Ice Point Depression in High Overrun Frozen Desserts

In high-overrun frozen dessert manufacturing, the equilibrium freezing behaviour of the serum phase determines both the rate of ice phase development in the scraped-surface heat exchanger and the final textural stability after blast hardening. The initial freezing point of a standard ice cream mix is depressed below 0°C by the collective molality of dissolved sucrose, lactose, milk salts, low-molecular-weight corn syrup solids, and any added monosaccharides; published values for typical full-fat mixes fall between -2.5°C and -4.0°C, while higher-solids high-sugar formulations may have initial freezing points as low as -5.0°C. At an overrun of 100%, roughly half the product volume is air, so the continuous serum phase occupies a significantly reduced volumetric fraction, yet its freezing point remains governed by the concentration of soluble solids per unit mass of water, not by the total product volume. The cryoscopic depression is conventionally expressed through the Blagden molality relationship for ideal dilute solutions, but industrial mixes deviate because of non-ideal solute interactions, high viscosity, and the colloidal contributions of fat globules and air. Continuous freezers such as scraped-surface units operating at dasher speeds between 300 rpm and 700 rpm remove latent heat at barrel wall temperatures between -30°C and -35°C, discharging product at draw temperatures from -5°C to -8°C. At those draw temperatures, the ice phase volume in conventional mixes is typically between 30% and 55%; the remaining water exists as a freeze-concentrated serum phase whose solute concentration has been elevated by the removal of pure ice from solution. This dynamic concentration effect means the actual freezing point of the serum phase at the outlet is substantially below the initial freezing point of the unfrozen mix, and it is this highly viscous, solute-rich fluid that must flow around air cells during the residence time allowed in the freezer. Failure to account for the shift in serum concentration during partial freezing routinely leads to inadequate dasher loading, inconsistent overrun, and accelerated ice crystal ripening in the hardening tunnel.

What governs the equilibrium freezing curve in a 110% overrun dairy emulsion?

For a frozen emulsion inflated to 110% overrun, the air phase occupies 52.4% of the total product volume, leaving 47.6% as concentrated matrix. In thermodynamic terms, the initial freezing point of the unfrozen mix is depressed below 0°C by the sum of soluble species. Applying the ideal cryoscopic equation ΔT = 1.86 K kg mol⁻¹ × osmolality gives only an approximation, because concentrated sugars exhibit significant non-ideality above 20% solids. Industrial formulation therefore relies on empirical freezing point depression factors: sucrose is assigned a factor of 1.0, dextrose 1.9, lactose 1.0, and 63 DE corn syrup solids 0.8 relative to sucrose on an equivalent mass basis. When lactose is hydrolysed to glucose and galactose, the molar concentration of small carbohydrates increases, and the depression factor roughly doubles for the hydrolysed fraction; a 70% lactose hydrolysis in a mix containing 6% lactose can lower the initial freezing point by up to 0.5°C. At 100% overrun, the freezing point of the serum phase remains governed by the same solute concentration, but the route by which heat is removed changes because air cells of 20–50 μm diameter create thermal resistances across the matrix. The initial freezing point for typical formulations can be measured by thermistor cryoscope in accordance with ISO 5764:2009, although the method was developed for milk and must be calibrated for mixes containing fat and stabilizers. During partial freezing, the serum phase concentration increases as pure water crystallizes; at -5°C, a mix with an initial freezing point of -2.8°C has already separated into roughly 30–40% ice and 60–70% freeze-concentrated serum. This residual serum exhibits a freezing point below -20°C and a viscosity many times that of the unfrozen mix, which is why freezer draw conditions cannot be predicted from the initial cryoscopic point alone.

Because air is a poor thermal conductor, reducing the density of the aerated mix below approximately 0.55 g/cm³ alters the rate at which latent heat can be removed during blast hardening. At -18°C equilibrium, the fraction of total water present as ice is governed by composition and is approximately 72–78% for a standard mix; however, in a 100% overrun product, the insulating air phase may delay core equilibration so that after an equivalent hardening interval the centre remains at -12°C to -15°C and the local ice fraction is only 65–70%. The remaining unfrozen serum phase is concentrated to a solute content that can exceed 60% by mass, producing a glass transition temperature often below -30°C for sucrose-dominated systems. This residual freeze-concentrated matrix governs the product’s tendency to undergo recrystallization; smaller ice crystals melt while larger crystals grow by Ostwald ripening, particularly when storage temperature fluctuates by more than ±2°C. High-overrun products with mean air cell diameters above 50 μm and serum lamellae below 15 μm exhibit more rapid heat penetration during intermittent warm excursions, increasing the mobile water fraction and accelerating coarsening. Production-scale audits from continuous freezers indicate that draw temperature variability of ±0.5°C is sufficient to change the median ice crystal size by 5–10 μm after 24 h of hardening, while variability above ±1.5°C often produces perceptible coarseness. This is why temperature control at the freezer outlet is usually specified to a tolerance tighter than ±0.3°C when the target median ice crystal diameter is below 35 μm. Published methods for consistent crystal sizing include optical microscopy adapted from ISO 13322-1 and image analysis of freeze-fractured samples; however, standardized protocols for aerated products are less developed than for non-aerated dairy systems.

