Direct oxidation of ethylene over silver supported on low-surface-area alpha-alumina is the dominant industrial route to ethylene oxide. The reactor is a shell-and-tube fixed-bed unit operating at pressures of 1.5–3.0 MPa and gas inlet temperatures of 220–270 °C. The selective partial oxidation C₂H₄ + 1/2 O₂ → C₂H₄O releases approximately 105–120 kJ mol⁻¹, while the complete combustion C₂H₄ + 3 O₂ → 2 CO₂ + 2 H₂O releases ~1323 kJ mol⁻¹. Selectivity is defined at the reactor exit as S_EO = [F_EO,out / (F_C2H4,in − F_C2H4,out)] × 100, and ethylene conversion per pass is deliberately limited to 8–14%. Modern silver catalysts achieve steady-state ethylene oxide selectivity in the range 78–90 mol% under optimized chloride moderation; the remaining ethylene is oxidized to CO₂ and water. The reactor feed typically contains 20–35 vol% ethylene, 5–9 vol% oxygen, and methane or nitrogen as ballast, maintaining the gas mixture below the flammable envelope. The epoxidation reaction is believed to proceed through adsorbed atomic oxygen and an oxametallacycle intermediate, while unselective carbon-carbon bond cleavage dominates on defect-rich silver surfaces.

On a production line, the reactor tubes are 20–40 mm internal diameter and 6–12 m long, cooled by boiling water on the shell side. Boiler feedwater chemistry and steam drum pressure control are critical because the boiling water saturation temperature sets the tube-wall temperature; the reactor pressure envelope is designed under ASME BPVC Section VIII Division 1. The recycle loop includes a CO₂ removal section, an ethylene oxide water scrubber, and a recycle compressor. The CO₂ removal section typically uses an activated hot potassium carbonate solvent or an amine-based solvent with downstream water washing or guard-bed protection to limit amine carryover to the silver bed. The gas hourly space velocity, calculated as standard volumetric flow divided by total tube volume, is typically 3000–6000 h⁻¹. At low space velocity the residence time of ethylene oxide increases and secondary oxidation to CO₂ is promoted; at excessively high space velocity the conversion per pass falls and the recycle ratio rises. Process gas chromatographs with thermal conductivity and flame ionization detectors measure methane, oxygen, nitrogen, CO₂, ethylene, ethane, ethylene oxide, and acetaldehyde at the reactor inlet and outlet. The oxygen analyzer in the recycle gas is a paramagnetic cell; the CO₂ analyzer is typically a non-dispersive infrared unit. A safety instrumented system in accordance with IEC 61511 maintains oxygen concentration below the limiting oxidant concentration for the methane-ethylene-oxygen mixture.

Selectivity loss appears first as a rising CO₂/EO molar ratio at constant ethylene conversion, followed by an increase in the inlet temperature required to hold oxygen conversion. In a stable commercial unit, the chloride moderator is adjusted continuously; a deficiency of 0.1–0.3 ppmv ethyl chloride in the recycle gas can produce a measurable shift of the CO₂/EO ratio within 8–24 h, though exact transient responses are catalyst-specific and often proprietary. The ethylene oxide product scrubbing loop also shows an increased load of glycols and acetaldehyde when EO is degraded in the reactor or in the hot gas path. Pressure-drop growth across the bed, measured by differential-pressure transmitters across the reactor, indicates carbon accumulation, local catalyst breakage, or support dusting. The key operational distinction is whether the selectivity loss is reversible by moderator adjustment or irreversible because of silver sintering, sulfur poisoning, or support damage.

What Operational Variables Accelerate the Loss of EO Selectivity?

Temperature, oxygen partial pressure, chloride concentration, and feed impurities interact. The most severe selectivity excursions occur when a hot spot pushes local tube-wall temperature above 300 °C, close to the Tammann threshold of silver near 344 °C, where surface silver atoms acquire sufficient mobility for crystallite coarsening. A single tube with a partial plug or uneven gas flow can develop a hot spot even when the average reactor inlet temperature remains within the normal range. The shell-side boiling water pressure and liquid level must be stable; a drop in heat transfer coefficient can allow a tube-wall temperature increase of 10–20 °C within minutes. Once silver crystallites coarsen, the number of selective low-index silver sites declines, and the relative contribution of defect-rich sites to complete oxidation increases. The loss of selectivity is not reversible by reducing temperature because the particle morphology has changed.

