Sodium Oxalate Stability in Alumina Digestion Liquor Makeup

Within the continuous closed-loop caustic circuit that constitutes the Bayer alumina refining process, the accumulation of sodium oxalate (Na₂C₂O₄) represents a persistent thermodynamic and hydrodynamic challenge to digestion liquor stability. Organic carbon species introduced with bauxite—primarily humic substances, lignin derivatives, and aliphatic carboxylates—undergo oxidative alkaline degradation in the presence of dissolved oxygen and high-temperature sodium hydroxide during the digestion stage (typically 140 °C to 280 °C depending on bauxite mineralogy, with boehmitic and diasporic feeds demanding the upper range). The degradation cascade passes through a series of colored intermediate compounds before terminating in the most refractory C₂ product: the oxalate anion, which couples with sodium ions to form a sparingly soluble salt exhibiting strong inverse temperature solubility behavior in concentrated aluminate liquors. This characteristic creates a processing paradox where oxalate remains largely dissolved under digestion conditions but precipitates aggressively as the liquor cools during pressure relief, heat exchange, and residue washing—depositing a dense, hard scale of anisotropic needle-like crystals (typically acicular, with aspect ratios exceeding 20:1) on tube walls, valve seats, and flash tank internals. In the specific context of liquor makeup—the sequence of unit operations through which evaporator condensate, wash water, and fresh caustic are combined and heated before being charged to the digestion train—sodium oxalate stability governs both the physical continuity of flow and the economic bleed rate of process liquor purged to control organic inventory. The solubility envelope, nucleation kinetics, and crystal habit of this solute are sensitive to multiple interdependent parameters: free NaOH concentration (expressed as Na₂O_c), alumina-to-caustic ratio (A/C), total organic carbon load, the presence of specific inorganic and organic impurities, and the thermal history of the liquor across the makeup circuit.

What Limits the Solubility of Sodium Oxalate in Synthetic Bayer Liquor Makeup Streams?

The equilibrium solubility of sodium oxalate in pure sodium hydroxide solutions follows a monotonic increase with rising caustic concentration and temperature; however, the addition of dissolved aluminate species profoundly suppresses this solubility through a common-ion effect and through the formation of complex ion pairs in solution. Published experimental data obtained using the isothermal saturation method with conductivity and ion-chromatographic verification (detection limits approximately 0.5 mg/L as C₂O₄²⁻) indicate that at a typical North American refinery liquor makeup condition of 120 g/L Na₂O_c and an A/C ratio of 0.65, the equilibrium oxalate concentration at 90 °C is in the range of 2.2 to 2.8 g/L, while at 50 °C the solubility collapses to less than 0.7 g/L. This steep gradient—amounting to a solubility loss of more than 70 % across a 40 K temperature interval—establishes the fundamental risk window for uncontrolled nucleation in shell-and-tube heat exchangers, particularly in the cold-side tube sheet region where laminar boundary layers permit localized supersaturation to develop even when the bulk liquor composition remains within metastable limits. The solubility is also non-linear with respect to free caustic: at a fixed temperature of 75 °C, increasing the Na₂O_c from 100 g/L to 180 g/L approximately doubles the oxalate capacity, yet the simultaneous elevation of aluminate concentration required to maintain a constant A/C ratio partially offsets this gain, resulting in a net solubility increase of only 40–60 %. These interdependencies mandate that liquor makeup temperature profiles be designed not simply to avoid the saturation limit, but to remain above the critical supersaturation boundary at which heterogeneous nucleation on existing scale surfaces or suspended particulates becomes kinetically favored.

In continuous make-up systems serving digestion units with a throughput exceeding 500 t/h of liquor, the thermal stability of sodium oxalate is further complicated by the recirculation of spent liquor from the residue washing circuit. This stream carries with it fine particulate matter—red mud tailings with a median particle diameter (D₅₀) below 10 µm—which act as active nucleation substrates, lowering the induction time for oxalate crystallization from several hours to mere minutes under certain thermal conditions. Plant operational data from Australian refineries processing Weipa bauxite, as reported in the peer-reviewed literature, have demonstrated that when the suspended solids load in the makeup tank exceeds 50 mg/L, the metastable zone width narrows by approximately 30 %, reducing the safe operating temperature margin. Consequently, makeup tank hydrodynamics must incorporate either a stilling well or a lamella clarification section to maintain suspended solids below this empirically derived threshold, or else the residence time distribution must be narrowed through baffling to prevent low-velocity dead zones where solids can accumulate and catalyze oxalate scaling.

