Clear Whey Isolate Drink Haze Control with Demineralized Whey Protein Isolate

In clear whey isolate beverage manufacturing, haze formation is measured as scattered light at 90° with detection angle per ISO 7027-1:2016, and particle size distribution is resolved by laser diffraction according to ISO 13320:2020. A demineralized whey protein isolate with ash content below 0.5 g/100 g dry matter and calcium below 20 mg/100 g dry matter reduces the concentration of insoluble calcium phosphate and calcium-bridged protein aggregates that nucleate during acidification and thermal treatment. On production-scale compounding lines, haze is not a single intrinsic protein property but a function of dissolution pH, mineral speciation, thermal history, and mechanical shear. Batch-to-batch variance in demineralized whey protein isolate originating from different spray-drying towers and membrane systems has been observed when the mineral reduction target is not verified on a dry-matter basis; incoming powder with residual calcium at 35 mg/100 g can produce visible haze after hot-fill at 85°C for 30 s, whereas a lot with calcium at 12 mg/100 g remains below 5 NTU under identical conditions. Published data for the exact haze threshold in every commercial formulation is limited because turbidity is also affected by flavor oils, acidulant type, and non-protein nitrogen. The operational specification for clear whey isolate drinks with a turbidity target below 5 NTU therefore includes a pre-production screen for mineral content, a laboratory-scale acidification test, and a pilot thermal cycle using a plate heat exchanger with hold tube dimensions matched to the production line. Without such screening, the first indication of haze may occur only after downstream filling, when rework and product loss become unavoidable.

What Role Does Residual Calcium and Phosphate Speciation Play in Heat-Induced Haze?

Residual calcium in demineralized whey protein isolate exists partly as free ions, partly as weakly complexed caseinomacropeptide-associated calcium, and partly as phosphate-bound species. At beverage pH values below 4.0, β-lactoglobulin and α-lactalbumin exhibit positive surface charge; neutralization of carboxylate groups by divalent calcium reduces interprotein electrostatic repulsion and increases the probability of collision-driven aggregate growth. Phosphate contributes an additional precipitation pathway because soluble calcium phosphate species can form hydroxyapatite-like nuclei with low solubility product. In a clear whey drink adjusted to pH 3.2 with citric acid, citrate chelation competes with phosphate for calcium, shifting the equilibrium away from insoluble calcium phosphate. The extent of chelation depends on citric acid concentration and ionic strength; a formulation with 0.5 g/L added citric acid provides approximately 2.6 mmol/L citrate, which can sequester a proportion of residual calcium equal to its stoichiometric binding capacity. When demineralized whey protein isolate contains calcium below 20 mg/100 g dry matter, the free calcium concentration in a 6% protein solution is typically below 12 mg/L. At this concentration, calcium phosphate nucleation is slow, but not absent. The residual phosphate concentration then becomes a limiting factor; phosphorus levels above 30 mg/100 g dry matter in the powder can still generate haze after thermal exposure because phosphate species precipitate as calcium phosphate or as protein-phosphate complexes during cooling. This is the processing conflict that demineralization partially resolves but does not eliminate.

Magnesium at residual concentrations below 5 mg/100 g dry matter does not drive haze independently, but it can compete with calcium for phosphate binding and modify the ionic strength of the continuous phase. Sodium and potassium are less directly involved in haze nucleation; however, they influence ionic strength and can compress the electrical double layer when total monovalent cation concentration exceeds 50 mmol/L in the final beverage. The mineral specification cannot be separated from the intended thermal process because a formulation that remains clear at pasteurization conditions may become turbid after ultra-high-temperature exposure at 121°C for 4 s. In such cases, the reaction order with respect to calcium concentration is often nonlinear; small increases in residual calcium produce disproportionately larger aggregate size because calcium acts as a bridging ion rather than a simple charge neutralizer. This bridging mechanism explains why demineralization to very low mineral content is more effective than adding chelating agents alone, although citric acid and sodium hexametaphosphate are sometimes used in downstream mixing to control residual free calcium.

