The Role of Enzyme Compatibility in Viscosity Cliffs of Heavy-Duty Liquid Detergents

In heavy-duty liquid detergent (HDL) formulations carrying 12–18 wt% anionic surfactants and a 3:1 linear alkylbenzene sulfonate (LAS) to sodium laureth sulfate ratio, the introduction of a subtilisin-type protease (EC 3.4.21.62) at active protein concentrations exceeding 0.2 wt% converts a structurally stable lamellar dispersion into a low-viscosity isotropic micellar phase within 48–72 h at 40 °C. This non-linear rheological collapse, recorded as a drop from 1,800 mPa·s to 180 mPa·s at 10 s⁻¹ on a Brookfield RVDV-II+ Pro viscometer fitted with a #3 spindle per ASTM D2196-20, arises from the enzyme’s hydrophobic surface patches competing with short-chain cosurfactants for the interfacial layer of multilamellar vesicles. Simultaneous small-angle X‑ray scattering (SAXS) data acquired on a laboratory beamline (Anton Paar SAXSpoint 2.0, Cu Kα 0.1542 nm) reveal a disappearance of the d‑spacing peak at q ≈ 0.35 nm⁻¹, confirming loss of long-range lamellar order. During scale‑up trials in a 5,000 L jacketed vessel agitated with a dual‑flight ribbon impeller (tip speed 1.2 m/s), a batch held at 37 °C exhibited a 92 % viscosity loss in 16 h, forcing immediate re‑qualification under the IKW method for hand‑dishwashing detergents because the product no longer met the 500 mPa·s minimum pump‑filling threshold on the volumetric piston filler (Sidel Matrix Combi 8/108/2). Protease‑compatible reformulation requires either a lamellar‑phase restructurer such as a C16-18 fatty alcohol ethoxylate (3EO) at 0.8–1.5 wt% or a covalent enzyme immobilization on a methacrylate carrier that shields hydrophobic domains, though the latter adds a pre‑dispersion step in a rotor‑stator mill (IKA Ultra‑Turrax UTL 1000/10) operating at 8,000 min⁻¹ under nitrogen blanket to preclude oxidative damage to the enzyme’s methionine 222 residue.

Why Does the Viscosity Plummet at 45 °C Storage in Amylase-Containing HDLs?

Long‑term storage of HDLs containing a thermostable α‑amylase (EC 3.2.1.1, Bacillus licheniformis, declared activity 300 KNU-T/g) at 45 °C often produces an abrupt viscosity decline between week 8 and week 12, a phenomenon that cannot be explained by simple thermal thinning or surfactant hydrolysis. Capillary breakup extensional rheometry on a HAAKE CaBER 1 instrument reveals that the filament lifetime shortens from 0.83 s to 0.15 s during this period, indicating a loss of the elastic component contributed by high‑molecular‑weight polyacrylate thickener (Carbopol® ETD 2020, 0.35 wt%). The root cause is the enzymatic cleavage of the α‑1,4-glucosidic linkages in the polyglucan backbone of a proprietary soil‑release polymer (SRP) based on modified starch‑g‑poly(vinyl alcohol). That SRP (0.7 wt%) serves not only as a fabric finish but also as a structure viscosity builder through entanglement with the Carbopol microgel network. When the amylase reduces the SRP weight‑average molecular weight from 1.8 × 10⁵ Da to below 3 × 10⁴ Da—verified by gel permeation chromatography (GPC) with a refractive index detector and pullulan standards (PSS Suprema columns, 30 °C, 0.1 M NaNO₃ eluent)—the entanglement node density falls below the critical value νₑ ≈ 1.4 × 10²³ m⁻³ required for a measurable plateau modulus, and the zero‑shear viscosity collapses from 2,400 mPa·s to 210 mPa·s. This mechanism was first suspected after a production run in a 15,000 L blending vessel equipped with an in‑line Mettler Toledo GPro 500 turbidity probe showed that the turbidity at 880 nm dropped from 120 NTU to 35 NTU concomitant with the viscosity cliff, consistent with dissolution of semi‑crystalline SRP aggregates. Reformulation trials conducted on a Doppel‑Sigma kneader (FrymaKoruma Dinex 700) demonstrated that substituting the starch‑based SRP with a purely petrochemical terpolymer (PET‑POET‑PET, sulfonated) eliminated the amylase‑induced viscosity cliff; however, the new SRP caused a 15 % reduction in particulate soil removal in wash tests run according to IEC 60456:2010 using WFK 10D soiled cotton, mandating an additional 0.2 wt% of PEG 4000 to restore performance.

