Dairy Science & Technology · Postgraduate & Research · also Advances in Dairy Chemistry, Physical Chemistry of Milk (PG)
Graduate-level protein, lipid and lactose chemistry, reaction kinetics and the physical chemistry of processing, as examined in ICAR AICE subject 34.
In the exams: ICAR AICE subject 34 (all units); ASRB NET Dairy Chemistry
Advanced dairy chemistry explains, at the molecular level, what happens to milk when it is heated, pressurised, acidified, concentrated, dried, stored and digested, and how its components can be measured, authenticated and exploited. It builds on the undergraduate subjects on milk composition, proteins, lipids, lactose and salts, and takes them to the depth expected in doctoral research, in industrial research and development, and in the specialist examinations that select research fellows and scientists.
The manual covers ten topics. It begins with chemical and enzyme kinetics in milk systems, the quantitative language of the subject. It then treats protein–protein interactions under heat and pressure, especially the reactions of β-lactoglobulin with κ-casein; the structure and function of casein micelles under processing; the chemistry of lipid oxidation and its flavour consequences; the pathways of the Maillard reaction and the formation of advanced glycation end products; and the modelling of mineral equilibria between the serum and colloidal phases. It continues with water activity, glass transition and state diagrams, which govern the stability of powders and frozen products; with proteomics, lipidomics and metabolomics of milk; with the testing of authenticity and geographical origin; and with bioactive components and their analysis.
Doctoral fellowship and research-service examinations in dairy chemistry draw on all of these topics, expecting candidates to explain mechanisms, interpret kinetic and analytical data and choose methods. The manual gives the established mechanisms and quantitative relationships, flags where models are still debated, such as the internal structure of the casein micelle, and points to the reference texts in which each topic is treated at greater length.
The rate of a chemical change in milk, such as protein denaturation, browning, vitamin loss or enzyme inactivation, is described by a rate equation, −dC/dt = kCⁿ, where C is the concentration of the reacting species, k the rate constant and n the reaction order. Integrated forms give zero-order (linear loss), first-order (exponential loss) and second- or higher-order behaviour, and the order is found by testing which transformed plot of the data is linear, or by non-linear fitting of the general equation. Many dairy reactions are first order, but whey protein denaturation is an important exception: the classical kinetic studies of the 1980s found an apparent order of about 1.5 for β-lactoglobulin and close to 1 for α-lactalbumin, and later work has shown that apparent orders depend on concentration, pH and the method of measurement.
The temperature dependence of k is described by the Arrhenius equation, k = A exp(−Eₐ/RT), or, in thermal processing, by the z value, the temperature increase that raises k tenfold. Over the narrow temperature ranges of processing the two are related approximately by z ≈ 2.303RT²/Eₐ. Reactions with high activation energies are very temperature-sensitive and have small z values. Microbial inactivation and many enzyme inactivations have z values of about 5 to 12 K, while chemical changes such as browning and loss of vitamins or lysine have z values of about 20 to 40 K. This difference is the basis of the high-temperature short-time principle: raising temperature and shortening time achieves the same microbial lethality with far less chemical damage, which is why UHT milk has better flavour and nutrient retention than in-container sterilised milk.
Denaturation of β-lactoglobulin follows a two-stage process: reversible unfolding of the native protein, exposing its free thiol group (Cys121) and hydrophobic regions, followed by irreversible aggregation through thiol–disulfide exchange and hydrophobic association. Arrhenius plots of the denaturation rate constant show a marked change of slope at about 85 to 90 °C. Below this, the apparent activation energy is very high, of the order of a few hundred kilojoules per mole, because the rate is limited by unfolding, which has a large positive enthalpy and entropy; above it, the apparent activation energy is much lower, because aggregation becomes rate-limiting. α-Lactalbumin denatures at lower temperatures in calorimetry but, being a calcium-binding protein without a free thiol, it refolds readily on cooling, so its irreversible denaturation in milk is slower and depends on reaction with other denatured proteins.
These kinetics are used to calculate the extent of denaturation for any time–temperature profile, by integrating the rate equation over the profile, as in the calculation of lethality. They predict, for example, that typical pasteurisation denatures only a small percentage of whey proteins, high-temperature heat treatment for yoghurt milk (such as 90 to 95 °C for several minutes) denatures most of the β-lactoglobulin, and direct UHT treatment denatures less than indirect UHT for the same sterilising effect because the heating and cooling phases are shorter. Undenatured whey protein nitrogen in milk powders, used to classify powders as low-, medium- or high-heat, is a direct application.
Enzymes in milk come from the cow (indigenous enzymes such as plasmin, lipoprotein lipase, alkaline phosphatase, lactoperoxidase and xanthine oxidoreductase), from micro-organisms and from deliberate addition (coagulants, lactases, transglutaminase, proteases and lipases for flavour). Their activity follows Michaelis–Menten kinetics under simple conditions, but in milk the substrate is often a colloidal particle, the enzyme may be bound to a particle or membrane, and inhibitors and activators are present, so apparent parameters depend strongly on the system.
The rennet coagulation of milk illustrates enzyme kinetics in a colloidal system. In the primary, enzymatic phase, chymosin specifically hydrolyses the Phe105–Met106 bond of κ-casein, releasing the hydrophilic caseinomacropeptide into the whey and removing the steric stabilising layer of the micelles. In the secondary, non-enzymatic phase, destabilised micelles aggregate, with calcium, once roughly 80 to 90% of κ-casein has been hydrolysed. The rennet coagulation time is inversely proportional to enzyme concentration over a wide range, a relationship known as the Storch–Segelcke rule, and it falls with temperature up to about 40 °C, with added calcium and with lower pH.
Plasmin, the main indigenous protease, illustrates the importance of activation and inhibition. Milk contains mostly the zymogen plasminogen, plus plasminogen activators and inhibitors of both plasmin and the activators. Heating inactivates the inhibitors more readily than plasminogen and its activators, so moderately heated milks can show increased plasmin activity during storage, and plasmin is sufficiently heat-stable to contribute, with bacterial proteases, to bitterness and age gelation of UHT milk.
