Molten Salt Corrosion Test Reveals Best Weld Filler for CFBC Boiler Super-Heater Pipes

…

The study of hot corrosion in welded superheater components sits at the intersection of metallurgy, electrochemistry, and plant engineering, and the findings from this investigation can be better appreciated when placed against the broader background of how molten salt attack develops on boiler steels. In circulating fluidized bed combustion systems, the flue gas carries ash particles rich in sodium, potassium, sulfur, and vanadium compounds. When these species deposit on superheater surfaces operating at temperatures near 770 degrees Celsius, they can form low-melting-point salt mixtures that remain liquid on the metal surface. The specific combination of sodium sulfate with vanadium pentoxide used in the thermogravimetric testing is a well-established laboratory surrogate for such deposits, because vanadium pentoxide reacts with sodium sulfate to form sodium vanadates whose melting points fall well below the metal surface temperature. Once a liquid film forms, it dissolves protective oxide scales through fluxing reactions, exposing fresh metal to accelerated oxidation in a cyclic process that can consume wall thickness far faster than dry oxidation alone would allow.

The choice of SA 335 P91 as the base material reflects its widespread adoption in modern thermal power stations. P91 is a modified 9 percent chromium, 1 percent molybdenum ferritic-martensitic steel that was developed to offer improved creep strength and oxidation resistance compared with earlier low-alloy grades, allowing higher steam temperatures and pressures and therefore better thermal efficiency. However, the welding of P91 introduces regions of very different microstructure within a single joint. The fusion zone solidifies from the molten filler, the heat affected zone experiences a thermal cycle that can transform the original tempered martensite, and the parent metal retains its normalized and tempered condition. Each of these regions has a distinct chromium content in solid solution, distinct precipitate distributions, and distinct grain sizes, all of which influence how a protective oxide scale nucleates and grows. This is why a comparative study that examines the parent metal alongside two different weldments is more informative than testing a single material condition in isolation.

The two filler metals compared in the work represent fundamentally different alloying philosophies. E9018 B9 is the covered-electrode counterpart of the P91 composition itself, designed so that the weld metal matches the base steel in chromium and molybdenum content and can sustain comparable creep performance. Its advantage is metallurgical compatibility, but its corrosion behavior is essentially that of a 9 percent chromium steel. ERNiCrMo-3, by contrast, is a nickel-based filler originally developed for demanding petrochemical and power generation service. With a matrix dominated by nickel and substantial additions of chromium, molybdenum, and niobium, it does not attempt to match the base steel composition; instead, it relies on the intrinsically superior oxidation resistance of nickel-rich alloys. Nickel oxides and nickel-containing spinels tend to form denser, more adherent scales than iron oxides, and nickel also has a lower tendency to form volatile or non-protective corrosion products in sulfate and vanadate environments.

The identification of nickel ferrite, NiFe2O4, as the protective spinel phase on the ERNiCrMo-3 fusion zone is consistent with a large body of high-temperature oxidation literature. Spinel oxides of the AB2O4 structure are valued in corrosion protection because their crystal structures can accommodate cation substitutions while remaining compact and slow-growing. Magnetite, Fe3O4, is itself a spinel, but when nickel substitutes for a portion of the iron, the resulting nickel ferrite has a lower concentration of cation vacancies and a reduced rate of outward cation diffusion, which slows scale growth. A thinner scale is mechanically more robust, less prone to cracking on thermal cycling, and presents a shorter diffusion path for inward penetration of corrosive species such as sulfur and vanadium compounds. The cyclic nature of the thermogravimetric testing, with fifty heating cycles, is particularly relevant here, because repeated thermal excursions are what cause brittle oxide scales to spall. A scale that survives fifty cycles with minimal mass loss and remains well adherent in cross-sectional examination has demonstrated genuine protective character rather than merely an initially fast-growing but non-adherent oxide.

The observation of blistering on the E9018 B9 weld zone deserves particular attention from a plant engineering perspective. Blistering in hot corrosion contexts generally indicates that corrosion products or trapped gas have accumulated at interfaces within the scale or between the scale and the substrate, generating local pressure and detachment. In vanadate-bearing environments, iron-rich oxides are susceptible to dissolution and reprecipitation cycles that disrupt scale integrity, and the resulting porous, layered scales can trap gaseous species. On an actual superheater tube, such blistered regions would be expected to spall during thermal transients such as boiler start-up and shutdown, progressively thinning the weld region. Because weld zones already carry residual stresses from the welding thermal cycle and may have slightly different creep properties than the parent tube, accelerated local corrosion at welds is a well-recognized failure location in superheater and reheater circuits. The finding that the SMAW joint degrades preferentially therefore aligns with field experience in which welds and heat affected zones are frequent sites of tube leaks.

