Meso-erythritol (CAS 149-32-6, molecular mass 122.12 g/mol) crystallises from both melt and aqueous solution in continuous cooling tunnels and on steel-belt coolers. The compound exhibits a melting point of 121 °C and a latent heat of fusion between 330 J/g and 340 J/g, which makes the solidification step energy-intensive and sensitive to undercooling. At the discharge of a cooling tunnel, insufficient nucleation produces a heterogeneous stream of glassy undercooled domains, sticky pastilles, and belt fouling, particularly when the melt film temperature falls below 45 °C before the crystalline fraction develops. The problem is amplified at belt speeds above 0.8 m/min and film thicknesses below 3 mm, where the residence time available for nucleation and growth is shortest. Because erythritol intended for food applications must meet the identity and purity provisions of 21 CFR 184.1444, and because phase-change composites are qualified by thermal cycling methods such as ASTM D3418-15, the acceleration of crystal nucleation at the cooling tunnel discharge must be accomplished without leaving residual additives, without broad particle size excursions, and without introducing metallic wear debris that would require post-treatment. The central process conflict is that rapid nucleation reduces liquid carryover but tends to generate fines, while slow nucleation improves crystal habit but risks amorphous discharge and unsafe belt release. Nucleation acceleration therefore requires controlled manipulation of seeding, surface roughness, thermal profile, and mechanical energy input at the point where the process stream approaches the discharge knife.
Classical nucleation theory describes the steady-state primary nucleation rate as an exponential function of the free-energy barrier for cluster formation. For homogeneous nucleation, the barrier depends on interfacial tension, molar volume, temperature, and the natural logarithm of supersaturation; for heterogeneous nucleation, the effective interfacial energy is reduced by a contact-angle function that accounts for the presence of a foreign solid. In erythritol melt crystallization, the driving force is expressed as undercooling relative to the melting point, and the critical radius is inversely proportional to that undercooling. Industrial experience on single-belt coolers indicates that undercooling of 40 K to 60 K can be required for spontaneous nucleation on electropolished stainless steel, whereas a properly roughened belt or an engineered seed population reduces the required undercooling to 5–15 K. The induction time, defined as the elapsed time from the onset of undercooling to the detection of a measurable crystalline fraction, must remain shorter than the belt residence time from the feed spreader to the discharge knife. On a cooling tunnel with 8–15 m active cooling length and belt speed of 0.5–1.8 m/min, the available residence time is 4–30 min. If the nucleation induction time exceeds this window, the discharge contains undercooled liquid that cannot be scraped cleanly. Methods that intensify primary nucleation in this context include addition of micronised erythritol seed, controlled surface cavitation, ultrasonic irradiation, air-knife evaporative chilling, and stepwise cooling water zoning. The selection of a method is constrained by thermal degradation risk above 140 °C, the food-grade requirement under 21 CFR 184.1444, and the mechanical limits of the belt material and discharge knife.
In continuous steel-belt cooling systems used for pastille or flake production, erythritol seed crystals with a median particle size of 5–25 µm are introduced at the feed spreader or through an air-assisted lance positioned immediately upstream of the cooling zone. A seed loading of 1–5 wt% on a dry-solids basis reduces the observed nucleation induction time at a belt-side melt temperature of 60–80 °C from several minutes to less than 30 s, depending on particle size distribution and dispersion uniformity. Seeds with D50 below 10 µm provide a high specific surface area, commonly above 2 m²/g when measured by nitrogen adsorption, but require deagglomeration through a high-shear mixer or an eductor using compressed air at 0.3–0.6 MPa to prevent cluster formation. Seeds with D50 between 25 µm and 75 µm disperse more easily but contribute fewer nucleation sites per unit mass, requiring higher loading to achieve the same discharge solids content. The seed addition point must be positioned before the melt film cools below the equilibrium melting point but after the feed temperature falls below the seed dissolution temperature. For erythritol, this window is typically 100–125 °C at atmospheric pressure. Addition above 121 °C causes partial dissolution of the seed crystals, while addition below 60 °C does not provide sufficient time for nuclei to grow before the discharge knife. The resulting crystal size distribution at discharge is often bimodal when seeding is not followed by a hold period, with a primary mode from seed-induced nucleation and a secondary mode from spontaneous nucleation in the remaining liquid film. The distribution is measured by laser diffraction according to ISO 13320:2020 and reported as span, defined as (D90−D10)/D50. A span below 1.2 is achievable only when the cooling water spray is divided into zones with inlet temperatures of 15–30 °C in the first zone and 5–10 °C in the final zone, maintaining a low supersaturation in the film after the initial nucleation burst.