At the natural pH of milk, β-lactoglobulin exists mainly as a non-covalent dimer. On heating above about 60 to 70 °C, the dimer dissociates and the monomer unfolds, exposing the free thiol of Cys121, buried in the native structure, and hydrophobic surfaces. The reactive thiol initiates thiol–disulfide exchange reactions with other proteins containing disulfide bonds or thiols: other β-lactoglobulin molecules, α-lactalbumin, bovine serum albumin, and the caseins that contain cysteine, κ-casein and αs2-casein. The products are covalently linked complexes held together by intermolecular disulfide bonds, with hydrophobic and other non-covalent interactions contributing.
Where the complexes end up depends strongly on the pH of heating. When milk is heated at a pH below about 6.7, most of the denatured whey protein associates with the casein micelles, binding to κ-casein at their surface; at a pH above about 6.9, more of the complexes remain in the serum as soluble aggregates of whey protein and κ-casein dissociated from the micelles. Calcium concentration, the ratio of whey protein to casein, and the heating rate also influence the partition. These details determine the properties of heated milk: whey protein attached to micelles increases the size of the micelles and changes their surface properties.
Several processing consequences follow. Heat-induced complexes impair rennet coagulation, because the attached whey proteins hinder chymosin's access to κ-casein and the aggregation of micelles, giving weak curds, a problem for cheese made from high-heat milk. In yoghurt, by contrast, denatured whey protein associated with micelles participates in the acid gel, increasing firmness and water-holding capacity, which is why yoghurt milk is heated intensively. Interactions on heat-exchanger surfaces drive protein fouling.
Milk is remarkably heat-stable compared with most protein solutions: it can usually be heated at 140 °C for many minutes before it coagulates, which is what makes UHT and in-container sterilisation possible. Heat stability is measured as the heat coagulation time, the time to visible coagulation at a set temperature, often 140 °C, plotted against pH. Most bulk cow milks show a type A profile, with a maximum at about pH 6.7, a minimum near pH 6.9 and increasing stability at higher pH; some individual milks show a type B profile, with stability rising steadily with pH and no minimum.
The mechanism involves several heat-induced changes acting together: a fall in pH during heating, caused by precipitation of calcium phosphate, which releases hydrogen ions, and by degradation of lactose to organic acids; dephosphorylation and hydrolysis of caseins; Maillard reactions; and, most importantly near the stability minimum, dissociation of κ-casein-rich complexes from the micelles, leaving κ-casein-depleted micelles that aggregate more readily. Concentration greatly reduces heat stability, which is why the stability of milk concentrates must be managed by forewarming before evaporation and by adding stabilising salts such as phosphates and citrates, which bind calcium and adjust pH. Raising the concentration of ionic calcium lowers heat stability; urea, present naturally and varying with the cow's diet, raises it.
Seasonal and individual variation in heat stability, linked to milk composition, mineral balance and urea, affects UHT and evaporated milk production, and plants test incoming milk for heat stability, often with the alcohol test as a rapid indicator.
High hydrostatic pressure, typically 100 to 600 MPa applied for minutes, is used to preserve foods with less heat damage, and it also changes milk proteins in ways different from heat. Pressure favours states of smaller total volume. It disrupts hydrophobic and electrostatic interactions, which involve volume changes, while leaving covalent bonds largely intact, so it denatures some proteins without breaking peptide bonds and without forming the Maillard products of heating.
β-Lactoglobulin is relatively pressure-sensitive, denaturing progressively above about 100 to 200 MPa and then forming disulfide-linked aggregates, partly with κ-casein, much as on heating. α-Lactalbumin, stabilised by its calcium-binding site and four disulfide bonds, is much more pressure-resistant, denaturing significantly only at about 400 MPa or more. Most indigenous enzymes are pressure-resistant at commercial pressures; alkaline phosphatase, in particular, survives treatments that inactivate many bacteria, so it cannot be used as an index of pressure treatment.
Casein micelles are disrupted by pressure above about 200 to 300 MPa. Pressure increases the solubility of colloidal calcium phosphate and weakens hydrophobic interactions, so micelles dissociate into smaller particles; skim milk becomes less turbid and more translucent, and its viscosity and the rennet coagulation and acid gelation properties change. On release of pressure some reassociation occurs, with partial recovery of size depending on time, temperature and pressure. These effects are exploited to modify cheese milk and yoghurt texture, and they must be understood when high-pressure treatment is used to extend the shelf life of dairy products.
Casein micelles are roughly spherical colloidal particles, about 50 to 500 nm in diameter with an average around 150 to 200 nm, present at about 10¹⁴ to 10¹⁵ per millilitre of milk. On a dry basis they consist of about 94% protein, the four caseins αs1, αs2, β and κ in approximate proportions of 4:1:4:1, and about 6% inorganic material, mainly calcium phosphate with some magnesium and citrate, called colloidal or micellar calcium phosphate. They are highly hydrated, holding roughly 2 to 4 g of water per gram of protein, and have a voluminosity of about 4 mL per gram.
The internal structure of the micelle has been debated for decades. The submicelle model, developed by Schmidt and others, proposed that the micelle is built of smaller protein subunits of 10 to 15 nm linked by calcium phosphate, with κ-casein-rich submicelles at the surface. The nanocluster model, developed by Holt and by de Kruif and others, proposes that calcium phosphate exists as many nanoclusters, about 2 to 5 nm across, each sequestered by the phosphoserine clusters of αs1-, αs2- and β-caseins, with the proteins forming a continuous network between them. The dual-binding model of Horne emphasises two kinds of linkage: hydrophobic interactions between protein segments and bridging through calcium phosphate nanoclusters, with κ-casein acting as a chain terminator because it has only one hydrophobic region and no phosphoserine cluster. Current evidence from small-angle scattering and electron microscopy favours a relatively homogeneous network with nanoclusters over discrete submicelles, although the details remain under study.
All models agree that κ-casein is concentrated at the surface, where its hydrophilic, negatively charged C-terminal segment, the caseinomacropeptide, projects into the serum as a hairy layer about 5 to 10 nm thick, stabilising the micelles sterically and electrostatically against aggregation.
Casein micelles are dynamic structures that respond to their environment. On cooling to refrigeration temperature, hydrophobic interactions weaken and part of the β-casein, together with some calcium phosphate, dissociates into the serum, reaching perhaps 20 to 30% of the β-casein at about 4 °C; this is reversed largely on warming. Cold-stored milk therefore has more soluble casein and slightly different rennet coagulation properties, and soluble β-casein is more accessible to proteases.
