Articles
Air oxidation of mesitylene-rich feedstocks above 180°C is conducted in a continuous stirred tank reactor fabricated from titanium Grade 2 or Hastelloy C-276, with an ASME BPVC Section VIII Div 1 design pressure of 2.5 MPa and a reflux condenser surface area sized for a heat removal duty of at least 1.8 kW per kg of mesitylene feed at full air rate. The stoichiometry C6H3(CH3)3 + 4.5 O2 → C6H3(COOH)3 + 3 H2O consumes 4.5 mol O2 per mol mesitylene, equivalent to approximately 4.0 Nm³ air per kg mesitylene at 0°C and 101.325 kPa when vent losses are excluded; actual air feed ranges from 1.05 to 1.35 times stoichiometric demand to maintain dissolved oxygen while venting spent nitrogen. Yield preservation in this temperature region depends on minimizing three simultaneous pathways: oxidative decarboxylation of the tricarboxylic acid to isophthalic acid, ring oxidation to carbon oxides, and incomplete methyl group oxidation to 3,5-dimethylbenzoic acid or 5-methylisophthalic acid. The condensation of acetic acid-water mixtures at the reactor head is used to control the exotherm; acetic acid having a normal boiling point of 118°C requires a system pressure of at least 0.6 MPa at 180°C to keep solvent liquid, but commercial oxidation reactors operate above 1.5 MPa to maintain oxygen solubility and to suppress vapor-phase flammable mixtures. The off-gas after acetic acid recovery is monitored by a paramagnetic oxygen analyzer calibrated against nitrogen-diluted oxygen standards traceable to ISO 17025; the signal is used to trim air flow and to hold oxygen in the vent below the limiting oxygen concentration determined according to ASTM E681-09(2015).
As the bulk temperature exceeds 180°C, the thermal decarboxylation of aromatic carboxylic acids in acetic acid-water media becomes pH- and redox-dependent; the protonated acid form is increasingly susceptible to electrophilic substitution at the ring carbon bearing the carboxyl group, leading to CO2 elimination and formation of isophthalic acid from trimesic acid. In the presence of Co(III) and Mn(III), decarboxylation may proceed either as a radical process through acyloxy radicals or as a metal-catalyzed oxidative pathway, with the former favored by high bromine radical concentrations and the latter by low water content. Published data for this specific configuration is limited, but analogous p-xylene oxidation to terephthalic acid indicates that acid-catalyzed degradation accelerates at temperatures above 180–185°C when the acetic acid-water ratio exceeds 9:1 w/w. The main impurity signature is a rising isophthalic acid/trimesic acid ratio in the mother liquor, measured by high-performance liquid chromatography with a sodium phosphate-methanol gradient and UV detection at 240 nm using a procedure adapted from ASTM D7883-20 for aromatic carboxylic acid impurities. Over-oxidation of the methyl groups beyond the acid stage becomes detectable as CO2 in the vent and as a reduction in carbon selectivity from 85–90% to below 75% when hot spots exceed 200°C. Yield preservation therefore requires that the bulk temperature never exceed 190°C for more than 15 minutes during continuous operation, and that the steam generation section of the reflux condenser maintain a condensate return temperature below 110°C. The temperature difference between reactor bulk and jacket or internal coil inlet should not exceed 30°C to avoid local film boiling and inadequate heat transfer at the wetted surface; published plant data on analogous p-xylene oxidizers show tube-wall temperatures above 230°C accelerate metal-catalyzed methyl aromatic combustion.
At oxygen partial pressures below 40 kPa in the vent header, radical-chain termination becomes significant because the propagation step requires reaction of a methylbenzyl radical with molecular oxygen to form a hydroperoxide; the rate constant for this O2 addition step is high but its driving force in a three-phase slurry depends on oxygen mass transfer through the gas-liquid boundary layer. The sparger in a 5 m³-scale titanium reactor is typically a ring-type sintered metal element with a pore size of 10–50 µm and a hole-free surface area of 0.15 m² per m³ liquid; kLa values in such equipment range between 0.10 and 0.25 s−1 at superficial gas velocities of 0.03–0.08 m/s. If superficial gas velocity drops below 0.02 m/s, the oxygen transfer rate falls and partially oxidized intermediates accumulate in the liquid phase. The result is a higher 3,5-dimethylbenzoic acid to 5-methylisophthalic acid ratio and a lower final trimesic acid yield because the third methyl oxidation is kinetically slower once the ring becomes more electron-deficient. A practical control strategy is staged air addition: 70% of the stoichiometric air is fed to the first oxidation zone, 20% to the second, and the remainder to the top of the slurry above the gas-liquid interface; this reduces local oxygen supersaturation and hot spots while preserving oxygen availability for difficult third oxidation. Off-gas oxygen concentration should be maintained in the range of 2 to 6 vol% dry basis after acetic acid condensation, with the lower bound set by catalyst reduction and the upper bound limited by vent flammability and solvent losses. Vent oxygen analyzers with a response time less than 5 seconds are interlocked to a proportional-integral air flow controller that adjusts air feed to keep the off-gas oxygen at a setpoint of 3.5 vol%; a deviation of more than 1 vol% for 60 seconds triggers automatic air reduction and an alarm in the control room.
