldol cleavage into a shower of small alpha-dicarbonyls: glyoxal, methylglyoxal (2-oxopropanal), diacetyl (2,3-butanedione, the buttery one) and 2,3-pentanedione, alongside hydroxyketones such as 1-hydroxy-2-propanone (acetol), which carries only one carbonyl and so is a fragment rather than a dicarbonyl. These two- to five-carbon fragments are the real engine of flavour: far more electrophilic than the sugar they came from, with the second carbonyl sitting exactly where it must to do what the sugar never could.

Strecker Degradation, Pyrazines and Melanoidins

An alpha-dicarbonyl meeting a free amino acid runs the Strecker degradation. The two condense to a Schiff base; the adjacent second carbonyl stabilises the charge left when the carboxyl departs, so the amino acid decarboxylates, releasing CO2. Hydrolysis of the resulting ylide returns an aldehyde one carbon shorter than the amino acid, plus an alpha-aminoketone.

That shortened aldehyde is the smell of cooking, and it is amino-acid-specific. Leucine gives malty 3-methylbutanal. Methionine gives methional, the boiled-potato note of over-processed milk. Phenylalanine gives honeyed phenylacetaldehyde. Proline — a secondary amine that cannot run a textbook Strecker — decarboxylates to 1-pyrroline, which captures methylglyoxal to give 2-acetyl-1-pyrroline, the bread-crust, basmati and popcorn molecule, perceptible at ~0.02–0.1 parts per billion in water.

The leftover alpha-aminoketone is not waste: two condense to a dihydropyrazine, which oxidises to a pyrazine — the roasted, nutty, coffee-and-cocoa class. Meanwhile the residue keeps condensing: dicarbonyls cross-link amines and the mixture polymerises into melanoidins, the brown, nitrogen-containing, high-molecular-weight pigment you actually see. No trace component — melanoidins account for up to ~25% of roasted-coffee dry mass — and they carry real antioxidant and metal-chelating activity.

Why 140 °C, Why Dryness: The Governing Numbers

Browning follows Arrhenius kinetics with an activation energy around 100–150 kJ/mol, high because the rate-limiting steps are enolisations and dehydrations, not diffusion. Take Ea ≈ 120 kJ/mol and compare a dry surface at 140 °C (413 K) with boiling water at 100 °C (373 K):

ln(k₁₄₀/k₁₀₀) = (Ea/R)(1/373 − 1/413) = (120 000 / 8.314) × (2.681 − 2.421) × 10⁻³ = 14 433 × 2.60 × 10⁻⁴ ≈ 3.75, so k₁₄₀/k₁₀₀ ≈ e³·⁷⁵ ≈ 42.

Forty times faster for forty degrees — and another 20 °C, 140 to 160 °C, multiplies the rate by about five again. This is why browning runs fast above ~140 °C; a wet surface pinned at 100 °C cannot brown. The threshold is physics, not chemistry: while liquid water boils off a surface, evaporative cooling clamps it near 100 °C however hot the pan is, and there the reaction is forty times too slow to finish before dinner. Pat the steak dry, or nothing happens until the surface dries itself. The penalty is paid in time, not forbidden outright — witness dulce de leche.

Water matters twice. It is a product of the condensation, so excess pushes the equilibrium back; but reactants must still diffuse and meet, which needs some. Hence a water-activity optimum ~0.6-0.8 for maximum browning rate — intermediate-moisture territory: bread crust, milk powder, jerky, fudge. A loaf shows both at once: in a 230 °C deck oven the crust dries, climbs to 150–180 °C and goes mahogany, while crumb three millimetres inside stays water-saturated at 100 °C and white.

How Browning Is Measured, Specified and Suppressed

Colour is measured instrumentally: CIE L*a*b* on a surface, or absorbance at 420 nm for melanoidins, with 294 nm tracking the colourless intermediates before them. Coffee roasters grade roast by near-infrared reflectance on the Agtron scale, whose SCA tiles run #95 (very light) to #25 (very dark); bean temperature is ~196 °C at first crack, ~224 °C at second.

Chemical markers beat colour because they appear first. HMF is honey's legal freshness index: EU Council Directive 2001/110/EC caps it at 40 mg/kg (80 mg/kg for declared tropical origin), measured by HPLC with UV detection near 284 nm — fresh honey sits under ~10 mg/kg and climbs with heating or storage. Furosine is the dairy equivalent: acid hydrolysis of epsilon-fructosyl-lysine liberates it in fixed proportion, so ion-pair HPLC reads out accumulated heat load. Raw milk is under ~5 mg per 100 g protein, UHT runs into the tens or low hundreds, in-container sterilised milk past ~250.