Acidification dissolves colloidal calcium phosphate progressively. As pH falls from about 6.7 toward 5.2, micellar calcium and phosphate pass into the serum, and below about pH 5.2 almost all the colloidal calcium phosphate has dissolved. The micelles do not fall apart, because hydrophobic and electrostatic interactions still hold the caseins together, but their structure loosens. Near the isoelectric point of casein, about pH 4.6, the net charge approaches zero, the hairy layer collapses, and the caseins aggregate into the acid gel of yoghurt and fresh cheeses, or precipitate as acid casein in industrial manufacture. The temperature and rate of acidification determine gel structure.
Changing calcium changes stability. Adding calcium chloride, as in cheesemaking, increases ionic calcium, lowers pH slightly and promotes rennet coagulation; removing calcium by chelators such as citrate, phosphate or EDTA, or by ion exchange, dissociates micelles, which is the basis of emulsifying salts in processed cheese and of the manufacture of some milk protein ingredients with modified functionality.
Many processes act on casein micelles. Rennet removes the hairy layer by cleaving κ-casein, and the destabilised micelles aggregate through calcium bridges and hydrophobic contacts, forming the curd. Heating attaches denatured whey proteins to micelles, as described earlier, and increases micelle size. Homogenisation leaves micelles largely intact but makes them adsorb at the new fat–water interface, where micellar casein forms much of the new fat globule membrane, so homogenised fat globules behave in part like large casein particles in acid and rennet gels.
Concentration by evaporation and membrane filtration brings micelles closer together. In evaporation, the serum becomes more concentrated in calcium and phosphate, more calcium phosphate associates with the micelles and pH falls, contributing to the lower heat stability of concentrates. Microfiltration with membranes of about 0.1 µm pore size separates micelles from whey proteins, producing micellar casein concentrates and native whey (serum protein) streams, ingredients widely used in cheese, beverages and nutrition products. Ultrafiltration concentrates both casein and whey proteins while removing water, lactose and soluble salts.
Drying, rehydration and storage also affect micelles. Micellar casein concentrate powders can dissolve slowly because the micelles at particle surfaces aggregate during drying and storage, forming a skin that resists water; adding monovalent salts or adjusting processing improves solubility. Ethanol stability, the concentration of ethanol that causes milk to coagulate, depends on micellar stability and ionic calcium, and is used as a rapid test of raw milk suitability for heat treatment. Understanding micelle structure, therefore, underlies the control of almost every dairy process.
Lipid oxidation is the reaction of unsaturated fatty acids with oxygen, producing compounds that cause off-flavours at extremely low concentrations and can reduce nutritional value. Autoxidation, the reaction with ordinary (triplet) oxygen, proceeds by a free-radical chain mechanism in three stages. In initiation, a hydrogen atom is abstracted from a fatty acid, usually from a carbon adjacent to a double bond and most easily from a bis-allylic carbon between two double bonds, forming an alkyl radical R•. Initiation is catalysed by transition metals such as copper and iron, by light, by heat and by pre-existing hydroperoxides. In propagation, R• reacts very rapidly with oxygen to form a peroxyl radical ROO•, which abstracts hydrogen from another fatty acid, forming a hydroperoxide ROOH and a new R•; this hydrogen abstraction is the slow, rate-determining step. In termination, radicals combine to form non-radical products.
Rates increase sharply with the degree of unsaturation: linoleic acid with one bis-allylic position oxidises roughly ten to forty times faster than oleic acid, and linolenic acid about twice as fast as linoleic. Milk fat is relatively stable because it is low in polyunsaturated fatty acids, but its phospholipids in the fat globule membrane are richer in them and are the main site of oxidation in fluid milk. Hydroperoxides themselves are odourless; they break down, especially in the presence of metals, by homolytic cleavage to alkoxyl radicals followed by β-scission, giving a range of volatile aldehydes, ketones, alcohols and hydrocarbons that carry the flavour.
Light accelerates oxidation in milk through photosensitisers that absorb visible or ultraviolet light and pass energy to other molecules. Riboflavin is the main photosensitiser in milk; porphyrins and chlorophyll derivatives contribute in some products. In type I photosensitisation, the excited sensitiser reacts directly with a substrate, abstracting an electron or hydrogen to form radicals that start chain reactions. In type II photosensitisation, the excited sensitiser transfers energy to triplet oxygen, forming singlet oxygen, a highly reactive form that adds directly to double bonds by the 'ene' reaction, forming hydroperoxides without a free-radical chain and reacting with linoleic acid on the order of a thousand times faster than triplet oxygen does.
Light-induced flavour in milk arises within minutes to hours of exposure. Its burnt, cabbage-like note comes mainly from methional, formed by photo-oxidation of methionine, and from dimethyl disulfide and other sulfur compounds; prolonged exposure adds lipid oxidation products with cardboard notes. Riboflavin itself is destroyed. Wavelengths around 400 to 500 nm, which riboflavin absorbs, are most damaging, so packaging that blocks this range, opaque or pigmented, gives the most protection; clear plastic bottles under fluorescent or LED lights in retail cabinets are a common cause of defects.
Carotenoids quench singlet oxygen, and tocopherols both quench it and scavenge radicals, so the antioxidant status of milk, influenced by the cow's diet, affects susceptibility.
The volatile products of lipid oxidation have very low odour thresholds and characteristic notes. Hexanal, from linoleic acid, is grassy and is the most widely used marker of oxidation in dairy products; trans-2-nonenal gives cardboard notes; 1-octen-3-one is metallic and mushroom-like and has an extremely low threshold, contributing to metallic off-flavour; unsaturated aldehydes such as 2,4-decadienal are deep-fried and fatty; and products of n-3 fatty acids, such as propanal and 2,4-heptadienal, give fishy and painty notes, important in fortified products containing fish oils. The combination of compounds, rather than any single one, determines the flavour perceived.