The soluble cobalt/manganese/bromide catalyst in continuous trimesic acid oxidation uses cobalt acetate, manganese acetate, and hydrogen bromide or sodium bromide in an acetic acid-water medium; the molar ratio of manganese to cobalt is usually held between 1 and 3 to accelerate the oxidation of the intermediate aldehyde to the carboxylic acid while cobalt promotes the initial methyl hydrogen abstraction. Bromide to cobalt ratio in the reactor liquid is kept between 0.5 and 1.0 because higher bromide levels above 180°C produce excessive methyl bromide and benzyl bromide byproducts and increase corrosion of titanium or Hastelloy wetted parts. The active catalytic cycle involves Br• abstraction of benzylic hydrogen, Co(III) regeneration by Mn(III), and Co(II) reoxidation by oxygen; kinetic measurements on analogous p-xylene systems show a rate dependence on oxygen partial pressure that follows a saturation law, so raising air pressure from 1.5 to 2.5 MPa increases rate substantially only below a dissolved oxygen concentration of about 25% of air saturation. Bromide is lost as HBr in the water-rich condensate and as methyl bromide in the off-gas; replenishment is carried out by injecting a dilute sodium bromide solution at a rate tied to the measured bromide concentration in the mother liquor by ion chromatography using a suppressed conductivity detector. Insufficient bromide causes accumulation of aldehyde intermediates and lower product yield, while excess bromide increases corrosion and can cause brominated ring impurities that are difficult to remove by recrystallization.
| Measured parameter | Method or standard | Operational boundary above 180°C |
|---|---|---|
| Off-gas oxygen concentration | Paramagnetic analyzer; ASTM E681-09(2015) | Alarm at 60% of limiting oxygen concentration; typical setpoint 3.5 vol% dry |
| Mother liquor water content | ASTM E203-16 Karl Fischer titration | Maintain 5–10 wt% water to promote acid dissociation and reduce decarboxylation |
| Free bromide ratio | Ion chromatography with suppressed conductivity per ASTM D4327-17 after aqueous dilution | Maintain 0.5–1.0 mol/mol Br/Co; replenish below 0.4 |
| Crude cake impurity profile | ASTM D7883-20 HPLC | 3,5-dimethylbenzoic acid below 2.0 wt%; 5-methylisophthalic acid below 1.5 wt% |
| Slurry supernatant pH at 25°C | ASTM D1293-18 pH electrode | Keep 2.5–3.5; lower pH accelerates decarboxylation |
| Vacuum dried moisture | ASTM E203-16 | Below 0.1 wt% before packaging |
The numerical boundaries in Table 1 are process-control starting points; published data for this specific configuration is limited, and each campaign requires revalidation with the actual feedstock and catalyst charge.
From the crystallizer to the filter feed tank, slurry density is a lagging indicator of solids hold-up and product recovery; for a continuous crystallizer operating at 180°C, the slurry density is typically controlled between 25 and 35 wt% solids because higher solids reduce oxygen mass transfer and increase impeller power draw, while lower solids increase residence time and allow dissolved trimesic acid to undergo solvent-mediated decarboxylation. The crystallizer is fitted with a retreat-curve impeller or a pitched-blade turbine with a tip speed of 2.5–3.5 m/s; measured slurry viscosity at 25°C by ASTM D2196 Method B ranges from 15 to 60 mPa·s depending on particle size distribution, and at reaction temperature the apparent viscosity is closer to water because the solids are suspended in acetic acid-water. A drop in agitator power draw during normal air feed often indicates gas cavity formation behind the blades, while a rapid rise indicates excessive crystallization in the impeller zone; both are avoided by maintaining the liquid level within 5% of design and by controlling the product withdrawal rate with a positive displacement slurry pump. Crystal mean particle size is measured by laser diffraction according to ISO 13320:2020; a median size below 15 µm increases filtration resistance and occluded mother liquor, while a median size above 200 µm creates handling difficulties and slower washing in the rotary vacuum filter. Particle size above 180°C is controlled by the cooling rate from the crystallizer to the filter feed tank; a cooling rate of 10–15°C/h is maintained by heat exchange in a scraped-surface cooler to avoid nucleation bursts that trap impurities. The filter feed is kept above 80°C to reduce mother liquor viscosity, but not above 100°C to limit evaporation and crust formation on the filter cloth.
