Mixed xylenes recovered from catalytic reforming or pyrolysis gasoline processing are not a single feedstock but a four-component C8 aromatic pool whose separation behaviour is controlled by close boiling points and widely separated freezing points. In a representative reformate-derived mixed xylene, m-xylene accounts for 45–50 wt%, p-xylene for 20–25 wt%, o-xylene for 20–25 wt%, and ethylbenzene for 10–15 wt%; pyrolysis gasoline streams can raise the ethylbenzene content to 30–40 wt%. The normal boiling points at 101.325 kPa are 136.2 °C for ethylbenzene, 138.3 °C for p-xylene, 139.1 °C for m-xylene, and 144.4 °C for o-xylene. The p-xylene/m-xylene binary exhibits a boiling-point difference of only 0.7 K, which corresponds to a relative volatility of approximately 1.02; the m-xylene/o-xylene binary exhibits a gap of 5.3 K, and the ethylbenzene/m-xylene binary a gap of 2.9 K. Direct fractional distillation therefore cannot perform the entire m-xylene isolation economically, and commercial purification is arranged as a sequence of upstream p-xylene removal, ethylbenzene rejection, m-xylene/o-xylene distillation, and downstream trace impurity verification. Gas chromatographic characterization by ASTM D7504-21, with flame ionization detection and effective carbon number correction, is the standard method used to quantify trace C8 aromatic impurities at mg/kg levels during pilot tests and production checks. Table 1 lists the physical constants that dominate the selection of unit operations.

Why does the m-xylene/p-xylene boiling gap impose a superfractionator limit?

At a relative volatility of 1.02, the p-xylene/m-xylene separation is not a simple close-boiling distillation; it is a high-stage-count superfractionation problem that becomes hydraulically and thermally unstable as the number of theoretical stages is increased. The Fenske total reflux relationship yields a minimum theoretical stage requirement of approximately 600–700 stages for a p-xylene overhead composition of 99.0 wt% and a m-xylene bottoms composition of 99.0 wt% when the feed contains 50 wt% m-xylene and 20 wt% p-xylene. Operation at a practical reflux ratio above minimum multiplies the required internal liquid and vapour flows; a reboiler duty of 15–25 GJ/t product is not unusual for such an arrangement, and the column pressure drop becomes a limiting factor rather than the heat-transfer surface. Industrial structured packing with specific surface area of 500–750 m²/m³ can reduce tray count and tower height by 30–40 % compared with high-performance sieve trays, but its capacity is limited by liquid distributor quality and feed distribution. Distributor level tolerances must remain within ±3 mm across the column cross-section, and the hydraulic F-factor should be held between 1.5 Pa⁰.⁵ and 2.0 Pa⁰.⁵ to avoid entrainment and loss of stage efficiency. Under these conditions, p-xylene/m-xylene distillation is not used commercially as the first purification step; the accepted practice is to remove p-xylene upstream by melt crystallization or simulated moving bed adsorption, leaving only the ethylbenzene/m-xylene and m-xylene/o-xylene binaries for distillation.

After the p-xylene-lean stream has been produced, the next process boundary is the two-step distillation of ethylbenzene and o-xylene from m-xylene. Because ethylbenzene is lighter than m-xylene, it is rejected as an overhead stream in a first column. The column is typically operated at a top pressure of 20–40 kPa to maintain a condenser temperature of 45–60 °C and to increase the relative volatility slightly. A column with 120–180 theoretical stages and structured packing having an HETP of 0.25–0.35 m can reduce the ethylbenzene content to below 0.10 wt% in the m-xylene bottoms. The second column separates m-xylene from o-xylene; m-xylene is taken overhead at 99.0–99.5 wt% purity by ASTM D7504-21 analysis, while o-xylene leaves the bottom. For a feed containing 70 wt% m-xylene and 25 wt% o-xylene, a reflux ratio of 6–10 and 150–200 theoretical stages are typical; the reboiler film temperature is kept below 175 °C to suppress oxidation products and trace styrene polymer. Process water entering the columns must be below 50 mg/kg because water changes relative volatility and can promote chloride-induced pitting in carbon steel. Chloride itself is controlled to below 1 mg/kg to protect structured packing and reboiler tubes. In actual production units, the main failure mode is not the separation thermodynamics but hydraulic maldistribution caused by fouled liquid distributors; a routine wash with hot aromatic solvent every 6–12 months is often required to restore stage efficiency.