Antioxidants slow oxidation by different mechanisms. Chain-breaking antioxidants such as tocopherols donate a hydrogen to peroxyl radicals, forming stable radicals that do not propagate the chain. Metal chelators, such as citrate and phosphates, and proteins such as lactoferrin and caseins, which bind iron and copper, reduce initiation. Singlet-oxygen quenchers such as carotenoids protect against photo-oxidation. Some compounds, including ascorbic acid at certain concentrations in the presence of metals, can act as pro-oxidants by reducing metal ions to their more active forms. Maillard reaction products formed in heated milk powders have antioxidant activity.
Oxidation is measured at different stages. Peroxide value measures hydroperoxides, the primary products, and conjugated dienes are measured by absorbance at about 234 nm; both rise and then fall as hydroperoxides decompose. Secondary products are measured by the thiobarbituric acid test, which detects malondialdehyde and related compounds, by the p-anisidine value for aldehydes, and most specifically by headspace gas chromatography of volatiles such as hexanal. Oxygen consumption in sealed packs and accelerated tests at elevated temperature estimate oxidative stability.
The Maillard reaction is a network of reactions between reducing sugars and amino groups, principally the ε-amino group of lysine residues in proteins. Hodge's scheme of 1953 divides it into three stages, still used to organise the chemistry. In the initial stage, the carbonyl group of the reducing sugar condenses with the amino group to form a Schiff base and then an N-glycosylamine, which undergoes the Amadori rearrangement to a 1-amino-1-deoxy-2-ketose, the Amadori product. In milk, where the reducing sugar is lactose, the Amadori product is lactulosyl-lysine; where lactose has been hydrolysed, as in lactose-free milk, glucose and galactose react faster than lactose, giving fructosyl-lysine and tagatosyl-lysine and much more rapid early Maillard reaction.
The extent of the early Maillard reaction is measured indirectly by furosine, ε-N-(2-furoylmethyl)-L-lysine, formed when the Amadori product is hydrolysed in strong acid during analysis. Only a fraction of the Amadori product is converted to furosine, so a conversion factor is needed to estimate blocked lysine. Furosine content increases with heat load, from a few milligrams per 100 g of protein in raw and pasteurised milk to much higher values in UHT and especially in-container sterilised milk and in stored powders; it is therefore used, with lactulose, to classify heat treatments and to detect the addition of reconstituted milk powder to fresh milk. Blocked lysine is nutritionally unavailable, so the early Maillard reaction reduces protein quality without visible browning.
In the intermediate stage, Amadori products degrade by several routes. Under acid to neutral conditions, 1,2-enolisation leads through 3-deoxyosones to furfurals such as hydroxymethylfurfural (HMF); under neutral to alkaline conditions, 2,3-enolisation leads through 1-deoxyosones to reductones and flavour compounds such as maltol, furaneol and other furanones; sugar fragmentation gives short-chain dicarbonyls such as glyoxal, methylglyoxal and diacetyl. These reactive dicarbonyls drive the later stages. In the Strecker degradation, an α-dicarbonyl reacts with a free amino acid, which is decarboxylated and deaminated to a Strecker aldehyde with one carbon fewer, such as methional from methionine, 3-methylbutanal from leucine and phenylacetaldehyde from phenylalanine, together with carbon dioxide and an α-aminoketone; aminoketones condense to pyrazines, which give roasted and nutty notes.
In the final stage, many reactive intermediates polymerise and react with proteins to form brown, nitrogen-containing polymers called melanoidins, whose structures are still not fully known, and a diverse group of advanced glycation end products (AGEs), protein modifications formed from dicarbonyls and from oxidation of Amadori products. Important AGEs in dairy products include Nε-carboxymethyl-lysine (CML), Nε-carboxyethyl-lysine (CEL), pyrraline and the cross-link pentosidine. Their concentrations increase with heat load and storage and are much higher in sterilised milk, powders and infant formulas than in pasteurised milk.
Heating milk under alkaline conditions also forms lysinoalanine, a cross-link produced by the reaction of lysine with dehydroalanine formed from phosphoserine or cysteine by β-elimination. It is not a Maillard product but is often measured alongside them as an indicator of severe processing.
The rate of the Maillard reaction depends on temperature, time, pH, water activity, the type of sugar and amino compound, and the presence of metals and other reactants. It increases with temperature with an activation energy typical of chemical reactions, so its z value is far higher than that of microbial death. It increases with pH, because the unprotonated amino group is the reactive form. It is fastest at intermediate water activities, about 0.6 to 0.8, where reactants are concentrated but still mobile; in very dry powders it is slow but continues during long storage, and in dilute solutions it is slowed by dilution. Monosaccharides react much faster than lactose.
In dairy products, the reaction is wanted in some and unwanted in others. It gives the colour and flavour of dulce de leche, caramelised milk confections, browned cheeses and many traditional heat-concentrated milk products. It causes browning, stale and gluey flavours and loss of available lysine in UHT milk, sterilised milk, milk powders and infant formulas during processing and storage, and it contributes to the reduced solubility of stored powders. Health aspects of dietary AGEs are debated: they are absorbed to a limited extent, and some studies associate high intakes with inflammatory markers, but evidence of harm at usual intakes is inconclusive. Acrylamide, formed from asparagine and reducing sugars at high temperatures in dry conditions, is a significant concern for fried and baked cereal and potato foods but is formed only at low levels in most dairy products.
Control relies on minimising heat load, as in direct rather than indirect UHT, low storage temperatures, low water activity in powders, and, in lactose-hydrolysed products, low storage temperatures and short shelf lives.
The salts of milk, though only about 0.7 to 0.8% of its mass, control the stability of the casein micelles, the heat stability of milk, rennet coagulation, the buffering of milk and many product properties. Typical concentrations in cow milk, varying with breed, stage of lactation, feed and health, are about 30 mM calcium, 20 to 21 mM inorganic phosphate, 5 mM magnesium, 9 mM citrate, 20 to 25 mM sodium, 35 to 40 mM potassium and 30 mM chloride.
These ions are distributed between the serum and the colloidal phase of the micelles. About two thirds of the calcium, about half of the inorganic phosphate, about one third of the magnesium and a small fraction of the citrate are colloidal, mainly as colloidal calcium phosphate bound to casein phosphoserine clusters; sodium, potassium and chloride are almost entirely in the serum. In the serum, calcium and magnesium are partly free ions and partly complexed with citrate, phosphate and other anions. Free ionic calcium is only about 2 to 3 mM, roughly a tenth of total calcium, but it is the form that affects protein interactions most directly.
