Recovery of durene (1,2,4,5-tetramethylbenzene, CAS 95-93-2) from C10 aromatic streams is accomplished by fractional crystallization under controlled cooling rather than by atmospheric distillation because the normal boiling point of durene, approximately 196.8 °C, lies too close to those of naphthalene and pseudocumene for a sharp split at justifiable reflux ratio. A typical feed stream withdrawn as a heartcut from a reformate rerun column or a steam-cracker heavy aromatic hydrodealkylation unit contains 40–55 wt% durene with 10–20 wt% naphthalene, 8–15 wt% pseudocumene, and 5–12 wt% isodurene and hemimellitene. This solution is fed to a scraped-surface crystallizer having 12–20 m² heat-transfer area per 10 kt/year of durene capacity, with a jacket coolant inlet at −5 °C to +15 °C and a controlled cooling rate of 5–10 °C/h. The resulting durene crystal slurry is concentrated to 30–40 wt% solids and discharged to a peeler centrifuge operating at 900–1,200 rpm; the cake is washed with chilled C6–C7 paraffinic naphtha at 0.5–1.5 kg wash solvent per kg cake to remove naphthalene and pseudocumene from the crystal surface. The isolation threshold is operationally defined as the minimum durene concentration in the wash-liquid-free cake that can be achieved without excessive crystal dissolution; a dried cake purity below 75 wt% durene is typically rejected from the oxidation feed network because non-durene C10 species become competing oxidizable substrates that increase the reactor hot spot and increase carbon oxide selectivity. Crystallizer jacket temperature and wash ratio are not independently optimized: increasing the wash ratio from 0.5 to 1.5 kg/kg reduces surface naphthalene concentration but dissolves 10–15% of the durene into the mother liquor, which must be recovered in a secondary scraped-surface crystallizer or by evaporation. Particle size distribution is measured by laser diffraction after slurry sampling; a mass-median crystal diameter below 150 µm reduces centrifuge dewatering capacity and raises residual cake moisture above 0.2 wt%. Durene crystal size is controlled by cooling rate and scraper tip speed; faster cooling rates above 10 °C/h generate fine particles and increase washed-away fines loss to the mother liquor. The washed cake is melted in a closed steam-traced melt tank maintained at 85–95 °C, blanketed with nitrogen, and transferred to the oxidation feed day tank. Storage below 25 °C causes solidification, while storage above 120 °C causes slow discoloration and accumulation of high-boiling oligomers; both conditions are treated as contamination events in the oxidation feed preparation. Purity is determined off-line by capillary gas chromatography following ASTM D5134, residual water by Karl Fischer titration following ASTM E203, and total sulfur by ultraviolet fluorescence following ASTM D5453.

What Purity Limits Are Applied Before Durene Enters the Oxidation Reaction Section?

PMDA-grade durene is specified less by a single purity figure than by the summed concentrations of catalyst poisons, combustibles, and high-boiling residues. A typical fixed-bed licensor specification lists a minimum durene content of 98.0 wt%, a maximum naphthalene content of 0.2 wt%, a maximum pseudocumene content of 1.0 wt%, a maximum total sulfur of 5 mg/kg measured by ASTM D5453, a maximum total nitrogen of 10 mg/kg measured by ASTM D4629, and a maximum ash content of 50 mg/kg measured by ASTM D482. At durene purity between 95.0 wt% and 98.0 wt%, oxidation may remain operable but requires a reduction in feed durene concentration from 1.2 mol% to 0.8 mol%, an increase in molten salt circulation rate of 10–20%, and a shortening of the operating cycle between catalyst regenerations. Published data for this specific configuration are limited; however, analogous phthalic anhydride fixed-bed operations demonstrate that impurity oxidation raises the peak bed temperature by 5–15 °C and shifts the hot spot upstream by 0.3–0.8 m. The threshold for rejecting a shipment is often tied to the sum of naphthalene and pseudocumene rather than durene purity alone because these two impurities have boiling points and condensation behavior close to durene and are not completely removed by crystallization. Naphthalene concentrations above 0.5 wt% in the melted feed are associated with naphthalene-free radical coupling and the formation of high-molecular-weight tars that deposit on the tube walls and raise pressure drop. Pseudocumene above 1.5 wt% competes for oxygen and leads to incomplete oxidation of the first durene methyl groups, shifting selectivity away from pyromellitic dianhydride and toward intermediate trimellitic anhydride and dimethyl phthalate species. Residual solvent in the durene cake is monitored by headspace gas chromatography and must be below 500 mg/kg because light naphtha vapors entering the oxidation reactor combust preferentially and create a sharp temperature spike near the tube inlet. The lower explosion limit of the durene-air mixture is controlled by maintaining the vaporised durene concentration below 1.4 mol%, but a segregated solvent flash drum is required to prevent localised solvent-rich slugs.

