In commercial fluidized bed propylene ammoxidation reactors, propylene, ammonia, and air are introduced through separate spargers to avoid premature gas-phase reactions and to maintain local oxygen-to-propylene ratios in the range 1.6–2.2. The selectivity to acrylonitrile, defined as moles C₃H₃N formed per mole C₃H₆ converted on a carbon basis, is determined by the competing rates of α-hydrogen abstraction, allyl intermediate formation, lattice oxygen insertion, and ammonia-derived nitrogen insertion. In bismuth molybdate-based catalysts promoted with iron, cobalt, nickel, potassium, cerium, and phosphorus on silica, the active phase consists of α-Bi₂Mo₃O₁₂, β-Bi₂Mo₂O₉, and γ-Bi₂MoO₆ phases distributed across a mechanically attrition-resistant silica matrix with surface areas between 10 and 60 m²/g. Industrial reactors typically operate at propylene conversions of 95–98% and acrylonitrile selectivities of 80–84%, with the remaining carbon distributed among acetonitrile (1–3%), hydrogen cyanide (2–5%), acrolein (0.2–1.0%), acrylic acid (0.1–0.5%), and carbon oxides (3–6%). The reaction network is exothermic, and the heat release in the dense bed is managed by internal vertical cooling coils operating with a heat transfer coefficient in the range 300–600 W/m²·K and by catalyst circulation through the reactor central draft tube, which prevents static bed zones and localized oxygen depletion. The selectivity control problem is further complicated by the fact that the catalyst surface undergoes continuous reduction and re-oxidation cycles, and the steady-state oxidation state of molybdenum and bismuth determines whether the allylic intermediate is oxidized to acrolein or ammoxidized to acrylonitrile. Because the reaction order in oxygen is near zero above a critical oxygen partial pressure, selectivity is less sensitive to bulk oxygen concentration than to the ratio of surface lattice oxygen vacancies to adsorbed ammonia species. Commercial units therefore monitor not only the outlet oxygen concentration—typically maintained above 0.2 vol%—but also the ratio of carbon monoxide to carbon dioxide, which serves as an indirect indicator of catalyst reduction depth and oxygen transfer efficiency. In addition, the presence of water vapor in the feed at 1–3 vol% has been reported to modify the surface hydroxyl coverage and suppress acrylic acid formation by accelerating desorption of oxygen-containing intermediates, although excessive water addition above 5 vol% reduces catalyst activity by competitive adsorption on bismuth sites. This combination of kinetic, mass transfer, and surface chemistry factors means that selectivity control in a commercial propylene ammoxidation unit is a multivariable problem rather than a single catalyst property, and each operating parameter must be interpreted within the context of the local catalyst oxidation state, gas velocity, and heat removal capacity.

How Does Lattice Oxygen Mobility Determine the Acrylonitrile Selectivity Ceiling?

For bismuth molybdate systems, the selectivity ceiling is governed by the relative rates of lattice oxygen insertion and ammonia-derived nitrogen insertion at the allylic intermediate. The Mars-van Krevelen redox cycle in propylene ammoxidation involves α-hydrogen abstraction from propylene at a bismuth or molybdenum site, forming an allyl surface intermediate, followed by either insertion of lattice oxygen to produce acrolein or reaction with activated ammonia species to produce acrylonitrile. The relative rates of these two pathways are controlled by the surface oxygen activity, which in turn is determined by the ratio of oxidized to reduced metal centers and the solid-state diffusion of oxygen from the bulk to the catalyst surface. In bismuth molybdate phases, the α-Bi₂Mo₃O₁₂ phase provides isolated molybdenum tetrahedra responsible for selective α-hydrogen abstraction, while the β and γ phases contribute oxygen mobility and re-oxidation capacity. Promoters such as cerium and cobalt alter lattice oxygen mobility by modifying the reducibility of the molybdate framework and by stabilizing specific molybdenum oxidation states. Cobalt and nickel ferrites are understood to enhance electron transfer between bismuth and molybdenum sites, while potassium suppresses strong acid sites that would otherwise catalyze cracking and coke formation. The selectivity ceiling is encountered when an increase in reaction temperature intended to raise lattice oxygen flux also accelerates the deep oxidation of surface intermediates. Commercially observed selectivity values above 84% are rare for propylene ammoxidation because the same oxygen species that re-oxidizes the active sites also attacks the allylic intermediate; therefore, the selectivity maximum occurs at an intermediate surface oxygen activity where lattice oxygen supply is sufficient to maintain the redox cycle but not so high that the surface becomes over-oxidized. The optimum degree of catalyst reduction has been expressed in the technical literature as a molybdenum oxidation state distribution where the average molybdenum valence is held slightly below the fully oxidized value, although published data for specific oxidation state distributions under industrial conditions are limited. Oxygen mobility is also influenced by the silica support because silicon migrates into the molybdate framework during calcination and creates isolated molybdenum sites that reduce the probability of adjacent oxygen atoms participating in nonselective oxidation. This structural isolation is one reason why silica-supported multicomponent molybdate catalysts give higher selectivity than bulk bismuth molybdate. In addition, the presence of fluidized bed attrition-resistant microspheres with average particle diameters between 40 and 80 µm ensures that the intraparticle diffusion path for lattice oxygen is short enough to avoid oxygen depletion in the particle core, which would otherwise lead to reduced catalyst selectivity and increased hydrogen cyanide formation. Process development units with single-cycle mass spectrometric analysis of reaction products have shown that transient lattice oxygen depletion after a hydrocarbon-rich feed disturbance can reduce acrylonitrile selectivity by 3–5 percentage points within 10–20 seconds of upset, with recovery occurring only after the oxygen-to-propylene feed ratio is increased to 2.0–2.5 for several minutes. Thus, lattice oxygen mobility is not an intrinsic constant but an operating variable that is managed through catalyst composition, particle design, and feed transient control.

