Mouthfeel Dependence on Erythritol Particle Size in No Sugar Added Chocolate

When erythritol is used as the bulk sweetener in a no sugar added chocolate, its particle size distribution becomes a processing and sensory variable with the same leverage as cocoa butter content or conching time. The mouthfeel of the finished mass is not governed by erythritol purity alone; it is governed by the fraction of particles exceeding the oral soft-tissue detection threshold near 25 µm, the rate of endothermic dissolution on the tongue, and the ability of the continuous cocoa butter phase to wet the high-surface-area powder. Dry-milled erythritol is commonly released for conching only after laser diffraction analysis according to ISO 13320:2020 confirms that the D90 falls within a product-specific target interval. If the D90 exceeds 30 µm, the mass tends to exhibit immediate sandiness and a delayed cooling onset. If the D90 is forced below 15 µm without reformulation, the chocolate can become excessively viscous and difficult to pump through downstream depositors. The particle size specification therefore defines a narrow processing window in which oral smoothness, cooling perception, and rheological stability coexist.

For products labelled no sugar added, the formulation constraints are stricter than for conventional chocolate because the bulk sweetener cannot be partially replaced with liquid sugar streams to adjust viscosity. Under 21 CFR 101.60(c), a no sugar added claim requires that no sugar or sugar-containing ingredient is added during processing, and the finished product must not be nutritionally inferior to a reference product. Erythritol replaces sucrose as a bulk solid, but its crystalline fracture behavior and solubility limit are not identical to sucrose. In European applications, erythritol is permitted as E968 under Regulation (EC) No 1333/2008, and no statutory particle size limit exists for polyol-sweetened chocolate. The manufacturer must therefore derive an internal acceptance limit from tribological, rheological, and sensory measurements rather than from a single regulatory number. This absence of a statutory particle size requirement places the burden on the production line to control the dry-milling step, the conching sequence, and the final moulding viscosity with sufficient documentation to support the label claim.

What Limits the Dry-Milling Fineness of Erythritol Before Conching?

The comminution limit for erythritol in a no sugar added chocolate line is set by three process conflicts rather than by the intrinsic hardness of the crystal alone. First, the classifier cut point in an air classifier mill controls the maximum particle size, but reducing that cut point also returns a larger mass fraction to the grinding chamber, increasing residence time and specific energy input per kilogram of milled erythritol. A pinned-disc mill with a stainless-steel dynamic classifier can reduce erythritol to a D90 below 20 µm, but throughput may fall because the aerodynamic equivalent diameter and particle density differ from sucrose. Second, the total surface area of erythritol powder increases as the D50 falls. A powder with a D90 of 18 µm can require more intensive dispersion during conching because the larger number of interparticle contacts immobilizes a greater proportion of cocoa butter. Third, fine erythritol particles are susceptible to electrostatic charging in pneumatic conveyance and can adhere to stainless-steel hopper walls at relative humidities below 30 %, producing bridging and irregular feed to the conche. These effects are observed on production-scale classifier mills when a single cut point change is made without a corresponding adjustment to the humidification and dosing equipment.

The choice of mill type also changes the shape of the particle size distribution. A pinned-disc mill with an integrated classifier tends to generate a broader distribution with a D90/D10 ratio that can exceed 3.5, while a jet mill with a classifier wheel can produce a narrower distribution with a D90/D10 ratio below 2.5. The broader distribution is not automatically inferior for mouthfeel because a controlled mass fraction of fine particles can occupy the voids between larger erythritol crystals and reduce the direct detection of oversize edges. However, a broad distribution also contains a larger oversize tail for the same D90, and that oversize tail dominates tactile response. Useful release parameters therefore include the D10, D50, D90, and the span calculated from (D90 − D10) / D50 according to ISO 9276-2:2014. A span above 2.0 may indicate excessive internal recirculation in the mill, while a span below 1.2 may indicate overgrinding and the generation of excessive fines that will later increase Casson yield stress.

Because the chocolate mass is a suspension of erythritol particles in a continuous cocoa butter phase, the Casson model is used as the standard engineering description of its flow behavior. The Casson equation relates the square root of shear stress to the square root of yield stress and plastic viscosity, and it is commonly measured with a rotational viscometer at 40 °C after a defined pre-shear period. A no sugar added chocolate containing erythritol at a D90 of 25 µm may show a Casson yield stress in the range of 2.0 Pa to 8.0 Pa and a Casson plastic viscosity of 2.0 Pa·s to 5.0 Pa·s. When the erythritol D90 is lowered to 15 µm, the yield stress can rise by 30 % to 70 % if the fat fraction remains constant, because smaller particles present a greater total surface area to the liquid fat and immobilize a larger proportion of the continuous phase. The same phenomenon alters mouthfeel: a high yield stress can suppress the initial flow of the mass over the tongue, delaying the perception of cooling and reducing the rate at which the chocolate melts. Published data for this specific configuration are limited, but the directional effect is reproduced in industrial conching records as a torque increase that requires either additional fat or a coarser refining endpoint.

