Extrusion Conditions and Hydrophilic Organic Acid Partitioning in Aquafeed

In co-rotating twin-screw extrusion of aquafeed, the addition of hydrophilic organic acids to preconditioned meal is not a dry-mix operation but a partition event distributed across steam, melt water, lipid traces, mineral carbonates, and protein-starch interfaces. Production-scale extruders for salmonid and penaeid feeds typically operate with screw diameter 45 mm to 70 mm, L/D 25:1 to 35:1, preconditioner moisture 18–25 g/100 g, barrel set points from 80 °C at the feed throat to 120–145 °C before the die, and die pressure 10–28 bar. Formic, acetic, propionic, lactic, and citric acids are added at 0.2–2.0 g/100 g for pH suppression and preservative activity, but their hydrophilic character does not automatically ensure retention in the extrudate. The undissociated fraction of a weak acid in the aqueous melt phase is governed by the Henderson-Hasselbalch relation: at melt pH values below the acid pKa minus 1.5, more than 970 g/kg of the acid exists in the neutral form, and the neutral form partitions into steam according to the vapor-liquid equilibrium of the acid-water system. Because aquafeed extrusion uses superheated water in the barrel and flash evaporation at the die, the neutral acid is removed in the same mass-transfer path as water vapor; this loss competes with retention through acid-base interaction with calcium carbonate, protonation of amine groups in fishmeal protein, and hydrogen bonding with starch. The resulting distribution of acid between pellet matrix, surface, and vented vapor is therefore an intrinsic function of extrusion water activity, temperature profile, screw fill, die pressure drop, and premix buffering capacity. Published data for this specific configuration is limited, but comparative data from feed and food extrusion indicate that retention can be as low as 60 g/100 g for formic acid when liquid acid is injected directly into a steam preconditioner at 85–95 °C, whereas citrate and lactate show higher retention because of lower vapor pressures and polycarboxylate interaction with minerals.

Why Do Hydrophilic Volatile Acids Follow the Vapor Fraction During Preconditioning?

The steam preconditioner is the first high-moisture thermal zone in which partitioning begins. A typical counterflow or co-current preconditioner operates at 85–95 °C, residence time 90–180 s, and moisture 18–25 g/100 g. At these conditions, the neutral fraction of a carboxylic acid follows steam volatility rather than aqueous solubility. Formic acid has pKa 3.75, acetic acid 4.76, propionic acid 4.87, lactic acid 3.86, and citric acid pKa1 3.13. If preconditioned meal pH after acid addition is 5.0, the undissociated fraction is 5.3 g/100 g for formic acid, 36.5 g/100 g for acetic acid, 42.6 g/100 g for propionic acid, and 6.8 g/100 g for lactic acid. Propionic acid therefore retains a significant neutral fraction and is readily steam-distilled even though its atmospheric boiling point is 141.2 °C, because water co-distillation lowers the vaporization temperature of partially miscible volatile acid-water systems. The vent from the preconditioner becomes a known loss point. Retention drops sharply when preconditioner temperature exceeds 90 °C and when injection occurs before rather than after steam injection; post-steam injection at the preconditioner discharge reduces vapor contact time and can increase retention by 10–20 percentage points compared with pre-steam injection. This is an operational control rather than a chemical modification, and it is applied when the target residual acid concentration in the final pellet must remain above 0.1 g/100 g for preservative activity under tropical storage conditions.

Barrel retention of propionic and sorbic acids is not adequately described by preconditioner loss alone; the screw pressure profile imposes a second partition stage in which the acid partitions between the compressed melt water and the vapor bubbles that form in partially filled screw elements. In a co-rotating twin-screw profile with kneading blocks at 40–60 % of barrel length and a compression ratio of 3.0:1 before the die, the local pressure can exceed 15 bar in fully filled sections and drop below 2 bar in vented sections. The neutral acid fraction transfers to the vapor phase in low-pressure zones and can recondense on cooler meal surfaces when the screw temperature drops. This internal reflux is governed by the acid Henry coefficient and the vapor-liquid surface area generated by screw elements. The net effect is that formic acid loss is accelerated by open-barrel venting, while lactic acid and citric acid remain largely in the aqueous melt phase. The difference is measurable by comparing acid recovery in pellets produced with atmospheric venting versus sealed-barrel operation; sealed-barrel operation can increase formic acid retention by approximately 8–15 percentage points when the die pressure is maintained above 12 bar, but it also raises the risk of acid-catalyzed starch hydrolysis if residence time exceeds 120 s.