Oxygen concentration affects the coverage of adsorbed atomic oxygen on silver. At high oxygen coverage, oxygen clusters and subsurface oxygen species become more abundant and promote C–C bond cleavage. At very low oxygen concentration, the epoxidation rate becomes mass-transfer-limited and the ethylene oxide that forms has a longer residence time in the hot zone, increasing secondary oxidation. Chloride moderator addition must be balanced against oxygen concentration; excess chloride poisons silver sites and suppresses conversion, while insufficient chloride allows unselective oxygen recombination and EO isomerization. The optimal chloride concentration is usually expressed as parts per million by volume of ethyl chloride or 1,2-dichloroethane in the reactor feed and is typically 0.5–4.0 ppmv. The actual required concentration shifts with catalyst age, feed CO₂, and reactor pressure.

Silver particle coarsening is governed by surface self-diffusion and is accelerated by local temperature excursions. Fresh catalysts typically contain silver crystallites with average diameters below 1 µm; after extended operation or hot periods, scanning electron microscopy of discharge samples shows growth beyond 2 µm and loss of fine silver particles. The specific surface area of silver, not total loading, controls the number of active sites. Because support surface area is intentionally low, silver particles are distributed as discrete particles on alpha-alumina. The support is selected for low acidity to minimize EO isomerization to acetaldehyde and hydrolysis to ethylene glycol; surface area measured by nitrogen physisorption per ISO 9277:2022 is generally 0.5–2.0 m²/g. Support particle size distribution measured by laser diffraction per ISO 13320:2020 is controlled to maintain uniform tube packing and low pressure drop. A low-surface-area support reduces parallel acid-catalyzed side reactions but also lowers the available area for silver dispersion, so catalyst manufacturing must balance silver particle size and support contact. Pore diffusion limitations become significant when the shell of the catalyst particle is depleted of oxygen while the interior contains higher ethylene oxide concentration; this leads to unselective internal oxidation. The effective diffusion path is shortened by using high-porosity alpha-alumina with large pores, but pores above 10 µm can lower mechanical strength.

Chloride redistribution across the bed is another reversible mechanism. When ethyl chloride is injected upstream of the reactor, it decomposes and releases chlorine species that adsorb on silver. The axial chloride coverage is not uniform because the reaction zone is hottest near the inlet. Slow changes in feed flow or reactor tube temperatures can strip chloride from some tubes and concentrate it in others; the resulting tube-to-tube selectivity distribution widens. Routine tube temperature scans from shell-side thermocouples or infrared thermography of the reactor head can identify tubes operating 5–10 °C above the average, but published data for individual tube selectivity is generally limited because commercial units rarely provide per-tube effluent measurements. Laboratory micro-reactor systems with differential recycle have been used to isolate chloride effects under isothermal conditions; these studies show that the selectivity gain from chloride is largest on small silver particles and stepped silver surfaces, while large particles respond less strongly. This is consistent with the industrial observation that older catalysts with coarsened silver require higher chloride concentrations to hold selectivity, but the increase in chloride also suppresses activity.

Thermal Degradation Pathways, Hot Spot Zones, and Silver Crystallite Growth

Hot spot formation is governed by the local rate of heat generation exceeding the shell-side heat removal capacity. The rate of ethylene oxidation is strongly exothermic, and the local temperature can be 20–40 °C higher than the bulk gas temperature. In the first 20–30% of tube length, oxygen partial pressure is highest and the silver surface is highly oxidized; this region is particularly sensitive to flow distribution. Uneven reactor packing, broken pellets, or fines accumulation can create low-resistance channels where gas velocity is low and residence time is high, causing localized oxygen depletion and an increase in ethylene oxidation to CO₂. Shell-side boiling water heat transfer is influenced by tube scale deposition and by the circulation rate of the steam drum. If boiler water chemistry deviates from conductivity and silica limits, scale deposits can form on the external tube walls, reducing the overall heat transfer coefficient. A rise in tube-wall temperature of 5–10 °C may be sufficient to shift the local surface oxygen coverage from the selective regime to the unselective regime.

