Author ORCID Identifier

https://orcid.org/0000-0003-2292-9470

Date Available

2-27-2021

Year of Publication

2020

Degree Name

Doctor of Philosophy (PhD)

Document Type

Doctoral Dissertation

College

Engineering

Department/School/Program

Chemical and Materials Engineering

First Advisor

Dr. Isabel C. Escobar

Abstract

Nonsolvent induced phase separation (NIPS) has been widely used to fabricate polymeric membranes. In NIPS, a polymer is dissolved in a solvent to form a dope solution, which is then cast on a substrate and immersed in a nonsolvent bath, where phase inversion occurs. Petroleum-derived organic solvents, such as N-Methyl-2-Pyrrolidone (NMP) and Dimethylacetamide (DMAc), have been traditionally used to fabricate polymeric membranes via NIPS. However, these solvents may have negative impacts on environmental and human health; therefore, using greener and less toxic solvents, preferably derived from biomass, is of great interest to make membrane fabrication sustainable. In this dissertation, two low-hazard solvents, Methyl 5-(Dimethylamino)-2-Methyl-5-Oxopentanoate (Rhodiasolv® PolarClean) and Gamma-Valerolactone (GVL), were investigated as sole- and as cosolvents to cast Polysulfone (PSf) membranes via NIPS.

In the first part of this project, Rhodiasolv PolarClean was studied. PolarClean is a bio-derived, biodegradable, nonflammable and nonvolatile solvent. From cloud point curves, PolarClean shows potential to be a solvent for polysulfone. Membranes prepared with PolarClean were investigated in terms of their morphology, porosity, water permeability and protein rejection, and were compared to membranes prepared with traditional solvents. The pores of polysulfone/PolarClean membranes were sponge-like, and the membranes displayed higher water flux values along with slightly higher solute rejection. On the other hand, PSf/DMAc membrane pores were finger-like with lower water flux and slightly lower solute rejection when compared to PSf/PolarClean membranes. Upon reverse-flow filtration to simulate membrane cleaning, it was observed that the pores of PSf/PolarClean membranes collapsed. To address this issue, GVL was investigated as a sole solvent and a cosolvent with PolarClean to fabricate PSf membranes. Membranes prepared using GVL as a sole solvent were observed to be gelatinous, hence not ideal for filtration. On the other hand, when GVL and PolarClean were used as cosolvents, viable membranes were cast with surface charge and hydrophicility not being significantly different from membranes made using PolarClean alone. Furthermore, the average pore size of membranes decreased as the weight percent of GVL in dope solutions increased. Therefore, the use of PolarClean/GVL as cosolvents shows promise for the fabrication of PSf membranes. With respect to operation, membranes cast from dope solutions containing equal amounts of PolarClean and GVL displayed the most similar flux curves and solute rejection to those made using the traditional solvent tested.

Once it was determined that membranes made using PolarClean and GVL as cosolvents were viable and showed similar morphological and operational characteristics to those made using DMAc, the use of PolarClean/GVL cosolvents was then researched at the production scale. In the last portion of this study, a slot die-roll to roll (R2R) system was used to fabricate polysulfone (PSf) ultrafiltration membranes using low-hazard solvents individually and as cosolvents at a production scale. Production-scale membranes were compared structurally, morphologically and operationally to laboratory-scale membranes made using a doctor’s blade. The chemical structure of membranes was not affected by the use of different solvents nor by the differences in scale. On the other hand, cross-sectional images showed that the structures of the membranes were different most likely due to differences in diffusion rates between the different solvents/cosolvents into the nonsolvent, water. Furthermore, it was observed that slot die and doctor’s blade casting methods produced membranes with different roughness values likely due to evaporation time differences between the methods. While to protein filtration, all membranes displayed similar operational parameters, i.e., flux decline, permeability and recovery. Overall, this dissertation shows that membranes fabricated using greener/less toxic solvent mixtures are comparable to membranes cast using petroleum-derived solvents, and are scalable using slot die-R2R.

Digital Object Identifier (DOI)

https://doi.org/10.13023/etd.2020.367

Funding Information

This research was supported by the National Science Foundation under Cooperative Agreement (grant number 1355438) and by the NSF Kentucky Established Program to Stimulate Competitive Research (EPSCoR) Program, 2015-2019.

Share

COinS