In the direct vicinity of live steam injection nozzles used to heat cold condensate returning from the evaporation plant, localized temperature spikes generate transient solubility maxima that, paradoxically, can aggravate downstream precipitation. The liquor exiting the immediate mixing zone may be at 105 °C while the bulk tank temperature is held at 70 °C. This local overheating temporarily dissolves residual oxalate nuclei that had formed in cooler condensate return lines, but as the heated plume disperses into the cooler bulk volume, a rapid cooling quench occurs at a rate often exceeding 2 K/min, a kinetic regime that favors the production of a fine, filterable oxalate crystal mass rather than slow, adherent scale growth. However, if the conveyance piping between the makeup tank and the digestion feed pump incorporates a section of uninsulated schedule 40 carbon steel pipe exposed to ambient air at 20 °C, the resulting heat loss of 3–5 K can shift the liquor from a metastable state into the labile region where spontaneous homogeneous nucleation occurs. This pipe segment inevitably becomes a chronic scaling point, necessitating mechanical descaling at intervals dictated by the oxalate deposition rate, which in severe cases may be as short as 14 days.

Thermodynamic Drivers of Precipitation and the Oxalate Scaling Window in Heat Exchange Equipment

The driving force for sodium oxalate precipitation is the difference between the actual solute activity product and the equilibrium solubility product (K_sp), a quantity that is strongly path-dependent in Bayer liquors due to the extensive hydrogen-bonded network formed between aluminate ions, hydroxide ions, and water molecules. The activity coefficient of the oxalate ion in a 6 M NaOH background electrolyte differs by a factor of approximately 3 from its value in dilute aqueous solution, as determined by vapor pressure osmometry and Pitzer ion-interaction model parameterization. This non-ideality means that solubility data generated in pure NaOH or NaCl brines cannot be directly transposed to authentic aluminate liquors without applying a correction factor derived from the Debye-Hückel limiting law extended with specific short-range interaction parameters for the Al(OH)₄⁻–C₂O₄²⁻ pair. The most accurate predictive models currently available in engineering practice, such as those embedded in proprietary process simulation packages used by major technology licensors, incorporate temperature-dependent binary and ternary interaction parameters validated against plant data spanning a free caustic range of 60 to 250 g/L and temperatures from 30 °C to 110 °C. Even with these advanced models, the root-mean-square deviation between predicted and measured oxalate concentrations during dynamic plant operation can exceed 15 % during transient events such as a bauxite changeover, underscoring the value of redundant online analytical measurements.

The scaling window—defined as the range of operating conditions within which the linear growth rate of oxalate scale on carbon steel surfaces exceeds 0.1 mm/month—is narrowest when the liquor’s total organic carbon (TOC) loading is low and when the specific surface area of the heat exchanger is minimized through the use of larger tube diameters. A comparative analysis of three shell-and-tube heat exchanger configurations in a South American refinery indicated that units with 19 mm OD tubes experienced oxalate scaling rates up to five times higher than those with 32 mm OD tubes, primarily due to the higher wall shear stress reducing the boundary layer thickness and thus the concentration polarization of oxalate at the tube wall. Service history records from this same facility show that the interval between chemical cleaning cycles (using inhibited sulfamic acid at 60 °C per ASTM G1-03 guidelines for cleaning procedure evaluation) extended from approximately 90 days for the smaller-diameter units to over 400 days for the larger-diameter exchangers, a difference which translates directly into reduced maintenance expenditure and higher annual availability of the digestion circuit.