Thermal Degradation and Steam Distillation as Competing Retention Mechanisms

Thermal degradation of hydrophilic organic acids in the extruder barrel cannot be separated from steam distillation, because both processes respond to the same temperature and water activity variables. Formic acid can decompose to carbon monoxide and water under strong acid catalysis, but in neutral-to-weakly acid aquafeed melts the dominant loss route is evaporation. Acetic acid has a boiling point of 117.9 °C and is steam-volatile; its losses in the barrel increase when the melt temperature exceeds 110 °C and when the moisture content remains above 20 g/100 g. Propionic acid has higher boiling point but a lower Henry coefficient sensitivity; it often shows retention losses that track die flash rather than barrel residence. Lactic acid can undergo intermolecular esterification to lactoyl lactate and minor lactide at barrel temperatures above 130 °C, especially under acidic conditions, and the resulting esters are less water-soluble and may partition into the lipid fraction or remain bound to the starch matrix. Citric acid is the least volatile of the common aquafeed acidulants, but its thermal sensitivity above 150 °C can produce aconitic acid, itaconic anhydride, and carbon dioxide through dehydration-decarboxylation pathways. These degradation products alter the acid value of the extrudate and interfere with ion-exclusion chromatography because they co-elute with residual citric acid when a simple refractive index detector is used. The process boundary for acid retention is therefore not a single barrel set point but a matrix of barrel temperature, moisture, and acid type. When barrel temperatures exceed 140 °C, formic and acetic acids should be replaced with calcium or sodium salts if final pellet acid content must exceed 80 g/100 g of the added dose; otherwise the unrecovered acid leaves through the vent and die flash. Published data for this specific configuration is limited, but the general behavior is consistent with vapor pressure data for dilute aqueous acid solutions.

In post-extrusion vacuum coating of salmonid feeds containing added lipids, hydrophilic organic acids may be injected as aqueous solutions into the vacuum coater after acid-neutralizing mineral addition. The vacuum coater operates at 600–800 mbar absolute pressure and 40–60 °C, with a liquid addition manifold that atomizes the acid solution onto tumbling pellets. Because the pellets are already formed, acid partitioning in this operation is mostly a surface adsorption and absorption event rather than a vapor-loss event. The hydrophilic acid solution does not penetrate the lipid coating completely; instead it distributes between the surface water film, the outer starch-protein matrix, and the lipid film. The partition coefficient between the aqueous phase and the lipid film is low for lactic acid and citric acid because of their negative log Kow values, but sorbic acid and benzoic acid show higher lipid affinity and can migrate into the oil film. This creates a concentration gradient in the pellet: the core remains lower in acid than the surface, and the surface acid is available for immediate dissolution when the pellet enters water. In shrimp feeds, this surface acid can reduce the pH of the pellet-water interface, which may affect attractability and water stability. The vacuum coater therefore provides a separate partitioning control point, but it is not a substitute for acid retention during extrusion when the target is uniform acid distribution throughout the pellet matrix.

When Calcium Carbonate Buffering Neutralizes Feed-Grade Organic Acids Before Die Expansion

The interaction between hydrophilic organic acids and calcium carbonate in aquafeed creates a process conflict at the die. Calcium carbonate is used at 0.5–3.0 g/100 g in many shrimp and tilapia feeds as a calcium source and crumble hardness modifier. When formic, propionic, or citric acid is added to the preconditioner, the acid reacts with calcium carbonate to release carbon dioxide, water, and the corresponding calcium salt. This reaction is rapid above 70 °C and is essentially complete when the melt reaches the die if the acid-to-carbonate stoichiometry is below the buffering capacity of the premix. The carbon dioxide generated inside the melt becomes a secondary blowing agent in addition to flash steam. At die pressure 12–22 bar, the carbon dioxide remains dissolved or compressed in the melt; when the melt exits the die, the pressure drop releases carbon dioxide and steam simultaneously, expanding the pellet. The expansion can be erratic if the carbonate reaction is incomplete at the die, because gas formation continues after the pellet exits the die and produces surface fissures. Die pressure fluctuations of ±0.4 MPa are observed in production lines when acid addition exceeds 1.0 g/100 g and the formulation contains 2.0 g/100 g calcium carbonate, because the gas phase in the melt disrupts the pressure profile. The pellet density can shift from a target of 1.05–1.10 g/cm³ for sinking shrimp feed to below 0.95 g/cm³, causing buoyancy failures in water stability tests. The acid buffering also raises melt pH after the carbonate reaction: the pH measured in a cooled extrudate sample may be 0.5–1.0 unit higher than the pH calculated from acid added to the dry mix, and this neutralization reduces the preservative activity of the acid. The process resolution is to add calcium carbonate in the preconditioner after the acid has been adsorbed onto the meal, or to use acid salts that do not generate free hydrogen ions until the pellet is rehydrated. The carbon dioxide path is also influenced by screw speed: at screw speeds above 400 rpm on a 45 mm extruder, the residence time in the final barrel section is too short for complete degassing, and the die pressure becomes more variable than at 250–350 rpm. The acid-mineral reaction therefore creates a narrow processing window: a die melt temperature of 110–125 °C, a preconditioner moisture of 22–25 g/100 g, and a screw speed of 250–350 rpm are often needed to prevent pellet collapse and acid loss simultaneously.

