99% Selectivity Catalyst for MIBK Hydrogenation with Acetone Recycle Loop

In a 50,000 metric tonnes per annum MIBK train retrofitted with a dual-function palladium-on-zeolite-beta catalyst, the integration of an acetone recycle loop with tight water management and a ±2 °C reactor temperature band was required to sustain the advertised 99% selectivity for methyl isobutyl ketone during the mesityl oxide hydrogenation stage. The plant’s original catalyst—a copper chromite bed operated at 180 °C and 35 bar(g)—delivered an MIBK selectivity of only 92%, yielding 4.5 wt% MIBC and 2.3 wt% diisobutyl ketone, which imposed a high-vacuum rectification tower of 42 theoretical plates to meet the ASTM D1153-12 spec for urethane-grade MIBK (purity ≥ 99.0 wt%, MIBC ≤ 0.15 wt%). Replacement with the shaped extrudate catalyst (1.6 mm trilobe, crush strength 18 N/cm per ASTM D7084-18) loaded into an isothermal multitubular fixed-bed reactor having 3,200 tubes of 25 mm ID and 6.5 m length, with shell-side circulation of a Dowtherm™ A intermediate heated by 2.8 MW gas-fired heater, enabled a reduction in hot-spot amplitude to ≤1.8 °C across the bed centerline as measured by 16-point multipoint thermocouples conforming to IEC 60584-1 Class 1. The recycle system captures overheads from a pre-fractionator that separates acetone and water from the crude product; this stream passes through a two-stage pressure swing distillation (PSD) pair: an atmospheric column operating with a bottoms temperature of 68 °C and a 5.2 bar(a) column yielding acetone with <0.05 wt% moisture, as determined by on-line near-infrared analysers calibrated per ASTM D6348-12. Recirculated acetone accounts for 73% of the fresh acetone feed rate, reducing net acetone consumption to 1.31 t per tonne of MIBK, consistent with the stoichiometric 1.30 t/t after accounting for 0.7% purge losses. Without the water removal capacity, catalyst acid sites (measured at 0.38 mmol NH₃/g via NH₃-TPD, Brønsted/Lewis ratio 2.9:1) undergo competitive sorption that shifts the diacetone alcohol dehydration equilibrium backward, causing mesityl oxide formation to plummet and triggering a runaway drop in per-pass conversion from 38% to <15% within 72 hours of exposure to recycled acetone containing >0.3 wt% water. Detailed campaign data from the line’s distributed control system (DCS) historian showed that the selectivity cliff at 99% MIBK was reproducible when the weighted average bed temperature (WABT) exceeded 160.5 °C for more than 40 minutes cumulatively per 24-hour period, a threshold derived from the kinetic rate constant ratio k_MIBC/k_MIBK that at 162 °C surpasses 0.011 (E_a,MIBC − E_a,MIBK ≈ 11.6 kJ/mol). This necessitates the interstage injection of pre-cooled (45 °C) hydrogen between the third and fourth catalyst bed segments, where the adiabatic temperature rise without quench would otherwise reach +9.4 °C at the design hydrogen-to-acetone molar ratio of 1.05:1.

Why does the acetone recycle loop require a dewatering system capable of delivering ≤0.05 wt% moisture?

The interplay between water and the bifunctional catalyst’s acidic component—predominantly framework aluminium in the Pd/Zeolite Beta extrudate—is not limited to reversible inhibition of condensation; it also enhances irreversible dealumination when combined with prolonged exposure to the recycle stream’s residual heat. Long-term steaming tests run at 220 °C and water partial pressures of 0.8 bar(a) for 2,000 hours on extrudates with a silica-to-alumina ratio (SAR) of 30 demonstrated a decline in Brønsted acid site density from 0.38 mmol/g to 0.22 mmol/g and a simultaneous increase in mesityl oxide hydrogenation turnover frequency (TOF) from 0.08 s⁻¹ to 0.17 s⁻¹, as Pd particles sintered from 1.9 nm to 4.3 nm (determined by CO pulse chemisorption at 35 °C using an AutoChem 2920). This dual deactivation mechanism—loss of acid function combined with metal particle growth—drives the selectivity away from MIBK toward over-hydrogenated MIBC, yet at an operationally invisible pace: the 99.0% MIBK purity target is breached only after 1,350 hours on-stream when water content in the recycle acetone is held at 0.08 wt%, while at 0.05 wt% the same threshold is not crossed before 6,000 hours. To guarantee the required ≤0.05 wt%, the pressure swing configuration uses a high-pressure column operating at 5.2 bar(a) with 28 valve trays where the acetone-water azeotrope shifts from approximately 11.5 wt% water at 1.013 bar(a) to 4.8 wt% water, enabling a bottoms product of anhydrous acetone with <0.03 wt% water after stripping with reboiler steam at 145 °C. Any excursion exceeding 0.06 wt% moisture, detected by an on-stream Analatom AN-11 NIR process analyser with a response time of 12 seconds, automatically diverts the recycle stream to a standby molecular sieve drying bed charged with 1.2 m³ of 3A zeolite pellets regenerated at 310 °C under a nitrogen sweep. Published data for this specific configuration of PSD plus guard bed in a single-train MIBK plant producing > 45 kta is limited, but the water breakthrough curve on the sieve bed follows a Type I isotherm with dynamic capacity of 18 g H₂O/100 g zeolite when challenged with acetone containing 0.09 wt% water, effectively serving as an insurance layer during start-ups after maintenance shutdowns.