If off-gas oxygen falls below 3 vol% during a continuous campaign, the catalyst redox cycle shifts toward Co(II) and Mn(II) accumulation because the gas-liquid oxygen flux is insufficient to reoxidize the metals; the bromide radicals then attack the aromatic ring rather than the methyl C–H bond, leading to brominated intermediates and higher solvent combustion. In this condition, the reactor temperature tends to fall because the radical initiation rate decreases, but the apparent oxygen consumption may rise as partially oxidized intermediates compete for the limited dissolved oxygen. The immediate corrective action is to reduce the mesitylene feed rate by 15 to 20% while holding air flow constant, increasing the oxygen-to-substrate ratio; the air flow should not be increased abruptly because that can carry acetic acid and corrosive bromide into the vent system and can push the off-gas oxygen above the flammability limit after the catalyst reoxidizes. The cobalt/manganese ratio in the liquid phase is then rechecked by atomic absorption spectroscopy or inductively coupled plasma optical emission spectroscopy using matrix-matched standards; if the dissolved cobalt concentration has decreased by more than 10% from its initial value due to precipitation as cobalt acetate or cobalt carbonate, a small amount of acetic acid or water is added to restore solubility. The off-gas oxygen setpoint is reset to 4.5 vol% for the following 30 minutes until the vent CO2 concentration measured by non-dispersive infrared stabilizes below 2 vol% and the measured 3,5-dimethylbenzoic acid concentration in the reactor filtrate falls below 3 wt% by ASTM D7883-20. Safety interlocks on the air compressor and the vent gas oxygen analyzer are verified against certified span gases before the campaign returns to peak feed; the lower flammability limit and limiting oxygen concentration data from ASTM E681-09(2015) and the vent header calculations per ISO 10156:2017 are revalidated with the actual off-gas composition.
Post-oxidation workup introduces an additional thermal load that must not exceed the decarboxylation threshold; recovered trimesic acid is discharged as a wet cake at 90–105°C with residual mother liquor containing acetic acid, water, cobalt/manganese/bromide salts, and isophthalic acid. The cake is washed countercurrently with acetic acid at 100°C and then with deionized water at 95°C using a displacement ratio of 1.2 to 1.5 kg wash liquor per kg dry cake; insufficient washing leaves sodium or bromide residues that interfere with downstream polyester or polyamide modifiers, while excessive washing increases wastewater treatment load. Wash efficiency is confirmed by acid number of the wash filtrate according to ASTM D664-18e2; the washed cake is dried in a conical vacuum dryer at 110–120°C under 70–80 kPa absolute pressure for 8 to 12 hours. A moisture content below 0.1 wt% as measured by ASTM E203-16 is required for sale as a polymer-grade intermediate. Drying temperatures above 130°C should be avoided because trimesic acid begins to sublime and thermally degrade at elevated temperatures; the vacuum system is equipped with a cold trap cooled to -10°C to recover sublimed acid and maintain vacuum. The dry powder is screened through a 150 µm sieve to remove aggregates and packed in polyethylene-lined paper bags under nitrogen. The overall recovered yield above 180°C can be preserved if the oxidation is stopped at 90–95% conversion and the residual intermediates are recycled to the oxidation feed after crystallization; operating at complete conversion increases the isophthalic acid impurity level because the desired trimesic acid remains in the hot acidic medium for the longest residence time. This recycle stream contains 3,5-dimethylbenzoic acid and 5-methylisophthalic acid, which are converted in subsequent passes because the methyl substituents remain available for further oxidation, but the stream must be evaluated for metal and bromide content to avoid catalyst overload.