Nickel Alloy Selection for Epichlorohydrin Caustic Dehydrochlorination

The caustic dehydrochlorination of dichloropropanol to epichlorohydrin proceeds in a reaction medium that simultaneously exposes process equipment to 10–20 wt% free NaOH, NaCl concentrations that can exceed 150,000 mg/L, and residual volatile organic chlorides at temperatures sustained between 80°C and 120°C. Carbon steel pressure vessels designed to ASME Boiler and Pressure Vessel Code Section VIII, Division 1 suffer catastrophic caustic stress corrosion cracking (caustic SCC) once the combination of temperature and alkalinity exceeds the thresholds documented in NACE SP0403; the safe operating envelope for post‑weld heat‑treated carbon steel is restricted to NaOH concentrations below 5% when the metal temperature reaches 65°C. Austenitic stainless steels such as 316L (UNS S31603) do not escape the hazard, because chloride stress corrosion cracking (Cl‑SCC) initiates above 60°C in the presence of even 10 ppm residual chlorides, a concentration trivial compared with the saturated brine that circulates through caustic stripping columns and static mixer reactors. The industrial response has been a transition to nickel‑based alloys whose passive films remain stable across the full alkalinity‑chloride pH‑temperature matrix, yet the qualification of a specific grade must reconcile the competing demands of uniform corrosion resistance, immunity to localised attack in stagnant zones, retention of toughness after multi‑pass welding, and economic fabricability in plate thicknesses up to 25 mm and welded tube lengths exceeding 12 m.

Data from plant audits conducted on five epichlorohydrin production lines operating in Southeast Asia and the US Gulf Coast reveal that the most frequent cause of unscheduled downtime is not general wastage but deeply penetrating intergranular attack (IGA) in the heat‑affected zone (HAZ) of weldments. In one documented case a 10,000 L agitated reactor built from 12 mm thick Alloy 600 (UNS N06600) plate per ASTM B168 was taken out of service after 18 months when dye‑penetrant inspection following a caustic wash identified a network of cracks extending 3–5 mm into the knuckle‑region weld. The process record indicated that the batch cycle included a 30‑minute hold at 110°C in 22% NaOH with 12% NaCl, and the vessel was steamed out between campaigns at 140°C. Metallographic sections prepared from the failed weldment, etched to reveal carbide precipitation according to ASTM A262 Practice A, exhibited a nearly continuous grain‑boundary network of chromium‑rich M₇C₃ carbides, and the nitric acid weight‑loss rate determined by ASTM A262 Practice C reached 0.45 mm/year on a specimen machined from the inner‑diameter HAZ. The low carbon content of Alloy 600 (≤0.15%) was insufficient to prevent sensitisation when the cooling rate through the 550–850°C range was slowed by the thermal mass of the 50 mm thick tubesheet to which the shell was welded. This failure mode underscores the necessity of selecting a nickel alloy whose grain‑boundary chemistry remains tolerant of fabrication‑induced thermal transients without the development of a chromium‑depleted zone exceeding 200 nm in width, a metric that directly governs IGA susceptibility in hot caustic chloride environments.

When hydrogen evolution and local cathodic alkalinity shift cause a severe pH depression beneath adherent deposits, the performance ranking of nickel‑chromium‑molybdenum alloys can invert relative to the general corrosion data generated in freely circulating autoclave tests. Alloy C‑276 (UNS N10276), which contains 15.0–17.0% Mo and 3.0–4.5% W, develops an air‑formed passive layer dominated by Cr(III) oxide under aerated 20% NaOH, yet the film becomes thermodynamically unstable when the interfacial pH falls below 3, a condition created inside tight crevices where hydrolysis of dissolved metal cations consumes hydroxyl ions. Accelerated crevice corrosion tests conducted according to ASTM G48 Method D in a 10% NaOH + 10% NaCl solution at 95°C give a critical crevice temperature (CCT) of approximately 75°C for Alloy C‑276 at an applied torque of 0.28 N·m on the multiple‑crevice assembly, but the same test on Alloy C‑22 (UNS N06022), with its higher chromium content (20.0–22.5%) and slightly lower molybdenum (12.5–14.5%), extends the CCT to above 90°C. In an operating plant this advantage translates directly into the service life of ring‑joint gasket seats on shell‑and‑tube heat exchangers that are exposed to splash‑zone evaporation cycles. A comparative trial at a European ECH facility replaced the Alloy C‑276 floating‑head cover with one machined from Alloy C‑22 plate conforming to ASTM B575, and after 26,000 hours of continuous operation at a tube‑side outlet temperature of 107°C, the average crevice penetration measured by ultrasonic phased‑array scanning was 0.08 mm versus 0.35 mm on the previously removed C‑276 cover that had seen an identical exposure period. The cost increment of the higher‑alloyed material was recovered within one maintenance cycle through the avoidance of a mid‑run gasket seat re‑machining.