When Acidification Occurs Before High-Pressure Homogenization

Acidification order determines whether demineralized whey protein isolate remains soluble or forms localized aggregate structures that later contribute to visible haze. When citric acid solution is added to a whey protein isolate dispersion under high shear using an in-line static mixer, the local pH near the acid injection point can transiently fall below 2.5 before the bulk reaches the target pH of 3.2. At a production-scale acid dosing skid, the use of a 2:1 recirculation loop with a centrifugal pump and pH transmitter response time of 10 s is insufficient to prevent short-duration excursions unless the acid stream is pre-diluted to 10% w/w and injected into the turbulent zone of a static mixer. Such excursions produce denatured protein films on the walls of the mix tank and create seed particles that persist through downstream filtration. The preferred sequence for low-haze clear whey drinks is to pre-adjust water pH to 3.0–3.4, then add demineralized whey protein isolate under moderate agitation, followed by pH trimming. This sequence avoids exposing protein powder particles to high acid concentration at the hydration surface, which can exceed the bulk pH by more than 1.0 unit during powder wetting. In high-pressure homogenization after acidification, the shear field can disperse soluble aggregates but cannot reverse covalent or electrostatic bridging that occurred during an acid excursion. Homogenization at 150–200 bar first stage and 30–50 bar second stage reduces the d90 of insoluble protein particles from above 5 µm to below 1 µm only when the particles are soft, hydrated flocs; hard calcium phosphate particles and thermal aggregates are not effectively comminuted by homogenization and may require microfiltration. The acidification sequence also affects the final beverage buffering capacity. Phosphoric acid addition before protein hydration can increase phosphate burden and reduce the benefit of demineralization, while citric acid provides chelation but can lower pH too quickly if not controlled. For these reasons, a pH correction protocol with pH, conductivity, and turbidity sensors on the acidification tank is used on manufacturing lines to terminate dosing and trigger recirculation when the turbidity exceeds 10 NTU or when the pH control deviation exceeds ±0.15 pH units.

Crossflow microfiltration with ceramic membranes at nominal pore size 0.1–0.2 µm is installed on several clear whey isolate beverage lines as a final clarification step before heat treatment. The process reduces visible haze by removing suspended protein aggregates, lipid residues, and calcium phosphate crystals, but it also introduces a pressure-dependent performance boundary. At a transmembrane pressure of 1.0–3.0 bar and crossflow velocity of 3–6 m/s, permeate flux for a 6% protein solution at pH 3.2 typically ranges from 50–120 L/m²/h; however, published data for this specific configuration is limited because flux is highly dependent on feedstock viscosity, temperature, and membrane fouling history. A production batch with elevated residual calcium above 25 mg/100 g dry matter can exhibit rapid flux decline from 110 L/m²/h to below 40 L/m²/h within 60 min due to calcium phosphate precipitation and protein-calcium bridging at the membrane surface. Backpulsing every 5 min at 0.5 s reverse pressure of 4 bar reduces the irreversible fouling layer but does not restore the original flux when the feed contains hard mineral scale. This is a practical bottleneck: the same residual minerals that cause haze also accelerate membrane fouling, so demineralization is both a clarity measure and a filtration process enabler. After microfiltration, the product may be cooled to below 4°C before storage or proceed directly to heat treatment; the hold time between filtration and thermal processing should be minimized because soluble aggregates can reassociate under low shear. Ceramic membranes are cleaned with 1% sodium hydroxide at 80°C followed by 0.5% nitric acid at 60°C, and the cleaned membrane is verified by clean water flux measurement according to the membrane manufacturer’s technical bulletin. Without this verification, incomplete cleaning can create a fouling layer that alters permeate turbidity and produces batch-to-batch clarity differences on the next production run.