Quantification of residual formic, acetic, propionic, lactic, and citric acids in extruded aquafeed requires aqueous extraction followed by high-performance liquid chromatography with ion-exclusion or reversed-phase separation. A representative extraction uses 10 g of ground pellet in 100 mL of 0.1 mol/L phosphoric acid, shaken for 60 min at 25 °C, filtered through a 0.45 µm membrane, and injected onto a sulfonated polystyrene column with a refractive index detector. Recoveries for lactic acid and citric acid in high-starch matrices can be 85–95 g/100 g, but recoveries for propionic acid and sorbic acid in lipid-coated pellets may be lower because of oil-phase partitioning. The analysis is not trivial after extrusion because acid esters, lactides, and degradation products can co-elute. Calibration curves are prepared from certified reference materials with purities above 99 g/100 g, and matrix spikes are used to correct for starch and protein adsorption. Sampling of pellets for acid analysis should follow the procedures in ISO 6497:2002 for animal feeding stuffs, and moisture determination should follow ISO 6496:1999 so that acid concentration is expressed on a dry matter basis. The analytical result is used to calculate retention efficiency: the mass of acid recovered in the pellet divided by the mass of acid added to the preconditioner or extruder, expressed as a percentage. Without such mass balance, the term “partitioning” cannot be verified, because losses to vent, degradation, and analytical interference cannot be distinguished. Process audits on production lines should therefore include acid recovery as a routine quality parameter, not as an offline research measurement.

Moisture, Specific Mechanical Energy, and Die Geometry Effects on Acid Retention

Extrusion moisture directly changes the steam partial pressure and the vapor-liquid partition of volatile acids. At preconditioner moisture below 20 g/100 g, the vapor-phase water fraction is lower, so acid volatilization can be reduced; however, the melt viscosity rises, specific mechanical energy increases, and the barrel temperature must be raised to maintain starch gelatinization. At moisture above 25 g/100 g, the plasticization is more uniform but steam distillation increases, especially for acetic and propionic acids. Specific mechanical energy values of 180–260 kJ/kg are common for floating aquafeed produced through a 3.0 mm die with open area of 250–350 mm², and these values rise to 300–400 kJ/kg for high-protein sinking shrimp feeds with lower moisture. Higher specific mechanical energy increases melt temperature at the die and promotes acid volatilization. The die pressure drop also drives acid loss: a pressure drop from 20 bar to atmospheric releases not only water vapor but also acid. A longer die land length of 10–15 mm increases die pressure and residence time under pressure, which can improve acid retention by preventing premature vaporization before the die exit, but it also increases specific mechanical energy and the risk of starch shearing. Die geometry therefore operates as a partition control: die insert open area and land length set the pressure at which the acid-water mixture flashes, and acid retention follows the same pressure-release curve as water evaporation. The practical consequence is that changing from a 3.0 mm die to a 2.0 mm die on the same extruder can shift formic acid retention downward by 5–12 percentage points because the higher pressure drop accelerates flash vaporization. Published data for this specific configuration is limited, but pressure-release mass transfer is well documented for steam-volatile food volatiles.

