The selective conversion of hydrogen peroxide to propylene oxide over titanium silicalite-1 (TS-1) is governed by the kinetic competition between heterolytic oxygen transfer to chemisorbed propene and non-selective hydrogen peroxide destruction to water and dioxygen. Peroxide selectivity is defined as the molar ratio of propylene oxide formed to hydrogen peroxide consumed, expressed as a percentage; reported commercial HPPO fixed-bed operations range from 90 mol% to 97 mol% hydrogen peroxide selectivity when the feed and recycle loops are maintained within a narrow chemical envelope. The active catalyst is an MFI-type crystalline titanosilicate in which isolated tetrahedral Ti(IV) occupies framework positions, ideally at Si/Ti ratios between 30 and 50, and the primary working solvent is methanol, which dissolves both hydrogen peroxide and propene while also participating in the stabilisation of the titanium hydroperoxo intermediate. Industrial reaction conditions are commonly reported as 40–60 °C, total pressure 2.0–4.5 MPa, hydrogen peroxide concentration in the methanol feed 5–10 wt%, water content 5–15 wt%, and pH 5.0–6.5. Under these conditions, hydrogen peroxide conversion exceeds 98%, but the selectivity margin is consumed by three competing side reactions: acid-catalysed hydrolysis of propylene oxide to propylene glycol, alcoholysis with methanol to 1-methoxy-2-propanol and 2-methoxy-1-propanol, and direct hydrogen peroxide decomposition to dioxygen and water. Each of these side reactions responds differently to water activity, acid-site concentration, temperature, and metal impurities; therefore the industrial control problem is not limited to selecting an intrinsic catalyst but extends to managing the full solvent-catalyst-impurity system across heat exchangers, methanol recovery columns, and propene recycle compressors. The presence of extra-framework TiO2 or silanol defects at the external crystal surface introduces Brønsted and Lewis acid sites that catalyse ring opening and peroxide decomposition, and these defects are often generated during catalyst shaping rather than during framework synthesis. Consequently, peroxide selectivity control in HPPO requires simultaneous control of framework Ti coordination, extrudate binder chemistry, feed pH, water accumulation in the methanol recycle loop, and the axial temperature profile of the fixed-bed reactor.

What Controls the Kinetic Partition Between Hydroperoxo-Mediated Epoxidation and Peroxide Decomposition?

The kinetic partition on TS-1 is determined by the structure and lifetime of the titanium hydroperoxo species formed when hydrogen peroxide coordinates to a framework Ti(IV) site. In the resting state, tetrahedral Ti(IV) in the MFI framework exhibits a ligand-to-metal charge transfer near 220–230 nm; upon addition of hydrogen peroxide, the coordination number increases and a hydroperoxo or peroxo intermediate appears, with reflectance or transmission UV-vis features commonly assigned between 260 nm and 280 nm. The desired epoxidation channel requires that this titanium hydroperoxo transfer an oxygen atom to propene in a concerted or near-concerted step, forming a Ti-bound propylene oxide precursor that releases propylene oxide and restores the Ti site. Non-selective decomposition proceeds when the hydroperoxo intermediate is protonated, when hydrogen peroxide reacts with extra-framework TiO2, or when redox-active metals such as Fe, Cu, or Cr cycle through Fenton-type reactions that generate hydroxyl radicals. Published kinetic studies of propylene epoxidation on TS-1 report apparent activation energies in the range 28–42 kJ/mol for the selective pathway and 54–88 kJ/mol for hydrogen peroxide decomposition, which means that high temperatures accelerate the non-selective channel more strongly than the selective channel. The same kinetic asymmetry sets the practical upper temperature at approximately 65 °C; above this threshold, peroxide decomposition becomes dominant, titanium leaching from framework defect sites increases, and the selectivity drop is no longer recoverable by raising propene pressure. At the lower temperature boundary, epoxidation rates decline and liquid hourly space velocity must be reduced to maintain hydrogen peroxide conversion above 98%, so commercial reactors select a compromise between 45 °C and 55 °C. Propene pressure is adjusted to maintain liquid-phase propene concentration and suppress epoxide retention in the catalyst pores; increasing total pressure from 2.0 MPa to 4.5 MPa typically improves propylene oxide selectivity by a few percentage points, but the effect saturates once the liquid phase is sufficiently enriched in propene. Methanol is not a passive diluent: it stabilises the titanium hydroperoxo through hydrogen bonding and reduces the effective water activity near the active site, thereby suppressing hydrolysis of the formed epoxide. The kinetic partition is therefore a strong function of solvent composition, and the same catalyst that gives high selectivity in methanol can exhibit poor selectivity in water-rich or acetone-rich media because the peroxo intermediate is destabilised and the epoxide undergoes faster solvolysis.