The temperature of 770 degrees Celsius selected for the testing is meaningful in relation to CFBC operation. Circulating fluidized bed boilers are often promoted for their relatively low combustion temperatures, typically around 850 to 900 degrees Celsius in the furnace, which limits the formation of certain molten deposits compared with pulverized coal firing. However, superheater surfaces can run hotter than the furnace bed temperature depending on steam conditions and heat flux distribution, and metal temperatures approaching or exceeding 750 to 800 degrees Celsius place the material squarely in the regime where sodium vanadates and complex alkali sulfates become molten. The salt mixture of sodium sulfate with 60 percent vanadium pentoxide has a melting behavior that produces a liquid phase at the test temperature, ensuring that the corrosion mechanism studied is genuinely molten-salt fluxing rather than dry oxidation. This makes the laboratory results directly relevant to the deposit chemistry that superheater tubes encounter when firing fuels with appreciable vanadium content, such as certain heavy fuel oils, petroleum coke blends, and some biomass or waste-derived fuels.

The analytical techniques employed in the study each contribute a complementary piece of the corrosion picture. Thermogravimetric analysis provides kinetic information, revealing whether the oxidation follows parabolic behavior characteristic of diffusion-controlled protective scale growth, or faster linear kinetics indicative of non-protective scaling. Repeated cycling in the thermobalance also captures mass changes from spallation, which a single isothermal exposure would miss. Scanning electron microscopy with energy dispersive spectroscopy maps the morphology and elemental distribution of the corrosion products, distinguishing between compact scales, porous layers, and internal attack along grain boundaries. X-ray diffraction identifies the crystalline phases, allowing the distinction between simple iron oxides such as hematite and magnetite, mixed spinels such as nickel ferrite, and sodium or vanadium-containing reaction products that signal interaction between the molten salt and the scale. Cross-sectional imaging is especially valuable because it reveals scale thickness, adherence, and whether attack penetrates along prior austenite grain boundaries into the substrate, which is the mechanism by which hot corrosion compromises mechanical integrity even before measurable wall loss becomes severe.

From a practical standpoint, the recommendation to use ERNiCrMo-3 filler with the gas tungsten arc welding process carries implications for fabrication and maintenance practice in CFBC plants. Gas tungsten arc welding offers precise control of heat input and dilution, which matters because the composition of the fusion zone in a dissimilar weld depends on how much base metal is melted and mixed with the filler. Excessive dilution of a nickel-based filler with iron-rich P91 parent metal would reduce the nickel content of the fusion zone and could erode the very advantage that the filler is meant to provide. Controlled dilution, typically achieved with careful current settings and technique, preserves a nickel-rich fusion zone capable of forming the protective spinel. Shielded metal arc welding, while more portable and economical for field repairs, involves a flux-covered electrode whose slag chemistry and greater dilution variability can complicate achieving an optimal fusion zone composition, and the observed blistering suggests that in this corrosive service the SMAW joint with the ferritic filler is the weaker link.

The economic dimension of these findings should not be overlooked. Superheater tube replacement in a 70 MW unit involves planned outages, and unscheduled failures of welds can force entire unit trips with substantial replacement power costs. If a filler upgrade extends weld service life even modestly in vanadium-containing fuel service, the incremental consumable cost of a nickel-based filler is small relative to the avoided outage risk. Utilities operating CFBC boilers on fuels with elevated vanadium and sodium content may find that specifying nickel-based fillers for superheater welds, combined with regular inspection of weld zones during overhauls, offers a cost-effective reliability improvement. The study also underscores the value of laboratory cyclic hot corrosion testing as a screening tool: by reproducing deposit chemistry and thermal cycling in a controlled setting, candidate consumables can be ranked before plant trials, shortening the path from metallurgical research to operational decision-making in thermal power generation.

Subject of Research: Materials Science

Article Title: A comparative investigation on molten salt hot corrosion of GTA and SMA-welded super-heater pipes of 70 MW CFBC boilers

Article References: Pillai, L. R., Aravind, A., Dhanushwaran, R. V., Praveen, S. K., & Prabu, S. S. (2026). A comparative investigation on molten salt hot corrosion of GTA and SMA-welded super-heater pipes of 70 MW CFBC boilers. Scientific Reports. https://doi.org/10.1038/s41598-026-70579-2

Image Credits: AI Generated

DOI: 10.1038/s41598-026-70579-2

Keywords: hot corrosion, molten salt, P91 steel, GTAW, SMAW, CFBC boiler, ERNiCrMo-3, E9018 B9, spinel oxide, super-heater tubes, comparative, investigation