The commercial manufacture of semi-moist pet food involves extrusion of a formulation containing cereal flour, meat and bone meal, humectants, salts, and plasticizing solutes to a terminal melt temperature of 110–140 °C, followed by die shaping, face cutting, and ambient or forced-air pre-drying to a target moisture content of 18–28 g/100 g and water activity of 0.65–0.85. The cut extrudate enters postextrusion coating not as an inert dry carrier but as a plasticized, thermodynamically unstable matrix in which low-molecular-weight solutes—glycerol, propylene glycol, sorbitol, sucrose, and sodium chloride—partition between the bulk and the surface boundary layer within minutes of die face cooling. Postextrusion coating operations therefore operate within a narrow window bounded on one side by surface tack sufficient to retain liquid fat and powder palatant, and on the other side by excessive moisture migration, clumping, and package adhesion. The coating sequence may include a liquid fat or hydrolysed protein digest precoat, a powdered palatant or mineral topcoat, and a final cooling or anti-caking dusting step, with batch or continuous rotary coaters, paddle mixers, and vacuum coaters used depending on throughput and porosity requirements. Water activity in the coated product is measured by ISO 18787:2017 or AOAC 978.18, moisture by AOAC 930.15, and surface free fat by extraction before and after coating. The coating window is not a single setpoint but a matrix of paired limits for product surface temperature, fat temperature, liquid viscosity, spray droplet size, drum speed, fill ratio, and post-coating cooling that must be re-established whenever formulation, ambient dew point, or coater scale changes.
The practical coating window for semi-moist extrudate is bracketed at the lower end by the temperature at which the applied lipid develops sufficient liquid fraction to pump and atomize, and at the upper end by the temperature at which the extrudate surface undergoes plasticization collapse. Fats used in pet food coating include chicken fat, beef tallow, palm stearin, and hydrogenated vegetable blends with solid fat content profiles that vary strongly between 20 °C and 40 °C; a typical chicken fat may have an iodine value of 70–85 g I₂/100 g and a complete melting range of 30–40 °C, whereas beef tallow may remain partially solid until 42–48 °C. Simultaneously, the surface of a semi-moist extrudate at 20% moisture and 35 °C behaves as a plasticized amorphous starch–protein matrix, with published differential scanning calorimetry data for similar formulations showing a broad glass transition from −5 °C to 35 °C depending on humectant ratio and thermal history. If a molten fat spray at 55–65 °C contacts a product surface whose local temperature exceeds the upper bound of that transition, the surface viscosity falls by several orders of magnitude, liquid fat penetrates the outermost pellet layers, and subsequent tumbling transfers the plasticized layer into contact with adjacent pellets, forming doubles and agglomerates that cannot be separated by cooling. For this reason, coating fats are ordinarily conditioned to 40–55 °C for chicken fat and 50–60 °C for tallow-containing blends, with the spray manifold held 5–10 °C above the fat reservoir to prevent line solidification while low product inlet temperature of 25–35 °C provides a sink for latent heat.
| Coating fat | Melting range | SFC at 20 °C | SFC at 40 °C | Viscosity at 60 °C | Reservoir set point | Spray manifold set point |
|---|---|---|---|---|---|---|
| Chicken fat | 30–40 °C | 8–18% | 0–3% | 40–60 mPa·s | 42–48 °C | 50–55 °C |
| Beef tallow | 42–48 °C | 20–35% | 6–12% | 50–75 mPa·s | 52–58 °C | 58–63 °C |
| Palm stearin blend | 38–52 °C | 45–65% | 15–25% | 65–90 mPa·s | 55–62 °C | 62–68 °C |
Solid fat content measured by pulsed NMR per AOCS Cd 16b-93 provides more useful prediction of pumpability than a single melting point because gear pumps and nozzle orifices experience shear-independent viscosity only when the liquid fraction exceeds roughly 85%. In production practice, rotary drum coaters of 2,000–5,000 kg batch size show a 5–12 °C temperature drop in uninsulated transfer lines, so the set points listed above are adjusted upward by exactly the measured line loss, which varies with ambient temperature and line length. Failure to compensate results in intermittent nozzle plugging at the 0.8–1.2 mm orifice, uneven fat distribution, and zones of uncoated product with reduced palatant adhesion. When the product surface is near the upper end of its glass transition, the fat must be applied as a dispersed spray rather than a stream; air-atomizing nozzles producing Dv50 droplet diameters of 200–500 μm permit rapid heat transfer to the cold extrudate and avoid local overheating, while hydraulic nozzles producing droplets above 600 μm can exceed the surface heat sink and melt the top layer of the bed.