Melt crystallization as p-xylene subtraction, not m-xylene direct recovery

Commercial p-xylene melt crystallization is operated at -5 to 5 °C, a window set by the p-xylene freezing point of 13.2 °C and the need to avoid freezing m-xylene at -47.9 °C. The crystallizer is a scraped-surface heat exchanger in which coolant is evaporated on the shell side; propylene refrigerant at evaporating temperatures of -15 to -5 °C removes heat through a film of stagnant hydrocarbon. In this operation, p-xylene crystals are formed, and the mother liquor becomes enriched in m-xylene; a typical mother liquor after washing contains 55–65 wt% m-xylene, 20–30 wt% o-xylene, 5–10 wt% ethylbenzene, and <1 wt% p-xylene. The slurry is transferred to a wash column, where residual p-xylene is removed from the crystals by countercurrent washing; the wash column diameter for a world-scale line is typically 1.0–2.0 m. Direct m-xylene crystallization is technically possible below -50 °C, but published data for this specific configuration is limited and indicates that the viscosity of the mother liquor rises sufficiently to reduce crystal growth and hinder slurry transport. The low density difference between m-xylene crystals and the liquid at that temperature further degrades washability. Therefore, crystallization in m-xylene purification is used as a p-xylene subtraction step rather than as a direct m-xylene recovery step. The main operating risk in p-xylene crystallization is not thermodynamic but mechanical: scraped-surface units experience wear of the scraper blades, and cooling surface fouling raises the required coolant temperature drop by 5–10 °C over a run length of 3–6 months; this is monitored by pressure drop across the crystallizer and the p-xylene purity of the crystal product.

Simulated moving bed adsorption and desorbent management for raffinate enrichment

Simulated moving bed adsorption using faujasite-type zeolites is the dominant industrial route for p-xylene recovery and therefore creates the principal m-xylene-enriched raffinate. The separation is achieved by adsorbing p-xylene preferentially within the zeolite pore structure while m-xylene, o-xylene, and ethylbenzene move as the raffinate stream. Industrial units operate at 130–180 °C and 0.5–1.0 MPa inlet pressure, with a liquid-phase desorbent such as p-diethylbenzene or toluene. The extraction column contains 12–24 compartments filled with a bound zeolite adsorbent, and a rotary valve manages the periodic movement of feed, desorbent, extract, and raffinate ports. Switch times are selected between 30 s and 120 s to match the adsorption front velocity to the internal liquid velocity. A key operational limit for m-xylene recovery is the raffinate purity: because para-xylene adsorption is favoured, the raffinate retains most m-xylene, o-xylene, and ethylbenzene, but it also contains some desorbent and trace para-xylene. The desorbent is separated from the raffinate in a first distillation column; it must be recycled at a water content below 100 mg/kg to avoid zeolite hydrothermal damage. Adsorbent aging typically appears as a loss of para-xylene selectivity of 0.1–0.3 units over 3–5 years and is monitored by extract purity decline rather than by direct pore inspection. Where an operator seeks to obtain m-xylene directly as the adsorbed extract, published data for a commercial SMB configuration is limited; the standard industrially proven operation is recovery of m-xylene from the raffinate, followed by distillation.