| Measurement | Standard or regulatory reference | Equipment type | Typical erythritol discharge range |
|---|---|---|---|
| Melting point and enthalpy of fusion | ASTM D3418-15 | Differential scanning calorimeter | 118–121 °C, 300–340 J/g |
| Crystal size distribution | ISO 13320:2020 | Laser diffraction analyser | D50 150–300 µm, span 0.9–1.5 |
| Moisture content | ISO 760:1978 | Karl Fischer titrator | 0.05–0.5 wt% |
| Belt surface roughness | ISO 4287:1997 | Stylus profilometer | Ra 0.8–1.2 µm |
| Discharge crystallinity | 21 CFR 184.1444 | Release knife inspection | Fully solid flakes, no liquid carryover |
After a scraped-surface heat exchanger discharges a supersaturated aqueous erythritol solution into a cooling tunnel, the dominant nucleation mechanism shifts from melt undercooling to solution supersaturation. The supersaturation ratio is defined as the ratio of actual concentration to saturation concentration at the discharge temperature. Erythritol solubility in water rises steeply with temperature; published data indicate approximately 370 g erythritol per 1000 g water at 20 °C and approximately 750 g per 1000 g water at 100 °C. When a solution saturated at elevated temperature is cooled rapidly, the discharge liquor can remain within a metastable zone width of 10–25 K before primary nucleation occurs, depending on cooling rate and impurity profile. Secondary nucleation at the discharge is enhanced by the mechanical action of the scraped-surface rotor, which generates nuclei through attrition of existing crystals and by contact of the rotor blades with the cooled wall. The rotor speed is typically 100–300 rpm, and the tip speed is 1–5 m/s; above this range, excessive fines are generated and the crystal habit becomes irregular. A downstream holding pipe with residence time of 30–120 s allows the generated nuclei to grow to a size retainable on a 200 µm sieve, reducing the load on drying and milling. The discharge temperature is commonly maintained at 25–35 °C with jacket coolant at 10–15 °C, and the outlet concentration is held between 45 wt% and 55 wt% erythritol. Under these conditions, the crystal mass fraction at the discharge is typically 10–25 wt%, and the crystal size distribution has a median diameter of 150–300 µm measured by ISO 13320:2020. The principal operational risks are encrustation on the cooled wall, which reduces the effective heat transfer coefficient from 800 W/(m²·K) to below 300 W/(m²·K), and product loss through amorphous carryover if the discharge temperature is not sufficiently below the saturation temperature. The scraped-surface unit must therefore be operated with a controlled wall temperature difference and a rotor speed that balances nucleation against crystal breakage.
| Parameter | Typical range | Effect on discharge nucleation | Limiting condition |
|---|---|---|---|
| Melt feed temperature | 125–135 °C | Avoids seed dissolution | Above 140 °C can cause yellowing |
| Belt speed | 0.5–1.8 m/min | Sets residence time | Above 1.8 m/min yields liquid carryover |
| First-zone cooling water temperature | 15–30 °C | Slow initial cooling reduces spontaneous burst | Below 10 °C may shock-cool and create fines |
| Final-zone cooling water temperature | 5–10 °C | Completes solidification | Below 3 °C risks condensation and frost |
| Seed loading | 1–5 wt% | Accelerates heterogeneous nucleation | Above 5 wt% may increase fines and viscosity |
| Seed median size | 5–25 µm | Controls nucleation site density | Above 75 µm reduces site count |
Belt adhesion and amorphous carryover at the discharge knife are the primary mechanical failure modes observed on production-scale single-belt coolers with stainless steel conveyor belts of 0.8–2.0 m width and 10–25 m cooling length. The discharge knife is usually set at a clearance of 0.5–2.0 mm from the belt surface; when nucleation is incomplete, a viscous undercooled layer adheres to the knife and accumulates on the return strand, requiring manual cleaning or high-pressure water jets operating at 5–20 MPa. A food-grade release agent may be employed, but its use is constrained by the requirement that the final erythritol product comply with 21 CFR 184.1444 and with a maximum residual moisture of 0.1 wt% when tested by ISO 760:1978 Karl Fischer titration. The recommended belt surface arithmetic mean roughness for pastille release is generally 0.8–1.2 µm. Polished surfaces with Ra below 0.2 µm delay nucleation, while surfaces above 2.5 µm mechanically entrap product and increase cleaning downtime. The cooling water spray system must deliver uniform heat flux; water film temperature variation across the belt width of more than 3 K produces side-to-side differences in nucleation induction time, causing curled flakes and discharge blockage. Air knives positioned before the discharge point, supplied with filtered air at 20–30 °C and velocities of 10–30 m/s, can accelerate surface crystallization by evaporative cooling, but the compressed air must meet ISO 8573-1:2010 Class 2 to avoid contamination. Published data for erythritol-specific belt adhesion at the exact combinations of belt speed and layer thickness used in production are limited; pilot trials on a belt cooler with minimum 4 m cooling length are required before transfer to full scale.