Across multiple HDL platforms, a frequently overlooked viscosity cliff trigger is the incompatibility between enzyme‑stabilizing calcium salts and the organic carboxylate chelators used for water hardness control. In a formulation containing 8 wt% sodium citrate, 2 wt% GLDA (tetrasodium glutamate diacetate), and a protease stabilized in a 15 % propylene glycol/ 4 % sodium formate matrix with 1.2 mM free Ca²⁺, the addition of 0.05 % calcium chloride dihydrate (CaCl₂·2H₂O) to improve thermal stability of the enzyme triggers a sudden drop in the storage modulus G′ from 45 Pa to 1.2 Pa at 1 Hz and 0.5 % strain on a TA Instruments DHR‑3 rheometer fitted with a 40 mm cone‑and‑plate geometry (1° angle, 28 µm truncation) per ISO 6721‑10:2015. The free Ca²⁺ ions competitively bind to citrate, precipitating as calcium citrate trihydrate nanocrystals that act as nucleation sites for the already supersaturated GLDA‑Ca complex, generating a colloidal suspension of needle‑like crystallites (length 2–5 µm, imaged by field‑emission SEM at 5 kV). These crystallites adsorb nonionic surfactant (C_12–14 ethoxylate 7EO) from the lamellar phase, thereby demoting the bilayer integrity and releasing interspersed wormlike micelles, leading to a fluidity increase of over 90 %. Rheo‑SALS (small‑angle light scattering) measurements conducted simultaneously on a Linkam CSS450 shear cell coupled to a Malvern Mastersizer 3000 confirmed that the viscosity cliff coincides with the disappearance of a four‑lobe scattering pattern characteristic of lamellar domains. The operational boundary was mapped with a full factorial DoE (Minitab 21) varying Ca²⁺ from 0.5–3.0 mM and citrate from 4–10 wt%, establishing a safe envelope only at Ca²⁺ below 0.8 mM when citrate exceeds 7 wt%. Outside this window, the product fails the self‑leveling requirement of ASTM D4359‑90 after 30 s, and the filling line (Ossid 2000 cup filler) registers a product rejection rate above 12 % due to short‑fills caused by erratic flow.

When Polyethyleneimine Ethoxylate Dispersants Replace Propylene Glycol as Enzyme Carrier

In an effort to reduce volatile organic compound content under the U.S. EPA Safer Choice Standard (Section 4.2.1), a plant trial replaced a propylene glycol‑based enzyme slurry with a water‑based delivery system containing 8 % polyethyleneimine ethoxylate (PEIE, M_n ≈ 2,000 Da, 80 % ethoxylation) as dispersant for a protease‑amylase blend. The finished product, mixed in a 12,000 L tank using a Lightnin A310 hydrofoil impeller (D/T = 0.4, 290 rpm), developed a transient gel phase at the impeller tip after 20 min of circulation, as the PEIE formed bridging floccules with the anionic LAS micelles through electrostatic interaction at the formulation pH of 8.7 (measured with a Mettler Toledo InLab Solids Pro ISM probe). Oscillatory time sweeps at 0.5 Hz and 0.2 Pa stress on an Anton Paar MCR 302 rheometer showed the tan δ dropping from 3.1 to 0.4 over 40 min, indicative of a sol‑to‑gel transition, with the complex viscosity η* peaking at 15,200 mPa·s before catastrophic fracture at a strain of 4.2 %. This viscosity cliff, which destroyed the product’s flowability, was reversed only when the impeller speed was increased to 620 rpm (tip speed 5.8 m/s) to rupture the floc network; however, that shear exposure reduced the protease residual activity from 100 % to 67 % after 4 weeks at 25 °C (detected via the azocasein assay according to ISO 22118:2011). The processing window for this carrier system is therefore constrained to a narrow agitation intensity range of Re 800–1,200 during the enzyme introduction phase. On the same line, a subsequent batch incorporating a high‑molecular‑weight branched PEIE (M_n ≈ 25,000 Da) generated irreversible gel particles that blinded the 100 µm mesh basket filter on the in‑line filling manifold (Krones Vipoll filler), causing a 45 min downtime. This outcome effectively bans cationic dispersants above a critical charge density of 2.5 meq/g in LAS‑rich HDLs.