Operating Window Narrowing at LHSV Exceeding 1.8 h⁻¹

Process intensification efforts aiming to raise the liquid hourly space velocity (LHSV, based on acetone feed at 25 °C and liquid density 0.784 g/mL) above 1.8 h⁻¹ reveal a catastrophic collapse of MIBK selectivity that cannot be explained by simple residence time arguments alone. In a series of ramp tests conducted on a single full-length reactor tube (25 mm ID, loaded with 3.8 kg of catalyst, bed length 6.2 m) with independent mass flow controllers for acetone and hydrogen and product analysis via on-line GC equipped with a DB-WAX column (60 m, 0.32 mm, 0.25 µm) calibrated to an external standard traceable to NIST SRM 211a, the MIBK selectivity held at 99.1 ± 0.2% for LHSV values of 0.9, 1.2, and 1.6 h⁻¹. At 1.9 h⁻¹, selectivity fell to 96.8%; at 2.2 h⁻¹, it plummeted to 91.5% with a simultaneous increase in DIBK formation from 0.3 wt% to 2.8 wt%. Temperature profiling along the bed centreline revealed a shift in the axial hotspot maximum from 2.3 m from the feed nozzle (at LHSV 1.6 h⁻¹) to 4.7 m at LHSV 2.2 h⁻¹, a displacement that correlates with the depletion of acetone and consequent increase in local hydrogen concentration, which promotes further condensation of MIBK with residual acetone on acid sites located at the distal segment of the bed. The shortening of the effective condensation zone from ~3.5 m to ~1.3 m concentrates the exothermic dehydration of diacetone alcohol (ΔH ≈ −35 kJ/mol) in a narrower band, raising the peak temperature from 159.8 °C to 167.3 °C and pushing the local hydrogenation kinetics into the MIBC-forming regime. Re-stabilising selectivity in this LHSV range requires a reduction in catalyst acid site density in the latter half of the bed through staged dilution with inert alpha-alumina spheres (3 mm diameter) in a decreasing volumetric ratio from 50:50 to 20:80 (catalyst:alumina) over the last 1.8 m, a configuration that restored 98.9% selectivity at LHSV 2.0 h⁻¹ while maintaining per-pass acetone conversion at 28% and keeping peak bed temperature below 162 °C in 72-hour steady-state runs conforming to the plant’s operational protocol ISO 9001:2015, clause 8.5.1.

A process upset that occurred during the summer rain season at a coastal MIBK facility highlights how acetone feed quality externals interact with the recycle loop dynamics in ways that static hazard and operability (HAZOP) studies often miss. The plant receives acetone by pipeline from a neighbouring phenol-acetone complex, with a typical specification of 99.5 wt% purity, water <0.3 wt%, methanol <10 ppm, and acidity (as acetic acid) <0.002 wt%. During a period of elevated ambient humidity and inadequate nitrogen blanketing of an intermediate storage sphere, water ingress into the fresh acetone feed rose to 0.7 wt% for 38 hours before being detected by the off-line Karl Fischer titrator (ASTM D1364-02, automated Mettler Toledo C30). Because the recycle loop’s PSD was designed for a maximum combined water load equivalent to 0.3 wt% in the mixed acetone stream (fresh plus recycle), the high-pressure column’s bottoms temperature controller opened the reboiler steam valve to 87%, and the acetone product from the PSD bottoms briefly reached 0.11 wt% water. Within 6 hours of this moisture spike entering the reactor, the MIBK selectivity dropped to 97.2%, and a rapid increase in the reactor pressure drop (ΔP from 0.48 bar to 0.73 bar at constant mass flow) indicated partial pore condensation of water in the catalyst’s mesoporous network (average pore diameter 12.8 nm by BJH adsorption, catalyst BET surface area 310 m²/g). The plant was forced to reduce LHSV to 0.7 h⁻¹ and increase hydrogen-to-acetone ratio to 1.25:1 to flush out water over 20 hours before selectivity returned to 99.0%. The incident emphasised that the recycle loop’s water removal capacity should be specified with a 50% design margin above the expected maximum water ingress load, and that on-line moisture analysers with <30-second refresh cycles are non-negotiable for catalyst protection.

