Antiplasticization and plasticization of Matrimid® asymmetric hollow fiber membranes—Part A. Experimental

https://doi.org/10.1016/j.memsci.2009.12.033Get rights and content

Abstract

The complex effects of highly sorbing feed gas contaminants such as toluene and n-heptane on performance of both annealed and non-annealed Matrimid® asymmetric fibers relevant to CO2/CH4 separation are reported. Membrane performance was quantified both during contaminant exposure and after removal of the contaminant from the feed stream. Exposure to either toluene or n-heptane during permeation reduces carbon dioxide permeance and the carbon dioxide/methane selectivity in non-annealed fibers. After exchange with a contaminant-free feed containing only CO2 and CH4 mixed gas, the carbon dioxide permeance and carbon dioxide/methane selectivity were affected, indicating a glassy state conditioning effect due to the prior contaminant exposure. Interestingly, the conditioning effect after simultaneous exposure to toluene and n-heptane (284 ppm toluene and 504 ppm n-heptane) was less than the conditioning observed for either toluene (293 ppm) or n-heptane (505 ppm) individually. Sub-Tg annealing reduced carbon dioxide permeance during actual contaminant exposure more severely than in non-annealed fibers. On the other hand, except for exposure to the highest n-heptane contaminant feed (2003 ppm), annealing significantly reduced the post-exposure conditioning observed in carbon dioxide permeance and carbon dioxide/methane selectivity. It appears that annealing allows the consolidation of segmental packing which stabilized the glassy matrix against swelling. At sufficiently high activities of even a relatively non-interacting penetrant like n-heptane, the annealing-induced stabilization can be reversed.

Introduction

Membrane-based separations of gaseous feed streams have been performed industrially for the past 25 years [1] and are expected to grow in the future. Especially, membrane-based CO2/CH4 separation has been explored by many researchers [2], [3], [4], [5]. It is well known that CO2 behaves as a plasticizer in CO2/CH4 separations at elevated pressures resulting in loss in selectivity. Bos et al. [6] demonstrated that heat treatment suppresses the undesired acceleration in CH4 permeability due to CO2-induced plasticization for the mixed CO2/CH4 permeation in polyimide Matrimid® 5218. In addition to CO2-induced plasticization, Visser et al. [7] showed that a subtle balance exists between competitive sorption and plasticization effects for a binary mixture of CO2/CH4. Separation of feed streams containing condensable components presents a large challenge for current generation membrane materials, which often experience substantial performance reductions, even at relatively low concentrations. Few studies of membrane performance in such situations are available [8], [9], [10], [11], and the mechanism of membrane performance degradation in the presence of aggressive feed streams, which is termed antiplasticization, is not completely understood.

Our previous work [12] explored the sorption of highly sorbing vapors such as toluene and n-heptane as model contaminants found in natural gas in the absence of CO2 and CH4. This investigation clarifies the effects of such contaminants on the permeation performance of hollow fiber gas separation membranes used for the removal of CO2 from CH4 in natural gases. The CO2 and CH4 permeances of both non-annealed and annealed Matrimid® hollow fiber membranes will be quantified during exposure to feeds containing toluene or n-heptane or their mixtures to assess effects on separation performance. Also, CO2 and CH4 permeances following removal of the contaminants from the feed stream were studied to evaluate longer terms conditioning effects from the exposures.

Section snippets

Materials

The membranes studied in this work were all formed from Matrimid® 5218, a wholly amorphous polyimide that is manufactured and marketed by Huntsman LLC. The backbone repeat structure of Matrimid® is shown in Fig. 1 and Table 1 [13] provides material and gas transport properties for Matrimid®. Matrimid® hollow fibers were spun using a dry-jet/wet-quench process [13], [14], [15], [16] using a polymer dope consisting of Matrimid®, N-methyl pyrrolidinone acting as a solvent, with ethanol acting as a

Gas permeation with low sorbing feeds

Nitrogen, oxygen, and helium permeances were determined before each membrane was exposed to conditioning agent. Measurement of the permeance for each of these inert gases was primarily made to ensure that the membrane module was initially defect-free (i.e. an oxygen/nitrogen selectivity >90% of the value for bulk Matrimid® given in Table 1). However, tracking of the gas permeances of these probes, prior to conditioning agent exposure, over the course of this experimental work also allows for

Conclusions

A standardized permeation protocol was used to investigate the effects of highly sorbing feed gas contaminants on membrane performance. The effect on membrane performance in non-annealed and annealed Matrimid® asymmetric hollow fiber during exposure to feed gas contamination such as toluene and n-heptane, as well as the sustained effect on its performance after contaminant removal was investigated in this work. Substantial reductions in the carbon dioxide permeance and the carbon

Acknowledgements

The authors would like to acknowledge financial support from The Coca Cola Company, Air Liqude, and Award no. KUS-I1-011-21 made by King Abdullah University of Science and Technology (KAUST).

References (30)

Cited by (45)

  • Organic solvent reverse osmosis using CuAAC-crosslinked molecularly-mixed composite membranes

    2021, Journal of Membrane Science
    Citation Excerpt :

    This is an estimate of the vol% in the final topcoat membrane layer calculated by converting from the mass% of polymer and cage used in the topcoat dope preparation. A polymer density of 1.2 g/cm3 [37] as well as the molecular volume and molecular weight of CC3 (319 Å3 and 1117 g/mol, respectively) [17] are used. While using the molecular dimensions of CC3 will induce inherent error in our calculation, we emphasize that the reported vol% is only an estimate intended to give the reader a relative comparison between membranes of different cage loadings.

  • Pure and mixed fluid sorption and transport in Celazole<sup>®</sup> polybenzimidazole: Effect of plasticization

    2019, Journal of Membrane Science
    Citation Excerpt :

    However, methanol mobility in PIM-1 decreases with increasing concentration, which is indicative of penetrant clustering or blocking [22,62,63]. Blocking refers to the progressive filling of the free volume available for penetrant transport [22,62,63]. Therefore, as concluded by Lively and co-workers, methanol does not plasticize PIM-1 [11].

View all citing articles on Scopus
View full text