When the liquor composition crosses the solubility boundary, the initial precipitate is not the thermodynamically stable anhydrous sodium oxalate but rather a metastable hydrate phase that gradually transforms, releasing waters of crystallization and altering the crystal lattice parameters. This ageing process—which can take 48 to 72 hours at 40 °C—generates internal stresses within the scale layer, leading to micro-cracking and partial spalling. In horizontal piping runs carrying partially cooled liquor at a velocity below 0.8 m/s, these spalled fragments accumulate in the lower quadrant of the pipe cross-section and sinter together under the overburden pressure of subsequent scale layers, eventually forming a dense, hard deposit that cannot be removed hydrodynamically and must be drilled out mechanically. The Mohs hardness of such aged scale specimens removed from a Caribbean refinery’s residue washing circuit was measured at 4.5, comparable to fluorite, necessitating the use of tungsten carbide-tipped cutting tools for removal.

How Does the Presence of Sodium Carbonate in Makeup Liquor Influence Oxalate Stability?

Sodium carbonate, inevitably present in Bayer process liquors due to the reaction of atmospheric CO₂ with caustic soda during storage, handling, and process exposure, exerts a measurable salting-out effect on sodium oxalate solubility. At a constant NaOH concentration and temperature, each 10 g/L increase in Na₂CO₃ (expressed as Na₂O_c equivalent) reduces the equilibrium oxalate solubility by 5 to 8 %. This effect is additive across the typical carbonate inventory range found in industrial circuits, which varies from 15 g/L to over 80 g/L depending on the extent of causticization practiced at the particular refinery. In facilities where a lime kiln and slaking system are employed to recausticize the carbonate bleed stream, the success of the oxalate control strategy is inextricably linked to the mass balance closure of the causticization loop: incomplete conversion in the slaker, due to particle size issues or short residence time, returns significant quantities of carbonate to the main liquor, progressively worsening the oxalate salting-out until the apparent solubility limit drops below the target operating envelope. The measurement of carbonate in the makeup liquor stream is typically performed by a double-endpoint titration method automatically executed by an inline titrator, with the first endpoint at approximately pH 8.3 (phenolphthalein) corresponding to the conversion of carbonate to bicarbonate, and the second at pH 4.5 (methyl orange) corresponding to complete neutralization. The precision of this measurement, traceable to the primary standard method ISO 3196:1975 for hydroxide and carbonate determination, must be maintained within ±0.5 g/L to ensure that the solubility model input parameters do not accumulate systematic error over repeated batch makeup cycles.

During makeup tank operation, where solid sodium hydroxide pellets or 50 wt% membrane-grade caustic solution is combined with process condensate and carbonate-laden spent liquor, the mixing sequence impacts the local pH gradient and thus the carbonate speciation. If concentrated caustic is added directly into a large volume of carbonate-containing solution without sufficient agitation (> 0.5 W/kg specific power input, per typical stirred tank design guidelines), a transient high-pH zone forms where bicarbonate and carbonate ions are rapidly interconverted, temporarily lowering the local carbonate activity and creating a false “window” of higher oxalate solubility. As convective and diffusive mixing subsequently homogenize the tank, this local anomaly dissipates and the equilibrium carbonate concentration is restored, possibly triggering a burst nucleation event of sodium oxalate that manifests as a sudden turbidity spike. To avoid this operational pitfall, best practice in modern refinery design specifies that the makeup tank be fitted with a side-entry agitator delivering a tip speed of at least 4 m/s and that the caustic addition point be located in the suction eye of a recirculation pump rather than at the liquid surface, thereby ensuring immediate dilution and heat dissipation.

Where the refinery configuration does not include an external oxalate removal side stream, the only mechanism for controlling oxalate inventory is the periodic purging of a fraction of the total liquor volume. The economic optimization of this purge stream—balancing the cost of caustic and alumina lost against the cost of oxalate-related scaling and availability losses—requires input from a kinetic model of oxalate accumulation that accounts for bauxite organic carbon input rate, degradation kinetics in the digester, solubility limitations in the lowest-temperature point of the circuit (often the primary security filtration step following residue separation), and the partition coefficient of oxalate between the liquor and the red mud solids (which adsorb oxalate at a loading of roughly 0.1–0.3 mg C₂O₄²⁻ per gram of dry mud). This level of complexity typically exceeds the capability of a simple steady-state spreadsheet and demands a dynamic process simulation linked to the plant distributed control system (DCS) historian for continuous recalibration against measured oxalate levels.

