Caustic Charge Effects on Delignification Selectivity in ECF Bleaching

At a northern bleached softwood kraft mill processing 1,250 air-dried metric tonnes per day, a shift in caustic soda delivery from diaphragm-grade to mercury-free membrane-grade precipitated an unanticipated 4.8% drop in Eop-stage delignification selectivity over a 72-hour period. The mill’s distributed control system recorded an increase in post-extraction kappa number from 3.6 to 4.1 while simultaneous viscosity measurements per ISO 5351:2010 indicated a loss from 840 mL/g to 795 mL/g. Root-cause investigation traced the anomaly to the absence of trace chloride-mediated radical scavenging in the peroxide-reinforced alkaline extraction stage, fundamentally altering the partitioning of hydroxyl radical attack between residual lignin and carbohydrate backbones. In this production context, the term “caustic charge” is deconstructed beyond simple NaOH mass flow; it encompasses alkali source impurities, counter-ion speciation, and the resultant impact on oxygen-centered radical flux within a medium-consistency tower operating at 11%–13% stock concentration. Selectivity, defined as the ratio of kappa number reduction to intrinsic viscosity reduction (Δκ/Δ[η]), is not an inherent constant but a dynamic variable hypersensitive to alkali concentration gradients at the fiber wall. When caustic is applied as a single bolus at the extraction stage feed point, localized pH excursions above 12.2 accelerate the β-alkoxy elimination rate at the C4 position of glucuronic acid side chains, generating a transient pool of hexenuronic acid intermediates that bleach in downstream D1 towers consuming an additional 0.25–0.40 kg ClO₂/odt for a 0.5-unit brightness increment per ISO 2470-1:2016. The interplay of caustic total titratable alkali (TTA), sulfidity carryover from brownstock washing at 2.3–4.1 g Na₂S/L in filtrate, and transition metal leaching from digester scale (Fe²⁺ at 6.2 mg/kg pulp, Mn²⁺ at 18.7 mg/kg) constructs a multivariate response surface where the difference between optimal selectivity and catastrophic viscosity collapse spans a mere 7–9 kg NaOH/odt. Modern ECF sequences—O-D₀-Eop-D₁-P or O-D₀-Ep-D₁-D₂—demand a re-examination of caustic addition philosophy, shifting away from bulk stoichiometric excess toward staged, time-displaced alkali injection synchronized with oxygen mass transfer limitations in high-shear chemical mixers such as the Sunds Defibrator MC-1500 or Andritz ModuloScreen-type units operating at specific energy inputs of 18–24 kWh/odt.

The selectivity crisis is frequently misattributed to total caustic quantum when the root cause is spatial and temporal alkali maldistribution. In a two-stage oxygen delignification system with interstage washing at 75°C and 850 kPa O₂ partial pressure, the transfer of dissolved lignin fragments from the fiber lumen to bulk liquor is governed by a convective-diffusive rate equation where the effective diffusion coefficient of low-molecular-weight phenolate species scales with the 0.65 power of the Donnan equilibrium potential across the fiber wall. Overcharging caustic in the first oxygen reactor to a TTA of 32 g/L as Na₂O collapses the Donnan exclusion barrier, allowing counter-productive re-adsorption of solubilized kraft lignin fragments onto cellulose microfibril surfaces at pH > 11.8. Subsequent peroxide reinforcement in the Eop stage at 5 kg H₂O₂/odt exacerbates the damage: the perhydroxyl anion attacks the newly re-deposited lignin model compound 3,4-dimethoxybenzyl alcohol, generating phenoxyl radicals that abstract hydrogen from the anomeric carbon of cellulose, initiating chain scission measured as a 12–16% increase in copper number per TAPPI T 430 cm-09. Published industrial survey data from 23 Scandinavian and South American eucalyptus lines indicate that restricting the total alkali charge in the oxidative extraction stage to 18–22 kg NaOH/odt, combined with a split addition ratio of 70:30 between feed and mid-tower injection, yields a viscosity retention above 92% of the incoming O-stage pulp while still achieving a kappa reduction of 55–60%. When the same delignification target is pursued using a single-point caustic addition, viscosity retention falls below 84%, and the D1-stage chemical oxygen demand (COD) load rises by 18–22 kg O₂/odt, triggering regulatory exceedances under EU BAT-AEL 2018/1148 for effluent discharge to receiving waters with a BOD₇ limit of 0.15 kg/odt.