That gap is engineered: spore inactivation has z ≈ 10 °C, Maillard damage z ≈ 25–45 °C, so hotter-and-shorter wins. UHT at 135–150 °C for 2–5 s sterilises far more than it browns; 115–120 °C for twenty minutes does the reverse.

Acrylamide is the Maillard product regulators police. Asparagine plus a reducing sugar, via the Schiff base and a decarboxylated Amadori intermediate, gives acrylamide above ~120 °C; the EU benchmark for potato crisps is 750 micrograms per kilogram (500 for ready-to-eat fries, 400 for roast coffee) under Regulation (EU) 2017/2158, quantified by LC-MS/MS. The same regulation mandates mitigation: store potatoes above ~6–8 °C so invertase does not sweeten them, blanch to leach surface sugars, fry to golden not brown, and dose with asparaginase, which converts asparagine to aspartate and cuts acrylamide by 50–90%.

History, Look-Alikes and the Ways It Goes Wrong

Louis-Camille Maillard published the reaction in 1912 (C. R. Acad. Sci. 154, 66), hunting protein synthesis and finding brown pigment instead; he never saw the mechanism. John Hodge did, in a 1953 review (J. Agric. Food Chem. 1, 928) sorting the chaos into initial, intermediate and final stages — still the map everyone draws.

It is confused with three things. Caramelization of sucrose needs ~160 °C and involves no nitrogen at all — sweet and buttery, but no pyrazines, no Strecker aldehydes, no nitrogen in the pigment. Enzymatic browning is polyphenol oxidase turning phenols into quinones at room temperature on a cut apple, stopped by lemon juice or a blanch, and chemically unrelated. And Strecker names two reactions: Adolph Strecker's 1850 amino-acid synthesis from an aldehyde, ammonia and cyanide, and the 1862 degradation he found when alanine met alloxan. Only the second belongs to browning.

The most durable myth is not chemistry at all: Justus von Liebig argued in 1847 that searing seals in juices. It does not — seared meat loses at least as much water as unseared, and searing earns its place by browning alone.

The reaction also fails. It destroys lysine, the limiting essential amino acid in cereals, because a blocked epsilon-amino group is nutritionally unavailable — hence furosine limits for infant formula and milk powder. It runs at 37 °C in blood, where glucose condenses with the N-terminal valine of hemoglobin's beta chain and the Amadori adduct is HbA1c, diagnostic at 6.5% (48 mmol/mol) and reporting the previous 8–12 weeks. Oxidation of Amadori products and dicarbonyls then yields advanced glycation end-products such as carboxymethyl-lysine. In a tablet press it is a formulation bug: lactose monohydrate plus an amine drug such as amlodipine besylate browns on the shelf, which is why preformulation screens lactose against every primary and secondary amine.

Four brownings that look alike on a plate and share almost no chemistry

Browning process	What actually reacts	Conditions	Telltale sign
Maillard reaction	Open-chain reducing sugar + a free amine, usually a lysine epsilon-NH2	Fast above ~140 °C; water activity ~0.6-0.8; accelerates as pH rises	Nitrogen in the aroma: roasty pyrazines, malty 3-methylbutanal, bread-crust 2-acetyl-1-pyrroline
Caramelization	Sugar alone — no nitrogen at all	Caramelization of sucrose needs ~160 °C (fructose from ~110 °C)	Sweet, buttery, faintly bitter; works on pure sugar syrup; no roast or meaty note
Enzymatic browning	Phenols + O2, catalysed by polyphenol oxidase (a copper enzyme)	Room temperature, seconds to minutes; killed by heat, acid or excluding O2	Cut apple, banana or avocado darkening in the fridge; lemon juice stops it dead
Glycation in vivo	Blood glucose + the N-terminal valine of hemoglobin's beta chain	37 °C, pH 7.4, weeks to months	HbA1c — diagnostic of diabetes at 6.5% (48 mmol/mol)
Pyrolysis / charring	Everything — sugars, protein, fat and cellulose decomposing	Above roughly 200-300 °C	Black, brittle, bitter carbon; visible smoke; the food loses mass