At What NaOH Concentration and Temperature Does the Balance Between Transpassive Dissolution and Caustic Embrittlement Shift for Ni‑Cr‑Mo Alloys?

Caustic SCC in nickel alloys is predominantly a film‑rupture‑dominated process that requires a critical concentration of adsorbed hydroxide ions to destabilise the passive film at the crack tip while the crack flanks remain protected, thereby preserving a sharp aspect ratio. Slow strain rate tests performed in accordance with NACE TM0198 on Alloy C‑276 tensile specimens exposed to de‑aerated 50% NaOH at 150°C exhibit a reduction in elongation to fracture from 45% in inert silicone oil to 28%, and scanning electron microscopy of the fracture surface reveals mixed intergranular‑transgranular quasi‑cleavage features. The threshold NaOH concentration below which no environmentally assisted cracking is detected in NACE TM0198 tests on solution‑annealed C‑276 plate (hardness ≤ 100 HRBW) is 30% at 140°C and drops to 15% when the temperature is raised to 180°C, based on data compiled in the alloy producer’s technical circular and cross‑referenced with the caustic soda service diagram published by the Nickel Institute. In contrast, low‑carbon nickel 201 (UNS N02201, ≤0.02% C), which derives its corrosion resistance from metallic nickel rather than a chromium‑oxide film, is immune to caustic SCC up to the boiling point of 75% NaOH but suffers rapid erosion‑corrosion when the stream contains entrained NaCl crystals at velocities above 1.8 m/s. This velocity limitation was quantified on a test loop built from 2‑inch Schedule 40 pipe following ASTM G73 liquid impingement practice; mass loss on Nickel 201 coupons exposed for 500 hours at 100°C in 15% NaOH with 5% suspended NaCl increased from 0.02 mm/year at 1.5 m/s to 0.31 mm/year at 2.5 m/s. Thus, in the main circulating loop of an epichlorohydrin plant where pump‑discharge line velocities routinely exceed 2.0 m/s, the nickel‑chromium‑molybdenum alloys remain the default choice, whereas Nickel 201 can be reserved for caustic storage tanks operating below 50°C and in the absence of abrasive solids.

Many epichlorohydrin licensees have retrofitted caustic stripping columns with titanium‑lined shells (ASTM B265 Grade 2) to cope with the aggressive overheads containing HCl and water, but the interface between the titanium cladding and the nickel‑alloy column internals introduces a galvanic hazard that is often underestimated during the design phase. Under freely corroding conditions at 85–110°C, the open‑circuit potentials of titanium and Alloy C‑276 measured versus a saturated calomel electrode in 10% NaOH + 8% NaCl brine differ by only 80–120 mV, but the presence of dissolved oxygen from an unblanketed caustic feed tank can polarise the titanium to a passive potential that is 300–400 mV noble to the nickel alloy. A zero‑resistance ammeter study conducted on-site at a US Gulf ECR plant recorded a galvanic current density of 22 µA/cm² on Alloy C‑276 tray hardware when electrically coupled to a titanium shell at an area ratio of 1:50 (anode:cathode), which translates, via Faraday’s law, to a metal dissolution rate of approximately 0.25 mm/year if concentrated entirely on fastener shanks. The corrective measure implemented there was electrical isolation of the tray support rings using 0.5 mm thick PTFE gaskets and a program of regular potential monitoring using permanent Ag/AgCl reference electrodes inserted through the column wall, a configuration described in NACE SP0286. Subsequent ultrasonic inspection after 8,000 hours of operation confirmed that the isolated trays lost less than 0.05 mm of cross‑sectional thickness, whereas an earlier inspection of non‑isolated trays had revealed localised metal loss in excess of 1.2 mm adjacent to bolting.