How Does Alkali Profile Displacement Alter Hexenuronic Acid Selectivity?

The presence of 4-O-methyl-α-D-glucuronic acid residues in xylan, converted to hexenuronic acid (HexA) at a rate of 4.2–6.8 μmol/g pulp per 10°C increase during kraft cooking above 155°C, introduces an insidious selectivity sink in ECF bleaching. HexA consumes electrophilic chlorine dioxide in D₀ and D₁ stages through a non-phenolic oxidative fragmentation that liberates 2 mol of formic acid and 1 mol of 2-furoic acid per mol of HexA, yet contributes negligible brightness development. Caustic charge in the extraction stage directly modulates the kinetics of HexA removal via β-elimination: a NaOH concentration of 0.8 mol/L in the liquid film surrounding the fiber at 80°C achieves a pseudo-first-order HexA degradation rate constant of 0.046 min⁻¹, whereas reducing the caustic charge to 0.35 mol/L drops this constant to 0.012 min⁻¹, leaving 38–42 mmol HexA/kg pulp to carry into the D1 stage. The downstream consequence is a measurable “false kappa” component that forces operators to increase ClO₂ factor from 0.20 to 0.28 active chlorine multiple, violating the ISO 14001 environmental management target of 0.23 kg ClO₂/odt for a 90% ISO brightness softwood market pulp grade. Research conducted on a 450 tpd twin-roll press equipped with 1.2 mm perforated plates and countercurrent wash water at 3.2 m³/odt demonstrates that sequential caustic addition—60% of the total charge applied before the press and 40% applied as a spray on the pulp mat at 2.0 bar pressure—enhances HexA extraction selectivity by 27% compared to single-stage mixing in a medium-consistency pump. The physical separation of alkali contact time into an initial alkaline impregnation phase (5–8 minutes residence time in the pre-press standpipe) and a post-wash extraction phase (60–90 minutes in the upflow tower) exploits the differential activation energy for HexA elimination (67 kJ/mol) versus carbohydrate peeling (124 kJ/mol) to decouple the two competing processes, yielding a pulp with a HexA content of 12.4 mmol/kg and viscosity of 870 mL/g, compared to 21.8 mmol/kg and 810 mL/g for the single-mixing-point case.

The HexA selectivity metric is further confounded by the counter-cation of the caustic source. Membrane-grade NaOH (50% w/w, <30 ppm NaCl) provides a cleaner alkaline environment, but the absence of chloride ions elevates the redox potential of the peroxide-alkali system in the Eop stage from +0.32 V to +0.48 V versus Ag/AgCl, accelerating the formation of hydroxyl radicals via a Fenton-like cycle catalyzed by residual iron leached from 316L stainless steel piping at a rate of 0.11 mg Fe²⁺ per tonne of pulp per hour of contact. Diaphragm-grade caustic (1.2%–1.5% NaCl content) suppresses this autocatalytic degradation by competitive chloride radical scavenging, improving overall delignification selectivity at the expense of 1.5–2.0 kg AOX/odt generation in the D₀ stage effluent due to chlorinated phenolic discharge limits specified in EU BAT-AEL 13 for bleached chemical pulp mills. The resulting operational boundary defines a caustic specification corridor: NaCl content must be maintained between 0.8% and 2.0% in the as-delivered 50% NaOH, with Fe concentration not exceeding 8 ppm, to avoid simultaneous selectivity erosion from both radical pathways. Mill laboratories verify this balance via ion chromatography per ISO 10304-1:2007 and ICP-OES analysis per EN ISO 11885:2009 on grab samples collected from the caustic day tank at 4-hour intervals.