The Tammann temperature for silver is approximately 344 °C, at which the mobility of surface silver adatoms becomes significant. Industrial reactors are normally operated with tube-wall temperatures below 290 °C, providing a margin of roughly 50–60 °C below the onset of rapid silver surface diffusion. However, process upsets such as oxygen feed flow fluctuations, temporary loss of recycle gas flow, or steam drum pressure drift can produce short-term excursions that exceed 300 °C in individual tubes. During such excursions silver particles coalesce, the selective Ag(111) facets and stepped silver surfaces diminish, and the CO₂/EO ratio increases irreversibly. The selectivity loss is often not immediately apparent in average reactor effluent because only a fraction of tubes experience hot spots; over weeks the effect spreads as tube-to-tube gas distribution shifts. Discharged catalyst samples from hot-spot tubes show localized silver depletion on the outer pellet surface and silver enrichment in zones of contact between adjacent pellets, consistent with surface migration and particle growth.

Feed purification is the first line of defense against selectivity loss. In a typical ethylene oxide plant, the fresh ethylene stream is routed through a selective hydrogenation unit for acetylene removal, a sulfur guard bed containing zinc oxide or promoted zinc oxide, and a molecular sieve dryer before mixing with oxygen and recycle gas. The guard bed removes hydrogen sulfide and light mercaptans but may not remove heavier organic sulfur species; therefore the fresh ethylene specification includes total sulfur below 1 ppmv. The methane ballast is preferably sulfur-free and is monitored for odorants if natural gas is used. Carbon monoxide in the fresh ethylene can be oxidized to CO₂ and is less severe than sulfur or acetylene, but it can slightly raise the CO₂ concentration in the recycle loop. Acetylene is particularly damaging because it can undergo oligomerization on hot silver surfaces, producing carbonaceous deposits that increase pressure drop and block selective sites. The recycle gas may also contain trace ethylene oxide, water, and CO₂; the water scrubber and CO₂ removal section must reduce these to controlled levels before the gas returns to the reactor. Incomplete water removal increases the steady-state water concentration in the feed, which can hydrolyze ethylene oxide on the support and reduce EO yield.

When Feed-Grade Ethylene Contains Sulfur, Acetylene, or Heavy Hydrocarbons

Sulfur is among the most damaging feed impurities for silver catalysts. Hydrogen sulfide, carbonyl sulfide, and organic mercaptans adsorb strongly on silver surfaces and block the sites required for selective epoxidation. The activity loss from sulfur is generally irreversible; thermal regeneration is not practical because the temperatures required to desorb sulfur would also sinter silver. Acetylene can undergo metal-catalyzed oligomerization and dehydrogenation, producing carbon-rich deposits on the silver surface and on the support. These deposits increase tube pressure drop, reduce heat transfer, and create local hot spots as they partially oxidize. Heavy hydrocarbons with carbon numbers above C₃ combust more readily than ethylene and release excessive heat, pushing local silver surfaces into the unselective temperature regime. Trace hydrogen can reduce surface silver oxide species and alter the electrophilic oxygen population; the effect at normal polymer-grade ethylene levels is small but can become measurable if hydrogen is introduced with recycled methane from units that have undergone methanation or hydrogenation.

Quantitative selectivity-loss rates for each impurity are catalyst-specific and generally not disclosed; published data for commercial silver catalysts under plant recycle conditions is limited. Laboratory microreactor experiments with accelerated impurity spike tests are used to estimate sulfur tolerance, but scale-up to multitubular reactors introduces complexities of axial flow distribution and tube-to-tube temperature differences. The operational response to sulfur breakthrough is typically to reduce inlet temperature, increase chloride moderation temporarily, and replace the sulfur guard bed; if breakthrough has already occurred, the activity loss is irreversible and the reactor must be operated at higher inlet temperature until catalyst change-out.

Chloride Moderation, Alkali Promotion, and Bed Pressure Drop

The silver surface is promoted by alkali metals and halogens. Alkali promoters such as cesium, potassium, or sodium are incorporated into the support or silver during catalyst manufacture. They modify the electronic density of silver, stabilize electrophilic oxygen species, and reduce the isomerization of ethylene oxide to acetaldehyde. Too high an alkali loading reduces ethylene conversion and can promote over-oxidation under oxygen-rich conditions. Chloride moderator is added continuously as ethyl chloride or 1,2-dichloroethane in the gas phase; it adsorbs as atomic chlorine and blocks the most unselective silver sites. The effective chlorine coverage is a balance between gas-phase moderator concentration, temperature, and the presence of water vapor. A shift from ethyl chloride to 1,2-dichloroethane alters the decomposition pathway and requires recalibration of the moderator injection rate. Batch-to-batch variation in silver particle size distribution and alkali promoter dispersion can change the optimal moderator setpoint by 0.2–0.5 ppmv, so each catalyst charge typically undergoes a start-up optimization sequence in which the moderator is ramped while the CO₂/EO ratio and oxygen conversion are tracked.