The serum is supersaturated with respect to several calcium phosphate phases, including hydroxyapatite, octacalcium phosphate and dicalcium phosphate dihydrate, yet does not precipitate them under normal conditions, because nucleation is slow and casein phosphopeptides inhibit crystal growth. The colloidal calcium phosphate itself is amorphous or poorly crystalline, its composition close to that of brushite-like or more basic calcium phosphates with associated magnesium and citrate. The ability of casein to sequester calcium phosphate in this form allows milk to carry far more calcium and phosphate than could otherwise stay in solution, a key nutritional function.
Mineral equilibria in milk are modelled by treating the serum as a solution of ions in equilibrium with each other through association reactions, and the colloidal calcium phosphate as a separate phase in equilibrium with the serum. The model writes, for each association reaction, an equilibrium constant, such as K = [CaCit⁻]/([Ca²⁺][Cit³⁻]) corrected for activity, together with mass balances for each component and the measured pH, and solves the resulting set of non-linear equations numerically for the concentration of every species.
Activities, not concentrations, enter the equilibrium constants. With an ionic strength of about 0.07 to 0.08 mol/L in milk serum, activity coefficients of divalent ions are well below one, roughly 0.3 to 0.4, and are estimated from extended Debye–Hückel or Davies equations, such as log γᵢ = −Azᵢ²(√I/(1 + √I) − 0.3I). Because activity coefficients depend on the ionic strength, which depends on the speciation, the calculation is iterative. Models developed by Holt, by Walstra and Jenness and by later workers, sometimes coupled with general chemical speciation software, predict serum composition and the amount of colloidal calcium phosphate under different conditions, and agree reasonably with measurements.
Measurements to test models require separating serum from the colloidal phase without disturbing equilibrium, by ultrafiltration, dialysis against a small volume of buffer, or ultracentrifugation at the temperature of interest, and analysing the serum for total calcium, magnesium, phosphate and citrate by atomic absorption, inductively coupled plasma spectrometry or ion chromatography. Ionic calcium is measured directly with a calcium ion-selective electrode, calibrated in solutions of similar ionic strength. Discrepancies between models and measurements arise from uncertain equilibrium constants, the effects of proteins and ion binding to caseins, and the slow kinetics of calcium phosphate changes.
Processing shifts the mineral equilibria, with important consequences. Heating causes calcium phosphate to precipitate onto the micelles, because its solubility falls with increasing temperature; the hydrogen ions released lower the pH of milk at high temperature, by more than half a pH unit at UHT temperatures, contributing to the reduction in heat stability. On cooling, these changes are largely but not entirely reversed, and reversal can take hours; part of the precipitated calcium phosphate remains in a form that alters rennet coagulation, one reason why high-heat milk coagulates poorly.
Acidification dissolves colloidal calcium phosphate, raising serum calcium and phosphate; at the pH of fresh cheese and yoghurt nearly all the calcium is in the serum, which affects whey composition and the calcium content of cheeses made by acid or rennet coagulation. Concentration by evaporation or membrane filtration raises the concentrations of all ions; in evaporation, more calcium phosphate becomes colloidal and pH falls, while in ultrafiltration soluble ions pass into the permeate in proportion to their serum fraction, so the retentate is enriched in colloidal minerals. Additions also shift the balance: calcium chloride raises ionic calcium and lowers pH; phosphates and citrates bind calcium and lower ionic calcium, the principle of stabilising salts in evaporated milk and emulsifying salts in processed cheese; carbon dioxide lowers pH and increases soluble calcium.
Cooling also has effects: at low temperature, some colloidal calcium phosphate dissolves together with β-casein. Mineral equilibria models thus provide a quantitative basis for predicting the behaviour of milk in heat treatment, concentration, cheesemaking and the formulation of milk protein ingredients.
Water activity, aw, is the ratio of the partial vapour pressure of water above a food to that of pure water at the same temperature; multiplied by 100 it equals the equilibrium relative humidity. It governs microbial growth, chemical reaction rates and physical properties more closely than water content does. A moisture sorption isotherm relates water content to aw at constant temperature; most dairy powders give sigmoid (type II) isotherms. The adsorption and desorption branches often differ, a hysteresis that reflects changes in the solid's structure.
Several models describe isotherms. The Brunauer–Emmett–Teller (BET) equation, valid up to aw of about 0.4 to 0.5, yields the monolayer moisture content, the water content at which the most active sites are covered by a single layer of water, often a few grams per 100 g of solids for milk powders. The Guggenheim–Anderson–de Boer (GAB) equation extends to aw of about 0.9 and is widely used for foods. The monolayer is often associated with maximum stability against lipid oxidation, because water at this level protects against metal-catalysed oxidation while reaction rates of most other processes remain low.
Milk powders containing amorphous lactose show a distinctive feature: as aw rises to about 0.4 to 0.5 at room temperature, the amorphous lactose crystallises, and because crystalline α-lactose monohydrate binds much less water than the amorphous form in total, the sorbed water content falls sharply at that point, and the released water can make the powder cake. The food stability map, developed by Labuza, summarises how rates of lipid oxidation, Maillard browning, enzyme activity and microbial growth vary with aw, with browning peaking at intermediate aw and oxidation minimal near the monolayer.
Rapid drying or freezing of solutions of sugars and proteins does not allow crystals to form, so the solids remain in an amorphous, non-crystalline state. Below a characteristic temperature, the glass transition temperature Tg, this amorphous material is a glass: a rigid, brittle solid in which molecular mobility is extremely low. Above Tg it becomes a rubbery, viscous material in which mobility rises steeply with temperature. Amorphous lactose, the main component of skim milk powder solids, has a Tg of about 100 °C when anhydrous, but water is a powerful plasticiser and lowers Tg sharply; at a water content corresponding to aw of about 0.4, Tg falls to near room temperature.