In the northern hemisphere winter operating months, when ambient temperatures around outdoor tanks drop below −10 °C, the thermal management of liquor makeup becomes particularly acute in the context of oxalate stability. Makeup water lines, if not heat-traced and insulated per ASME B31.3 process piping code with a minimum insulation thickness corresponding to a heat loss of less than 15 W/m², can cool the water feed to near-freezing. When this chilled water contacts warm returning spent liquor at 70 °C to 80 °C, not only does the bulk mixed temperature drop but, critically, the local undercooling at the mixing interface promotes instantaneous oxalate nucleation before the thermal mass of the tank can equilibrate the temperature. The resulting fine crystalline suspension, with a particle size distribution centered around 5–10 µm, is highly effective at settling in low-velocity areas of downstream equipment such as the tube passes of barometric condensers, where it undergoes Ostwald ripening into larger, blocky crystals with a terminal settling velocity exceeding 0.5 mm/s, eventually plugging the tube cross-section.

Operational experience at a European alumina refinery processing diasporic bauxite identified a critical threshold: when the temperature difference between the returning spent liquor stream and the combined condensate/makeup water stream exceeded 25 K, the turbidity of the mixed liquor measured by an inline nephelometric turbidity meter calibrated to 0.1 NTU resolution increased by an order of magnitude within 30 seconds of contact, indicating massive instantaneous nucleation. The corrective action implemented involved splitting the cold condensate addition into multiple points distributed along the length of the recirculation loop, each at a flow rate that ensured a local temperature differential of no more than 10 K. This distributed addition scheme required the installation of a multi-port sparger fabricated from duplex stainless steel (UNS S32205) to withstand the corrosive alkaline environment containing 0.5 g/L dissolved silica, which is known to induce caustic stress corrosion cracking in conventional 300-series stainless steels.

The above tabulated values demonstrate the non-linear interaction between caustic strength and aluminate loading. In makeup operations where the target A/C ratio varies seasonally due to gibbsite precipitation circuit constraints, the operator must anticipate the consequential shift in the oxalate solubility floor and adjust the target oxalate inventory accordingly, using the table as a forward-looking reference rather than a trailing indicator. The standard deviations reported reflect the inherent variability arising from trace organic impurities in the synthetic liquor and are representative of the best currently attainable reproducibility without sophisticated multistage purification of the starting reagents.

When sodium oxalate does precipitate within the makeup circuit, its crystal morphology is not a simple function of supersaturation alone but is profoundly influenced by the cooling rate and the presence of growth-modifying anionic species. Nuclear magnetic resonance spectroscopy (¹³C NMR at 100 MHz) of the oxalate ion in alkaline solution indicates that the two carboxylate groups are in rapid exchange with the surrounding hydration sphere, and that specific adsorption of aluminate oligomers onto the fastest-growing crystal faces can retard growth along the crystallographic c-axis, producing a more equant crystal habit that is less prone to interlocking and forming a cohesive scale network. This observation has motivated the deliberate addition of crystal habit modifiers, both organic (such as sodium gluconate at dosages of 50–100 mg/L) and inorganic (such as dilute concentrations of calcium ion, typically below the lime solubility limit of approximately 3 mg/L at 90 °C in 140 g/L NaOH), to alter the deposition characteristics of any oxalate that does nucleate. However, the efficacy of these additives must be weighed against their potential downstream interference with gibbsite precipitation, where even sub-ppm levels of certain polyol additives can severely inhibit crystal growth and reduce product particle size.

