Nonequilibrium Sorption of Volatile Petroleum Hydrocarbons by Surfactant-Modified Zeolite

We determined the effect of concentration and competing solutes on nonequilibrium sorption of benzene, toluene, ethylbenzene, and xylenes (BTEX) by surfactant-modified zeolite (SMZ). SMZ has been shown effective in removing BTEX from produced water generated in petroleum operations. In order to optimize the use of SMZ for removal of petroleum hydrocarbons from water, the kinetics of BTEX sorption must be understood.

BTEX sorption kinetics were determined by fitting a two-site model to batch sorption data. The kinetic sorption parameters derived from the batch data were then used to parameterize a nonequilibrium transport model to predict BTEX sorption through packed columns of SMZ. The predicted breakthrough curves were compared to the measured curves.

BTEX sorption by SMZ consisted of a near-instantaneous step followed by a rate-limited step. The equilibrium sorption coefficient (K_d) was not affected by the initial BTEX concentration nor the presence of competing solutes, and was strongly correlated with each compound's octanol-water partition coefficient (K_ow). However, both the fraction of instantaneous sorption sites (F) and the sorption rate coefficient (k₂) decreased with increasing BTEX concentration or the presence of competing solutes.

A conceptual model relating the fraction of instantaneous sorption sites to the fraction of zeolite covered by a surfactant bilayer was confirmed by performing sorption experiments with SMZ treated to only monolayer surfactant coverage. The fraction of instantaneous sorption sites was 3-10 times greater for sorption by monolayer compared to bilayer SMZ. The K_ds for sorption of BTEX by monolayer vs. bilayer SMZ were virtually identical, in agreement with results for other nonpolar solutes.

The fraction of instantaneous sorption sites and the kinetic rate coefficients derived from batch experiments agreed with those observed in dynamic column experiments. The equilibrium sorption constant, however, was significantly higher in the column experiments compared to the batch experiments.