Across the feed mixing zone, composition disturbances propagate through the reaction network in ways that are not captured by bulk molar ratios, and ammonia partial pressure exerts one of the strongest influences on byproduct distribution. The ammonia-to-propylene molar ratio in commercial operation is maintained between 1.05 and 1.20 because excess ammonia suppresses acrolein and acrylic acid formation by shifting the allylic intermediate toward nitrogen insertion, but excessive ammonia above 1.25 increases hydrogen cyanide and acetonitrile production and reduces catalyst activity through competitive adsorption on molybdenum sites. The oxygen-to-propylene molar ratio is controlled in the range 1.6–2.2, with air or oxygen-enriched air introduced through a sparger system that maintains a minimum outlet oxygen concentration of 0.2–0.5 vol%. When the oxygen-to-propylene ratio is allowed to fall below 1.5, catalyst re-oxidation becomes rate-limiting and the selectivity shifts toward acrolein and hydrogen cyanide because the surface vacancy concentration increases and ammonia decomposition becomes favored. Conversely, increasing the oxygen-to-propylene ratio above 2.4 raises carbon oxide selectivity through over-oxidation of the allylic intermediate and can create flammable gas mixtures near the oxygen injection point if the air is not sufficiently diluted with nitrogen or steam. Water addition at 1–3 vol% in the feed has been used to moderate surface hydroxyl coverage and reduce acrylic acid formation, but water above 5 vol% reduces conversion and accelerates catalyst attrition in the fluidized bed because of hydrothermal attack on the silica support. The feed system in a commercial unit includes separate propylene and ammonia vaporizers, flow control loops with Coriolis mass flow meters, and gas chromatographic analysis of feed composition at intervals no greater than 15 minutes to ensure that the ammonia-to-propylene ratio remains within the prescribed control band. In addition, the presence of trace iron, sulfur, or heavy metal contamination in propylene feedstock can poison the catalyst surface and shift selectivity by 1–2 percentage points even when the bulk feed ratios are unchanged. Therefore, selectivity control requires simultaneous adjustment of ammonia-to-propylene ratio, oxygen-to-propylene ratio, steam co-feed, and reactor temperature, and the response surface is not monotonic; a change in oxygen-to-propylene ratio that improves selectivity at one ammonia-to-propylene ratio can reduce selectivity at another because the surface concentrations of adsorbed ammonia and lattice oxygen are coupled. Plant data from commercial fluidized bed units have shown that the sensitivity of acrylonitrile selectivity to ammonia-to-propylene ratio is approximately 0.5–1.0 percentage points per 0.1 unit change in the ratio within the range 1.05–1.20, while the sensitivity to oxygen-to-propylene ratio is smaller in the oxygen-rich regime but becomes abrupt near the oxygen-lean threshold. These sensitivities vary with catalyst age, promoter formulation, and reactor hydrodynamics, and they are typically determined experimentally during catalyst performance evaluation campaigns using a combination of online mass spectrometry and condensation-based product recovery.