In the context of critical threshold risks, the margin between stable operation and incipient scaling in the makeup system narrows dramatically when the free caustic concentration drops below 90 g/L—a condition that can occur during startup after a maintenance shutdown if the initial caustic charge is incorrectly calculated or if the condensate return line contains a significant dead leg of diluted liquor. At such reduced causticity, the solubility of sodium oxalate may be less than 0.5 g/L at the 65 °C to which the makeup tank is typically heated by residual heat exchange from nearby operating vessels. If the standing inventory of oxalate in the plant exceeds this value—and it almost invariably does, with typical process liquors carrying 2.5–4.0 g/L—the entire volume of liquor in the makeup tank will become instantly supersaturated upon mixing, and unless the agitator is started immediately and the caustic concentration is rapidly increased by addition of fresh caustic, the tank internals and the suction strainer of the transfer pump will be coated with a thin, tenacious layer of oxalate scale within a matter of hours. Recovery from such an event requires either a full chemical cleaning of the tank using hot inhibited acid, or, in mild cases, a prolonged soak with a highly caustic, oxalate-undersaturated solution at 95 °C circulated at high velocity through the affected piping, a procedure that may take 24–36 hours to restore full flow capacity.

Instrumentation and Control Strategies for Real-Time Oxalate Management

The transition from periodic laboratory reporting to continuous online monitoring of oxalate concentration has been enabled by advances in Raman spectroscopy coupled with chemometric modeling, specifically the application of partial least squares (PLS) regression to the spectral region between 900 cm⁻¹ and 1500 cm⁻¹ where the symmetric and asymmetric C–O stretching modes of the oxalate ion produce distinguishable bands. A robust industrial Raman probe, constructed with sapphire windows and a Hastelloy C-276 immersion body to withstand the alkaline environment, can be installed directly in the recirculation loop of the makeup tank via a 2-inch flanged retractable fitting, providing a measurement update every 30 seconds with a detection limit of approximately 0.2 g/L and a precision of ±0.15 g/L after calibration against a series of reference liquors whose oxalate content has been independently verified by ion chromatography according to the procedures established in ASTM D4327-17 for anions in water by chemically suppressed ion chromatography. The critical success factor for reliable long-term operation of such a probe is not the optical stability—which is excellent, as the Raman signal arises instantaneously from the inelastically scattered light—but the maintenance of a clean optical interface free of scale or particulate fouling. Automated retraction and cleaning mechanisms that use a high-pressure flush of filtered process liquor synchronized with the plant distributed control system (DCS) have demonstrated mean time between maintenance of greater than 6 months in pilot installations.

Closed-loop control of oxalate stability can be implemented by coupling the online Raman analyzer output to the speed control of the condensate return pump and the setpoint of the steam injection control valve, thereby maintaining the temperature of the mixed liquor above the critical supersaturation limit as defined by the real-time composition. The control algorithm, typically a model-predictive controller (MPC) with a linear dynamic matrix relating the manipulated variables (steam flow, condensate return rate, fresh caustic addition rate) to the controlled variable (calculated supersaturation index), requires a robust state estimator because the oxalate concentration itself exhibits a significant time delay of several minutes between a change in makeup composition and the equilibration of the tank inventory. The state estimator is usually implemented as an extended Kalman filter that ingests the Raman measurement, the tank level, and the flow rates and temperatures of all inlet streams, and provides a filtered estimate of the current oxalate concentration and its rate of change. This estimate is then used by the MPC to move the operating point within a defined safe region, avoiding the oxalate precipitation boundary while simultaneously observing limits on the maximum steam flow to prevent localized boiling and on the maximum fresh caustic addition to prevent overshooting the target Na₂O_c setpoint, which would impose an unnecessary chemical cost and increase the purge liquor volume required to control the total Na₂O inventory.

