Reports: UNI656896-UNI6: Using Ionic Liquid Mixtures for the Extraction of Organosulfur Compounds from Petroleum Streams
Andrew S. Paluch, PhD, Miami University
This project is focused on the use of ionic liquid and ionic liquid blends as replacement solvents for extraction processes using a combination of molecular simulation and thermodynamic modeling. There are a large number of ionic liquids that have previously been synthesized, and an even larger number of potential ionic liquids and ionic liquid mixtures. The use of molecular simulation alone would be impracticable. A central feature of this work is on the development of thermodynamic models to guide molecular simulations. Specifically, we are working to extend the model MOSCED to ionic liquids and their mixtures. The use of MOSCED is advantageous because it can not only be used to predict limiting activity coefficients, but it can additionally give insight into the underlying phase behavior. The work of the past year has focused on three areas involving MOSCED.
MOSCED takes the general form:
where
is the limiting activity coefficient of the solute (2) in a solvent
(1), and the first three terms result from comparing solute and
solvent intermolecular interactions. A limitation of MOSCED is it can
only predict limiting activity coefficients. In order to predict
composition dependent activity coefficients which may be necessary
for phase-equilibria calculations, MOSCED must be used to
parameterize an excess Gibbs free energy model. In this study, we
assessed the ability of MOSCED to predict composition dependent
activity coefficients by predicting binary isobaric azeotropic
vapor-liquid equilibrium and binary liquid-liquid equilibrium (or
mutual solubilities). The excellent performance achieved for these
highly non-ideal systems is encouraging as we move on to complex
ionic liquids.
As written in equation (1), MOSCED is applicable for a solute in a
solvent. As we ultimately desire to model solutes in ionic liquid
mixtures, we desire to extend MOSCED to multicomponent solvents. One
way to accomplish this is to use MOSCED to predict limiting activity
activity coefficients for each binary pair making up the solution,
and using this to parameterize an excess Gibbs free energy model.
Unfortunately, we lose the physical insight of MOSCED when looking at
solvent mixtures. We found that we could extend MOSCED to a solute in
multicomponent solvent by using volume fraction average solvent
MOSCED parameters. This allowed us to compute the limiting activity
coefficient of the solute as a function of the solvent composition.
The methodology was verified by using MOSCED to predict solubility in
binary solvents. As a consequence of our mean solvent parameters, we
were able to develop MOSCED based rules for how to select solvents to
achieve solubility enhancement, which results when a solute is more
comfortable in a solvent mixture as compared to any of the pure
solvents.
A limitation of MOSCED is that parameters need to be available for
a solute/solvent pair in order to make predictions. Currently, MOSCED
is parameterized for 130 organic solvents and water. We therefore
developed a group contribution method to predict MOSCED parameters.
As we seek to study solutes in ionic liquid and mixtures of ionic
liquids with conventional solvents, this will allow us to obtain
accurate parameters for missing species.
The proposed method works well to predict parameters of organic
solvents and relatively small solid solutes. However, parameters are
missing for large, multi-functional solid solutes. We have therefore
investigated the use of molecular simulation and electronic structure
calculations in the SMD and SM8 continuum solvent models to predict
MOSCED parameters. Specifically, both molecular simulation and
electronic structure calculations may be used to generate a set of
reference solvation free energies, which may then be used to regress
MOSCED parameters. As compared to using molecular simulation or
electronic structure calculations alone, by parameterizing MOSCED,
predictions may be made in additional solvents, in solvent mixtures,
and at additional temperatures. Between this and our group
contribution method, we are confident that we can obtain accurate
parameters for any solute and organic solvent that we may encounter.
After the first year, we are well poised to accomplish the goals
of the proposal. In the second year we will move on to parameterize
MOSCED for ionic liquids, and to use this to identify systems to
study using detailed molecular simulations.
Three undergraduate students were primarily supported by this PRF
grant in the first 2016-2017 grant year, which encompassed Winter and
Summer 2017, and one additional undergraduate student worked on this
project supported by the Undergraduate Summer Scholars Program from
the Miami University Office of Undergraduate Research with additional
support from the Miami University College of Engineering and
Computing. Students were paid during break times (Winter and Summer),
and earned academic credit during the academic year for their work.
An additional two undergraduates worked on the project solely during
the academic year for credit. The experience afforded the students
the opportunity to use state-of-the-art computational methods and
additionally to use a physical based model for process design
applications. Additionally, the students have had the opportunity to
write-up research articles with the guidance of the PI, and to
additionally present their work as poster and oral presentations at
professional meetings. Based on his work, one of these students
placed second in the paper competition at the AIChE North Central
Student Regional Conference. Of the undergraduate students who worked
on the project, three have graduated. Two of these students will
continue to work on the project in the combined BS/MS program at
Miami, where they will be supported next year by the grant as MS
students. Without financial support to stay at Miami, graduate
studies may not have been an option to them.
The results of this project have been disseminated through
presentations at regional and national meetings. Both the PI and
students have presented at meetings including the Ohio Supercomputing
Center Statewide User Group bi-annual conference, the Midwest
Thermodynamics and Statistical Mechanics Conference, the American
Institute of Chemical Engineers (AIChE) Annual Meeting, and the AIChE
North Central Student Regional Conference.