Dimerisation of isobutene on acidic ion-exchange resins

Dimerisation of isobutene produces diisobutenes that can be hydrogenated to isooctane (2,2,4-trimethyl pentane). Isooctane can be used as a high octane gasoline component. The aim of this work was to study the selective production of diisobutenes through the dimerisation of isobutene on ion-exchange resin catalysts and to construct kinetic models for the reactions in the system for reactor design purposes. High selectivities for diisobutenes were obtained in the presence of polar components such as methanol and tert-butyl alcohol (TBA). The effects of these polar components were found to be basically the same, i.e. the selectivity for diisobutenes increases with the content of the polar component while the activity of the catalyst decreases. Because TBA does not react with isobutene and because small amounts are sufficient to obtain high selectivities, TBA was concluded to be a good selectivity-enhancing component and was used in further studies of the system. Different ion-exchange resin structures in the dimerisation of isobutene were studied and it was observed that good accessibility to the active sites gives high isobutene conversions and medium crosslinking high selectivies for diisobutenes. When linear butenes were added to the feed of the system, they were found to react with isobutene to codimers especially at low TBA contents and at high temperatures. In these experiments some 2-butanol was formed from the linear butenes and the water from TBA dehydration. In the kinetic modelling, it was observed that a basic Langmuir–Hinshelwood-type kinetic model describes the diisobutene formation rates well. Similar model was constructed for the codimerisation of isobutene and 2-butenes. The trimerisation of isobutene was found to proceed via diisobutenes and an additional free catalyst active site participates in the reaction. TBA dehydration was well described by Langmuir–Hinshelwood-type kinetics where the formed isobutene does not adsorb on the catalyst and only one active site is needed. In the simulation of miniplant-scale reactors these models were found to represent the experimental data well and thus they can be used in reactor design.reviewe