The Gordon–Taylor equation, Tg = (w₁Tg₁ + k w₂Tg₂)/(w₁ + k w₂), estimates the Tg of a mixture of solids (1) and water (2) from their weight fractions, the Tg of each component and an empirical constant k. Above Tg, rates of physical changes follow the Williams–Landel–Ferry (WLF) equation rather than Arrhenius kinetics: log(η/ηg) = −C₁(T − Tg)/(C₂ + T − Tg), with universal constants C₁ = 17.44 and C₂ = 51.6 K often used as a first approximation, so that viscosity and related times fall by orders of magnitude within a few tens of degrees above Tg.
The consequences for dairy powders are direct. When a powder's temperature exceeds its Tg by roughly 10 to 20 K, the viscosity at particle surfaces falls far enough for particles to stick together, the sticky point, causing deposits in spray dryers and caking in storage; further above Tg, amorphous lactose crystallises, releasing water and promoting caking, browning and loss of solubility. Spray dryer outlet conditions are chosen to keep particle surfaces below the sticky point, and powders are stored dry and cool and packed in moisture barriers.
A state diagram plots temperature against the solids concentration of a system and shows the equilibrium curves for ice melting and solubility together with the non-equilibrium glass transition curve. It maps which physical state a product will be in under given conditions and how processing paths, such as drying or freezing, move it between states.
In freezing, water crystallises as ice and the remaining unfrozen phase becomes more concentrated, its freezing point falling along the melting curve. If cooling continues, the unfrozen phase eventually becomes so concentrated and viscous that no more ice forms in practical times; this maximally freeze-concentrated state is characterised by Tg′, its glass transition temperature, and Cg′, its solids concentration. For lactose, Tg′ is about −28 °C, and for sucrose about −32 to −34 °C. Below Tg′, the unfrozen phase is a glass and changes such as ice recrystallisation, lactose crystallisation and chemical reactions are extremely slow.
Ice cream and frozen desserts are stored at temperatures, typically about −18 to −25 °C in distribution, that are above the Tg′ of their mixes, which lie around −30 °C or lower depending on the sugars and stabilisers. The unfrozen matrix is therefore rubbery, ice crystals grow by recrystallisation during temperature fluctuations, causing coarse texture, and lactose may crystallise, causing sandiness. Stabilisers slow these changes by increasing viscosity, and formulation with sugars of higher molecular weight raises Tg′. Freeze-dried cultures and other freeze-dried dairy products rely on maintaining the glassy state during and after drying, by keeping product temperature below the collapse temperature, near Tg′, during primary drying and storing the dried product below its Tg.
Proteomics is the large-scale identification and quantification of the proteins in a sample. Milk's major proteins, the four caseins and the main whey proteins, account for over 95% of its protein, so classical methods long overlooked the hundreds to thousands of minor proteins, many in the milk fat globule membrane and the whey, with roles in immunity, nutrition and the biology of the mammary gland. Two-dimensional gel electrophoresis, separating proteins by isoelectric point and then mass, was the first proteomic tool used for milk; it showed genetic variants, phosphorylation and glycosylation forms of caseins and many minor proteins.
Mass spectrometry now dominates. In bottom-up proteomics, proteins are digested, usually with trypsin, the peptides are separated by liquid chromatography and analysed by tandem mass spectrometry (LC-MS/MS), and proteins are identified by matching peptide spectra to sequence databases. Top-down proteomics analyses intact proteins and is better for characterising modifications and variants, such as the many phosphorylated and glycosylated forms of κ-casein. Quantification uses label-free methods, isotopic labelling and isobaric tags, or targeted methods, multiple- or selected-reaction monitoring, in which specific peptides are measured very precisely; targeted methods are used to quantify allergens, species markers and bioactive proteins.
Because the major proteins dominate the signal, fractionation or depletion of caseins and abundant whey proteins is often needed to reveal minor ones. Proteomic studies have catalogued milk proteomes of many species, followed changes through lactation and in mastitis, identified proteins released by heating and processing, and characterised the peptides formed during cheese ripening and digestion, the peptidome.
Lipidomics is the comprehensive analysis of lipid molecular species. Milk fat is the most complex natural fat known: over 400 different fatty acids have been identified, varying in chain length, branching, number, position and geometry of double bonds, and these are combined into hundreds of triacylglycerol species. Fatty acid profiles are measured by gas chromatography of fatty acid methyl esters, with long, highly polar columns and silver-ion chromatography needed to separate the many cis and trans isomers of 18-carbon acids, including the conjugated linoleic acids.
Triacylglycerol species and the positional distribution of fatty acids are analysed by high-temperature gas chromatography, by liquid chromatography with mass spectrometry, and by stereospecific methods that determine which fatty acid sits at each glycerol position. Polar lipids, the phospholipids and sphingolipids of the fat globule membrane, including phosphatidylcholine, phosphatidylethanolamine, sphingomyelin and gangliosides, are separated by normal-phase or hydrophilic interaction liquid chromatography and identified by mass spectrometry; their profiles are of interest for infant nutrition and for ingredients enriched in milk fat globule membrane.
Shotgun lipidomics infuses a lipid extract directly into a mass spectrometer and identifies species from characteristic fragments, giving fast profiles of hundreds of species. Lipidomic studies relate milk fat composition to breed, diet, season and health, track oxidation and lipolysis products, and provide fingerprints for authenticity testing. Extraction, a critical step, uses methods such as the Folch or Bligh and Dyer procedures, and incomplete extraction of polar lipids is a common error.
Metabolomics analyses the small molecules, metabolites, of a sample: sugars and oligosaccharides, organic acids, amino acids, nucleotides, fatty acids, ketone bodies and many others. In milk, the metabolome reflects the animal's metabolism, health, diet and genetics, as well as microbial activity and processing. Nuclear magnetic resonance spectroscopy gives highly reproducible, quantitative profiles of the more abundant metabolites with minimal sample preparation; gas chromatography–mass spectrometry and liquid chromatography–mass spectrometry give greater sensitivity and coverage. Untargeted metabolomics seeks all detectable features and patterns; targeted metabolomics quantifies predefined compounds.
Metabolomic studies have identified milk markers of subclinical ketosis, such as β-hydroxybutyrate and acetone, markers of mastitis, of feeding systems such as pasture versus confinement, of heat treatment, and of the species and origin of milk. Milk oligosaccharides, a distinct glycomic field, are far more abundant and diverse in human milk, at several grams per litre with hundreds of structures, than in cow milk, where they are present at much lower levels; bovine oligosaccharides such as sialyllactoses, and human milk oligosaccharides such as 2′-fucosyllactose produced by fermentation, are now ingredients of infant formulas.