At a peroxide selectivity threshold of 95 mol%, the shaping of TS-1 powder into industrial fixed-bed extrudates creates a second kinetic regime that intrinsic kinetic measurements on submicron crystals do not resolve. Commercial TS-1 is crystallised as particles of 0.2–0.4 µm, but fixed-bed reactors require formed bodies of 1.6 mm to 3.0 mm diameter to avoid excessive pressure drop and fluidisation. Shaping with alumina binders is incompatible with peroxide selectivity because alumina introduces acid sites that hydrolyse propylene oxide to propylene glycol and accelerate hydrogen peroxide decomposition; silica sol binders are therefore preferred, but silica binder formulation must balance crush strength and pore accessibility. Extrudate crush strength above 10 N measured according to ASTM D4179 is a typical mechanical acceptance criterion, and lower values lead to catalyst attrition, fines release, and downstream fouling of the methanol recovery column. Silica binder levels between 20 wt% and 40 wt% of the formed body are common, but increasing binder content beyond this range dilutes the active phase and increases intraparticle diffusion barriers because the binder matrix surrounds TS-1 crystals and reduces the effective diffusivity of propene and methanol. The local selectivity loss inside formed bodies is caused by retention of propylene oxide in liquid-filled micropores; the longer the epoxide remains near acid sites or water, the higher the probability of hydrolysis or alcoholysis. Egg-shell profiles, in which TS-1 is concentrated at the extrudate surface, reduce diffusion path length but increase the risk of binder exfoliation during thermal cycling. Production-scale extrusion lines with twin-screw mixers and shot-belt dryers introduce batch-to-batch variability in pore volume, average crush strength, and residual ammonium or alkali content, and these variations are frequently detected only after the catalyst is loaded into a multi-tubular reactor and the peroxide selectivity falls below 90%. Silica-bound extrudates are also sensitive to over-calcination: calcination above 550 °C causes silanol condensation and reduces mechanical strength, while calcination below 450 °C leaves organic residues that foul the titanium sites and promote hydrogen peroxide decomposition during start-up. The industrial control response is to specify strict limits for binder sodium, aluminium, iron, and sulfate content, because each of these impurities converts a selective titanium hydroperoxo site into a non-selective decomposition centre.

Solvent Composition, pH, and the Maximum Allowable Water Concentration in Continuous Recycle Operation