Hot spot control in a durene oxidation reactor is governed by the radial heat-transfer path from the catalyst bed through the tube wall into the molten salt bath. In a commercial multitubular fixed-bed reactor, the tubes are commonly 21–25 mm inner diameter and 3.0–3.5 m in length, arranged in a triangular pitch to give a tube-side heat-transfer area of 0.20–0.28 m² per tube. The tube-side feed is a vaporised mixture of durene at 0.8–1.2 mol% in air, because the lower explosion limit and dew point considerations constrain the feed concentration; the air-to-durene molar ratio is therefore on the order of 80:1 to 125:1. The shell side is filled with a potassium nitrate/sodium nitrite eutectic salt circulating at 1.0–2.0 m/s and entering at 340–400 °C. The peak bed temperature is usually observed 0.5–1.2 m below the tube inlet where the oxygen partial pressure and the unreacted durene concentration remain high. The design maximum deviation between peak bed temperature and salt-bath inlet temperature is held at ≤15 °C; excursions above 20 °C cause accelerated sintering of the vanadium pentoxide/titanium dioxide catalyst, loss of surface area, and a downstream migration of the hot spot. Inert dilution with low-surface-area fused alumina spheres of 4–6 mm diameter is applied in the upper 0.8–1.5 m of the tube, reducing volumetric heat release per tube length by 25–40% compared with undiluted catalyst. The dilution is not uniformly distributed: a common three-zone loading uses 60 vol%, 35 vol%, and 15 vol% inert in successive 0.5 m increments from the inlet, followed by undiluted catalyst to the tube outlet. This staging flattens the axial temperature profile and shifts the peak away from the tube wall, but it also reduces the average catalyst inventory and increases the pressure drop per unit conversion. Tube diameter is the strongest geometric lever on hot spot control; reducing the inside diameter from 25 mm to 21 mm shortens the radial conduction path and increases the bed-to-wall heat-transfer coefficient by 10–15%, but it raises the tube count by 20–25% for the same production capacity and increases the air-side pressure drop. Consequently, reactor mechanical design is fixed by an optimisation between hot spot margin and compressor power.

When Feedstock Sulfur Exceeds 5 mg/kg, Catalyst Deactivation Accelerates and the Hot Spot Moves