Hydrolysed protein digests and liquid palatants for semi-moist pet food are delivered at 50–70% solids content with viscosities between 100 mPa·s and 1,000 mPa·s at 40–60 °C, requiring positive-displacement gear pumps, heated jacketed hoses, and full-cone or flat-fan nozzles rather than simple centrifugal transfer. The pressure drop across a 0.8 mm nozzle at 0.8–1.5 MPa generating a Dv50 droplet size of 200–500 μm is matched to batch coater geometry so that the expanding spray cone covers the moving product bed without coating the drum wall. If the viscosity exceeds 1,200 mPa·s or if protein particles larger than 600 μm survive the in-line strainer, nozzle pressure rises erratically and the spray pattern shifts from a continuous fan to pulsating jets that produce localized wet spots. On a production line using a 2,000 kg twin-shaft paddle coater with 12 air-atomizing nozzles, the liquid delivery rate is typically 12–30 kg/min for a 2–6% fat precoat, and the coating uniformity target is a coefficient of variation below 10% across 15 top, middle, and bottom product samples. Batch records show that variation doubles when the return loop temperature falls below 45 °C for chicken fat and below 55 °C for tallow, because the resulting increase in viscosity changes nozzle discharge coefficient and shifts droplet size toward 600 μm or greater. Shear heating in a gear pump operating at 10–40 rpm may add 1–3 °C to the liquid, which is advantageous for nozzle discharge but cannot compensate for a cold transfer line; therefore the return loop is maintained at 55–65 °C with steam tracing or electrical resistance heating and the in-line strainer is installed immediately before the manifold with a mesh size of 600–850 μm to remove digest particles without creating excessive backpressure.
Moisture migration during the first 20 min after coating is the dominant variable controlling agglomeration on semi-moist products. The wet extrudate leaving the dryer at 20–26% moisture and 30–38 °C possesses a water activity between 0.70 and 0.82, whereas the surrounding coating room often operates at 20–25 °C and 45–60% relative humidity, equivalent to an ambient water activity of 0.45–0.60. The resulting water activity gradient drives evaporative cooling of the product surface and simultaneous moisture loss of 0.1–0.5 g/100 g during the coating cycle. A fat precoat applied at 40–55 °C functions as a partial moisture barrier only after the latent heat of crystallization has dissipated; during the first 3–8 min, the film remains liquid enough to allow water vapour and low-molecular-weight humectants to migrate to the surface. If the coater discharge is transferred directly to a bulk bin or packaging hopper without forced-air cooling, condensation on the bin wall and inter-pellet capillary bridging increase the number of fused clusters by a factor of three to five compared with product that passes through a 15–25 °C fluid-bed cooler for 4–10 min. Surface moisture also determines the adhesion of a subsequent powder coating; when surface free moisture exceeds 0.8 g/100 g as measured by filter-paper contact or Karl Fischer titration of the surface scrapings, fine palatant powders dissolve into a gummy layer rather than adhering as discrete particles.
When postextrusion coating is performed on a semi-moist product with a water activity above 0.85 or in an unventilated coater, the transient moisture flux can exceed 0.5 g/100 g·h, at which point conventional atmospheric rotary coating becomes unstable. The product surface remains plasticized and tacky, fat absorption is uneven, and the tumbling action promotes transfer of a sticky humectant film from the pellet exterior to the drum baffles and spray nozzle guard. Under these conditions the lower temperature boundary of the coating window must be raised to keep the surface above the glass transition but below the melting point of the least stable fat fraction, a range that can be as narrow as 5–10 °C for a product containing 20% glycerol and 8% propylene glycol. Dehumidified air at 15–20 °C and 30–40% relative humidity is introduced into the coater discharge chute at 15–25 m/s face velocity to strip the boundary layer and reduce the package dew point below 12 °C. Production trials on a continuous rotary drum with a 1,500 kg/h feed rate show that an increase in product surface temperature from 32 °C to 38 °C reduces the melt viscosity of the fat film at the pellet surface, but an increase to 42 °C in the same formulation causes clumping even when the average fat application is held constant at 4.0% by mass. These observations indicate that the coating window is not controlled by average product temperature alone but by the localized surface temperature during the first 60–90 s after the spray contact.