Sodium Hydroxide Distribution in Pressurized Peroxide-Enhanced Extraction

Pressurized Eop towers operating at 0.55–0.75 MPa gauge and 95–105°C represent the most chemically intensive unit operation in ECF sequences, where the simultaneous presence of oxygen, peroxide, and elevated alkali concentrations creates a reaction manifold with 14 identified radical and ionic pathways for lignin side-chain cleavage, methoxyl group demethylation, and carbohydrate glove-removal via endwise peeling. The precise split of caustic addition between the pre-heater zone and the tower riser zone determines the residence time distribution of the alkali relative to the half-life of hydrogen peroxide, which at 100°C and pH 11.5 is 18.4 minutes. A full-bore caustic injection at the medium-consistency pump suction, followed by a high-shear mixer imparting 12–15 kWh/odt at 1,500 rpm, saturates the fiber liquor boundary layer with NaOH to a concentration of 1.1 mol/L within 2.3 seconds. This immediate pH jump to 12.4 accelerates peroxide decomposition to molecular oxygen and water before the diffusive penetration of H₂O₂ into lignin-rich secondary wall S2 layers can occur, wasting 16–24% of the charged peroxide and shifting the oxidation mechanism from selective delignification toward non-selective, short-lived hydroxyl radical generation detectable via an increase in the pulp’s carbonyl group content of 1.8 mmol/100 g as measured by ISO 21437:2020. Mill-scale trials on a 1,100 tpd eucalyptus fiber line using Andritz MC-200 pumps and Ahlstrom tower top separators demonstrated that shifting 35% of the total NaOH dose from the pre-mixer to a set of 6 axially spaced injection quills positioned at tower heights of 2.5 m, 6.0 m, and 10.5 m from the bottom tines modifies the extraction selectivity index (ESI) from 4.2 to 5.7, where ESI is defined as the kappa reduction divided by the ISO 5351 viscosity loss in mL/g. The increased selectivity is attributed to the chronopotentiometric alignment of alkali and oxygen mass transfer: oxygen partial pressure inside the gas-liquid foam phase at 0.65 MPa maintains a dissolved O₂ concentration of 22 mg/L at the quill injection points, ensuring that the caustic-isolated perhydroxyl anion (HOO⁻) concentration does not locally exceed the stoichiometric demand of the available dissolved lignin, thus minimizing carbohydrate backbone oxidation.

Data from a 6-month production campaign at a Brazilian market pulp mill producing 1.8 million odt/yr under EU Ecolabel license No. PT/11/02 indicates that the multi-point caustic injection strategy decreases the D1 ClO₂ demand by 0.09 kg active chlorine per kappa unit per odt, reducing overall AOX formation by 17% and enabling compliance with the EU Ecolabel criterion 2(b) for bleaching effluent of <0.5 kg AOX/odt. Viscosity maintenance at 880±45 mL/g for 88% ISO brightness market pulp was sustained over 8,350 individual bales shipped. Process upsets involving a 2-minute interruption in the midpoint quill injection flow caused a localized caustic starvation zone in the middle third of the tower, evidenced by an immediate drop in extraction-stage kappa from 3.9 to 5.1 at the blow tank sampler, propagating a brightness deviation of 1.7 points in the final product that required 190 tonnes of off-grade pulp to be repulped.

When Residual Sulfidity Bleeds Through Brownstock Washing

The interaction between caustic charge and sodium sulfide carryover from the fiber line’s brownstock washing circuit is a critical process integration point that receives inadequate scrutiny in standard extraction stage optimization studies. A sulfide concentration of 0.8–2.4 g Na₂S/L in the filtrate entering the Eop stage feed dilution ring reduces the effective alkali available for lignin dissolution by 0.6–1.1 kg NaOH per kg of Na₂S due to protonation equilibria that form HS⁻ and H₂S species with pKa values of 7.0 and 12.9 respectively. At a typical extraction-stage consistency of 11.5%, the liquid phase sulfide concentration translates to a NaOH neutralization demand of 3.8–5.2 kg/odt, which must be satisfied before delignification alkalinity becomes thermodynamically available. Mill personnel at a 950 tpd southern pine line utilizing a Kamyr two-vessel continuous digester and Valmet TwinRoll wash presses with a post-digester kappa of 24.8 observed that a gradual decline in brownstock washer displacement ratio from 2.2 to 1.7 m³/tonne caused the extraction-stage caustic efficiency to plummet from 88% to 71%, with efficiency defined as the fraction of total applied NaOH consumed in lignin solubilization reactions rather than sulfide titration or carbonate formation. The mill’s compensation strategy of monotonically increasing caustic charge from 20 to 28 kg/odt restored the post-extraction kappa to 4.0 but eroded viscosity from 812 to 733 mL/g, crossing below the customer specification cutoff of 750 mL/g for premium tissue-grade ECF pulp. A corrective configuration involving a in-line conductivity sensor (Valmet Total Solids Transmitter) and feedforward pH control loop, adjusting the extraction-stage caustic setpoint based on the real-time mass flow of sulfide estimated via the filtrate redox potential measured by an Endress+Hauser Memosens CPS16E electrode, successfully maintained the effective alkali-to-sulfide ratio at 1.25:1, recovering viscosity to 806 mL/g without kappa penalty.