Low-ash high-starch salmonid feeds differ from high-protein penaeid shrimp feeds in the degree to which acid partitioning responds to screw speed. In a salmonid grower formula with 35–40 g/100 g crude protein and 20–25 g/100 g wheat starch, the continuous starch melt acts as a vapor barrier, and the acid is retained more effectively in the melt core. In a shrimp feed with 38–42 g/100 g protein, 8–12 g/100 g starch, and 4–6 g/100 g mineral ash, the melt is more discontinuous, and the vapor phase has more pathways to escape. At screw speeds above 450 rpm on a 45 mm twin-screw extruder, the residence time distribution narrows, but the shear rate increases, and the temperature rise can offset the benefit of shorter residence time. At screw speeds below 200 rpm, the melt may remain in the barrel long enough for acid-mineral reactions to complete, but throughput drops and the barrel fill becomes insufficient to maintain die pressure. The operating window for acid retention in shrimp feed is consequently narrower than for salmonid feed: 250–350 rpm, 22–25 g/100 g moisture, and die melt temperature 115–130 °C. Outside this window, acid retention becomes highly variable between batches, with coefficients of variation above 15 % on formic acid residue. Experience from production lines indicates that batch-to-batch variance is often driven by fishmeal buffering capacity, which varies with fish species, freshness, and calcium carbonate content. Fishmeal with total volatile nitrogen above 120 mg/100 g may have higher amine buffering, which increases acid consumption and reduces the free acid available for vaporization. This matrix effect must be measured before adjusting acid dose; otherwise the same added acid concentration can produce different final pellet pH and acid residue depending on the fishmeal lot.

What Limits Gravimetric Retention of Lactic and Citric Acids in Water-Stable Shrimp Feeds?

Lactic and citric acids are classified as low-volatility acidulants, but their retention in water-stable shrimp feeds is not unlimited, because both acids can react with calcium and magnesium ions released from mineral supplements and fishmeal ash. Calcium lactate and calcium citrate are soluble to moderately soluble in water, and their formation reduces the free acid concentration in the extrudate. The gravimetric recovery of total lactate or citrate therefore depends on whether the analytical method detects only the free acid or also the salt. When the method uses ion-exclusion chromatography without acid hydrolysis, the salt is not always released, and the measured recovery can be artificially low. In water-stable shrimp feeds manufactured with 3.0 g/100 g calcium carbonate, the addition of 1.0 g/100 g citric acid produces calcium citrate and carbon dioxide, and the free citric acid residue can drop below 0.2 g/100 g while total citrate remains near 0.9 g/100 g. This discrepancy creates a compliance problem if the feed specification requires free acid as a preservative. The second limitation is water stability itself. Hydrophilic acids are leached from the pellet surface when the pellet is immersed in water. In a water stability test at 28 °C for 120 min, the leached acid fraction can reach 20–40 g/100 g of the residual acid for formic and acetic acids, whereas lactic acid and citric acid leach less because they are bound to minerals and starch. The water stability of the pellet also changes because acid addition lowers the pH of the extrudate and can reduce the degree of starch retrogradation after extrusion. Pellets with added citric acid at 0.5–1.0 g/100 g may show lower water absorption index and higher pellet hardness when measured by diametral compression with a texture analyzer equipped with a 50 N load cell, because the acid promotes starch hydrolysis and produces low-molecular-weight dextrins that form a denser matrix. Published data for this specific configuration is limited, but the trend is consistent with acid-catalyzed starch modification at extrusion temperatures above 120 °C.

Operational boundaries for extrusion acidification are defined by relative humidity, mineral content, and acid salt selection. At relative humidity above 60 %, dry acid powders and acid-treated meals absorb moisture and can form lumps in the preconditioner feed screw, causing mass flow fluctuations. Pre-drying of acid powders is required before blending when the storage environment exceeds 60 % relative humidity. Acid salts such as calcium propionate, sodium lactate, and sodium citrate reduce vapor loss but also reduce the immediate pH drop in the preconditioner, so they are less effective when rapid acidification is desired for microbial control. The use of free acid in combination with amine-based additives should be avoided when the amine group can react with the acid and form amides or salts that alter melt rheology. In particular, free formic acid should not be blended with high levels of choline chloride in the same preconditioner because the acid can degrade choline and release trimethylamine, which is volatile and produces off-odor. The final acid concentration in the pellet must be interpreted with the analytical method and the salt-form fraction, and the feed label must match the authorized acid or acid salt in the formulation. For feed additive compliance, organic acids used as preservatives or acidifiers must be authorized under EC 1831/2003 and listed in the European Union Register of Feed Additives with the relevant functional group and maximum content. The production record should include acid recovery, pellet water stability, die pressure, and melt temperature for each batch, because these parameters are not independent when hydrophilic organic acids are introduced into aquafeed extrusion.