Filter Media Selection for Dissolved Lignin Carryover Control in Oxygen Delignification

In oxygen delignification, medium-consistency softwood or hardwood pulp is contacted with molecular oxygen under alkaline conditions at 80–110 °C and 0.4–0.8 MPa oxygen partial pressure, with alkali charge adjusted to maintain terminal pH between 10 and 12; the reaction depolymerizes residual lignin through free-radical pathways, generating alkali-soluble lignin fragments that partition into the liquid phase as dissolved organic carbon and as colloidal lignin aggregates. Downstream washing and filtrate reuse create a carryover stream in which dissolved lignin, fiber fines, pitch droplets, oxalate salts, and partially degraded hemicelluloses are returned to the recovery cycle or discharged to treatment. Selection of filter media for this stream is not governed solely by nominal particle retention but by simultaneous demands of alkaline hydrolysis resistance, oxidative stability at elevated temperature, resistance to fouling by deformable pitch and lignin colloids, and tolerance for variable solids loading across batch digester cycles. Standard laboratory filtration tests under neutral water conditions consistently underestimate life because they do not replicate the combined effects of pH 11 hydrolysis, dissolved carbonate/bicarbonate buffering, and residual peroxide generated in oxygen-stage filtrate. Filtration objectives must be defined separately for protective duty—preventing shower nozzle plugging and heat exchanger fouling—and for selective removal of colloidal lignin before membrane fractionation or biological treatment. In both cases, media pore size must be correlated with particle size distributions measured by laser diffraction under alkaline conditions rather than with dry-sieve classifications, because lignin colloids and fiber fines hydrate and swell at pH > 10.

What Limits Media Life in Hot Alkaline Oxidative Filtrate?

Chemical compatibility of filter media in oxygen delignification filtrate is dominated by alkaline hydrolysis of polyester and polyamide fibers, oxidative attack on polypropylene at sustained temperatures above 80 °C, and stress cracking of rigid polymer housings when exposed to dissolved sulfur species and tall oil soaps. Polyethylene terephthalate monofilament cloth hydrolyzes at measurable rates when continuous exposure exceeds 70 °C at pH 10; material supplier chemical resistance data generally exclude PET from continuous service at pH > 9 at temperatures above 70 °C. Polyamide fabrics degrade more rapidly through amide bond scission and are generally excluded from oxygen-stage filtrate duty unless the stream is cooled below 45 °C and pH is reduced below 9. Polyphenylene sulfide needlefelt exhibits acceptable hydrolytic stability in alkaline media up to 190 °C and has been used in bag filters on weak black liquor, but its oxidative resistance is limited when residual peroxide exceeds 5 mg/L as H₂O₂ at temperatures above 90 °C. Polyvinylidene fluoride and ethylene chlorotrifluoroethylene membranes provide broader chemical resistance, with ECTFE rated for continuous service to 150 °C in strong alkaline streams, but membrane modules require prefiltration to 200 µm or finer to prevent irreversible fouling by fiber bundles. Stainless steel 316L wedge wire screens resist alkali at oxygen-stage temperatures but suffer crevice corrosion when chloride concentrations exceed 500 mg/L under oxygen-saturated conditions; duplex 2205 screens are specified where chloride accumulation is expected. Seal and gasket materials are often the life-limiting component rather than the medium itself; ethylene propylene diene monomer elastomer performs acceptably at pH 11 up to 150 °C, whereas nitrile and chloroprene seals harden and leak within weeks. Chemical immersion testing according to ISO 175:2010 should be performed on candidate media in synthetic oxygen-stage filtrate at the maximum expected temperature rather than in water alone. Bench-top tests with synthetic latex particles overestimate retention and underestimate fouling because they do not reproduce the deformability and adhesive character of oxygen-stage lignin colloids.

The fiber bundle structure of needlefelt media introduces depth filtration, but deformable lignin colloids and pitch droplets penetrate the surface and undergo shear-induced coalescence, producing a sticky internal foulant layer that cannot be removed by conventional backwashing. Woven monofilament fabrics act as surface filters and release cake more readily, but their lower dirt-holding capacity requires larger filtration area or shorter cycles when solids loading exceeds 100 mg/L. The selection sequence should therefore begin with a particle size analysis under alkaline conditions, followed by chemical compatibility screening, then pilot-scale flux decline testing using actual filtrate.