Rheopexy Onset in Lipase-Stabilized Microemulsions at pH 9.5

Certain liquid lipases (EC 3.1.1.3, Thermomyces lanuginosus) formulated in a water‑continuous microemulsion stabilized with 4 wt% sodium di‑hexyl sulfosuccinate (AOT) and 0.5 wt% allyl‑capped alkyl polyglucoside exhibit a time‑dependent viscosity increase under constant shear—a rheopectic response—that culminates in a yield‑stress cliff when the system is rested. Cone‑and‑plate rheometry (TA Instruments AR‑G2, 60 mm, 1°) under a steady shear rate of 0.1 s⁻¹ shows the viscosity rising from 180 mPa·s to 1,350 mPa·s over 800 s, after which the sample seizes and the stress diverges beyond the transducer limit of 200 N·m. The phenomenon is attributed to the lipase’s calcium‑binding loop adsorbing onto the sulfosuccinate headgroups, causing a gradual conversion from spherical to wormlike micelles, as evidenced by cryo‑TEM micrographs (JEOL JEM‑2100 Plus, 200 kV, vitrified on Quantifoil R2/2 grids) that show an increase in micellar contour length from 15 nm to over 200 nm after 10 min of incubation. This structural evolution is critically pH‑dependent: at pH 9.0 the thixotropic recovery dominates and no cliff occurs, while at pH 9.5 the rheopexy is fully expressed. Manufacturing operations in a continuous pin‑mill mixer (Silverson L5T‑A with square‑hole high‑shear screen) require immediate pump‑out within 240 s of lipase addition to avoid gel‑like clogs in the gear pump (Viking SG‑4184) that handles the transfer to intermediate storage. The product specification was therefore revised to mandate a resting viscosity test after 1,800 s according to a modified DIN 53019‑1:2017 protocol, with a rejection criterion set at η > 800 mPa·s.

In production campaigns aimed at ultra‑concentrated HDLs with water content below 30 wt%, the anionic‑nonionic surfactant phase still retains sufficient polarity to solvate enzyme particles, but the absence of a continuous aqueous pseudophase eliminates the bulk‑water‑mediated protease autolysis that normally stabilizes the low‑shear Newtonian plateau. A batch formulated with 38 wt% LAS‑MEA salt (90 % active), 12 wt% Neodol 25‑7, 10 wt% propylene glycol, and 0.8 wt% Savinase Ultra 16L demonstrated a classic viscosity cliff during air‑pressurized filling: the product flowed turbulently through the 12 mm ID piping at 1.2 L/min (Reynolds number 2,000) but solidified within the nozzle of the pneumatic piston filler (Odenberg Mach‑3 actuator) when the shear rate dropped to near zero. In‑line rheometry using a Brookfield AST‑100 viscometer installed in a ¾‑inch slip‑stream showed that the time to reach a dynamic yield stress of 100 Pa was 11 s, leaving no margin for the 15 s cycle time of the filler. Differential scanning calorimetry (Mettler Toledo DSC 3+, ‑50 to 120 °C, 10 K/min, N₂ purge) on the gelled plug revealed a glass transition (T_g) of –28 °C for the surfactant matrix, but the enzyme‑rich domains exhibited a melting endotherm at 52 °C, suggesting formation of protein‑LAS crystalline co‑acervates. The risk was mitigated by incorporating 3 wt% of a PEG 200-di‑ester as a kinetic diluent, extending the no‑flow latency to 42 s, sufficient for the packaging line to operate at its rated 120 bottles/min. Published data for this specific enzyme‑solvent‑free configuration remains limited, though analogous build‑up phenomena with subtilisin in 90 % active surfactant pastes have been reported in trade literature with no disclosed rheometric protocol.

Enzyme‑induced viscosity cliffs are not exclusively catastrophic; a controlled cliff is deliberately engineered in the “single‑dose, water‑soluble pouch” segment, where a freshly introduced protease must maintain a stable 800–1,200 mPa·s rheology in a non‑aqueous suspending base until the pouch is sealed, after which the base must degrade within 30 s upon pouch dissolution in wash water. A formulation comprising 45 wt% sulfonated polyester polyol (Degalan P24N), 15 wt% propylene carbonate, 8 wt% tetraacetylethylenediamine (TAED), and a granulated protease (Savinase 16.0 L type impregnated onto sodium sulfate carrier, 200–400 µm particle size) exhibits a yield‑stress plateau of 85 Pa at 0.05 s⁻¹ that holds for over 48 h in a sealed glass vial (stability test per ASTM D2196‑20). Introduction of 0.2 % free moisture from ambient humidity (RH > 60 %) triggers an immediate collapse: the polyester polyol hydrolyzes under enzyme‑catalyzed and base‑catalyzed pathways, the pH drops from 7.5 to 4.8, and the suspension liquefies to 45 mPa·s. A manufacturing team utilizing a Karl Fischer titrator (Metrohm 870 KF Titrino plus) linked to an in‑hood atmosphere monitor rejected 17 % of a pre‑production run because the moisture content exceeded 0.15 wt%. The filling line, a Cycle‑to‑Cycle machine operating at 60 pouches/min, is therefore enclosed in a dry‑air glovebox with a dew point below ‑25 °C to ensure compliance.