Fluidized Bed Hydrodynamics and Hot Spot Suppression

Within the dense bed, the fluidized bed reactor imposes a characteristic residence time distribution that ranges from near plug flow in the freeboard to strongly backmixed in the dense bed, and this distribution directly influences acrylonitrile selectivity because backmixed catalyst particles contact product olefin and ammonia at multiple oxidation states. Superficial gas velocity in commercial units is maintained in the range 0.3–0.9 m/s, corresponding to a bubbling or turbulent fluidization regime where gas bubbles coalesce and rise through the dense bed while solid catalyst particles circulate through a central draft tube or annular zone. The catalyst particle size distribution is controlled with a mean particle diameter near 50–70 µm and a fines fraction below 10% below 20 µm to balance fluidization quality against cyclone losses. Internal vertical cooling coils remove the heat of reaction, which for a commercial acrylonitrile reactor can exceed 15–25 MW for a 100,000 t/year unit, and the local heat transfer coefficient at the coil surface is typically in the range 300–600 W/m²·K. Hot spots in the dense bed reduce acrylonitrile selectivity because high local temperatures accelerate carbon oxide formation and ammonia decomposition; therefore, cooling coil spacing and the distribution of feed air across the bottom distributor are designed to keep the radial temperature gradient below 10–15 °C. The distribution plate typically contains a large number of perforated nozzles or bubble caps with pressure drop in the range 10–25 kPa to ensure uniform gas distribution and prevent solids backflow into the feed lines. The cyclone recovery system returns entrained catalyst to the reactor, but fines generated by attrition are continuously lost and must be replaced with make-up catalyst; attrition rates in commercial units are typically below 0.5 wt%/day for silica-supported microspheroidal catalysts, although poor sparger design or excessive gas velocity can raise this value above 1.0 wt%/day and reduce the average particle residence time. When catalyst holdup is reduced by cyclone losses, the dense bed height changes and the gas residence time distribution shifts toward bypassing, which in turn increases the formation of nonselective intermediates and reduces acrylonitrile selectivity. The reactor is typically operated with a dense bed height between 5 and 10 m and a freeboard height sufficient to reduce entrainment velocities below 0.2 m/s. Hot spot suppression is further aided by staged air injection, in which a portion of the total air is introduced at a secondary elevation above the distributor to flatten the oxygen concentration profile and avoid re-oxidation exotherms in the lower dense bed. The staged air fraction is commonly 10–25% of the total air, with the remaining air introduced through the main distributor; this configuration reduces the peak dense bed temperature by 5–10 °C and improves selectivity by maintaining a more uniform surface oxygen activity. Measurement of dense bed temperature is performed with multiple thermowells at different elevations and radial positions, and the selectivity response to a 10 °C temperature excursion can be a decrease of 1–2 percentage points in acrylonitrile selectivity if the excursion persists for more than 30 minutes because catalyst reduction and phase segregation begin to occur. Therefore, fluidized bed hydrodynamics is not simply a solids handling issue but a direct selectivity control variable that determines the contact between gas-phase intermediates and catalyst particles at different oxidation states.

Iron Antimonate Catalysts Shift the Optimum Temperature Window by 30–50 °C.