Homogenization Pressure, Shear History, and Aggregate Particle Size Distribution

High-pressure homogenization is applied to clear whey isolate drinks to reduce the mean particle diameter of suspended protein aggregates and to disperse any remaining fat globules. A typical two-stage homogenizer operates with first-stage pressure between 150 bar and 200 bar and second-stage pressure between 30 bar and 50 bar. In an acidified beverage at pH 3.2, the combination of turbulent flow, impingement, and cavitation reduces d90 from approximately 3–6 µm to 0.5–0.8 µm when the aggregate population consists primarily of soft, freshly formed protein flocs. The shear stress generated in the valve gap can be estimated from pressure drop and gap geometry; however, exact local shear rates are machine-specific and are not routinely measured on production lines. Laser diffraction per ISO 13320:2020 is used to track the particle size distribution after a fixed number of passes, because a single pass may not reach the minimum achievable d90 if the feed temperature exceeds 20°C and protein unfolding occurs. The processing conflict arises when homogenization is performed after thermal denaturation: the same mechanical energy that disrupts aggregates can expose hydrophobic regions and generate new aggregation surfaces. In demineralized whey protein isolate solutions, the absence of calcium bridging limits reaggregation after shear, but the addition of flavor oils or fat-based clouding agents increases the available hydrophobic interface and can destabilize the system. If the beverage contains 0.1–0.2% citrus oil emulsion, the homogenization pressure should be validated by bottle ring and sedimentation tests because oil droplets below 1 µm may remain suspended but can coalesce and scatter light. A particle size specification of d90 below 1 µm is not sufficient to guarantee clarity unless the particle shape and refractive index are also considered; protein aggregates with d90 below 0.5 µm can still produce visible haze if their concentration exceeds 0.1 g/L because turbidity is proportional to both particle concentration and scattering cross-section.

Enzymatic hydrolysis of demineralized whey protein isolate with a food-grade serine endopeptidase reduces the average peptide molecular weight and intermolecular hydrophobic association that contributes to haze after prolonged storage. A limited degree of hydrolysis between 2–5% is typically used for clear beverages because extensive hydrolysis above 8% produces bitter peptides and decreases foaming capacity, which may be undesirable in still and carbonated clear whey drinks. The hydrolysis is carried out at pH 6.0–7.0 and 50–60°C for 30–120 min before acidification, then the enzyme is inactivated by heating at 85°C for 10 min or by low pH hold. After hydrolysis, the particle size distribution shifts to d90 below 0.4 µm, and turbidity remains below 5 NTU even after accelerated storage at 35°C for 14 days. This is a process-specific result; the benefit depends on the substrate mineral content, the enzyme-to-substrate ratio, and the inactivation hold time. On manufacturing lines, incomplete enzyme inactivation leads to continued peptide breakdown in the finished product, causing changes in viscosity, taste, and clarity that are not detected by initial turbidity measurement. Hydrolysates also interact with flavor compounds and can develop grassy or brothy off-notes if the degree of hydrolysis is not tightly controlled. Regulatory compliance for hydrolysates used in clear whey isolate drinks is established under 21 CFR 184.1979c for whey protein isolate and under 21 CFR 170.3(o)(28) for enzyme preparations where applicable, with limits for heavy metals and residual solvents as specified by the supplier. The benefit of hydrolysis must be weighed against the loss of intact protein functionality; in thermally processed clear whey drinks, intact demineralized whey protein isolate already provides low haze when mineral and pH limits are observed, so hydrolysis is generally reserved for formulations with high flavor oil content or low pH below 2.8 where intact protein solubility is compromised.

Thermal Denaturation Pathways in Demineralized Whey Isolate Solutions Are Suppressed below pH 3.4

Thermal treatment of acidified demineralized whey protein isolate is performed in plate-and-frame or tubular heat exchangers ranging from pasteurization at 85°C for 30 s to ultra-high-temperature at 121°C for 4 s. At pH 3.2, the major whey proteins carry a net positive charge, and intermolecular electrostatic repulsion reduces aggregation even when unfolding occurs. The denaturation of β-lactoglobulin is not prevented by low pH; instead, the unfolded monomer remains more soluble because the strong electrostatic repulsion reduces the collision efficiency for irreversible aggregate formation. This is a limiting condition: if the thermal hold time is extended beyond the regulatory minimum, the population of soluble aggregates increases and can later assemble during chilled storage. In a plate heat exchanger, the temperature differential between the heating wall and the bulk liquid can create a thermal boundary layer in which the local temperature exceeds the target by 5–10°C; this boundary layer is a source of nucleation for haze-forming aggregates that are then carried into the hold tube. The use of a tubular heat exchanger with a higher wall shear rate and lower temperature differential is preferred when haze control is critical, but the installed equipment often dictates the available temperature profile. Exhaustive technical elaboration of aggregation kinetics in this specific demineralized whey isolate matrix is limited in the peer-reviewed literature; however, the relevant processing principle is that haze after thermal treatment depends on the product of unfolding rate, free calcium concentration, phosphate concentration, and hold time. Demineralization reduces the calcium and phosphate terms, while pH reduction increases the charge repulsion term. A formulation that shows no visible haze immediately after thermal treatment may still exceed 10 NTU after 24 h at 4°C because soluble aggregates undergo further association. This delayed haze development is measured by storing the heat-treated sample at 4°C and 25°C and recording turbidity at 0 h, 24 h, 72 h, and 7 days with ISO 7027-1:2016. The process control limit for heat treatment is therefore not only the lethal rate or enzyme inactivation but also the post-thermal aggregate size and turbidity in the cooled product.