Dissolved lignin in oxygen delignification filtrate is not a single molecular population but a continuum from low-molecular-weight phenolic monomers and dimers below 1 kDa to high-molecular-weight lignin-carbohydrate complexes exceeding 100 kDa. The truly dissolved fraction cannot be removed by conventional mechanical filtration because its hydrodynamic diameter is below 0.01 µm; only colloidal lignin aggregates and lignin adsorbed on fiber fines can be captured by media with pores in the 0.1–10 µm range. Ultrafiltration membranes with molecular weight cut-off between 10 kDa and 50 kDa can retain the high-molecular-weight fraction, but flux is strongly dependent on dry solids content and shear rate. Published data for spiral-wound polymeric ultrafiltration of oxygen-stage filtrate at 10–15% dry solids remain limited, and most reference installations operate at dry solids below 8% with crossflow velocities above 3 m/s to control gel-layer formation. Therefore, filter media selection for carryover control in the washing loop should explicitly distinguish between particulate carryover—which mechanical filtration can address—and dissolved lignin carryover, which requires either chemical precipitation, ultrafiltration, or process reconfiguration. Failure to make this distinction leads to installation of 25 µm bag filters that remove fiber fragments but no measurable dissolved lignin, creating a false sense of protection while downstream evaporator fouling continues.

Filter Media Pore Geometry and Lignin Aggregate Retention

Woven monofilament media with plain weave geometry provide a narrow pore size distribution and are preferred where cake release is critical, whereas twill and satin weaves improve backwashing efficiency at the expense of lower burst strength. For oxygen-stage filtrate, a retention rating of 25–50 µm absolute is commonly specified for shower protection, but the actual removal of lignin colloids at this rating is limited because the colloidal particles are typically 0.1–1 µm; depth media with graded fiber density capture a larger fraction of these submicrometer particles at the cost of rapid irreversible blinding. Needlefelt media constructed from polyphenylene sulfide staple fiber with a basis weight of 500 g/m² and a nominal air permeability of 200 L/dm²/min at 200 Pa have been used on plate-and-frame filter presses for weak black liquor clarification, but cycle times decrease sharply when the feed contains more than 50 mg/L of pitch or soap. Multilamellar media combining a woven scrim of 150 µm polyamide-free polyester with a fine meltblown polypropylene face layer can achieve 10 µm nominal retention at lower cost, but the polypropylene face layer softens above 80 °C and must be derated to 60 °C for continuous duty. Stainless steel sintered fiber media in the 5–20 µm range are attractive for high-temperature alkaline streams because they can be cleaned with hot alkali or acid and withstand differential pressure up to 1.0 MPa, but the capital cost per unit filtration area is approximately five to ten times that of synthetic fabric media. The choice of pore geometry should be tied to the filtration objective: protective duty demands high throughput and long life with only moderate fine-particle capture, while clarification duty demands low effluent solids and accepts higher pressure drop and shorter cycles.

When Carryover Solids Are Not the Dominant Fouling Species

In many oxygen delignification filtrate systems, the measured fiber fines concentration is below 50 mg/L, yet filter media blind within days. The dominant fouling species are often pitch droplets, fatty acid soaps, and calcium oxalate precipitates that form when black liquor is subjected to temperature and pH shifts after the oxygen reactor. Tall oil soaps precipitate when filtrate temperature falls below their phase separation point, producing a tacky deposit that bridges pores and binds otherwise recoverable fiber fines. Calcium oxalate forms as a crystalline scale when dissolved calcium encounters oxalate released from hemicellulose degradation, and the crystals grow preferentially in stagnant zones, filter pores, and piping dead-legs. Media selection under these conditions should prioritize surface smoothness and release characteristics over fine-particle retention, because a 25 µm woven monofilament screen may maintain hydraulic capacity longer than a 5 µm depth medium that captures more solids but cannot release the sticky agglomerate. Filter aid precoating with diatomaceous earth or perlite at 2–5 kg/m² of filter area is standard on pressure leaf filters and rotary vacuum drum filters handling soap- and pitch-laden liquors; the precoat protects the base medium from direct contact with adhesive foulants and allows periodic knife discharge without damaging the substrate. In such configurations, the base medium is not the primary retention surface, and its selection criteria change from pore size to support strength, dimensional stability under vacuum, and compatibility with the precoat discharge mechanism. On a rotary vacuum drum filter handling oxygen-stage filtrate with pitch content above 200 mg/L, the drum speed is typically reduced to 0.1–0.3 rpm and the doctor blade gap adjusted to 0.05–0.15 mm to remove the precoat layer without scoring the underlying cloth. Published data for specific filter media life under these conditions is limited, and pilot trials should therefore be conducted on side-streams with actual process liquor before capital purchase.