A separate but complementary approach to stabilizing sodium oxalate in the makeup circuit involves the deliberate addition of a side-stream oxalate crystallization reactor that operates on a slipstream of the circulating liquor and is maintained at a temperature of 35–45 °C, well below the main liquor circuit temperature. This cold crystallizer, typically a draft-tube baffled (DTB) unit fabricated from rubber-lined carbon steel to resist caustic attack, provides the residence time (on the order of 4–6 hours) necessary for the precipitation of coarse oxalate crystals that can be separated by filtration and removed from the system as a solid, rather than building up in the liquor inventory. The successful operation of such a unit depends on precise temperature control—within ±1 K of the setpoint—to ensure that the supersaturation driving force is maintained within the metastable zone where crystal growth on existing seed crystals dominates over nucleation, thereby producing a filterable product. The seed crystal inventory is maintained by recirculating a fraction of the classified product back to the crystallizer as seed slurry, with a seed mean particle size of approximately 150 µm and a seed loading of 50–80 g/L. Under these carefully tuned conditions, the oxalate removal rate can be matched to the input rate from bauxite organics, stabilizing the circuit inventory at a target value determined by the liquor temperature profile and the caustic concentration, without the need for excessive liquor purging.

One of the most frequently overlooked operational parameters affecting long-term sodium oxalate stability in makeup systems is the quality of the evaporator condensate used as makeup water. Condensate from multi-effect evaporators operating on caustic aluminate solutions inevitably contains traces of volatile organic compounds (VOCs) such as methanol, ethanol, acetone, and acetone degradation products, which are not captured by the standard conductivity-based quality checks. These VOCs, even at total concentrations below 50 mg/L TOC, can participate in a secondary oxidation cycle when the condensate is reheated in the presence of dissolved oxygen and residual transition metal ions (particularly iron leached from carbon steel piping, typically present at 0.2–1.0 mg/L). The oxidation of these VOCs generates further oxalate ion directly within the makeup tank, contributing an incremental load that is difficult to quantify because the oxidation rate depends on the catalytic activity of the dissolved iron, which in turn varies with the age and condition of the upstream piping. Published data for this specific VOC-to-oxalate pathway in the condensate return system of alumina refineries is limited; however, plant mass balance calculations at a Queensland refinery indicated that the condensate-derived oxalate increment could amount to as much as 15 % of the total oxalate input during winter operation when the organic content of the bauxite feed was low, making it a non-negligible source. Mitigation, where required, involves either the installation of a condensate polishing unit using activated carbon adsorption (granular activated carbon with a specific surface area exceeding 1000 m²/g) or the substitution of a fraction of the condensate with demineralized water from a ion-exchange polishing plant, though the latter option significantly increases the process water cost.

All boundaries listed in Table 2 are derived from a synthesis of published plant case studies, equipment manufacturer recommendations, and fundamental thermodynamic and kinetic considerations. The lower temperature boundary of 70 °C for the mixed liquor is not a fixed constant but context-dependent: in refineries where the oxalate inventory is maintained below 1.0 g/L by aggressive purge or side-stream crystallization, the safe temperature floor can be lowered to approximately 55 °C without inducing scaling, as validated by a multi-year data set from a low-oxalate Eastern European plant. Conversely, a refinery carrying an oxalate burden above 4.0 g/L must hold a minimum temperature of at least 85 °C even during turndown operation, imposing a higher energy overhead. These interdependencies emphasize that oxalate stability cannot be isolated from the overall organic carbon management strategy of the refinery; a myopic focus on the makeup tank alone, without simultaneous attention to the bauxite feed organic assay, the digestion air injection rate (which influences the oxidative degradation rate), and the performance of the organic removal side stream, will result in frequent excursions outside the safe operating envelope and chronic production losses attributed to unscheduled caustic cleaning cycles.

During the detailed design of a new digestion liquor makeup facility, the specification of the tank geometry and the heat exchanger configuration must undergo a hazard and operability (HAZOP) review that specifically addresses oxalate scaling as a credible process deviation. The guide words “less temperature” and “more caustic” applied to the makeup tank node will generate risk scenarios that are best mitigated by a combination of engineering controls: a dedicated steam sparger with a modulation valve fail-closed to prevent overheating in the event of control signal loss, a high-level alarm combined with an independent high-high-level shut-off to prevent overfilling and dilution of caustic concentration, and a low-flow alarm on the recirculation loop to ensure that the tank contents are never stagnant when heated—a condition that promotes thermal stratification and localized supersaturation. The materials of construction for the tank and attached piping are typically carbon steel (ASTM A36 for plates, ASTM A106 Grade B for pipes) with a specified corrosion allowance of 3 mm to account for the slow thinning caused by alkaline attack in the absence of adequate inhibitor (the aluminate ion itself acts as a mild corrosion inhibitor, but at the low A/C ratios typical of makeup liquor, typically 0.60–0.70, the corrosion rate of carbon steel can be on the order of 0.05 mm/year at 80 °C according to long-term coupon exposure data). Stress corrosion cracking is not a significant risk at the temperatures and NaOH concentrations encountered in the makeup system, provided the steel is post-weld heat treated (PWHT) to relieve residual stresses in accordance with the requirements of ASME Boiler and Pressure Vessel Code Section VIII, Division 1, for welded pressure vessels.