Methanol recycle composition determines the steady-state water activity at the active site more strongly than the fresh feed composition, because water is a reaction co-product and accumulates in the methanol recovery loop unless a purge or side-draw distillation is continuously operated. When the recycle methanol water content exceeds 15 wt%, the selectivity to propylene oxide declines in a non-linear manner: water competes with methanol for hydrogen bonding to the titanium hydroperoxo, increases the dielectric constant of the medium, and directly participates in propylene oxide hydrolysis to propylene glycol. At water concentrations above 20 wt%, the selectivity loss is commonly greater than 5 percentage points, and the additional propylene glycol generated in the reactor further accelerates catalyst fouling because glycols and their oligomers accumulate in the recycle loop and deposit on TS-1 external surfaces. pH is controlled between 5.0 and 6.5 because this range minimises both acid-catalysed and base-catalysed epoxide ring opening while preserving hydrogen peroxide stability. Alkaline conditions above pH 7.0 promote base-catalysed hydrolysis to propylene glycol and destabilise hydrogen peroxide; acidic conditions below pH 4.0 protonate the epoxide oxygen, accelerate methanolysis to methoxypropanols, and increase corrosion of stainless steel feed lines, releasing Fe and Cr ions that catalyse peroxide decomposition. Sodium hydroxide is avoided as a pH-adjustment agent because sodium can exchange with framework or defect Ti and reduce long-term catalyst activity; sodium-free basic agents and buffer systems are used instead, and sodium content in the circulating methanol is maintained below 1 mg/kg. Hydrogen peroxide commercial stabilisers such as sodium stannate, phosphoric acid, or aminopolycarboxylates are similarly managed because they introduce non-volatile species that accumulate in the recycle loop and alter catalyst surface chemistry. The methanol recovery column is operated to remove water overhead, glycol ethers as a side draw, and heavy polyols as bottoms, but the separation is energy-intensive because methanol-water distillation requires high reflux ratios and because propylene glycol and methoxypropanols are heavier than methanol and water. The compliance matrix in Table 1 summarises the feed and catalyst acceptance limits that are applied when peroxide selectivity above 95 mol% is required over a full catalyst cycle.

Quantifying Framework Titanium Coordination and Silanol Nest Formation in Deactivating TS-1 Beds

Selectivity control cannot be sustained without periodic or in-situ characterisation of the active titanium environment, because deactivation mechanisms that reduce peroxide selectivity often manifest as changes in framework Ti coordination, extra-framework TiO2 formation, and silanol defect density before the corresponding selectivity loss is visible in routine reactor data. Diffuse reflectance UV-Vis spectroscopy of fresh and spent TS-1 distinguishes tetrahedral framework Ti at 220–230 nm from octahedral or extra-framework Ti at 260–280 nm and from bulk anatase-like species above 330 nm; a spent catalyst that shows a rising shoulder above 280 nm is typically accumulating the extra-framework Ti that catalyses hydrogen peroxide decomposition. Raman spectroscopy with excitation below 365 nm and X-ray absorption near-edge structure at the Ti K-edge provide complementary information on the coordination number and the degree of Ti-O-Si condensation. Fourier-transform infrared spectroscopy of adsorbed pyridine or ammonia is used to quantify Brønsted and Lewis acid sites on shaped extrudates, but pyridine adsorption itself can deactivate TS-1 unless performed on sacrificial samples. Temperature-programmed ammonia desorption and pyridine-IR give acid-site concentrations per gram; an increase in Lewis acid density above fresh-catalyst baseline by more than 20% correlates with higher propylene glycol and methoxypropanol formation in the reactor. X-ray fluorescence and inductively coupled plasma optical emission spectroscopy according to ISO 11885 are used to measure Ti/Si ratio, sodium, aluminium, iron, and chloride in fresh and spent samples. Surface area and micropore volume are measured according to ISO 9277 and ASTM D4365, with reduction in micropore volume indicating retained glycols, oligomeric carbon, or framework collapse. These analytical indicators are linked to selectivity through the same acid-site and mass-transfer mechanisms that govern the reactor, so fresh-catalyst acceptance limits and spent-catalyst deactivation analyses are written against the same quantitative benchmarks.

