Solvent swing adsorption for the separation of acrylonitrile from process water

C. Wegmann

    Research output: ThesisPhd Thesis 1 (Research TU/e / Graduation TU/e)

    1472 Downloads (Pure)


    Adsorption is a commonly applied method when organic components need to be removed from aqueous streams. The recovery of the adsorbed components can however be challenging and therefore in many cases, the sorbent material is regenerated without recovering the adsorbed components (e.g. thermal regeneration of activated carbon). The combination of two or more separation techniques—which results in so called hybrid technologies—offers new possibilities for efficient separation processes which allow not only the removal of organic components from a water stream but also the recovery of the organics. Solvent Swing Adsorption is a hybrid separation technology which is based on the combination of adsorption with extraction and which allows the separation of dissolved organic components from water. This thesis describes the development of the Solvent Swing Adsorption technology for the recovery of dissolved acrylonitrile from an aqueous process stream. Solvent Swing Adsorption is based on the shift in the adsorption equilibrium of a sorbate when the composition of the liquid phase is changed. An organic component which needs to be separated from a water stream is in a first step adsorbed onto a solid sorbent in a fixed bed. In a second step, the component is desorbed from the sorbent material by feeding a solvent to the column which has a high affinity for the adsorbed component. During the desorption step, very high concentrations of the organic component are obtained in the solvent stream at the column outflow. As a last step, the solvent is removed from the fixed bed by flushing the column with water. The liquid mixtures which are produced in the Solvent Swing Adsorption process are treated by use of distillation. For the separation of dissolved acrylonitrile from a process water stream by use of Solvent Swing Adsorption, a suitable sorbent material was selected. Dowex Optipore L-493, Amberlite XAD-1180, activated carbon, zeolite H-BEA-150, and silica gel were preselected based on a literature research. These five adsorbents were evaluated in batch experiments to determine the acrylonitrile loading on the sorbent materials and — by varying the temperature in the batch experiments — the reversibility of the acrylonitrile adsorption. Based on these experiments, Dowex Optipore L-493 was selected as sorbent material due to its high capacity for acrylonitrile, due to the reversibility of the acrylonitrile adsorption, and due to the fast intraparticle mass transfer of acrylonitrile. The selection for an efficient desorption solvent was made based on a combination of computer aided molecular design (CAMD), in which the Hansen model was applied, and batch equilibrium experiments. Acetone, acetaldehyde, methanol, ethanol, and trimethylamine were preselected by use of CAMD and tested by use of equilibrium experiments. Because of its high affinity for acrylonitrile, its miscibility with water, its low boiling point, and its good environmental, health and safety properties, acetone was chosen as a desorption solvent. The acrylonitrile loading on Dowex Optipore as a function of the liquid phase composition was determined experimentally using batch experiments and described using three different models: the Langmuir isotherm, an activity based isotherm, and the model of Minka and Myers. It was found that the acrylonitrile loading is strongly dependent on the acetone concentration in the liquid phase. The acrylonitrile equilibrium loading at a concentration of 10 kg m-3 in water is 0.29 kg kg-1. This value drops to 0.02 kg kg-1 when the solvent is pure acetone. The reasons for this decrease are the low bulk phase acrylonitrile activity when the solvent contains a large fraction of acetone and the competition for adsorption sites between acrylonitrile and acetone. By applying the theory of Minka and Myers, acrylonitrile loadings for concentrations which are much higher than used in the experiments were predicted from the binary isotherms (acrylonitrile adsorption from water, acrylonitrile adsorption from acetone, acetone adsorption from water). The intraparticle mass transfer of acrylonitrile in the Dowex Optipore particles was described by applying the theories of Fick and of Maxwell-Stefan. The intraparticle diffusion coefficients were found by comparing the results of kinetic experiments (zero-length column experiments) with kinetic models. Two models were used which are based on Fick’s diffusion: the homogeneous surface diffusion model and the pore diffusion model. A third diffusion model was based on pore diffusion according to Maxwell-Stefan. Intraparticle pore diffusion coefficients of acrylonitrile in a water solution and in an acetone solution were found which are in the same range as the bulk diffusion coefficient of acrylonitrile in