Electrolytic effects beneath exfoliated corrosion product or polymer‑derived scales generate pH canyons as low as 1.5–2.0 measured by micro‑pH probes on cut‑out sections of under‑deposit pitting. In the bottom head of a caustic dehydrochlorination reactor operating at 100°C with a 5‑rpm anchor agitator, a stagnant layer of NaCl crystals and degraded epoxy‑phenolic coating fragments accumulated to a depth of 10‑15 mm and served as a diffusion barrier for dissolved oxygen, creating a differential aeration cell. Three pits with a maximum depth of 4.8 mm were found on Alloy 625 (UNS N06625) plate, which had been specified to ASTM B443 with a pitting resistance equivalent number (PREN = %Cr + 3.3 × %Mo + 16 × %N) of approximately 50 based on the actual mill certificate composition. The pits were concentrated in the region where the calculated wall shear stress from the agitator was below 0.5 Pa, a value that is insufficient to resuspend settled salts. After the vessel was returned to service with a polished surface finish of Ra 0.4 µm and with the agitator speed increased to 8 rpm to guarantee a minimum near‑wall velocity of 0.15 m/s, follow‑up eddy current examination after one year found no measurable propagation of the remaining pits. This operational history reinforces the importance of maintaining a surface finish finer than 0.5 µm Ra on all interior surfaces exposed to caustic brine, as recommended in the fabrication guidelines of ASME PCC‑2 for caustic service repairs, and of incorporating computational fluid dynamics into the vessel design to eliminate dead zones where the local chloride concentration can become hypersaline.

Selection of a filler metal for joining nickel alloys in epichlorohydrin caustic loops demands more than matching the base metal composition; the weld metal must tolerate dilution‑induced compositional shifts that locally lower the PREN below the resistance threshold for crevice corrosion at the operating temperature. For Alloy C‑276 base metal, the standard gas‑tungsten arc welding (GTAW) filler ERNiCrMo‑4 (AWS A5.14 ERNiCrMo‑4) yields a deposited weld chemistry close to the nominal base metal when the dilution ratio is maintained below 15%, yet in a single‑V groove joint with a root gap of 3 mm the root pass can be diluted up to 35% with elements from the parent material. Electron probe microanalysis of the diluted root pass on a 6 mm thick test coupon revealed a molybdenum content of only 12.8%, which lowered the calculated PREN to 43, below the 48 that had been empirically correlated with immunity to crevice corrosion in the plant’s specific brine chemistry. The fabricator therefore substituted the root pass with ERNiCrMo‑10 filler wire (AWS A5.14 ERNiCrMo‑10), whose nominal molybdenum content of 13.0–15.0% and chromium of 20.0–22.5% provided a compositional buffer that kept the diluted PREN at or above 50. This practice was incorporated into the welding procedure specification (WPS) that was qualified by test coupons subjected to the ASTM G48 Method D crevice corrosion test at 85°C for 72 hours, with the acceptance criterion of no crevice attack deeper than 0.025 mm when examined metallographically at 200× magnification. Production welds were post‑weld pickled in a 10% HNO₃ – 2% HF solution at 45°C for 30 minutes to remove heat tint, and the final surface was verified by portable X‑ray fluorescence to ensure that the integrated molybdenum and chromium values at the weld toe stayed within the tolerance band established during WPS qualification.

The process conditions that produce epichlorohydrin do not normally contain hydrogen sulfide, eliminating the need to follow the sour‑service restrictions of NACE MR0175/ISO 15156‑3, yet several plants that co‑produce allyl chloride in an adjacent unit have encountered cross‑contamination with 5–15 ppm H₂S in the recycle dichloropropanol stream. When H₂S is present at these trace levels in a high‑pH brine, the passive film on nickel‑chromium‑molybdenum alloys can incorporate sulfide inclusions that act as pit initiation sites under the combined influence of chloride and caustic. Cyclic potentiodynamic polarisation scans performed per ASTM G61 on Alloy C‑22 in simulated ECH brine dosed with 10 ppm H₂S showed that the pitting potential dropped from +850 mV_SCE to +460 mV_SCE, and the repassivation potential fell below the free‑corrosion potential, indicating that stable pitting could initiate spontaneously. The facility’s response was to impose a feed specification of ≤1 ppm H₂S in the dichloropropanol and to install an activated carbon guard bed upstream of the caustic reactor, which brought the H₂S concentration consistently below the detection limit of the lead‑acetate tape analyzer (0.1 ppm). This example illustrates how a seemingly minor process impurity can invalidate the material selection criteria derived from pure reagent corrosion tables and must be accounted for in the early front‑end engineering design (FEED) stage through a detailed stream composition matrix that includes all potential upsets.

^a Nickel 201 crevice attack is predominantly mechanistic under deposit; ^b Oxygen content above 0.5 ppm required for SCC initiation.