Disc Filters Require Seam-Stable Synthetic Cloths at Oxygen-Washer Temperatures

Disc filters used on oxygen-stage washer filtrate typically employ sectors covered with synthetic filter bags. The media synthetic cloth is specified by air permeability and retention rating, with common bags rated at 25–100 µm and constructed from polypropylene or polyester depending on filtrate temperature. Mill experience indicates that polypropylene bag life falls sharply when the filtrate temperature consistently exceeds 75 °C, with seam failures occurring before body fabric degradation. Polyester bags withstand higher temperature but lose tensile strength rapidly at pH > 9; installations operating at pH 10–11 and 85 °C therefore require either PPS needlefelt bags or PTFE-coated polyester bags with the coating providing temporary chemical protection. Pressure screens in the same service use wedge wire or punched plate elements with slot widths from 50 µm to 250 µm; the open area percentage, typically 20–35% for wedge wire, determines the pressure drop and backwash effectiveness. Unlike filter bags, pressure screens can be cleaned by mechanical rakes or differential backflushing, but scale adhesion to slot edges can narrow the effective opening by 50% within 72 hours when calcium oxalate is present. Selection of screen slot width must therefore account not for the nominal opening but for the worst-case scaled opening.

Crossflow membrane filtration of oxygen delignification filtrate shifts the retention target from particulate carryover to dissolved and colloidal lignin fractionation. Ceramic ultrafiltration membranes with 15 kDa cut-off and a channel diameter of 3–4 mm have been evaluated on weak black liquor at dry solids up to 15%, with reported steady-state flux in the range of 30–70 L/m²/h when crossflow velocity is maintained above 4 m/s and temperature is held at 80–90 °C. Polymeric spiral-wound membranes are limited by maximum operating temperatures of 95 °C for polysulfone and 135 °C for PVDF; however, spacer fouling by fiber fines and pitch requires a prefilter with absolute rating 200 µm or finer. The membrane filtration approach removes a portion of high-molecular-weight lignin and lignin-carbohydrate complexes, but the low-molecular-weight fraction below the cut-off passes into the permeate, so the dissolved lignin carryover is reduced only by incremental amounts unless the membrane process is designed for fractionation rather than complete removal. Published data for specific dissolved lignin rejection rates in oxygen-stage filtrate are limited, and the economic viability is highly sensitive to the cost of concentrated retentate incineration in the recovery boiler.

How Backpulse Pressure Determines Recovered Flux in Soap-Laden Filtrate

Constant-pressure filtration of oxygen-stage filtrate follows a gel-layer or cake-filtration model in which the reciprocal of flux is proportional to cumulative filtrate volume per unit area, with the specific resistance of the cake increasing as a function of pH and colloidal lignin content. Field measurements on pressure leaf filters show that specific cake resistance can exceed 1 × 10¹² m/kg when the cake contains precipitated soap and calcium oxalate, whereas a fiber-fines cake from well-washed pulp exhibits values one to two orders of magnitude lower. Backwashing frequency is determined by the maximum allowable pressure differential, commonly 0.15–0.25 MPa for bag filters and 0.5–1.0 MPa for sintered metal elements, beyond which the medium is at risk of mechanical deformation or bag bursting. Automatic backwash filters with wedge wire elements in the 25–200 µm range can be programmed for sequential backpulse based on timer or differential pressure, but the backpulse volume must be sufficient to disrupt the adhesive pitch layer; low-pressure backpulses below 0.3 MPa are frequently inadequate for oxygen-stage soap deposits. Filter aid conditioning with DE or perlite modifies the cake structure by providing permeable, rigid particles that reduce the compressibility of the combined cake, but the addition rate must be controlled within 0.5–5 g/L of feed to avoid excessive solids loading and rapid pressure build-up. Standardized filtration tests for hydraulic filter elements according to ISO 16889:2022 measure particle retention and pressure drop using synthetic test dust in hydraulic oil, which does not reproduce the deformable and swelling character of oxygen-stage filtrate; therefore, the results should be used only for initial comparative screening, not for final media selection.

Validation of filter media for oxygen delignification filtrate requires a pilot side-stream that receives actual filtrate from the oxygen-stage washer, not a synthetic surrogate. The side-stream should include temperature control, a feed tank with gentle agitation to avoid shear-induced precipitation of soap, and continuous monitoring of differential pressure and filtrate turbidity. Samples of media should be removed at intervals of 100, 500, and 1000 hours and subjected to tensile testing per ISO 13934-1:2013 to quantify strength loss from hydrolysis and oxidation. Filtrate should be characterized for total solids, ash, lignin content by ultraviolet absorbance at 280 nm, and particle size distribution. Media that show acceptable performance at 100 hours but catastrophic failure at 500 hours demonstrate the limitation of short laboratory screening and the need for long-duration exposure to the hot alkaline oxidative environment.