Regarding the interaction of sodium oxalate with other scale-forming species encountered in Bayer liquors—specifically aluminum hydroxide (gibbsite scale), sodium aluminum silicate (desilication product scale), and calcium carbonate—simultaneous precipitation can produce composite scale layers of extreme hardness and chemical resistance. In the specific temperature window of 60–80 °C characteristic of the cooler sections of the makeup circuit, sodium oxalate and aluminum hydroxide can co-precipitate if the liquor is also supersaturated with respect to gibbsite due to a sudden drop in temperature following the dissolution of freshly precipitated gibbsite seed in the residue washing circuit. The resulting scale is a laminated structure of oxalate needles and pseudo-boehmite gel, which binds the oxalate crystals into a cohesive matrix that is exceptionally resistant to mechanical removal by high-pressure water jetting at 70 MPa (typical maximum for rotating nozzle cleaners). In such cases, chemical cleaning with a sequential application of warm inhibited hydrochloric acid (to dissolve the aluminum hydroxide matrix) followed by a hot caustic/oxidant solution (to remove the residual organic component) is required, with a total cleaning time approaching 12–18 hours per affected pass. This operational reality underscores the need to maintain a minimum temperature margin above the gibbsite and oxalate co-saturation envelope, a margin that can be as narrow as 5 K in borderline liquor compositions and requires stringent control to hold.

The startup sequence following a major scheduled turnaround, when the entire digestion liquor circuit is initially filled with a fresh charge of caustic solution and condensate, represents a uniquely vulnerable period for oxalate instability. Because the fresh liquor contains essentially no dissolved oxalate, the saturation index is intentionally low, and dissolution of residual oxalate scale from pipe walls and tank surfaces occurs rapidly, enriching the liquor with oxalate in an uncontrolled manner. The dissolution rate of oxalate scale in hot, turbulent caustic solution follows a first-order rate law with an activation energy of approximately 45 kJ/mol, meaning that the dissolution half-life decreases by a factor of about 4 for every 20 K increase in temperature. During the initial heat-up phase of the startup, when the liquor temperature is ramped from ambient to operating temperature at a rate of 15 K/h, the oxalate concentration can rise from 0 g/L to 1.5 g/L within the first 4 hours simply by scavenging scale from internal surfaces. This enrichment can push the liquor into a condition where, as the temperature stabilizes at the normal operating point, it is already supersaturated relative to the stable solubility limit, triggering a precipitation episode that defeats the purpose of the lengthy cleaning outage. Experienced plant operators anticipate this phenomenon by dosing a small quantity of a chelating agent—sodium ethylene diamine tetraacetate (EDTA) at a concentration of 100 mg/L—into the initial fill to complex dissolved calcium and transition metal ions that might catalyze oxalate oxidation, and by incorporating a high-flow recirculation loop that promotes turbulent mixing and minimizes dead zones where dissolution rate is limited by mass transfer. The final temperature ramp is paused for a dwell period of 2–3 hours at 55–60 °C to allow any precipitated oxalate to form in a controlled, filterable morphology before the liquor is forwarded to the digestion feed pumps, thus preventing transport of seed crystals into the digester where they would exacerbate scaling. Published data for the effectiveness of this dwell-period strategy at varying ramp rates and final target temperatures is not comprehensively documented in the open literature, but internal technical reports from multiple operating companies confirm its utility as a best practice for mitigating startup oxalate surges.