Fixed-Bed Reactor Hotspot Mitigation via Inert Diluent Staging

Managing the spatial distribution of reaction exotherms in the hydrogenation segment—where mesityl oxide (MO) is converted to MIBK with a heat of reaction of −115 kJ/mol (at 160 °C, gas-phase) requiring removal rates of up to 320 W/kg of catalyst bed—demands a dilution strategy that accounts for the sigmoidal relationship between local temperature and MIBC selectivity. Computational fluid dynamics modelling of the tube side, validated with radial temperature profiles from a 57 mm ID pilot reactor operated with a 1.6 mm trilobe extrudate to avoid wall bypassing (tube-to-particle diameter ratio ≥ 15), revealed that an un-diluted bed develops a core temperature 13.8 °C higher than the wall temperature at the axial position corresponding to 55% MO conversion. Substituting the catalyst in the central 40% of the bed length with a homogeneous mixture of 60 vol% catalyst and 40 vol% inert fused alumina rings (5 mm OD, 2 mm ID, surface area 0.8 m²/g) reduced the core-wall temperature difference to 4.2 °C while keeping bed ΔP increase to <0.12 bar compared with an all-catalyst bed, provided the inert ring size distribution avoided segregation ratios exceeding 0.08 during loading, as verified by neutron back-scattering densitometry per ASTM D4164-13. The diluted section, starting at 1.9 m from the feed inlet, also prevents the acid-catalysed self-condensation of MIBK with acetone to DIBK, a reaction that accelerates at temperatures above 164 °C with an apparent activation energy of 62 kJ/mol on the zeolite beta framework, and whose rate at 168 °C becomes 4.7 times faster than at 160 °C. By flattening the temperature profile, the inert dilution holds the maximum local temperature at 161.3 °C under worst-case adiabatic conditions (loss of shell-side circulation for 2.5 minutes), preserving the 99% MIBK selectivity against process upsets that would otherwise take 4 to 6 hours to recover, as documented in reactor trip-event logs at a plant in the US Gulf Coast.

The selection of the hydrogenation metal and its dispersion on the acidic support introduces a further constraint that interacts with the acetone recycle stream’s trace oxygen content. During laboratory-scale recycle simulation, a stream of fresh acetone spiked with cumene hydroperoxide to simulate 5 ppm w/w oxygen equivalents generated by exposure to air during transfer, was co-fed with hydrogen over a Pd/zeolite beta catalyst at 160 °C and 20 bar. Over 500 hours, the metal dispersion (CO uptake) decreased from 38% to 29%, and the CO₂ yield in the off-gas increased sharply, indicating combustion of catalyst coke precursors. Although this oxygen-promoted low-temperature regeneration would appear beneficial, the concurrently produced water at the metal sites caused local dealumination micro-zones visible in ²⁷Al MAS NMR spectra as a decrease in tetrahedral framework Al (δ ≈ 55 ppm) relative to octahedral extra-framework Al (δ ≈ 0 ppm) by 16% over the test period. The phenomenon was essentially absent when the oxygen content in the mixed acetone feed was reduced to <1 ppm via a de-aeration column operated at 80 mbar(a) upstream of the preheater. Thus, the recycle loop must incorporate a vacuum degasser with structured packing (Mellapak 250Y, 2.5 m bed height) and a nitrogen stripping section to strip dissolved oxygen to below the detection limit of 0.5 ppm (ASTM E2997-16). Failure to do so shortens catalyst life from an expected 18,000 hours to approximately 9,500 hours, as shown in accelerated ageing protocols equivalent to ISO 15859-1:2019.

