Equilibrium shift, poisoning prevention, and selectivity enhancement in catalysis via dehydration of polymeric membranes

Selective removal of reaction byproduct water can shift reaction equilibria toward products (Le Chatelier’s principle), prevent catalyst poisoning, and suppress undesirable side reactions. Industrially relevant reactions such as RWGS, methane combustion, and Fischer–Tropsch synthesis are hindered by H2O accumulation. Inorganic membranes (zeolites, silica, metal oxides) lose H2O permselectivity at high temperatures due to diminished hydrophilicity and pore-blocking effects, while MOF-based membranes can suffer from limited thermal/chemical stability and processability. The research question is whether a robust, high-temperature polymeric membrane can selectively and continuously remove H2O to enhance conversion, stability, and selectivity across diverse catalytic systems. This study develops and tests a thermally rearranged poly-benzoxazole (TR-PBO) hollow fiber membrane designed for high H2O permselectivity and stability up to 440 °C, and evaluates its impact on equilibrium shift, poisoning prevention, and side reaction suppression in membrane reactors.

Prior work includes sorbent-assisted reactors and additional reactions that consume water, which require regeneration or add complexity. Inorganic hydrophilic membranes achieve H2O selectivity via pore blocking by condensed water, but this selectivity collapses at elevated temperatures. MOF membranes often face limited thermal/chemical stability and processing challenges for reactor integration. Recently, modified polyimide membranes demonstrated moderate durability up to 300 °C with retained H2O selectivity and hollow fiber processability. However, there remains a need for polymeric membranes with higher thermal limits and sustained water permselectivity. The present TR-PBO approach builds on thermally rearranged polymers known for rigidity and stability, aiming to surpass 300 °C operational limits while retaining high H2O/gas selectivity.

Membrane fabrication: Hydroxyl polyimide (HPI) hollow fibers were prepared via dry-jet/wet-quench spinning (home-built setup). Heat treatment was optimized from TGA of HPI. Fibers were first treated at 300 °C for 1 h (to remove residual solvent and complete imidization), then heated to 425 °C at 5 °C/min, dwelled 30 min under N2, and cooled. Thermal rearrangement of HPI to PBO generates CO2, producing a bimodal microcavity structure (d_small ≈ 0.3 nm; d_large ≈ 0.8 nm), enhancing H2O selectivity via size exclusion and diffusion. TR-PBO fiber dimensions: outer diameter ~487.2 µm; wall thickness ~48.4 µm.

Module assembly: A 1/2-inch stainless steel housing with two three-way connectors contained a bundle of 40 TR-PBO fibers (effective membrane area ~210 cm2). Fiber tips were presealed to prevent epoxy intrusion; epoxy potting cured 12 h.

Characterization: ATR-IR confirmed rearrangement (disappearance of O–H stretch of HPI at ~2971 cm−1; appearance of C=N stretch of benzoxazole at ~1063 cm−1). In situ XRD of hollow fibers showed interchain spacing peak at 2θ ~15° unchanged up to 400 °C, indicating rigid, thermally stable structure. DSC showed no Tg up to 400 °C. TGA indicated stability up to ~500 °C under N2 and air; 400 °C isothermal results corroborated high thermal stability. Gas permeation (250–400 °C) measured permeance of H2O, H2, CO2, O2, CO, CH4; permeance increased with temperature and inversely with kinetic diameter; H2O/gas permselectivity increased with T. Mixed-gas permeation showed some deviation from single-gas permselectivity due to competitive sorption/diffusion.

Reactor studies: Membrane reactors combined TR-PBO hollow fibers with packed-bed catalysts; sweep gas flowed in bore side, reaction in shell side.

• RWGS (equilibrium shift): 6.4 g CuZn/Al2O3 catalyst; feed and sweep each 24 sccm, H2/CO2=1 (1:1), atmospheric pressure, GHSV 225 mL g_cat−1 h−1; temperature ramped 250→440 °C over ~350 h. Sweep composition matched reaction gas unless otherwise stated.
• Methane combustion (poisoning prevention): 1.0 g Pd/Al2O3 catalyst; 300 °C, 0.1 MPa, 150 h TOS; feed 83.3 sccm with 0.4% CH4, 4% O2 in Ar; sweep N2 at 83.3 sccm in membrane mode.
• Fischer–Tropsch to olefins (side-reaction suppression): 1.0 g K4Fe100Cu6Al16 Fe-based catalyst; 320 °C, 0.1 MPa; feed 30 sccm, 30% CO/60% H2 in Ar balance; sweep identical composition at 30 sccm. GC with internal standards monitored both sides for all cases. Comparative tests were run with and without the membrane under identical conditions. Reactor design/sweep composition sensitivity explored; additional sweep compositions tested in supplementary work.