Shelf-life haze progression in clear whey isolate drinks is evaluated under refrigerated, ambient, and accelerated conditions because the physico-chemical state at day zero does not always predict clarity at month six. A product held at 4°C may develop reversible chill haze from weak hydrophobic association, while a product held at 25°C may develop permanent haze from oxidative crosslinking, calcium phosphate recrystallization, or headspace oxygen ingress. The contribution of packaging to haze is frequently underestimated; polyethylene terephthalate bottles with oxygen transmission rate above 0.1 mL/bottle/day/atm permit dissolved oxygen above 0.5 mg/L, which promotes sulfhydryl oxidation and disulfide-linked protein aggregates. Glass bottles reduce oxygen transmission but introduce light transmission unless amber or coated glass is used. Light exposure at 400–500 nm can generate riboflavin-mediated photooxidation in residual riboflavin from whey, producing small molecular weight compounds that bind to proteins and increase turbidity. Accelerated storage at 35°C for 14 days is a practical screening test, but it does not fully reproduce the kinetic pathway at 4°C because reaction order and precipitation rates differ. A clear whey isolate drink containing 6% protein, pH 3.2, and residual calcium below 20 mg/100 g dry matter can remain below 5 NTU for 6 months at 4°C when the headspace oxygen is below 0.5 mg/L and the bottle is protected from light. If the same formulation is stored at 25°C in clear PET with high oxygen transmission, turbidity can exceed 10 NTU within 30–60 days. These boundaries are formulation-specific and must be verified with each acidulant, flavor, and stabilizer combination. The use of nitrogen flushing and oxygen scavengers reduces oxygen-mediated haze but adds cost and may not address mineral-driven precipitation. Shelf-life studies should therefore include turbidity, visible sediment, particle size, pH, and dissolved oxygen measurement at predefined intervals, and the specification should include an upper turbidity limit and a lower particle size limit rather than visual clarity alone.

At Acidified pH, Turbidimetric and Laser Diffraction Data Diverge

Analytical verification of clear whey isolate beverages is complicated by the fact that turbidimetric measurements per ISO 7027-1:2016 and laser diffraction particle size distribution per ISO 13320:2020 report different physical phenomena. Turbidity is an ensemble scattering measurement that responds to particle concentration, size, shape, and refractive index contrast, whereas laser diffraction reports volumetric particle size distribution and may not detect low-concentration submicron aggregates if the scattering signal is below the instrument detection limit. A beverage can exhibit turbidity below 5 NTU while still containing a particle population with d90 above 1 µm if the total particle volume fraction is low; conversely, a high concentration of 0.2 µm particles can produce visible haze without exceeding a d90 specification. This divergence is especially pronounced when acidified whey protein isolate contains residual lipid droplets, which have different optical properties than protein aggregates. The correct analytical approach is to use both methods together, supplemented by pH, zeta potential, and mineral content. The following verification matrix is used for low-haze clear whey isolate drinks. The table is not a regulatory checklist but a compilation of method designations and typical specification ranges derived from dairy analytical standards and membrane filtration technical bulletins.

Calcium chloride fortification is incompatible with demineralized whey protein isolate when residual phosphate exceeds 30 mg/100 g dry matter because calcium phosphate precipitation occurs at pH above 3.5 and can defeat the clarity specification. Certain natural antioxidants and polyphenol-rich botanical extracts can bind to whey proteins and form haze under heat treatment. Addition of sodium hexametaphosphate at 0.05–0.1% w/w can reduce free calcium but may contribute to sodium load and alter the taste profile. The operational boundary for this matrix is therefore defined by mineral content, pH, thermal load, and packaging oxygen ingress; published data for every combination of these variables is limited, so production-scale validation with the actual heat exchanger and filling line is required before a clear whey isolate drink can be released with a defined shelf-life clarity specification.