Solution‑annealed plate stocked to ASTM B575 for Alloy C‑276 is routinely delivered in the 1040–1120°C annealed condition, but plate thicknesses greater than 16 mm may develop centre‑line segregation of molybdenum during solidification of the original ingot, leading to bands of lower PREN that persist even after hot rolling and annealing. When such a plate was used for the lower shell course of a vertical separator operating at 125°C and 14% NaOH, preferential grooving occurred along the segregation bands, creating troughs up to 1.8 mm deep after 34,000 hours. The specification for the replacement shell was therefore tightened to require that the finished plate possess a minimum molybdenum content of 15.5% at the mid‑thickness location, verified by optical emission spectroscopy on a cross‑sectional coupon taken from each mother plate in accordance with the sampling plan of ISO 4950‑2. The re‑fabricated vessel exhibited uniform corrosion rates no greater than 0.015 mm/year at the first and second in‑service inspections, confirming that the seemingly small increase in minimum molybdenum specification—from the 15.0% allowed by the ASTM standard to 15.5%—was decisive in suppressing the micro‑galvanic attack that had been initiated by the compositionally depleted bands. This case illustrates a recurring theme in nickel alloy selection for aggressive caustic‑halide environments: compliance with a recognized material standard alone is not sufficient; supplementary mill‑certificate chemistry limits tailored to the specific degradation mechanism must be written into the purchase order, and a receiving inspection must verify the key alloying elements at the critical section of the fabricated component.

Post‑weld heat treatment (PWHT) of nickel‑chromium‑molybdenum alloys for the purpose of stress relief is rarely practiced in ECH caustic service, because slow cooling through the 550–950°C sensitisation range can nucleate topologically close‑packed (TCP) phases such as mu (μ) and P‑phase that embrittle the alloy and preferentially dissolve in hot alkaline chloride media. One epichlorohydrin plant opted to perform a stress‑relief treatment at 620°C for 2 hours on a large Alloy C‑22 reactor that had been heavily cold‑formed from 18 mm plate into a conical bottom. During the subsequent hydrostatic test, leaks developed at the toe of the welds, and metallographic investigation showed that the grain boundaries within 50 µm of the fusion line contained a semi‑continuous precipitate identified by energy‑dispersive spectroscopy as rich in molybdenum and chromium. When this precipitated microstructure was subjected to the ASTM G28 Method A boiling ferric sulfate‑50% sulfuric acid test, the corrosion rate accelerated to 1.4 mm/year, compared with 0.3 mm/year for the as‑welded condition. The lesson was clear: if cold‑forming exceeds 5% outer‑fibre elongation and the process stream contains chlorides at any temperature above 60°C, the distortion must be accommodated by mechanical straightening or by a full solution anneal at 1120°C followed by rapid water quenching to dissolve all secondary phases, rather than by a sub‑critical stress‑relief treatment. The replacement cone was solution‑annealed after forming in a gas‑fired furnace with a quench delay of less than 60 seconds, and its corrosion rate in the subsequent production campaign, monitored via 0.5 mm diameter electrical resistance probes, stabilised at 0.012 mm/year.

Long‑term operational data from a caustic dehydrochlorination unit that switched its entire hot‑loop piping from butt‑welded Alloy 625 (Schedules 40 and 80) to Alloy C‑276 revealed a reduction in pipe‑wall replacement frequency from every 4 years to beyond the 10‑year inspection interval, as documented in the site’s mechanical integrity report. The original Alloy 625 piping, although qualified by initial corrosion coupons recording 0.08 mm/year, had suffered progressive under‑deposit pitting at horizontal runs where flow velocities dropped below 0.5 m/s during turndown operations. Detailed pit‑depth statistics collected by manual ultrasonic scanning on 1,500 conditioning‑plate locations gave a Gumbel extreme‑value extrapolation that predicted a 5% probability of exceedance of the 3.2 mm corrosion allowance within 42 months, compelling the replacement. The Alloy C‑276 piping, fabricated to ASTM B622 seamless tube, was installed with a minimum slope of 1:100 towards drain points to prevent pooling, and its in‑service inspection interval was extended from 36 months to 72 months based on the negligible pit‑propagation rate measured during the first two cycles. This shift illustrates how the higher molybdenum content of Alloy C‑276 directly diminishes the probability of stable pitting initiation in the low‑velocity regions that are inevitable in any complex piping network subjected to seasonal throughput variations.