The sensory correlate of these rheological values is not captured by a single particle size number. Descriptive sensory panels trained according to ISO 5492:2008 usually separate grittiness, sandiness, chalkiness, cooling onset, cooling intensity, and residual cooling. Grittiness is associated with hard particles above the oral tactile threshold near 25 µm, while sandiness may be perceived with angular particles even if the D90 is below that threshold. Chalkiness often arises when the D50 falls below 8 µm and the powder surface is not sufficiently wetted by cocoa butter; the mass then absorbs saliva rapidly and produces a dry, adherent film. A no sugar added chocolate with a D90 of 22 µm and a D50 of 9 µm can therefore have an acceptable total particle size distribution but an undesirable mouthfeel if the fraction between 8 µm and 15 µm is too low or too high. The operational target is not to minimize particle size, but to keep the oversize tail below the oral detection threshold while preserving enough coarse-fine bimodality to limit Casson viscosity increase.

When D90 Falls Below 20 µm During High-Shear Refining

When the erythritol D90 is driven below 20 µm in a ball mill or air classifier mill, the chocolate mass enters a high-surface-area regime in which small changes in particle size produce disproportionately large changes in conche motor load and moulding behavior. In a conventional roll refiner, the gap pressure and roll speed are set to achieve a specific film thickness on the roll surface. Erythritol crystals, because of their high hardness and distinct fracture mode, can produce clean shards with sharp edges rather than the rounded fragments typical of sucrose after roll refining. These shards remain detectable in the mouth even at D90 values below 20 µm if the shape factor is high. Consequently, particle shape measurement by static image analysis is used alongside laser diffraction. A circularity below 0.85 or an aspect ratio above 1.5 may signal the need for a low-shear polishing step before final liquefaction. Published data for this specific configuration are limited, but the shape effect is visible in scanning electron micrographs of milled erythritol after dry classification.

The processing window narrows because the fine fraction competes for the available fat. In a no sugar added formulation, cocoa butter is the continuous phase, and any increase in particle surface area immobilizes additional cocoa butter. The result is a higher Casson yield stress and a higher plastic viscosity at a given fat level. If the mass is over-refined, the conche may be unable to develop the shear necessary to wet all particles, or the conche motor may trip when the torque exceeds the drive limit. The correction is not always as simple as adding more cocoa butter, because the no sugar added label and the target nutritional profile restrict the total fat increase. Emulsifiers such as lecithin or polyglycerol polyricinoleate can reduce yield stress, but their concentration is limited by legal and sensory constraints. In practice, the lowest practical D90 for erythritol in a no sugar added chocolate is often between 15 µm and 18 µm, depending on the fat fraction and the lecithin dosage. Below this interval, the mass becomes too sensitive to minor variations in moisture and particle size distribution, and the risk of batch-to-batch viscosity drift increases sharply.

Dissolution Enthalpy, Oral Cooling Onset, and Residual Erythritol Crystals

Erythritol is distinguished from other polyols by its strong endothermic dissolution event, which is reported in food-grade material as an enthalpy of solution near 180 J/g. This value means that when a crystal of erythritol contacts saliva, the local temperature at the tongue surface is reduced as the solid dissolves. Particle size distribution changes the time course of this cooling. A large crystal with a diameter above 30 µm dissolves slowly and creates a delayed cooling event, while a fine particle below 10 µm dissolves within a shorter oral residence time and may produce an early but less persistent cooling sensation. The perceived intensity of cooling is therefore not a function of total erythritol content alone; it is a function of the mass fraction of erythritol that can dissolve within the first 5 s of mastication. If the D90 is too low, the entire cooling event may occur before the consumer recognizes it, and the product may taste flat. If the D90 is too high, the cooling event arrives after swallowing, and the product may be described as gritty or sandy rather than cooling.

The dissolution kinetics are conventionally described by the Noyes-Whitney equation, in which the dissolution rate is proportional to the specific surface area and the concentration gradient. As the specific surface area is inversely related to particle size, reducing the erythritol D50 from 25 µm to 12 µm can increase the initial dissolution rate by a factor that scales approximately with the increase in surface area. However, the oral environment is not a well-stirred vessel; the saliva flow rate, the viscosity of the chocolate bolus, and the migration of fat to the particle surface all limit the effective dissolution rate. This is why a chocolate with the same D90 can show different cooling behavior depending on the emulsifier package. A higher lecithin concentration melts the fat phase more rapidly and delays water contact with erythritol, while a lower lecithin concentration allows the hydrophilic erythritol surface to be exposed to saliva more quickly. The result is a formulation-dependent shift in cooling onset that cannot be predicted from particle size alone.

On a production-scale moulding line, the particle size of erythritol also influences the viscosity at the tempering stage and the final demoulding shrinkage. Tempering units operate at specific cooling-water temperatures and retention times. A no sugar added chocolate with a high yield stress may not flow uniformly into the mould cavities, and the vibration cycles may need to be extended. The tempering temperatures for a no sugar added chocolate containing erythritol are often adjusted to account for the lower viscosity stability of the mass: the first cooling stage may be held near 27 °C, the reheating stage near 31 °C, and the final cooling tunnel at 12 °C to 14 °C. If the erythritol D90 exceeds 25 µm, solid particles can nucleate a portion of the cocoa butter in an unwanted polymorphic form, but this effect is highly dependent on the surface energy of the milled particles. Published data for this specific configuration are limited, so each formulation requires a polymorphic stability assessment by differential scanning calorimetry during scale-up. The operational boundary is that pre-drying of erythritol is required when the ambient relative humidity exceeds 60 %, and pneumatic conveyance systems should be designed with a dew point below −20 °C to prevent surface moisture adsorption and subsequent agglomeration.