Shear-Induced Enzyme Desorption from Silicone Antifoam Droplets Creates Localized Gel Patches

Silicone antifoam emulsions (PDMS, 20 % active, droplet size 10–40 µm, stabilized with an EO‑PO block copolymer) are routinely post‑dosed into HDLs at 0.05–0.2 wt% to suppress foam during high‑speed bottling. Certain proteases, however, adsorb irreversibly at the PDMS‑water interface, and the intense shear field within the filling nozzle (estimated shear rate 5,000 s⁻¹ across a 1.5 mm annular gap) can strip aggregated enzyme‑antifoam complexes into the bulk, where they act as crosslinking nodes for multifunctional carboxylic acid polymers (Sokalan® PA 25 CL PN, 4 wt%). This generates discrete, semi‑solid gel lumps measuring 0.2–0.8 mm in diameter that are detected by an in‑line laser diffraction particle sizer (Sympatec HELOS/KR, dry dispersion at 0.5 bar) and lead to a transient blockage of the filling valve, causing a “spit‑stop‑spit” flow pattern. A pilot trial on a Hibar syringe filler (6‑station) demonstrated that the viscosity in the valve chamber surged from 320 mPa·s to 2,800 mPa·s over 15 cycles before the fill weight drifted outside the ± 2 g tolerance for a 1.5 L bottle. Elimination of the PDMS antifoam entirely solved the gelation but created a foam‑over problem in the bottle neck, violating the leave‑one‑day‑after check test of the European Cleaning Industry Association (A.I.S.E.) protocol; instead, a non‑adsorbing polyether‑based defoamer (Pluronic® PE 10400 at 0.08 wt%) was introduced after verification that its cloud point of 62 °C remained above the formulation’s maximum processing temperature of 45 °C.

When an HDL plant transitions from a non‑enzyme control formula to an enzyme‑built version on the same compounding system, the cleaning procedure between batches becomes a decisive factor in viscosity cliff prevention. Residual borosilicate glass‑scale deposits on heat‑exchanger plates (Alfa Laval M15‑BFM, 0.5 mm channel gap) can harbor denatured enzyme films that seed crosslinking of the fresh batch’s carbohydrate‑based stabilizers upon contact. In one instance at a contract manufacturing site, a rinse with 0.5 % NaOH at 80 °C for 20 min failed to remove a protease‑gellan gum complex, and the subsequent batch gelled within the plate heat exchanger, reaching a pressure drop of 4.2 bar—exceeding the 3.0 bar safety limit of the centrifugal feed pump (Grundfos CRNE‑HS, 5.5 kW). The batch was rejected, and the cleaning protocol was revalidated with a 1.0 % peracetic acid step at 25 °C followed by a controlled rinse measured by ATP bioluminescence (Hygiena UltraSnap surface swab, RLU < 10), according to the internal SOP aligned with 21 CFR 211.67 for equipment cleaning. This experience underscores that enzyme compatibility cannot be assessed solely through formulation chemistry; the process engineering and sanitation sequences must be integrated into the stability guarantee.

In the realm of consumer‑driven cold‑water washing, HDLs are increasingly fortified with a cocktail of protease, amylase, mannanase, and pectate lyase (each dosed at 0.1–0.25 wt% active). The four‑enzyme mixture imposes a rheological challenge because each protein possesses a distinct charge distribution at the product pH of 7.8—protease (pI ~9.0) is net positive, amylase (pI ~5.2) is net negative, mannanase (pI ~4.5) is strongly negative, and pectate lyase (pI ~9.5) is highly positive. In the absence of a polymeric charge‑shielding hydrotrope, electrostatic hetero‑aggregation creates soluble complexes (diameter 30–60 nm by DLS, Malvern Zetasizer Ultra, 173° backscatter detection) that eventually phase‑separate as coacervate droplets, yielding a sediment layer at the bottle bottom. The supernatant viscosity drops from the target 900 mPa·s to 220 mPa·s, a silent cliff that is only detected by consumer complaints of “water‑thin” product. Incorporation of 0.6 wt% sodium xylene sulfonate (SXS, hydrotrope) and 0.15 wt% polymeric charge‑balance agent (Lupasol® FG, PEI modified with 0.5 ethylene oxide per NH) maintained a single‑phase fluid for 24 weeks at 25 °C with viscosity deviation below ±8 %. The packaging line (KHS Innoket 500 filler) confirmed uniform fill weights within ±1.5 g across a 10,000‑bottle lot, validating the combinatory enzyme compatibility strategy.