At reactor temperatures above 440 °C, iron antimonate catalysts operate in a higher temperature regime than bismuth molybdate catalysts and are used in configurations where higher ammonia partial pressures or feed impurities make the molybdate system unsuitable. The active phase in these catalysts is FeSbO₄, which is structurally related to rutile and exhibits lower oxygen mobility than bismuth molybdate, requiring reactor temperatures in the range 440–480 °C compared with 410–450 °C for promoted bismuth molybdate. The higher operating temperature increases the rate of lattice oxygen supply but also raises the deep oxidation rate, so iron antimonate catalysts generally exhibit acrylonitrile selectivities in the range 78–82% at propylene conversions of 96–99%, with carbon oxide selectivities typically 2–4 percentage points higher than bismuth molybdate under similar feed ratios. The selectivity advantage of iron antimonate is the substantial suppression of acetonitrile formation to below 1%, which is attributed to the weaker surface acidity of the antimonate phase and the lower concentration of strong Brønsted acid sites; this reduces the downstream purification load for acetonitrile removal from the crude acrylonitrile stream. Hydrogen cyanide selectivity, however, is not necessarily reduced and can be in the range 3–6% under commercial conditions. The temperature window is narrower for iron antimonate because the rate difference between selective ammoxidation and nonselective oxidation has a higher apparent activation energy than the corresponding rate difference for bismuth molybdate. Published data from fixed-bed laboratory reactors indicate that the selectivity maximum for iron antimonate shifts to lower temperatures by 10–20 °C when tungsten or molybdenum promoters are added, while vanadium promotion raises the oxygen activity and reduces hydrogen cyanide formation. Commercial experience with iron antimonate in fluidized beds is less extensive than with bismuth molybdate, and published data for specific fluidized bed configurations is limited because most acrylonitrile capacity is based on promoted molybdate catalysts. Nevertheless, the higher thermal stability of iron antimonate makes it suitable for reactors processing propylene streams containing sulfur compounds or heavier hydrocarbons that would reduce the selectivity of bismuth molybdate. The catalyst is typically prepared by coprecipitation of iron and antimony oxides followed by calcination at temperatures above 700 °C to form the rutile phase, and the resulting microspheres have surface areas near 5–20 m²/g, which are significantly lower than those of silica-supported molybdate catalysts. The low surface area reduces the number of nonselective surface sites but also places a higher demand on intraparticle oxygen diffusion; therefore, iron antimonate catalysts are generally manufactured with smaller particle diameters or with macroporous supports to avoid diffusion-limited re-oxidation. The selectivity response to temperature in iron antimonate is approximately 0.3–0.6 percentage points per 10 °C near the optimum, but the response becomes asymmetric with a steeper decline on the high-temperature side because carbon oxide formation accelerates. This asymmetry is one reason why reactor temperature is controlled within a narrow band of ±5 °C in commercial units using these catalysts.

In multicomponent molybdate catalysts, promoter selection modifies selectivity by altering surface acidity, oxygen binding energy, and bismuth dispersion, and the effects are not additive because the promoters interact with the molybdate framework and with each other during calcination. Potassium, cesium, and rubidium cations neutralize strong Brønsted acid sites on the silica support that would otherwise catalyze oligomerization, cracking, and coke formation, thereby reducing acetonitrile and heavy byproduct formation and increasing the useful life of the catalyst. The optimum alkali promoter concentration is typically in the range 0.1–1.0 wt% relative to the total catalyst mass, with higher concentrations reducing propylene conversion by blocking active molybdenum sites and increasing the selectivity to hydrogen cyanide. Cerium and lanthanum modify lattice oxygen mobility by introducing oxygen vacancies that facilitate bulk-to-surface oxygen transport, but excessive rare-earth loading above 2 wt% can destabilize the molybdate phases and reduce hydrothermal stability. Cobalt and nickel ferrite components enhance electron transfer from bismuth to molybdenum and improve the re-oxidation rate of reduced molybdenum centers, but they also increase the deep oxidation activity if their concentration exceeds 5–8 wt% as transition metal oxide. Phosphorus, when added at 0.1–0.5 wt%, stabilizes the α-Bi₂Mo₃O₁₂ phase and reduces surface acidity, but higher phosphorus levels lead to the formation of bismuth phosphate phases that are nearly inactive for propylene ammoxidation. The promoter package is incorporated into the catalyst during spray drying of the silica sol slurry, and the resulting microspheres are calcined in multiple stages to control the distribution of bismuth molybdate phases and the migration of silicon into the active phase. Promoter effects on selectivity are evaluated through laboratory fixed-bed reactors with online gas chromatography, and the results are compared with pilot fluidized bed tests that replicate the temperature, gas velocity, and catalyst attrition conditions of commercial units. Published data from industrial catalyst development programs indicate that the selectivity improvement from alkali promotion can be as high as 2–4 percentage points compared with unpromoted bismuth molybdate, while rare-earth promotion can improve oxygen mobility but only within a narrow loading window. The interaction between potassium and cerium is particularly important because potassium reduces surface acidity but can also suppress oxygen mobility if it substitutes into bismuth sites; therefore, the potassium-to-cerium mass ratio is controlled within a range determined by laboratory redox titration and catalytic testing. Catalyst manufacturers do not publish the exact promoter ratios for commercial catalysts, and the information available in the open literature is limited to broad ranges and patent examples. From a manufacturing perspective, batch-to-batch variation in promoter dispersion can alter acrylonitrile selectivity by 0.5–1.5 percentage points even when the overall elemental composition is unchanged, and this variation is controlled by standardizing the pH, temperature, and mixing time during slurry preparation and by using spray dryer inlet and outlet temperatures of 250–350 °C and 100–130 °C, respectively, to ensure uniform microsphere formation.