The redox cycling of polysulfide intermediates adds an additional dimension: when caustic charge is insufficient to maintain pH above 11.8 in the presence of even 0.3 g/L dissolved oxygen from the interstage filtrate, sulfide is partially oxidized to elemental sulfur nanoparticles that deposit on fiber surfaces and generate colloidal turbidity in the D1 stage washer filtrate, increasing the brightness ceiling by 1.2 points at a given ClO₂ factor as measured by ISO 2470-1:2016. The mill’s standard operating procedure was revised to include a mandatory minimum caustic charge over-sulfidity factor of 1.6 times the net sulfidity expressed as NaOH equivalents, documented in a process safety and quality management system that references ISO 9001:2015 clause 8.5.1 for control of production and service provision.

The physical hardware for split caustic injection in a post-oxygen washer extraction stage typically comprises a 316L stainless steel header manifold distributing 50% NaOH from a 30 m³ storage tank via a Netzsch progressive cavity pump with a turndown ratio of 10:1 and a magnetic flowmeter calibrated to ±0.5% accuracy per ISO 10790:2015. Injection quills fabricated from alloy C-276 (UNS N10276) are inserted through 1-inch flanged nozzles at the tower shell, terminating in a 3-millimeter orifice designed to produce a jet velocity of 12 m/s to promote turbulent mixing with the pulp plug ascending at 0.8 m/min. Erosion-corrosion of the quill tips due to the impingement of sand- and grit-laden pulp at 105°C and pH 12.5 occurs at an observed rate of 0.18 mm/month, requiring scheduled replacement every 4 months to prevent caustic streaming that manifests as ±3.2 kappa unit variation in the blow tank sample standard deviation.

The direct relationship between caustic split ratio and selectivity index in Table 1 is not linear but exhibits a diminishing-returns behavior above 60% split to downstream injection points. The incremental improvement in viscosity for each additional 10% shift of caustic from pre-heater to mid-tower and top zones falls from +30 mL/g to +10 mL/g as the post-extraction kappa asymptotically approaches a floor value of 3.4 that is controlled by the inaccessible condensed lignin structures (5-5′ and 4-O-5′ bonds) resistant to alkaline peroxide attack. The D1-stage ClO₂ factor reduction from 0.26 to 0.21 translates into an annual chemical cost saving of approximately €1.1 million for the 1.8 million tpy mill at €0.65/kg ClO₂ equivalent, but these savings must be balanced against the capital cost of the quill injection system estimated at €380,000 and the maintenance burden of quill tip replacement at €12,000 annually.

Process Control Window Narrowing at High-Temperature Alkaline Peroxide Extraction

At Eop tower temperatures above 105°C, the caustic charge operating window constricts to a perilously narrow ±1.8 kg NaOH/odt around the setpoint, beyond which the pulp undergoes either insufficient delignification (kappa > 4.5) or catastrophic viscosity collapse (below 700 mL/g for eucalyptus). This thermal threshold is a direct consequence of the Arrhenius activation energy divergence between lignin ether cleavage (62.7 kJ/mol for β-O-4 bond homolysis under alkaline conditions) and the combined initiation-propagation steps of the peeling reaction at the reducing end group of cellulose (88–92 kJ/mol). When temperature increases from 100°C to 110°C, the delignification rate constant rises by a factor of 2.3, while the peeling rate constant increases by a factor of 3.8, narrowing the selectivity window proportionally. A southern hemisphere market pulp mill operating a 650 tpd line with a Metso (now Valmet) Eop tower of 42 minutes retention time reported that a single deviation to 108°C during a steam sparger control valve malfunction, combined with a caustic charge at the upper end of the normal range of 24 kg NaOH/odt, produced a 45-minute batch of pulp with viscosity 623 mL/g and copper number 3.8 (TAPPI T 430), rendering the pulp unfit for the tissue market and necessitating its diversion to low-grade fluff pulp inventory at a €45/tonne price discount. The incident prompted installation of a triple-redundant temperature interlock system with a safety integrity level of SIL 2 per IEC 61511, which automatically reduces caustic flow by 15% when the tower top temperature exceeds 103.5°C for more than 90 seconds.