Sulfur-bearing compounds in durene are not removed by crystallization alone because thiophenic C10 homologs can co-crystallize with durene and remain in the washed cake. A sulfur concentration above 5 mg/kg in the vaporised feed has been reported in analogous fixed-bed aromatic oxidation systems to shift the peak bed temperature toward the reactor outlet by 10–20% of the tube length because the active sites in the upper bed are partially neutralised. The oxidation section is therefore designed with a liquid feedstock hydrodesulfurisation guard vessel or an activated carbon adsorber upstream of the vaporiser, and the durene sulfur is monitored online with ultraviolet fluorescence following ASTM D5453. If the upstream hydrotreating reactor is designed for 1 wt% sulfur in the C10 heartcut, the outlet can be held at <1 mg/kg; however, hydrotreating also saturates a proportion of the durene to tetramethylcyclohexanes, which are not oxidised to pyromellitic dianhydride and must be purged before crystallization. This trade-off creates a process conflict: hydrotreating improves catalyst life but reduces durene yield through ring saturation and cracking to lighter aromatics. Published data for this specific configuration are limited. Sulfur removal upstream of the crystallizer is preferred to post-melt adsorption because the crystallizer feed contains naphthalene and other sulfur heterocycles that compete for adsorption sites; therefore, the adsorber bed life is shortened if placed downstream of crystallization. The adsorption guard vessel is specified with a bed void fraction of 0.35–0.45 and a liquid hourly space velocity of 0.5–2.0 h⁻¹, but actual breakthrough depends on the sulfur speciation and the moisture content of the melted durene. Moisture above 0.2 wt% displaces adsorbed sulfur species and causes premature breakthrough. Avoid combination with amine-based additives in the isolation train because basic nitrogen compounds neutralise surface acidity and increase the temperature required for a given durene conversion.

Each reactor tube of the multitubular bundle is not individually instrumented; instead, axial temperature profiles are measured in 8–12 representative tubes fitted with multipoint thermocouples at 0.5 m intervals. The overall control loop cascades the molten salt inlet temperature to the highest of the measured peak bed temperatures, with a sampling time of 1 s and a response time of the salt heater limited to 30 s by the thermal inertia of the eutectic. In the event of a hot spot exceeding 25 °C above setpoint, the safety instrumented system initiates a hydrocarbon feed trip and nitrogen purge at 1.5–2.0 Nm³/h per m³ of catalyst bed, while maintaining the molten salt pump in service to remove stored heat. Pressure relief is designed in accordance with API 521 for the fire case and runaway exotherm case; the relief load considers the decomposition of durene and the formation of non-condensable gases. Reactor tubes are fabricated to ASME SA-213 TP316L for salt-side corrosion resistance, and the tubesheet is protected with ceramic ferrules at the inlet. The safety instrumented system is classified according to IEC 61511 with a Safety Integrity Level target of SIL 2 for the hot spot trip because the consequence of a thermal runaway includes tube rupture, salt/hydrocarbon contact, and loss of containment. Tube wall thermocouples are welded under ferrules at 0.5 m intervals, not inserted into the catalyst bed, to avoid creating preferential flow paths. The molten salt pump is interlocked to the reactor feed valve so that loss of salt circulation automatically trips the durene vaporiser and initiates an inert gas purge before the tube wall temperature exceeds the material limit of 550 °C.

Molten Salt Recirculation and Tube-Side Heat Transfer Resistance

The limiting heat-transfer resistance in a durene oxidation tube is not the molten salt film but the packed-bed gas film and the conduction path through the catalyst pellet and tube wall. With 4–6 mm fused alumina spheres and a superficial gas velocity of 1.5 m/s, the gas-side heat-transfer coefficient in the packed bed is typically 80–150 W/(m²·K), while the salt-side coefficient under forced circulation at 1.5–2.0 m/s is 400–800 W/(m²·K); the thermal conductivity of TP316L tube material at 400 °C is approximately 18–21 W/(m·K). This means the overall heat-transfer coefficient is dominated by the gas-side coefficient: increasing salt-bath velocity from 1.5 to 2.0 m/s raises the overall coefficient by only 5–10%, whereas increasing the gas mass flux by 20% can raise the gas-side coefficient by 15–25%. The salt circulation system is nevertheless designed for a high turndown ratio because molten salt stagnation develops localised shell-side temperature stratification, producing tube-to-tube temperature differences of 5–10 °C and hot spots in tubes exposed to stagnant zones. Baffle design in the shell side follows the same cross-flow principle as a conventional shell-and-tube exchanger but must avoid dead zones behind the baffle cut; computational fluid dynamics and residence-time distribution tests on full-scale loops are used during commissioning. The salt pump is specified with a centrifugal or axial flow configuration depending on the loop pressure drop; the impeller material is selected for molten nitrate service, and the pump casing is steam-jacketed to prevent freezing below 142 °C, the eutectic melting point of the potassium nitrate/sodium nitrite mixture. Salt-bath temperature measurement is performed with three independent thermocouples per inlet and outlet header, and the temperature difference across the salt side is maintained below 8 °C to ensure a uniform shell-side environment.