When Oxygen Partial Pressure Falls Below a Critical Threshold, Acrylonitrile Selectivity Deteriorates Nonlinearly

When oxygen partial pressure drops below the stoichiometric requirement for catalyst re-oxidation, the selectivity decline is nonlinear because the catalyst surface becomes reduced and the allylic intermediate partitions toward oxygen-containing partial oxidation products and hydrogen cyanide rather than acrylonitrile. The critical oxygen-to-propylene ratio depends on the catalyst formulation, reactor temperature, and fluidized bed backmixing, but commercial operation generally avoids values below 1.5 under normal bismuth molybdate operation. In the oxygen-lean regime, the average molybdenum oxidation state decreases, the concentration of surface lattice oxygen vacancies increases, and the rate of ammonia oxidation to nitrogen and water becomes competitive with nitrogen insertion. The product distribution shifts to acrolein, acrylic acid, and hydrogen cyanide, with acrolein selectivity increasing by 1–3 percentage points for a 0.1 unit decrease in oxygen-to-propylene ratio below the threshold. This response is not captured by linear models developed from oxygen-rich data because the kinetic order in oxygen changes from approximately zero to positive and the surface vacancy concentration becomes a controlling variable. Commercial reactors monitor the outlet oxygen concentration as a primary constraint, with a minimum of 0.2 vol% required for bismuth molybdate catalysts and 0.5 vol% for iron antimonate systems, and any excursion below these limits for more than 5–10 minutes can cause a measurable selectivity loss that persists even after the oxygen feed is restored. The recovery time is determined by the re-oxidation rate of the reduced catalyst, which is controlled by gas-phase oxygen mass transfer to the catalyst surface and solid-state oxygen diffusion within the molybdate framework. In a commercial fluidized bed, the re-oxidation rate is sufficient to restore the catalyst oxidation state within 10–20 minutes after normal oxygen feed is re-established, but the selectivity can remain depressed for several hours because the surface molybdenum site distribution requires a longer period to return to the optimal oxidation state. In some plants, the oxygen-to-propylene ratio is temporarily increased to 2.2–2.5 for 30–60 minutes following an oxygen-lean excursion to accelerate surface re-oxidation, but this strategy must be balanced against the risk of over-oxidizing the catalyst and increasing carbon oxide production. The oxygen-lean threshold is also influenced by the presence of reducing feed impurities such as propane, which consumes oxygen without contributing to acrylonitrile production; propane concentrations in polymer-grade propylene are typically below 0.5 vol%, but refinery-grade propylene can contain 5–15 vol% propane and requires a higher oxygen-to-propylene ratio or a separate propane removal step. The nonlinearity of the selectivity response in the oxygen-lean regime means that the control system cannot rely on single-loop air-to-propylene ratio control alone; it must incorporate outlet oxygen measurement and a catalyst reduction index derived from the carbon monoxide-to-carbon dioxide ratio. The carbon monoxide-to-carbon dioxide ratio in the reactor effluent is maintained below 0.5 for bismuth molybdate catalysts to ensure that the catalyst is not operating in a deeply reduced state, although the exact target varies with catalyst age and promoter package. Published data for specific commercial oxygen-lean excursions are limited because operating companies treat such incidents as proprietary performance events, but the general response pattern is consistent with redox kinetic models.

Which Analytical Standards Govern Selectivity Calculation in Commercial Propylene Ammoxidation?

