The separation of phenolic compounds from neutral oils and nitrogen bases

[摘要] ENGLISH ABSTRACT: Coal pyrolysis liquors are a major source of phenolic compounds. The separation of thephenolic compounds from the neutral oils and nitrogen bases also present in thepyrolysis liquors is difficult due to low relative volatilities and the formation ofazeotropes. The desired phenolic recovery and phenolic product purity of 99.5 % cantherefore not be achieved by means of conventional separation processes.Alternative processes such as liquid-liquid extraction with various low-boiling solvents,mixtures of high-boiling solvents and extractive distillation have been investigated.Disadvantages of these processes include the high solvent ratios required, low recoveryof the higher substituted phenolic compounds, inability to treat a wide-boiling feedstockin one process step and complex post-purification of the phenolic product. A solventsystem consisting of a selective solvent, water as a co-solvent, and hexane as acountersolvent, is proposed.An industrial heavy naphtha stream was analysed and the most prevalent phenoliccompounds, neutral oils and nitrogen bases identified. Three synthetic feed streamswere compiled to represent the industrial stream, namely:1. phenol + benzonitrile + aniline + mesitylene + 5-et-2-me-pyridine2. m-cresol + o-tolunitrile + o-toluidine + pseudocumene + undecane + indene3. 2,4-xylenol + 3,5-xylenol + 3,4-xylenol + indane + dodecane + naphthaleneThe stream containing phenol was used as a basis for solvent selection, with emphasison the separation of phenol from benzonitrile. A variety of molecules containinghydroxyl and ether functional groups were identified as potential solvents by means ofcomputer-aided molecular design using a genetic algorithm. Of the commerciallyavailable solvents tested on batch extraction scale, triethylene glycol achieved thehighest phenol-benzonitrile, phenol-aniline and phenol-5-et-2-me-pyridine separationfactors as well as the highest phenol recovery.It was concluded from the solvent selection process that effective solvents for theproblem under investigation were those containing hydroxyl groups positioned on themolecule backbone in such a way as to facilitate hydrogen bonding with more than onephenolic molecule at a time. Two commercially unavailable solvents, 1,3-(ethoxy-2-hydroxy)-propane-2-01 and 1,3-(diethoxy-4-hydroxy)-propane-2-01 were thereforesynthesised from ethylene glycol and diethylene glycol respectively. The molecularstructures of these two solvents are analogous to that of triethylene glycol, and containan additional hydroxyl group. The performance of the synthesised solvents wasevaluated and compared to that of triethylene glycol on the basis of m-cresol-otolunitrile,2,4-xylenol - o-tolunitrile, and 2,4-xylenol - o-toluidine separation factors andphenolic recoveries achieved by means of batch extraction tests. 1,3-(Diethoxy-4-hydroxy)-propane-2-01 yielded higher phenolic recoveries, but lower separation factorsthan did triethylene glycol. Triethylene glycol was therefore selected for further processdevelopment as it is commercially available.A series of batch extractions were carried out on each of the synthetic feed streamsusing the proposed solvent system. For phenol and m-cresol, recoveries in excess of99% were obtained in a single stage. Recoveries in excess of 98% were obtained forthe xylenol isomers. It was found that the recoveries of the xylenol isomers were moresensitive to changes in the solvent ratios.The separation of phenolic compounds from paraffins, naphthalene, indene, indane andthe alkyl-substituted benzenes was trivial using the proposed solvent system. Highlysatisfactory separation of the phenolic compounds from pyridines and aromatic nitrileswas achieved. The separation of phenol from aniline, although satisfactory, was not asgood.The optimum solvent to feed, water to solvent and hexane to feed ratios were identifiedas being 3.0, 5.0 and 0.25 respectively.Binary interaction parameters for the NRTL equation were obtained by regression of theequilibrium data from the batch extraction tests. The NRTL model fitted the equilibriumdata satisfactorily.The proposed solvent system was tested on pilot plant scale. The performance of theextraction column was optimised using a synthetic feed stream consisting of m-cresol,p-cresol, aniline and o-tolunitrile. The optimum solvent ratios and operating parameterswere then implemented in further tests on an industrial heavy naphtha stream. Aphenolic product purity of 99.75% was achieved for this stream. The correspondingphenolic recovery was in excess of 91 %.The proposed separation process, including solvent recovery was simulated using theNRTL model with the experimentally determined interaction parameters. A singlestream consisting of all the components used in the batch extraction tests was specifiedas the feed stream to the simulated process. A final simulated phenolic product purityof 99.5% and recovery in excess of 94% was obtained after solvent recovery. Theoptimum solvent to feed, hexane to feed and water to solvent ratios were determined asbeing 3.0, 5.0 and 0.25 in both the pilot plant tests and the simulated extraction process.It can be concluded that the proposed separation process is successful in recoveringhigh purity phenolic compounds from tar liquors. Further development of the processhas commenced in industry.

[发布机构] Stellenbosch University