Omics datasets are large and high-dimensional, with many more variables than samples, so their analysis relies on multivariate statistics: principal component analysis to explore structure, partial least squares discriminant analysis and its orthogonal variant to classify samples, and machine-learning methods. The danger of overfitting is severe: a model can perfectly separate groups by chance. Rigorous studies therefore use independent validation sets, cross-validation with permutation tests, correction for multiple testing, and confirmation of putative markers by targeted analysis and, where possible, identification with authentic standards.
Dairy products are frequent targets of economically motivated adulteration because of large price differences between products, species and origins. Species fraud is common: cow milk, cheaper and more plentiful, is added to sheep, goat or buffalo milk products, such as cheeses made traditionally from those species. Detection methods target species-specific molecules. Isoelectric focusing of γ-caseins after plasmin hydrolysis is a classical reference method for detecting cow milk in cheeses of sheep, goat and buffalo milk, adopted in official control in the European Union; polymerase chain reaction (PCR) detects species-specific DNA, usually mitochondrial genes such as cytochrome b, though DNA content does not translate simply into milk proportions; immunochemical methods (ELISA) detect species-specific proteins; and mass spectrometry detects species-specific peptides from caseins and whey proteins, allowing quantification of mixtures.
Other adulterations are detected by targeted tests. Vegetable fats in milk fat are detected through phytosterols such as β-sitosterol, which milk fat lacks, through triacylglycerol profiles compared with reference equations, and through fatty acid ratios. Whey added to milk or milk powders is detected through caseinomacropeptide, released only by rennet action, measured by liquid chromatography. Reconstituted milk powder in fresh milk is detected through furosine and lactulose. Nitrogen-rich adulterants added to inflate apparent protein, notoriously melamine in 2008, are detected by liquid chromatography–tandem mass spectrometry; Codex has since set maximum levels for melamine in infant formula and other foods. Water, starch, sugar, urea and neutralisers in milk are detected by simpler tests.
Fraud prevention also relies on management systems: food-fraud vulnerability assessments, now required by several food-safety certification schemes, analyse where and how fraud could occur in a supply chain and set controls such as supplier approval and targeted testing.
Geographical origin matters for products protected by geographical indications, such as protected designations of origin, and for claims about local, regional or national origin. Stable isotope ratio analysis is the main tool. The ratio of carbon isotopes, δ¹³C, reflects the cow's diet: maize and other C4 plants give less negative values than grass and other C3 plants, so it distinguishes maize-based from grass-based feeding. Nitrogen isotopes, δ¹⁵N, reflect fertilisation, legumes and soil conditions. Oxygen and hydrogen isotopes, δ¹⁸O and δ²H, in milk water reflect drinking water and climate, varying with latitude, altitude and distance from the sea. Sulfur and strontium isotopes, ⁸⁷Sr/⁸⁶Sr, reflect local geology. Combined with trace element profiles and multivariate statistics, isotopes can discriminate regions, but they require reference databases of authentic samples from each region and season.
Production systems leave chemical signatures too. Grass-fed and pasture-based milk tends to have more α-linolenic acid, conjugated linoleic acid, carotenoids and certain terpenes, and different δ¹³C, than milk from cows fed maize silage and concentrates; organic milk, with more grazing and no synthetic nitrogen, often differs in δ¹³C, δ¹⁵N and α-linolenic acid. Mountain and alpine cheeses carry terpenes and other compounds from local flora. Such markers support claims about feeding but overlap between systems, so they indicate rather than prove.
Spectroscopic fingerprints from near-infrared, mid-infrared, Raman, fluorescence and NMR spectroscopy, analysed with chemometrics, are used for rapid screening of authenticity, with suspicious samples confirmed by specific methods. Genetic tests confirm claims such as A2 milk, by genotyping cows for β-casein variants or measuring A1 and A2 peptides in milk by mass spectrometry.
Authenticity also concerns processing. Products labelled as raw, pasteurised, extended shelf-life or UHT must have received the corresponding treatment, and fresh milk must not be partly reconstituted from powder where this is not allowed. Indicators include alkaline phosphatase activity, which is absent after adequate pasteurisation; lactoperoxidase activity, which survives pasteurisation but is destroyed by higher heat treatments; undenatured whey proteins such as β-lactoglobulin, which fall with heat load; furosine, an early Maillard marker; and lactulose, an isomerisation product of lactose. Combinations of these markers distinguish pasteurised, high-pasteurised, UHT and sterilised milks and detect powder in fresh milk, and proposals for their use in regulation have been made in several jurisdictions.
Authenticity methods must be validated for their purpose. Key performance characteristics are selectivity for the target, limit of detection, for example the smallest percentage of cow milk detectable in buffalo cheese, quantification accuracy, robustness across breeds, seasons and processing, and transferability between laboratories. Reference materials of known composition are essential and often scarce; species mixtures, for example, behave differently in fresh milk and ripened cheese because of proteolysis.
Untargeted fingerprinting methods face an extra problem: they detect deviation from a reference population, so their reliability depends on how representative the reference database is, and a new feed or breed may appear as fraud. Legal enforcement requires robust, validated confirmatory methods, while untargeted methods are best used for screening and risk-based targeting of inspections. Authenticity testing is therefore an active field of research at the interface of analytical chemistry, statistics and law.
Milk contains proteins with biological activities beyond nutrition. Lactoferrin, an iron-binding glycoprotein of about 80 kDa, binds two ferric ions with high affinity, depriving bacteria of iron, and has direct antimicrobial, antiviral and immunomodulatory effects; its concentration is about 0.02 to 0.2 g/L in mature cow milk, much higher in colostrum, and around 1 to 2 g/L or more in human milk. It is isolated from whey by cation-exchange chromatography because of its high isoelectric point and added to some infant formulas. Lactoperoxidase catalyses the oxidation of thiocyanate by hydrogen peroxide to hypothiocyanite, an antimicrobial agent, the basis of the lactoperoxidase system for raw milk preservation. Lysozyme, much less abundant in cow than in human milk, hydrolyses bacterial cell walls.