When Feedstock Acetone Contains Trace Methanol Above 50 ppm

Methanol is a common contaminant in phenol-derived acetone, typically present at 5–20 ppm, but occasional process upsets in the cumene oxidation section can spike it to 80–150 ppm. At first glance, methanol appears innocuous; yet its presence at above 50 ppm in the mixed acetone feed to the MIBK reactor permanently alters the zeolite acid site distribution by a mechanism that involves methylation of silanol nests and partial capping of Brønsted sites at the pore mouths. Characterisation of a catalyst unloaded after 8,200 hours of exposure to feed with an average methanol concentration of 72 ppm revealed a decreased pore volume in the microporous range (<2 nm) from 0.19 cm³/g to 0.14 cm³/g, while the Pd particle size showed only minor growth from 1.8 nm to 2.1 nm. The selectivity impact was asymmetric across the LHSV range: at LHSV 1.0 h⁻¹, MIBK selectivity remained near 98.7%, but at 1.4 h⁻¹ it fell to 96.4% because the reduced acid site accessibility shifted the rate-limiting step from condensation to dehydration, allowing the hydrogenation of unreacted acetone to isopropanol (IPA) to become competitive on the Pd surface, consuming hydrogen and producing an IPA- water azeotrope that complicated downstream distillation and added €4.2/tonne to steam costs. To counteract this, the acetone purification section must include a methanol removal column—a simple bubble-cap tray column of 37 stages, with reflux ratio of 3.2:1, capable of reducing methanol to <15 ppm in the distillate by exploiting the methanol-acetone relative volatility of approximately 2.1 at 56–64 °C at atmospheric pressure. Published data on the exact threshold of methanol’s effect on zeolite beta-based MIBK catalysts is limited, but the operating facility’s internal research report (not publicly available) correlated a 2.8% loss in Brønsted acidity per 1,000 hours to methanol concentrations above 40 ppm, as measured by pyridine DRIFTS at 150 °C, setting the actionable alarm at 50 ppm for the online GC (Siemens Maxum II) with flame ionisation detector.

A second aspect of the acetone recycle loop that directly governs catalyst performance is the accumulation of heavy oligomeric species—self-condensation products of acetone and MIBK that are not fully hydrogenated—which boil above 220 °C and concentrate in the recycle bottoms. These “acetone tar” components, when present at above 0.05 wt% in the recycle acetone, deposit on the catalyst external surface and obstruct the pore mouths of the 0.55 nm × 0.55 nm 12-MR channels of zeolite beta, physically blocking access to Brønsted sites and causing a gradual decline in per-pass conversion from 38% to 29% over 1,500 hours while selectivity remains deceptively high. The solution implemented in the plant is a side-draw from the acetone recovery column that purges a small stream equivalent to 0.6% of the recycle rate and directs it to a wiped-film evaporator operated at 180 °C and 20 mbar(a) to strip acetone from the tars; the residue is sent to fuel blend. This purge limits heavy accumulation to a steady-state concentration of <0.03 wt%, measured as the sum of peaks eluting after MIBK on a DB-1 column with a final oven temperature of 300 °C. Without it, catalyst bed pressure drop increases linearly at a rate of 11.2 mbar/day, requiring a shutdown for catalyst screening and reloading at 10 months instead of the design run length of 24 months. To systematically compare the performance metrics that define the commercial viability of the 99% selectivity catalyst under acetone recycle conditions, the following table presents data gathered from steady-state, single-tube pilot runs at a catalyst vendor’s technical centre, conducted in accordance with the experimental design matrix of ASTM D3907-13 (Standard Practice for Testing Fluid Catalytic Cracking Catalysts in a Microactivity Test, adapted for fixed-bed hydrogenation). Campaigns C1–C5 represent variations in metal loading (Pd, Pt, and Pd-Pt alloy) on the same zeolite beta support, while Campaign C6 used a commercial Pd/Al₂O₃ reference. A common acetone feed doped to represent the mixed fresh + recycle stream (water 0.04 wt%, methanol <10 ppm) was used at an LHSV of 1.4 h⁻¹, H₂/acetone molar ratio 1.05:1, and WABT 160 °C ± 1 °C.