When extractive distillation becomes technically and economically defensible for this separation

Extractive distillation has been evaluated for the m-xylene/p-xylene binary because the added polar solvent can alter relative volatility through differential solvation of the p-xylene molecule. Solvents described in technical literature include sulfolane, N-methyl-2-pyrrolidone, N-formylmorpholine, and dimethyl sulfoxide. The observed relative volatility shifts are smaller than those obtained in aliphatic/aromatic extractive distillation; a solvent-to-feed mass ratio of 2:1 to 6:1 may increase the p/m relative volatility from 1.02 to only 1.03–1.08, depending on solvent purity and water content. The resulting column still requires a high stage count, and the main cost moves from the distillation column to the solvent recovery section. Solvent degradation products, oxygenates, and chloride ions can accumulate in the reboiler and accelerate corrosion; sulfolane-based systems must be operated at reboiler temperatures below 180 °C and with oxygen ingress limited to <10 mg/kg to prevent acid formation. In patent literature, coordinated chemical agents based on HF-BF₃ or silver nitrate have been proposed to form selective complexes with m-xylene; however, the corrosive and toxic handling burden has prevented commercial adoption at world-scale m-xylene capacity. Because of these constraints, extractive distillation is not the standard first-line route for m-xylene recovery; it is considered only when a site has spare solvent recovery infrastructure and cannot justify a dedicated simulated moving bed train.

Xylene isomerization reactors are generally configured to convert residual ethylbenzene and a portion of m-xylene to p-xylene in p-xylene production; for m-xylene purification, the isomerization loop is smaller because m-xylene is already the thermodynamically favoured isomer above normal isomerization temperatures. At 380–420 °C over a Pt/zeolite catalyst, the C8 aromatic equilibrium contains m-xylene at roughly 50–55 wt%, p-xylene at 20–25 wt%, o-xylene at 20–25 wt%, and ethylbenzene depending on dealkylation. A m-xylene producer can therefore operate the isomerization reactor to replenish m-xylene from the p-xylene-lean and o-xylene-lean streams rather than to destroy m-xylene. The reactor effluent is cooled and sent to a stabilizer, where light hydrocracked gas is removed. Hydrogen partial pressure is maintained at 0.5–1.5 MPa to limit coke deposition; moisture and chloride in the feed are controlled to <1 mg/kg and <0.5 mg/kg, respectively, to protect the zeolite acid sites. In industrial units, a shift in the para/meta ratio in the reactor effluent is an early indicator of chloride breakthrough into the catalyst bed. The reactor loop raises the overall m-xylene yield per tonne of mixed xylene feed but also increases the concentration of nonaromatic impurities that must be removed in the ethylbenzene column.

Not every purity specification transfers across jurisdictional boundaries

Analytical verification of m-xylene purity depends on the intended downstream conversion. For isophthalic acid production, m-xylene is fed to a liquid-phase air oxidation using cobalt/manganese/bromide catalysts; trace p-xylene and o-xylene are tolerable only up to limits that prevent off-spec isophthalic acid colour and by-product formation. Typical limits are 0.10 wt% p-xylene, 0.15 wt% o-xylene, 0.10 wt% ethylbenzene, and 0.05 wt% total nonaromatics. Analysis by ASTM D7504-21 provides a common gas chromatographic basis for these limits. Compliance with occupational-exposure and transport classification is governed by REACH EC 1907/2006 and GHS Regulation (EC) No 1272/2008; the low flash point of m-xylene, 25 °C closed cup, requires process equipment rated for flammability under IEC 60079-10-1 zone classification. In storage and handling, m-xylene containing trace water above 200 mg/kg can exhibit haze and phase separation at low ambient temperature; nitrogen blanketing at gauge pressure 2–5 kPa prevents oxygen ingress and limits peroxide formation. Antioxidants or amine-based inhibitors must not be introduced into m-xylene storage if the material is to be sent to catalytic oxidation; traces above 5 mg/kg may retard the radical chain and reduce isophthalic acid yield. These boundary conditions are integrated into the separation train rather than deferred to tank-farm practice.