Immunoglobulins provide passive immunity. Bovine colostrum is very rich in them, with immunoglobulin G1 dominant, at tens of grams per litre in the first milking, falling rapidly to well under 1 g/L in mature milk; in human milk, secretory IgA predominates. Bovine colostrum and immunoglobulin concentrates are sold as supplements. Growth factors such as insulin-like growth factor 1 and transforming growth factor β, and hormones, occur at low concentrations, higher in colostrum.
The milk fat globule membrane contains bioactive proteins and lipids, including butyrophilin, xanthine oxidoreductase, mucins, lactadherin, sphingomyelin and gangliosides, studied for roles in infant development, gut health and cognition. Ingredients enriched in membrane material, from buttermilk and butter serum, are produced for infant formulas and nutritional products.
Bioactive peptides are fragments of milk proteins, inactive within the parent sequence, that are released by digestion, fermentation or enzymatic processing and then exert biological effects. Casein phosphopeptides, containing clusters of phosphoserine, bind calcium and other minerals and are studied for enhancing mineral absorption and for dental remineralisation. Angiotensin-converting enzyme (ACE)-inhibitory peptides, such as the tripeptides valine–proline–proline and isoleucine–proline–proline released from caseins in milk fermented by certain strains of Lactobacillus helveticus, lower blood pressure in some human trials; regulators have assessed the evidence differently, with such products approved for health claims in some countries but the claim judged not substantiated in others. Opioid peptides, the casomorphins, include β-casomorphin-7, released more readily from the A1 than the A2 variant of β-casein because of histidine at position 67; its health relevance remains debated. Lactoferricin, released from lactoferrin by pepsin, is strongly antimicrobial. Caseinomacropeptide, released by chymosin, contains no phenylalanine or other aromatic amino acids, making it useful as a protein source for people with phenylketonuria.
Other bioactive components include conjugated linoleic acids, butyric acid, sphingolipids, milk oligosaccharides such as sialyllactoses, nucleotides and vitamins, each with proposed functions supported by evidence of varying strength.
Claims of bioactivity require a chain of evidence: release of the active compound from the food during digestion, survival through the gut, absorption or local action, a demonstrated mechanism, and an effect in well-designed human trials at realistic doses. Many claimed activities rest only on in vitro assays and remain unproven in humans.
Analytical methods for bioactives must identify and quantify specific molecules in complex matrices. Bioactive proteins are measured by immunochemical methods, such as ELISA and radial immunodiffusion for immunoglobulin G and lactoferrin, which detect native, immunoreactive protein; by chromatography, such as cation-exchange or reversed-phase liquid chromatography for lactoferrin; by surface plasmon resonance biosensors, which measure binding in real time; and increasingly by targeted mass spectrometry of signature peptides. Methods that measure native protein and methods that measure total protein can give different results after heat treatment, which denatures immunoglobulins and lactoferrin; the choice must match the claim.
Bioactive peptides are identified by liquid chromatography–tandem mass spectrometry, often after enrichment, and quantified with isotope-labelled internal standards. Their activity is tested by bioassays: ACE inhibition in vitro expressed as the concentration giving 50% inhibition (IC₅₀), antioxidant assays, antimicrobial assays, cell-culture assays of immune or barrier function, and animal and human studies. In vitro assays are cheap and fast but often poorly predictive of effects in people.
Simulated gastrointestinal digestion links food composition to what reaches the intestine. The standardised static in vitro digestion protocol developed by the INFOGEST international network, published in 2014 and refined in 2019, specifies oral, gastric and intestinal phases with defined enzymes, concentrations, pH and times, so that results from different laboratories can be compared; dynamic models simulate gastric emptying and changing conditions. Peptides released in such digestion of dairy products are identified by peptidomics, and their survival and absorption are assessed with intestinal cell models before human studies. Combining these methods is now the standard route from discovery to substantiated claim.
Work through the quantitative relationships until you can use them: integrate a denaturation rate over a heating profile, estimate z from activation energy, calculate ionic strength and an activity coefficient, fit a GAB isotherm, and estimate a mixture's Tg with Gordon–Taylor. Read the original reviews on casein micelle structure, compare the models, and be able to explain which observations each explains. In the laboratory, measure heat coagulation time against pH, peroxide value and hexanal during storage, and furosine in pasteurised and UHT milk.
A research or technical scientist five years into the job uses advanced dairy chemistry to solve product problems: diagnosing the cause of age gelation in UHT milk, specifying heat treatment to control whey protein denaturation for yoghurt or cheese, setting storage conditions for powders from their glass transition and sorption data, tracing oxidised flavour to light exposure or metals, and applying authenticity tests for incoming milk and ingredients. Typical decisions concern choosing the right analytical method and interpreting it against mechanism rather than relying on rules of thumb.
A research and development director or chief scientist with 25 years' experience sets the scientific direction of a business or institute: deciding which processing technologies, ingredients and claims to develop, how to protect against fraud in the supply chain, and which analytical capabilities to invest in. They judge evidence for bioactive and health claims, guide regulatory submissions, and represent the company or country in international standard-setting on methods and specifications, where sound chemistry must be weighed against commercial and political pressure.
A professor or research leader works at the frontier: resolving the internal structure of casein micelles with advanced scattering and microscopy, modelling mineral equilibria and protein interactions during processing, applying omics to milk biology and authenticity, and establishing the bioactivity of milk components through digestion models and human trials. They train doctoral students in rigorous kinetic analysis, method validation and statistics, and contribute to the international methods and reference materials on which the industry's analytical chemistry depends, and they publish negative results too.
Numerical constants (activation energies, association constants, Tg values, mineral concentrations) are typical literature values that vary with method and milk; research use requires the primary sources. Detailed reaction mechanisms of melanoidin formation, the full structure of the casein micelle, and the health effects of dietary AGEs and casomorphins remain unresolved and are presented as such. Instrument operation, specific official methods and their validation data are not reproduced. Regulatory status of bioactive claims and authenticity methods differs between jurisdictions and changes; only examples are given.
Original practice questions written to the syllabus and pattern of the exams named in each question's examTags. They are not past papers and do not reproduce any board's questions.
These notes are free to read here, with no account.
What the Laisha app adds:
Free for ten days, then a subscription. We would rather write that here than have you find it later.