Sonication at 20 kHz to 40 kHz with an acoustic power density between 5 W/L and 50 W/L has been applied to organic melts and solutions to reduce induction time through cavitation-induced local pressure fluctuations and heterogeneous nucleation on imploding bubble surfaces. In erythritol melts, the viscosity near the melting point is reported in the range of 20–60 mPa·s at 100–120 °C, which attenuates the ultrasonic field and restricts the effective cavitation zone to a radial distance of 20–50 mm from a horn tip. The sonotrode must therefore be located in a flow cell or a narrow discharge channel rather than in the bulk melt on the belt, with melt residence time in the sonication zone of 5–30 s. An industrial ultrasonic processor rated at 1 kW with a titanium horn of 25 mm diameter can treat a melt stream of 50–200 kg/h under continuous flow, but the horn surface is subject to cavitation erosion and metallic contamination. Hard chromium or diamond-like carbon coatings are required if the product is intended for food use under 21 CFR 184.1444. The acoustic power is usually reported as the average electrical power delivered to the transducer, not as the actual power dissipated in the melt; process transfer therefore requires calorimetric calibration using a jacketed vessel and a thermal method such as ASTM D3418-15. For sugar alcohols of similar molecular mass, published laboratory studies report induction time reductions of 40–70% with 20 kHz ultrasound at 30 W acoustic power; erythritol-specific published data for continuous belt cooler configurations is limited. The ultrasonic treatment must not raise the melt film temperature by more than 5 K, because excess heat reduces the undercooling and offsets the nucleation benefit. When applied to an already partially crystallized slurry, ultrasonication can increase the fines fraction below 63 µm through particle attrition; therefore, the sonication point should be located upstream of the belt discharge and downstream of any scraped-surface device while the stream is still fully liquid.
Stainless steel surfaces with an arithmetic mean roughness of 1.0–1.5 µm promote heterogeneous nucleation by lowering the contact angle of the solidifying phase and by providing re-entrant cavities that stabilise critical nuclei. On a continuous cooling belt, the surface topography is usually generated by glass bead peening or by a longitudinal brushed finish, and the resulting Ra value is measured with a stylus profilometer according to ISO 4287:1997. A surface with Ra below 0.2 µm behaves as a near-smooth substrate and delays the onset of nucleation, whereas a surface above 2.5 µm can cause mechanical interlocking of the crystalline layer and reduce the cleaning efficiency of the discharge knife. The optimum roughness for erythritol melt solidification is therefore specific to the belt material, layer thickness, and cooling rate. For a 1.2 m wide stainless steel belt with a 4 mm thick melt film cooled from 121 °C to 30 °C in 6–10 min, a roughness of 0.8–1.2 µm is typically specified. The nucleation density on a roughened surface is higher by approximately one to two orders of magnitude than on a polished surface under the same undercooling; this value is derived from classical heterogeneous nucleation theory and from analogous polyol melt studies, while published data for erythritol-specific surface nucleation density on stainless steel belts are limited. The belt surface must be passivated before first use according to ASTM A967-17, because residual iron oxide and weld-scale can act as uncontrolled nucleation sites and produce black specks in the final product. The discharge knife itself should be made of a material with lower surface energy than crystallized erythritol, such as polytetrafluoroethylene or hard-coated stainless steel, to reduce adhesive failure. The knife angle is maintained between 20° and 35° from the belt plane to shave the solidified layer without crushing large crystals or spreading undercooled liquid.
Process analytical technology deployed at the discharge of an erythritol cooling tunnel includes focused beam reflectance measurement for chord length distribution and an inline near-infrared probe calibrated against ISO 13320:2020 laser diffraction. The focused beam reflectance probe is inserted into a slip stream drawn from the discharge hopper at a flow rate of 0.2–1.0 L/min and provides a real-time particle size distribution trend. The near-infrared probe predicts moisture content in the range 0.05–0.5 wt% with a standard error of calibration of 0.02 wt% when validated against ISO 760:1978 Karl Fischer titration. These measurements allow closed-loop control of seed addition and belt speed, with the seed feeder adjusted in increments of 0.1 wt% to maintain discharge crystallinity above 95% as determined by ASTM D3418-15 differential scanning calorimetry. The control strategy is bounded by the maximum belt speed at which the cooling tunnel can deliver sufficient heat flux; above 1.8 m/min, the available cooling length is insufficient for complete solidification even with seed addition, and the discharge fraction becomes sticky. The lower bound is set by economic throughput and by the risk of excessive crystal growth that produces flakes thicker than 10 mm, which fracture and create fines. The use of a nucleation accelerator must be integrated with the upstream evaporator and the downstream milling and sieving circuit to avoid shifting the particle size distribution outside the specification for direct compressible tableting or for phase-change composite encapsulation.