Accurate overrun determination is inseparable from ice point depression control because the volume of air modifies the measured density and the effective thermal mass of the product. Overrun is calculated as the volume of frozen product minus the volume of unfrozen mix, divided by the volume of unfrozen mix, then multiplied by 100. A 100% overrun product therefore weighs approximately 4.5–4.8 lb/gal when the liquid mix weighs 9.0–9.6 lb/gal; this weight range meets the minimum weight standard of 4.5 lb/gal under 21 CFR 135.110, which also requires not less than 1.6 lb total food solids per gallon. Density measurement of aerated products is typically performed at the freezer outlet using a fixed-volume cup and an electronic scale, and production-scale overrun controllers adjust air intake through feedback from product density rather than direct volume. Because the presence of air does not change the intrinsic cryoscopic properties of the serum phase, an aerated product and its parent mix may have identical initial freezing points; however, their rates of ice crystallization diverge during dynamic freezing because air cells create local regions of reduced thermal conductivity and because the scraped-surface freezer must work against the insulating effect of the air phase. This distinction is frequently overlooked when the freezer is operated only to a target draw temperature with no measurement of ice phase volume. A product discharged at -7°C with 80% overrun can contain 10–15% more ice than the same formulation at 120% overrun, depending on heat transfer intensity and dasher speed. The resulting hardening time in a blast tunnel set to -35°C air and 3–5 m/s air velocity may differ by 25–40%. Published correlations for heat transfer in aerated dairy systems are limited, so production trials typically rely on a combination of thermocouple centre-line measurements and ice crystal microscopy rather than predictive models.

A 4°C processing window separates acceptable ice crystal diameters from coarse fracture

In commercial high-overrun production, the difference between a fine-grained structure and a coarse, crumbly texture is controlled by a narrow thermal and mechanical window. The scraped-surface freezer must transform approximately 30–50% of the water into ice while simultaneously dispersing injected air into cells with diameters of 20–50 μm; this occurs in a residence time of 30–90 s, depending on throughput and dasher speed. The barrel is commonly cooled with ammonia or R-404A refrigerant at evaporation temperatures from -32°C to -38°C, and the internal barrel wall is maintained at -25°C to -30°C. Dasher tip speed, rather than shaft speed alone, governs shear stress at the wall; commercial units operate at tip speeds between 4 m/s and 11 m/s. In parallel, air injection is controlled by mass flow meters to maintain overrun within ±5% of target. A draw temperature elevation from -7°C to -6°C may reduce ice phase volume by 5–10% absolute, leaving excessive free water for hardening and producing visible ice banding. Conversely, lowering draw temperature from -7°C to -8°C increases viscosity sharply, often above 10 Pa·s at 10 s⁻¹, causing pump cavitation and reduced overrun stability. Thus the operating window for a typical 12% butterfat, 16% sucrose-equivalent mix is roughly 4°C wide, from -6°C to -10°C measured at the outlet, with tighter limits when overrun exceeds 110%. This window narrows further with high dextrose equivalence corn syrups, because low-molecular-weight sweeteners depress the freezing point and increase the unfrozen water fraction; replacement of 10% of sucrose with 63 DE corn syrup solids can lower the draw temperature required for equivalent ice phase volume by 1–2°C. Operational failure modes observed on production lines include dasher vibration above 50 Hz rotational frequency, back-pressure fluctuations exceeding 0.5 bar, and air coalescence when serum viscosity drops below 0.2 Pa·s at the outlet. Each of these shifts the ice crystal size distribution toward coarser fractions. Published data for the exact shear-rate dependence of ice nucleation in aerated dairy systems under commercial freezer conditions is limited, but the observed process boundaries are reproducible across continuous freezers with barrel diameters from 300 mm to 600 mm.

Table 1 compiles representative freezing behaviour across a formulation gradient for high-overrun mixes, based on published ice cream freezing curves and industrial cryoscope data.