| Coater configuration | Batch size | Product inlet temperature | Fat/palatant temperature | Vacuum level | Residence time | Maximum surface moisture change | Typical liquid uptake |
|---|---|---|---|---|---|---|---|
| Atmospheric rotary drum | 2,000–6,000 kg | 28–35 °C | 40–55 °C | Atmospheric | 3–8 min | −0.3 to +0.4 g/100 g | 2–6% |
| Twin-shaft paddle coater | 500–2,000 kg | 30–38 °C | 42–58 °C | Atmospheric | 2–6 min | −0.2 to +0.3 g/100 g | 3–8% |
| Vacuum coater | 1,000–3,000 kg | 32–42 °C | 45–60 °C | 50–80 mbar absolute | 5–12 min | −0.5 to +0.2 g/100 g | 6–15% |
The upper moisture flux boundary is particularly sensitive to fill ratio in rotary equipment. A fill ratio of 25–35% of the drum volume provides sufficient bed motion to expose new surfaces to the spray while allowing the product to shed moisture to the vapour space; fill ratios above 40% reduce the effective headspace and increase the equilibrium relative humidity inside the drum, which slows evaporative cooling and narrows the coating window by an additional 2–4 °C. The drum speed is normally set to 6–12% of critical speed, or 8–14 rpm for a 2.4 m diameter drum, so that the product bed rolls rather than centrifuges. With high-humectant semi-moist formulations, the spray interval is divided into 3–5 pulses separated by 30–60 s of tumbling, a technique that allows the surface to cool between spray additions and prevents cumulative heat build-up from exceeding the plasticization limit.
Powdered palatants, dried liver digest, yeast cell wall fractions, and mineral premixes applied after a liquid fat precoat require a coherent lipid film of 2–5 μm average thickness on the pellet surface. The liquid fat precoat is typically limited to 1–3% by mass for a product that will receive 1–2% powder, because excess free fat creates a slick surface from which powder particles detach during packaging and transport. Sieve analysis of the powder stream is controlled by ISO 3310-1:2016; the fraction passing 75 μm may constitute up to 50%, but particles below 20 μm create dusting losses and particles above 150 μm are not retained by the lipid film under normal tumbling. The powder addition is made 30–60 s after liquid spray shutdown in a twin-shaft paddle coater, when the product surface temperature has fallen to 30–35 °C and the fat film is partially solidified. In this interval the contact angle of the molten fat against the product surface decreases with increasing surface moisture; at 0.75–0.80 water activity the fat spreads into the surface roughness but does not penetrate deeply, whereas at 0.85 and above the film migrates into the pellet interior and powder adhesion falls below 80% of the applied mass after one week of packaged storage. Production-scale batch records for a 5,000 kg rotary drum show that the powder retention after a 4.0% chicken fat precoat and 1.5% dried digest topcoat is 90–95% when the fat reservoir is held at 46 °C and the product enters at 30 °C, whereas retention drops to 70–80% when the product enters at 25 °C because the fat crystallizes before the powder contacts the surface. The incompatibility of highly hygroscopic powder palatants with high relative humidity in the coating room should be recognised; at relative humidity above 65%, dried digest particles absorb moisture from the air during storage in the hopper, bridge, and discharge as clumps that exceed 150 μm even when the original sieve profile was within specification.
Vacuum coating intensifies liquid penetration into porous extrudate by removing occluded air and compressing the gas phase within the pellet, but in semi-moist products the high water content occupies much of the available pore volume, so the attainable uptake is lower than in dry expanded kibble. In a vertical vacuum coater operating at 50–80 mbar absolute, the product at 32–38 °C and 18–22% moisture is charged, liquid fat or digest is sprayed under vacuum, and the vessel is restored to atmospheric pressure over 2–5 min so that the pressure differential drives the liquid into surface fissures and cut faces. The expected total liquid uptake is 6–15% by mass, compared with 2–6% for atmospheric coating of the same formulation. Surface free fat after vacuum coating is frequently 1.5–3.0% lower than after atmospheric coating at the same total fat content, because the liquid is forced below the outer surface rather than remaining as a sticky external film. However, the rapid vacuum drop can flash water from the product, reducing the moisture content by 0.2–0.8 g/100 g and lowering surface water activity by 0.02–0.05 units; this moisture loss is beneficial for flowability but may shift the texture of a semi-moist product toward a chewier, firmer bite. Published data for very high-humectant semi-moist matrices under vacuum is limited, so direct scale-up from dry kibble coating is not recommended without pilot trials at the target vacuum ramp rate and fill level.
When vacuum coating is applied to a semi-moist extrudate with internal gas cells from die expansion, the pressure differential must be released stepwise over 3–5 min to avoid explosive bubble expansion at the surface. A release rate exceeding 150 mbar/min has been observed on pilot equipment to create surface blisters and local delamination of the outer layer, which then absorb fat unevenly and release the fat as surface smearing during packaging. For this reason, a programmable vacuum valve with three discrete hold steps at 400 mbar, 200 mbar, and 80 mbar is preferred over a single-step evacuation. The product fill level in vacuum coating is usually 40–55% of vessel volume to provide adequate bed movement during the liquid injection phase, and the liquid is sprayed through a centrally mounted rotating manifold to avoid droplet expulsion during vacuum release.