When Catalyst Deactivation and Thermal Runaway Risk Coincide with Selectivity Loss

In fixed-bed HPPO operation, the reactor is exposed to a process conflict in which the same high hydrogen peroxide concentrations that maximise throughput also increase the adiabatic temperature rise from decomposition and push the axial temperature profile toward hot spots that degrade selectivity. Hydrogen peroxide decomposition to water and dioxygen releases approximately 98 kJ/mol of hydrogen peroxide decomposed, and in a methanol-water mixture a 1 wt% hydrogen peroxide concentration corresponds to an adiabatic temperature rise of about 6–7 °C when decomposition is complete. For a commercial inlet hydrogen peroxide concentration of 8 wt%, complete decomposition in an isolated adiabatic zone would raise the temperature by roughly 55 °C, transforming a 45 °C inlet stream into a local temperature near 100 °C unless shell-side coolant removes the heat at the same rate. Multi-tubular reactors with tube inner diameters near 25 mm typically rely on shell-side cooling water or vaporised methanol to maintain a tube wall temperature below 50 °C, but radial temperature gradients inside the tube can still exceed 10 °C when hydrogen peroxide concentration is high or when catalyst activity is elevated at the inlet. A hot spot above 70 °C immediately accelerates the non-selective decomposition channel because its activation energy is higher than that of epoxidation; it also strips Ti atoms from framework defect sites, producing extra-framework Ti species and silanol nests that are inactive for epoxidation but active for peroxide destruction. The resulting deactivation pattern is spatially non-uniform: the inlet zone loses selectivity first, and the active epoxidation front moves down the tube, increasing pressure drop and redistributing heat release. Thermal runaway risk is highest during start-up, shutdown, and water-wash recovery because residual hydrogen peroxide can accumulate in stagnant zones and because the catalyst surface may contain adsorbed organic impurities that consume peroxide exothermically. The control response is to restrict inlet hydrogen peroxide to the 5–10 wt% range, to preheat the feed only after the catalyst bed is uniformly wetted, and to interlock the hydrogen peroxide feed pump with shell-side coolant flow, reactor outlet temperature, and oxygen concentration in the vent gas. Hydrogen peroxide injection is typically split across multiple feed points or diluted in methanol before entering the reactor, reducing the local concentration and flattening the axial temperature profile. Catalyst deactivation and hot-spot formation are therefore inseparable from peroxide selectivity control: a local temperature excursion that reduces selectivity by 2 percentage points also shortens catalyst life by leaching framework titanium and increasing the rate of silanol nest formation, so the same feed limits and heat-transfer constraints are written into both the safety documentation and the catalyst performance warranty.

Across a fully integrated HPPO train, the propylene feed, hydrogen peroxide quality, methanol recycle inventory, and catalyst formation are treated as a single impurity-limited system rather than as independent unit operations. Polymer-grade propylene is specified at ≥99.5 mol% purity according to ASTM D2163, but trace components such as methylacetylene and propadiene at concentrations above 100 ppm can poison TS-1 active sites or participate in oligomerisation reactions that deposit carbon on the catalyst surface, while CO, ethylene, and sulfur compounds alter the redox chemistry of the titanium hydroperoxo intermediate. Hydrogen peroxide used in HPPO is typically prepared by anthraquinone autoxidation and then purified to remove transition metals and organic stabilisers; iron content below 50 ppb is required because Fe(II)/Fe(III) cycles decompose hydrogen peroxide even at trace levels and also generate hydroxyl radicals that attack the epoxide product. Methanol recovery is designed with a water removal column, a propylene oxide purification train, and a recycle compressor that returns unreacted propene to the reactor inlet; the recycle loop accumulates dimethyl ether, methyl formate, acetaldehyde, and light hydrocarbons in small amounts, and these components are bled off to prevent solvent quality drift. Propylene oxide product specifications impose strict limits on water, aldehyde, and chloride impurities, but published data for this specific HPPO configuration is limited and individual licensors apply proprietary internal limits. The fixed-bed catalyst charge is expected to operate for several thousand hours depending on feed impurities and temperature excursions, and spent catalyst analysis typically shows titanium loss in the first bed length, carbon deposition in the middle zone, and silanol condensation near the outlet.