The kinetics of the peeling reaction are further influenced by the calcium concentration in process water, which, when exceeding 80 mg/L as CaCO₃, precipitates as calcium hydroxide or carbonate onto the fiber surface at a local pH of 12.0, creating a solid-state buffering environment that sustains alkali activity within the fiber wall even after bulk-phase caustic has been consumed. The calcium scale layer, identified via SEM-EDX analysis of fiber cross-sections, exhibits a thickness of 0.3–0.7 μm and functions as a cation-exchange reservoir, releasing Na⁺ ions into the S1 layer over an extended period of 15–25 minutes beyond the tower discharge. This latent alkalinity reduces the effective delignification selectivity by promoting post-extraction carbohydrate degradation in the blow tank and early D1 stage where temperature remains at 75°C–85°C. The mitigation strategy involves continuous injection of a chelant—diethylenetriaminepentaacetic acid (DTPA)—at 0.8 kg/odt into the Eop stage pre-heater using a dosing pump controlled by a calcium-specific ion electrode (Orion 9720BNWP) that maintains a free Ca²⁺ concentration below 25 mg/L in the extraction filtrate, as per EPA Method 215.2.

Recovery Cycle Integration Constraints

Alkali charge optimization in the bleach plant is inextricably linked to the sodium-sulfur balance in the kraft recovery boiler and causticizing plant. For every 1.0 kg of NaOH introduced as makeup caustic to the Eop stage in excess of the mill’s pulping chemical balance closure, the electrostatic precipitator ash sulfation index increases by 0.17 units, as defined by the ratio of Na₂SO₄ to Na₂CO₃ in the recovery boiler dust collected at 320°C. This shift elevates the first melting temperature of the deposit from 745°C to 770°C, raising the risk of superheater tube fouling and eventual smelt-water explosion hazard. Mills operating ECF sequences with Eop-stage NaOH charges above 22 kg/odt on softwood routinely report a sulfidity imbalance that drives the green liquor sulfidity from an optimal range of 36%–40% to 43%–47%, attributable to the net import of sodium relative to sulfur. The viscosity penalty of caustic overcharging thus acquires an additional dimension: the corresponding increase in recovery boiler operating risk and steam generation efficiency loss of 2.1–2.8 tonnes of steam per tonne of excess NaOH charged, as the recovery boiler’s reducing capacity is partially diverted to reduce sulfate to sulfide rather than producing carbonate for causticization via the Andritz or ANDRITZ-type slaker-classifier loop operating at a TTA of 105–115 g/L as Na₂O.

At a 1,450 tpd integrated mill with a 5,200 m² recovery boiler firing 2,400 tds/day, the accumulation of excess sodium from bleach plant caustic over a 12-month campaign caused a 4.8% increase in reduction efficiency variance, forcing three unplanned water washes of the superheater sections that each incurred 72 hours of lost production and €275,000 in chemical and downtime costs. The mill implemented a closed-loop chemical balance control algorithm based on the GEMS (General Energy and Material balance System) model, which enables daily adjustment of the causticizing plant’s lime feed rate and bleach plant extraction caustic setpoint to maintain a Na/S ratio of 1.95±0.05 in the green liquor. Delignification selectivity, in this integrated context, is no longer a standalone bleach plant optimization problem but a multi-departmental constraint problem solvable only by dynamic pulp mill-wide simulation software certified under ISO 14040:2006 life cycle assessment protocols.