Catalyst deactivation in durene oxidation is observed first as an increase in the measured inlet-to-outlet pressure drop from 0.15–0.25 bar at start-of-run to 0.35–0.55 bar at end-of-run, caused by deposition of high-boiling oligomers and pyromellitic dianhydride fume in the upper bed. The hot spot moves downstream as the upper catalyst loses activity, and the salt-bath inlet temperature is raised by 2–5 °C per month to keep outlet conversion at 98–99%. When the pressure drop exceeds 0.55 bar, the air compressor discharge pressure can no longer maintain the required oxygen-to-durene ratio and the unit requires catalyst screening or regeneration. Pressure drop for spherical catalyst pellets follows the Ergun equation; a bed of 4 mm spheres with a void fraction of 0.40 at a superficial velocity of 1.5 m/s and bed length 3.0 m produces a calculated start-of-run pressure drop near 0.18 bar, consistent with field data for analogous fixed-bed aromatic oxidation reactors. Regeneration is carried out by controlled air decoking at 400–450 °C with oxygen concentration gradually increased from 0.5 vol% to 5.0 vol%, but repeated regeneration reduces catalyst activity by 3–8% per cycle due to sintering and vanadium oxide redistribution. During regeneration, the molten salt flow is reduced to prevent overcooling of the front section, and the reactor outlet is maintained above the dew point of water and PMDA to avoid condensation corrosion.

When Durene Feedstock Contains More Than 30 wt% Naphthalene, Fractional Crystallization Must Be Preceded by Melt Crystallization or Solvent Dewaxing

High-naphthalene C10 streams alter the durene solid–liquid equilibrium and require a prepurification step before the scraped-surface crystallization train. At naphthalene concentrations above 30 wt% in the C10 heartcut, the mother liquor viscosity increases sharply at crystallizer temperatures, and the durene crystals assume a plate-like habit with an aspect ratio near 1:3 that entrains naphthalene-rich liquid in the cake. A falling-film melt crystallizer or static crystallizer operated at −10 °C to +5 °C can produce a preliminary durene-enriched fraction of 85–90 wt% durene, after which the scraped-surface crystallizer and centrifuge achieve the final 98.0 wt% specification. The melt crystallization step is limited by the low crystal growth rate of durene and the need to drain naphthalene-rich mother liquor from the crystal bed; a post-crystallization sweating phase at 0–5 °C above the initial crystallization temperature removes trapped mother liquor but reduces overall yield by 5–10%. For streams containing more than 40 wt% naphthalene, solvent dewaxing with a polar aprotic solvent such as N-methyl-2-pyrrolidone is used to reduce the naphthalene concentration below the eutectic threshold, but solvent recovery adds distillation capacity and introduces a nitrogen-containing solvent that must be removed to below 1 mg/kg to avoid catalyst deactivation in the oxidation reactor. This is one of the strict operational boundaries: recovered NMP or amine-bearing wash solvents must never contact the oxidation feed because basic nitrogen compounds neutralise the surface acidity of the vanadium pentoxide/titanium dioxide catalyst and increase the temperature required for a given durene conversion. Activated carbon guard beds downstream of the solvent-recovery column are specified with a bed life of 6–12 months at 5 mg/kg breakthrough concentration, and the beds are regenerated off-site under inert atmosphere.