For selectivity calculation in commercial units, a carbon balance closure of at least 95% is required, and the analytical system is configured to quantify propylene, ammonia, oxygen, nitrogen, carbon monoxide, carbon dioxide, acrylonitrile, acetonitrile, hydrogen cyanide, acrolein, acrylic acid, and water in the reactor effluent. Online gas chromatographs are the standard analytical platform, with thermal conductivity detectors used for permanent gases and flame ionization detectors used for hydrocarbons and nitriles. The gas chromatograph is calibrated with gravimetrically prepared gas mixtures according to ISO 6142-1:2015, with the calibration range covering the expected outlet concentrations and with a minimum of five calibration points spanning the operating range. The precision of the analytical method is established through interlaboratory or round-robin programs following the approaches described in ASTM E691 and ISO 5725-2:2019, with repeatability limits for acrylonitrile measurement typically below 0.5 mol% in the concentrated reactor effluent. Water is determined separately by condensation and Karl Fischer titration, with the condensate collected from a quench tower operated at 5–15 °C. The carbon balance is calculated from the molar flow rates of carbon-containing species at the reactor outlet divided by the carbon molar flow at the inlet, after correcting for liquid product recovery and quench water absorption. When the carbon balance falls below 95%, the analytical data are considered unreliable for selectivity calculation, and the likely sources of error include incomplete recovery of acetonitrile and hydrogen cyanide in the quench system, calibration drift in the thermal conductivity detector, or leakage in the sampling system. Selectivity is reported on a carbon basis as the ratio of acrylonitrile carbon to converted propylene carbon, and this definition avoids distortion by water or nitrogen balance errors. The analytical standards used for online gas chromatograph performance verification include ASTM E594 for flame ionization detector linearity and ISO 10723:2012 for evaluation of online gas analysis system performance, although the latter is adapted for the higher reactivity and polarity of nitriles. The use of mass spectrometry as a secondary analytical technique has increased in recent years because it provides continuous response for multiple species and can detect transient oxygen-lean conditions faster than gas chromatography; however, mass spectrometers require daily calibration against gas chromatograph results or certified gas standards. The compliance checklist for a commercial selectivity measurement system includes verification of sampling system passivation, because acrylonitrile and hydrogen cyanide can adsorb on stainless steel surfaces if the sample lines are not heated above 80 °C. The sampling probe is installed in the freeboard region above the dense bed to avoid catalyst fines entrainment, and the sample line is heat-traced and purged with nitrogen between analyses to prevent condensation of acrylic acid and water. Published data for the absolute accuracy of online selectivity measurements in commercial acrylonitrile plants are limited, but the repeatability and carbon balance requirements described above are consistent with established industrial practice.

Under commercial operating conditions, the boundaries for selectivity control are defined by the lower flammability limit of propylene in air, the ammonia partial pressure constraints imposed by ammonium salt deposition in downstream equipment, and the thermal stability limit of the molybdate phases. The lower flammability limit of propylene in air is approximately 2.0 vol% at 20 °C, but the hot reactor environment and the presence of ammonia and water vapor shift the flammability envelope, and the feed composition is managed to remain outside the explosive region under all operating temperatures and pressures. Ammonia partial pressure upstream of the quench tower must be kept below the ammonium acrylate and ammonium carbonate deposition thresholds, which can vary between 0.1 and 0.5 kPa depending on the presence of acrylic acid and carbon dioxide; when these thresholds are exceeded, ammonium salt deposits accumulate in the quench exchanger and cause pressure drop increases and selectivity measurement artifacts due to liquid hold-up. The catalyst itself is incompatible with reducing atmospheres at temperatures above 300 °C; extended exposure to hydrogen or carbon monoxide at elevated temperature reduces molybdenum to lower oxides and causes irreversible phase segregation. Halides and sulfur compounds are catalyst poisons at part-per-million levels, and the feed purification section typically includes adsorbent beds for sulfur removal to below 0.1 mg/kg and for arsenic and mercury removal to below 0.01 mg/kg. The fluidized bed catalyst is also sensitive to hydrothermal degradation, and the addition of steam to the feed is limited to 1–3 vol% to avoid accelerated silica support sintering and attrition. During catalyst changeovers and reactor start-ups, the catalyst is pre-oxidized with air at 350–400 °C for 24–48 hours before propylene is introduced, and the ammonia-to-propylene ratio is initially held at the lean ammonia end of the control range to avoid excessive hydrogen cyanide formation during the catalyst conditioning period. The operational boundary for catalyst bed temperature is 480 °C for bismuth molybdate and 520 °C for iron antimonate, beyond which irreversible phase transformation and selectivity collapse occur. These limitations are not theoretical; they are observed as pressure drop increases, cyclone losses, and rapid selectivity fade in commercial units when feed purification, steam co-feed, or temperature control are allowed to drift outside the specified windows. Published data for specific commercial catalyst deactivation rates under each boundary condition remain limited, but the ranges given here represent the compilation of operating guidelines and catalyst supplier recommendations.