• TR-PBO membrane exhibits high thermal stability and rigidity: XRD peak at 2θ~15° unchanged up to 400 °C; no Tg up to 400 °C; TGA shows no degradation up to ~500 °C in N2 or air.
• Bimodal microporosity (≈0.3 and 0.8 nm) from thermal rearrangement yields high H2O permselectivity that increases with temperature (250–400 °C). Among reported polymeric and inorganic membranes, TR-PBO hollow fibers show the highest H2O permselectivity at 250–400 °C.
• Gas transport governed by diffusion; permeance increases with temperature; mixed-gas tests show maintained H2O selectivity though with competitive sorption/diffusion effects.
• RWGS (CuZn/Al2O3): Membrane operation increased CO2 conversion beyond thermodynamic equilibrium. H2O permeation increased with temperature; ΔQ_CO (extra CO beyond equilibrium) scales proportionally with permeated H2O, confirming equilibrium shift by water removal. Performance stable from 250 to 440 °C over ~350 h. Without appropriate sweep composition, reactant crossover can offset benefits; using sweep matching reactant composition avoids loss and enables continuous water removal.
• Methane combustion (Pd/Al2O3, 300 °C): Without membrane, CH4 conversion declined from ~95% to ~80% over 150 h due to H2O-induced PdO reduction/poisoning. With TR-PBO membrane, conversion remained ~96% over 150 h; ~40% of produced water was removed. XPS: membrane-run catalyst retained Pd oxide states similar to fresh; non-membrane catalyst showed metallic Pd indicative of irreversible PdO reduction.
• Fischer–Tropsch to olefins (Fe-based, 320 °C): In situ H2O removal increased CO conversion and hydrocarbon selectivity; CO2 selectivity from WGS side reaction dropped from 47.9% (no membrane) to 15.1% (with membrane). Olefin-to-paraffin ratios for C2–C4 increased with the membrane due to reduced H2 formation from WGS, enhancing olefin selectivity.
• Process intensification: For commercial-scale CO production (supplementary), membrane reactor volume is lower than that of a packed-bed reactor at the same feed rate.

The study demonstrates that a high-temperature-stable polymeric membrane (TR-PBO) can selectively remove water in situ, directly addressing three pervasive challenges in catalytic reaction engineering: equilibrium limitations, water-induced catalyst deactivation, and water-driven side reactions. By sustaining high H2O permselectivity up to 400–440 °C, TR-PBO enables equilibrium shifting in RWGS, resulting in conversions beyond the thermodynamic limit. The proportionality between additional CO formation and permeated water quantitatively links separation to conversion enhancement. In methane combustion, selective water removal prevents PdO reduction to Pd0, maintaining active sites and stabilizing performance over long times on stream. In Fischer–Tropsch synthesis, eliminating water suppresses the WGS reaction, decreasing CO2 byproduct and hydrogen co-production that hydrogenate olefins, thereby boosting olefin selectivity. Compared to inorganic membranes, whose hydrophilicity and pore-blocking selectivity diminish at high temperatures, the rigid TR-PBO architecture maintains size-selective pathways for water permeation. Hollow fiber configuration affords high area-to-volume ratios, facilitating scalable process intensification. Reactor-level factors (sweep composition, flow configuration) influence performance through potential reactant crossover, but matching sweep to feed composition mitigates losses and supports continuous water removal, suggesting favorable translation to industrial scales with appropriate reactor design.

This work introduces a thermally rearranged poly-benzoxazole hollow fiber membrane that sustains high H2O permselectivity and thermal stability up to ~440 °C, enabling effective in situ dehydration across diverse high-temperature catalytic reactions. The TR-PBO membrane reactor: (i) shifts RWGS conversion beyond equilibrium by selective H2O removal; (ii) prevents Pd-catalyst poisoning in methane combustion, maintaining ~96% conversion for 150 h; and (iii) suppresses WGS during Fischer–Tropsch synthesis, reducing CO2 selectivity from 47.9% to 15.1% and enhancing light-olefin selectivity. The membrane maintains structural integrity and performance over extended operation, and the hollow fiber format supports high area per volume and reduced reactor size. Future work should optimize membrane area, flow configuration, sweep composition and rate, and incorporate gas–liquid separation and sweep recirculation to maximize efficiency. Scale-up studies to minimize dead volumes and crossover, along with long-duration mixed-gas stability and fouling assessments, will further advance industrial deployment.

• Sweep gas composition and reactor design can strongly affect performance. Using an unmatched sweep (e.g., pure N2 in RWGS) can reduce conversion due to reactant crossover above the catalyst bed, especially in lab-scale modules with non-negligible dead volumes. This is expected to diminish at scale but needs careful design.
• Mixed-gas permeation exhibits nonideal behavior due to competitive sorption/diffusion, leading to differences from single-gas selectivities; process conditions must be tuned accordingly.
• The study primarily demonstrates performance under specific lab-scale flow rates and catalyst bed configurations; broader operating envelopes (pressure, varying GHSV, long-term multi-cycle thermal histories) require further validation.
• Quantitative permeance/selectivity values under industrially relevant mixed feeds are presented qualitatively in the main text; comprehensive long-term fouling and chemical stability tests in complex feeds (e.g., impurities) remain future work.