water and in acetone, respectively. For the diffusion of acrylonitrile in acetone, a Maxwell-Stefan diffusion coefficient of 1.2·10-9 m2 s-1 was found. For the diffusion of acetone in water, the Maxwell-Stefan diffusion coefficient is 1.4·10-9 m2 s-1 and for the diffusion of acrylonitrile in water 4.5 · 10-9 m2 s-1. These diffusion coefficients are remarkably large for liquid phase intraparticle diffusion which is advantageous for the Solvent Swing Adsorption process as relatively high superficial velocities can be applied. The findings from the experiments and from the modeling concerning adsorption equilibrium and kinetics were used in the development of a column model. A model was built based on plug-flow with axial dispersion. The liquid film mass transfer coefficients and the intraparticle diffusion coefficients were combined which resulted in overall mass transfer coefficients. The acrylonitrile breakthrough curve, the acrylonitrile peak during the desorption step, and the acetone removal from the column by use of water were determined experimentally. For this purpose, columns with a length of 0.112 m and 0.430 m filled with Dowex Optipore L-493 were used. The liquid was pumped over the columns with velocities of 1.70 · 10-4 m s-1 to 1.89 · 10-4 m s-1. The experimental results of the acrylonitrile adsorption and the acetone removal from the column can be described well with the developed model. The acrylonitrile desorption is described fairly well. Reasons for the difference between model and experiments are mainly the inaccuracies in the description of the acrylonitrile adsorption equilibrium and in the description of the mass transfer. The axial dispersion coefficient has only a small influence on the shape of the concentration curves in the column outflow. Based on the column model, two process schemes for the Solvent Swing Adsorption process were developed. The relatively simple non-recycle process consists of an adsorption step, a desorption step, and an acetone removal step. Two distillation columns are necessary in order to separate the resulting liquid mixture. The recycle process is more complex. A few recycle streams are included in the process, in order to increase the acrylonitrile concentration in the outflow. In this way, the separation goals can be reached by use of only one distillation column. However, the total volume of the adsorption/desorption columns is larger than for the non-recycle mode and the length of one adsorption/desorption cycle is longer. It was found — experimentally and by use of the column model — that the acrylonitrile is desorbed from the solid sorbent efficiently and it appears in the outflow of the column in a narrow and high peak. The acetone removal from the column by use of water is however much slower. An acetone/water mixture is produced with a volume which is larger than the volume of the acetone/acrylonitrile/water mixture which originates from the acrylonitrile desorption. Not only the acetone/acrylonitrile/water mixture needs to be distilled but also the acetone/water mixture which makes the process less energy efficient. Table 1: Comparison of cycle length and distillation costs for the non-recycle mode and the recycle mode. Non-recycle mode Recycle mode Length of adsorption/desorption cycle, h 5.2 15.6 Volume distilled, m3 h-1 0.87 + 2.40 = 3.27 2.60 Cost for distillation, e h-1 36 24 For a column length of 2 m, the cycle length for one adsorption/desorption cycle, the volume which needs to be distilled, and the energy costs for the distillation are given in Tab. 1. For both operating modes, a benefit of 114 e h-1 is generated by the recovery of the acrylonitrile which means that both modes can be economically feasible. The length for one adsorption/desorption cycle is three times shorter for the non-recycle mode than for the recycle mode which makes the non-recycle mode much more flexible. Additionally, for the non-recycle mode the total volume of the Solvent Swing Adsorption columns is smaller than for the recycle mode and only two storage tanks are required whereas six storage tanks are used if the recycle mode is applied. In order to make the Solvent Swing Adsorption process more energy efficient, the use of acetaldehyde as desorption solvent could be taken into consideration. Acetaldehyde has a higher affinity for acrylonitrile than acetone and a lower boiling point. However, it is more flammable, more reactive, and more dangerous to health than acetone
    Original languageEnglish
    QualificationDoctor of Philosophy
    Awarding Institution
    • Chemical Engineering and Chemistry
    • Kerkhof, Piet, Promotor
    • de Haan, André, Promotor
    Award date17 Jan 2013
    Place of PublicationEindhoven
    Print ISBNs978-90-386-3295-7
    Publication statusPublished - 2013

    Fingerprint Dive into the research topics of 'Solvent swing adsorption for the separation of acrylonitrile from process water'. Together they form a unique fingerprint.

    Cite this