The high water activity of semi-moist pet food does not arrest lipid oxidation because the surface fat phase is continuous and contains dissolved oxygen from the coating process. Postextrusion coating adds 2–8% fat by mass, increasing the total lipid surface area available for oxidation and the concentration of pro-oxidant iron from meat-based ingredients. Antioxidant strategies include adding 200–500 mg/kg mixed tocopherols or 0.02–0.05% rosemary extract to the coating fat, with chelation by citric acid at 0.1–0.3% where permitted. Package oxygen transmission rate should be below 30 cm³/m²·day at 23 °C and 50% relative humidity when measured by ASTM D3985-17, but for high-fat semi-moist products the practical limit is often 10–20 cm³/m²·day because the product is packaged at 0.65–0.80 water activity and the film becomes partially hydrated. The fatty acid profile of chicken fat, tallow, and palm stearin blends should be monitored by ISO 12966-4:2015 and free fatty acid content by AOCS Ca 5a-40; coatings with free fatty acid above 1.5% are associated with poor palatant adhesion and increased off-note formation during shelf life. Batch-to-batch variance in peroxide value of incoming coating fat above 2 meq O₂/kg correlates with a 30–50% reduction in oxidative induction time, so the coating fat should be sampled at the tote and in-line after the heated filter to confirm that no oxidation occurs during hold time.
The product surface after fat coating is not static; oxidative reactions are accelerated by the intimate contact between the lipid phase and the aqueous phase containing dissolved iron and copper. For this reason, metal chelators are added to the liquid palatant rather than to the dry formulation, ensuring that the chelator remains at the exact lipid–water interface where oxidation initiates. The use of ethoxyquin in pet food is governed by national feed additive registers and may be subject to residue maxima; where used, it is typically added to the coating fat at the lowest effective concentration and verified in the finished product because the surface concentration can be 2–4 times the bulk average. In formulations that avoid synthetic antioxidants, a combination of mixed tocopherols and rosemary extract at the upper end of the stated range is required to achieve comparable induction time, and the package oxygen barrier must be verified under the actual product water activity rather than dried-film conditions.
Surface application of acidulants and mould inhibitors to semi-moist pet food is constrained by the water activity reduction and the legal maximum residues in the finished product. Potassium sorbate applied as a 5–10% aqueous solution at 0.05–0.2% of product mass reduces the effective surface pH and delays mould growth when the product water activity is between 0.70 and 0.85, but the same solution can cause local swelling of the starch–protein surface if its water content increases the boundary-layer moisture above 2%. Lactic acid and phosphoric acid are used at 0.1–0.5% to reduce surface pH below 4.5; because they are corrosive to mild steel, the spray system must be constructed of 316L stainless steel with PTFE seals, and the line must be rinsed within 2 h to prevent corrosion pitting. In the United States, potassium sorbate is listed as GRAS for food use in 21 CFR 182.3640; for animal feed, AAFCO ingredient definitions and state feed laws govern quantitative use. For European production, the substance must be registered under REACH and its use in feed materials is subject to Regulation (EC) No 1831/2003; the maximum allowable residue in the final feed must be verified by batch analysis rather than assumed from the addition rate.
Because acidulant solutions carry water, they shift the local surface water activity upward immediately after spraying, and the coating window narrows until the applied water is absorbed and redistributed. A 0.2% potassium sorbate spray at 10% concentration introduces 1.8 kg of water per 1,000 kg product, which is sufficient to raise the average moisture by 0.18 g/100 g but can raise the surface layer by 1–2 g/100 g before equilibration. The product therefore must be held in a ventilated finishing conveyor for at least 5–10 min after acidulant application to allow the surface to return to the target water activity before packaging. If the product is packed immediately, the package headspace humidity increases and condensation occurs on the film, which can dissolve sorbate crystals and create a local pH zone where mould inhibition is lost despite a compliant bulk average.
On lines running dual-belt coolers and drum coaters in series, the coating window shifts by 3–6 °C for every 1% change in residual moisture. The heat load imposed by a 4% fat addition at 50 °C is sufficient to raise the bulk temperature by 1–3 K if no forced-air cooling is applied, and this temperature rise is larger when the fat is applied in a single pulse than when it is split into two or three spray intervals. A change in ambient relative humidity from 45% to 65% reduces evaporative cooling and can push a formulation that was stable at 38 °C surface temperature into agglomeration at the same mechanical settings. The use of a dew-point transmitter in the coater exhaust and an infrared pyrometer aimed at the tumbling bed provides feedback, but the response loop is slower than the residence time of the product in the drum; therefore the setpoints are adjusted manually or by feed-forward control based on the upstream dryer moisture and ambient conditions rather than by closed-loop temperature control alone. Coater sanitation is a further practical limit for semi-moist lines because the high sugar and humectant content of the dust adhering to drum baffles and spray nozzles supports rapid microbial growth if the line is not dry-cleaned and sanitised between flavour changeovers; this boundary is particularly strict when the next production run is a dry expanded product, because residual sugars and fats can transfer to the dry product and alter its surface free fat and water activity beyond the declared specification.