Bed pressure drop is an indirect indicator of selectivity loss. Fresh catalyst loads are designed for a pressure drop of 0.1–0.5 bar across the reactor at design GHSV. As carbon deposits accumulate or silver particle movement shifts particle packing, pressure drop rises, causing gas flow maldistribution and lower space velocity in some tubes. High pressure drop also increases recycle compressor energy and may force a reduction in plant throughput. Fines from mechanical attrition of alpha-alumina support can migrate downstream and block the lower tube sheet, leading to localized channeling. A pressure-drop increase of more than 0.2 bar above baseline often triggers a reduction in oxygen concentration to protect against hot spots, but this reduction may itself lower selectivity because the surface oxygen coverage enters a less selective regime. In extreme fouling, the plant must reduce feed rate or shut down for catalyst screening and reloading. The spent catalyst is discharged under inert atmosphere to avoid silver dust exposure and to prevent adiabatic oxidation of carbon deposits during air exposure.

Quantifying Selectivity Loss Through Effluent Carbon Oxides and EO Yield Decline

The primary analytical indicator of selectivity loss is the molar ratio of CO₂ to EO in the reactor outlet. In a digital control system, the ratio is calculated from the process gas chromatograph data and corrected for CO₂ entering with the recycle stream. An increase in CO₂/EO molar ratio from 0.15–0.25 to 0.30–0.40 under constant feed conditions indicates a selectivity loss of several percentage points. The absolute selectivity is calculated using the ethylene consumption and EO production flows; custody-transfer-grade flow meters are not required for selectivity, but the online process gas chromatograph must be calibrated with certified gas standards at least daily. The uncertainty of the calculated selectivity is usually ±0.5–1.0 mol% depending on analyzer cycle time and flow meter accuracy. In addition to CO₂, the presence of acetaldehyde in the reactor outlet or in the EO scrubber bottoms indicates EO isomerization. Elevated glycols in the scrubber water indicate EO hydrolysis in the hot gas path or in the absorber. These byproducts are measured by liquid chromatography or by gas chromatography after derivatization.

When a shift in selectivity is detected, the control strategy first adjusts chloride moderator in small steps of 0.1–0.5 ppmv and observes the CO₂/EO response over 6–24 h. If selectivity does not recover, the next diagnostic step is a reduction in inlet temperature or oxygen concentration to reduce hot spots, accepting a temporary loss of conversion. If the selectivity remains depressed after these moves, the cause is likely irreversible silver sintering or sulfur poisoning. Parallel analytical checks include tube temperature scans, pressure-drop trending, and spent catalyst sampling from selected tubes. A tube with a long-term hot spot may show selectivity loss only when the reactor is operating near full capacity, because the increased heat load exposes the local cooling limitation. The production boundary is therefore not a single temperature or pressure limit but a combination of average bed temperature, peak tube-wall temperature, oxygen conversion, and CO₂/EO ratio.

Operationally, free amine carryover to the silver bed should be prevented because amines can displace chloride and form nitrogen-containing residues that alter silver surface chemistry. Amine-based CO₂ removal solvents require an amine guard bed or water wash to limit carryover to the reactor. The catalyst support should remain dry during loading; moisture adsorption at relative humidity above 60% can cause pore filling and subsequent pellet fracture during heat-up. The maximum allowable oxygen concentration must not be exceeded even if selectivity is declining; increasing oxygen above 9 vol% without compensating methane ballast can move the mixture into the explosive region. The reactor pressure envelope, relief valves, and safety instrumented system must follow ASME BPVC Section VIII Division 1 and IEC 61511 requirements. Silver catalyst discharges should be handled as hazardous waste under applicable environmental regulations. The documented selectivity trend, tube temperature records, and moderator addition logs provide the technical basis for distinguishing reversible chloride redistribution from irreversible silver morphological change.