Campaign	Metal (wt%)	Acid site density (mmol NH₃/g)	Acetone conversion per pass (%)	MIBK selectivity (mol%)	MIBC (mol%)	DIBK (mol%)	ICP Pd particle size (nm) post-run	Time-on-stream selectivity half-life (h)
C1	0.3% Pd	0.38	35.8	98.9	0.6	0.3	2.2	8,200
C2	0.5% Pd	0.38	41.2	97.4	1.8	0.5	2.4	7,100
C3	0.3% Pd, 0.05% Pt	0.35	38.9	99.2	0.3	0.3	1.9	8,900
C4	0.3% Pd	0.25 (dealuminated)	28.1	99.0	0.5	0.2	2.0	>10,000
C5	0.3% Pd	0.52 (high Al)	44.7	94.2	1.0	3.9	2.9	4,300
C6 (ref.)	0.5% Pd/Al₂O₃	<0.05	17.6	92.8	4.9	0.2	4.8	2,800

The data demonstrate that the MIBK selectivity target of 99% is a delicate function of both the metal function (hydrogenation activity must be moderated to avoid MIBC formation) and the acid function (sufficient to drive condensation and dehydration but not so high as to promote DIBK). Campaign C3 with 0.3% Pd, 0.05% Pt alloy combined with acid site density 0.35 mmol/g achieves 99.2% MIBK selectivity at a commercially attractive per-pass conversion of 38.9%, while the Pt inclusion reduces Pd sintering as revealed by the smallest post-run particle size of 1.9 nm. Campaign C4 illustrates that sacrificing acid site density to 0.25 mmol/g preserves selectivity but drops conversion below 30%, requiring larger reactor volume and increased recycle rates. The commercial Pd/Al₂O₃ reference (C6) fails utterly on both conversion and selectivity because the absence of acid sites precludes the condensation pathway entirely, resulting in direct acetone hydrogenation to IPA which follows a radical-based route yielding high MIBC upon subsequent condensation under metallic influence. This table, therefore, codifies the necessary property envelope for the catalyst to function in a fixed-bed reactor paired with an acetone recycle loop, and serves as a selection guide for toll manufacturers considering a catalyst change-out under a technology license that specifies metal-acid balance with a control range of 0.33–0.38 mmol/g for acid sites and Pd crystallite size of ≤2.5 nm.

The final compliance checkpoint table correlates the product MIBK quality obtained with the 99% selectivity catalyst against statutory and industry standards, ensuring the material is fit for use in coatings, adhesives, and chemical intermediates without further purification beyond a simple flash to remove residual lights. All values are from a 60-day composite sample collected at the distillation column product rundown, tested by an ISO/IEC 17025:2017 accredited laboratory.

Property	Test Method	Required Limit	Measured Value	Unit
MIBK purity	ASTM D1153-12	≥99.0	99.67	wt%
Water	ASTM D1364-02	≤0.05	0.032	wt%
MIBC	ASTM D1153-12 (GC)	≤0.15	0.09	wt%
DIBK	ASTM D1153-12 (Appendix)	≤0.3	0.14	wt%
Acidity (as acetic acid)	ASTM D1613-17	≤0.002	0.0011	wt%
Colour (Pt-Co)	ASTM D1209-05	≤10	6	APH A
Non-volatile residue	ASTM D1353-13	≤0.001	0.0006	wt%
Distillation range (initial to dry point)	ASTM D1078-11	114–118	115.2–116.7	°C
Methanol	ASTM D1153-12 (specific impurity)	≤20	<5	ppm
Particle count (>10 μm)	ISO 4406:2021	≤ 12/10/8	11/9/7	code
Food contact compliance (indirect)	FDA 21 CFR 175.300	Not to exceed migration limits	Passes	—

Note that the MIBC level of 0.09 wt% is comfortably below the 0.15 wt% ceiling for urethane-grade MIBK, ensuring no interference with isocyanate reactivity in coating formulations, as demonstrated by gel time measurements per ASTM D2471-99 on a common 2K polyurethane clearcoat where a 0.10 wt% MIBC addition extended pot life by only 4 minutes versus 19 minutes for 0.50 wt% MIBC. The suitability of MIBK from this catalyzed route for adhesives under FDA 21 CFR 175.105 is also confirmed, provided the film thickness does not exceed 0.05 mm and the residual solvent level after curing is below 100 ppm, as per the standard’s extractives protocol. The acetone recycle loop’s contribution to this purity profile is direct: without the <0.05 wt% moisture control, the water content of the final MIBK would violate the ASTM D1153 specification, and the elevated MIBC from catalyst deactivation would push the product out of the high-value electronic cleaning solvent tier, reducing the sales price by approximately €180/tonne based on market assessments from the ICIS pricing report Q2 2024.