When overrun exceeds 120%, air serum interfacial area becomes rate-limiting

The geometry of the air serum interface changes the apparent freezing kinetics because the average serum lamella thickness approaches the same order of magnitude as the target ice crystal size. At 120% overrun, the air fraction is 54.5%, and the specific air–serum interfacial area for a mean air cell diameter of 40 μm is approximately 8.2 × 10⁴ m²/m³. A 20 μm radius air cell under a serum surface tension of 0.045 N/m experiences a Laplace pressure of roughly 4.5 kPa; smaller cells experience proportionally higher internal pressure, so gas diffuses from small cells to larger cells at a rate governed by the solubility and diffusivity of nitrogen or air in the freeze-concentrated serum. This pressure-driven disproportionation is more pronounced at high overrun because the lamellae between adjacent air cells are thin, often 5–10 μm, and ice crystals nucleated within these lamellae are geometrically constrained to grow normal to the interface. The result is a bimodal or skewed ice crystal size distribution in which a few large crystals grow at the expense of numerous small ones. Thermal diffusivity is also reduced in high-overrun products: the thermal conductivity of air at -25°C is approximately 0.024 W/m·K, compared with 2.2 W/m·K for ice and 0.5 W/m·K for unfrozen serum, so the effective thermal resistance increases as air volume fraction rises from 50% to 55%. Production-scale blast tunnels must therefore be operated with lower product bed heights or longer dwell times when overrun exceeds 120%; otherwise the centre of the package may remain at -10°C for hours while the surface has already reached -30°C. This thermal lag promotes recrystallization and can produce a hard, icy outer shell around a softer core. Published data for combined heat and mass transfer at overrun levels above 130% is limited, and direct on-line measurement of ice crystal size remains uncommon outside research lines.

Sucrose equivalence, lactose hydrolysis, and capillary freezing point shift

Formulation decisions that alter sucrose equivalence also shift the entire freezing curve, not only the initial freezing point. Replacing 10% of sucrose with 42 DE corn syrup solids raises the average molecular weight of the sweetener fraction, reduces the freezing point depression per gram, and increases the ice fraction at a given draw temperature; replacing with 63 DE corn syrup solids does the opposite. Lactose hydrolysis in low-lactose or lactose-free products doubles the molar contribution of the hydrolysed lactose and may require a reduction in added sugars of 1–2% to maintain the same draw consistency. High-intensity sweeteners have negligible cryoscopic effect, so formulations in which bulk sucrose is replaced by maltodextrin and non-nutritive sweeteners often show a higher initial freezing point and firmer texture at storage temperature unless the maltodextrin DE is selected to restore the desired depression. In parallel, capillary freezing point depression arising from the curved ice–serum interface follows the Gibbs-Thomson relation; ice crystals with diameters below 10 μm exhibit a melting point suppressed by a few tenths of a Kelvin relative to bulk ice. This is beneficial for immediate sensory smoothness, but it makes small crystals thermodynamically unstable relative to larger crystals, so the rate of ripening increases when the median crystal size is pushed below approximately 20 μm by aggressive freezing. Stabilizers such as locust bean gum, guar gum, and carboxymethylcellulose at total levels of 0.15–0.30% do not significantly change the initial freezing point, but they increase the viscosity of the freeze-concentrated serum and reduce the mobility of water and solute during temperature fluctuation. Emulsifiers, typically mono- and diglycerides at 0.1–0.2%, alter fat globule aggregation and air cell adsorption but have no direct cryoscopic effect; their influence on ice point depression is limited to changes in interfacial structure that affect ice crystal spatial distribution. Thus a formulation designed for 120% overrun cannot be optimised on the basis of freezing point depression alone; it must simultaneously satisfy the serum rheology, air cell stability, and thermal history constraints imposed by the continuous freezer and hardening tunnel.

The following compliance and test methods are applied to verify serum freezing point, rheology, and ice crystal structure when qualifying high-overrun frozen desserts.

Control of ice point depression on production lines requires synchronisation of freezer outlet temperature, overrun, and hardening capacity. A batch-to-batch variance in mix freezing point of 0.2°C, which can result from normal seasonal milk composition shifts, is sufficient to alter the ice fraction at draw by 3–5% absolute. On continuous freezers with barrel pressure control, an unexpected reduction in serum osmolality increases ice phase volume, raises apparent viscosity, and may lift the dasher motor load by 10–15%. Conversely, an increase in serum osmolality reduces ice phase volume at the same draw temperature, leading to a wetter product that may collapse under its own weight after packaging. Therefore, mixes are often standardised to a target freezing point rather than only to total solids or fat content. In-plant verification combines cryoscope readings of the pasteurised mix with density-based overrun measurement and off-line ice crystal microscopy. The operational boundary for most high-overrun products is a draw temperature between -6°C and -8°C; above -5°C, the product is too fluid to retain air, and below -10°C, the product may be too viscous to flow through the transfer pipe. Hardening must then reduce the centre temperature to below -18°C within 4–6 h to limit ripening; longer hardening cycles allow the growth of ice crystals above 50 μm and produce perceptible coarseness. These limits are not universal; published data for specific non-dairy fat systems and for overrun levels above 140% is limited, and direct adaptation to such systems requires pilot-scale trials with the actual freezer geometry and air injection system.