The chemical analysis required for maintaining this balance involves the titration of green liquor total titratable alkali (TTA), active alkali (AA), and sulfidity according to SCAN-N 30:85 and SCAN-N 33:94, completed at a frequency of every 2 hours during stable operation and every 30 minutes following a liquor cycle upset. Bleach plant caustic consumption is directly monitored via a Coriolis mass flowmeter (Endress+Hauser Promass F 300) on the 50% NaOH supply line, with data integrated into the mill’s PI system for material balance closure at a sampling rate of 0.5 Hz. The historical trend data reveals a sinusoidal seasonal pattern in caustic demand with an amplitude of ±1.4 kg/odt, correlated with the incoming wood chip moisture content variation from 38% in dry-season Eucalyptus urophylla to 52% in wet-season harvests.

The analytical limits in Table 2 are derived from the reference document for Best Available Techniques in the pulp and paper industry (JRC Science for Policy Report EUR 29427 EN). The DTPA limit of <0.20 kg/odt indirectly constrains the caustic charge because higher NaOH doses require proportionally higher chelant addition to sequester the increased dissolved metal load from the pulp, potentially exceeding the permit threshold. The chlorate limit of <0.15 kg/odt is relevant because caustic overdosing in the extraction stage raises the pH of the D1 stage filtrate from 3.8 to 4.2, slowing chlorine dioxide decomposition to chlorate and thus reducing the process’s intrinsic chlorate destruction mechanism; this forces the installation of a biological chlorate removal stage (anaerobic selector reactor) with a hydraulic retention time of 4.5 hours, adding €0.75/odt in operating cost.

A production-scale case study from a 1,200 tpd mixed hardwood line processing Acer rubrum, Quercus alba, and Betula papyrifera at a 55:30:15 ratio reports that the caustic charge sensitivity of the oxidative extraction stage is compounded by the heterogeneous anatomical composition of the chip furnish. Vessel elements from red oak (Quercus alba) with a diameter of 180–340 μm exhibit faster alkaline liquor penetration than libriform fibers, creating microscopic caustic concentration gradients that result in a measured 6.7% lower viscosity in the vessel-rich fraction after screening on a 0.50 mm slot screen (Bauer McNett classifier, TAPPI T 233 cm-15). The mill’s process engineering group compensated by installing an inline refining step (a single-disc refiner at 0.5 kWh/odt) immediately after the Eop blow tank to homogenize fiber wall damage distribution, which improved the pulp’s tensile index by 1.8 N·m/g at the expense of a 2.3% increase in fines generation. The acceptability of this trade-off is evaluated against the end-use specification for woodfree uncoated paper grades requiring a drainage resistance (Schopper-Riegler) of 28–32 °SR per ISO 5267-1:1999.

The final selectivity of an ECF sequence is also influenced by the physical form of the caustic addition hardware under conditions of high organic scaling. When a 50% NaOH solution is mixed with extraction-stage filtrate containing 18–24 g/L dissolved organic carbon (DOC) of ligneous origin, calcium- and barium-lignosulfonate complexes precipitate as sticky, brown-black deposits on the inside walls of the alkali feed pipe, narrowing the effective diameter from 50 mm to 34 mm within 6 weeks of operation. This fouling induces a pressure drop increase from 0.8 to 2.3 bar, causing the magnetic flowmeter’s signal-to-noise ratio to degrade below the manufacturer’s specification of 10:1, necessitating a shift to a Coriolis meter insensitive to coating build-up (Rotamass Total Insight from Yokogawa). The resulting uncertainty in caustic flow measurement of ±1.8% of span translates into a real selectivity fluctuation of ±0.12 kappa units and ±22 mL/g viscosity, undermining the ability to certify the pulp grade as prime quality under ISO 11475:2017 for diffuse blue reflectance factor.

When a caustic charge increase is mandated by a shift in incoming chip species from pure Eucalyptus nitens to a E. nitens–E. globulus blend at 60:40 ratio, the extraction-stage alkali demand rises from 18 to 24 kg NaOH/odt due to the 3.1–4.0 percentage unit increase in initial kappa number and the higher syringyl-to-guaiacyl (S/G) ratio of E. globulus lignin, which forms more alkali-resistant carbon-carbon bonds during pulping. The mill’s response using a Valmet Kappa Q analyzer on the post-oxygen washer outlet pulp and a feedforward caustic control loop with a setpoint of 3.2 post-extraction kappa reduced the output viscosity standard deviation from ±51 to ±28 mL/g over a 14-day transition period.