Enhancing gas permeability in mixed matrix membranes through tuning the nanoparticle properties

Highlights

• Dense mixed matrix membranes fabricated using Matrimid and a range of nanoparticles.

• Images of membrane morphology strongly correlated with nanoparticle hydrophobicity.

• Permeability not correlated with morphology changes.

• Permeability readily described by a simple free volume relationship.

Abstract

Mixed matrix membranes containing a variety of nanoparticles were fabricated by the solvent casting method using a commercial aromatic polyimide as the base polymer. All gas permeabilities increased with increasing particle loading with no reduction in the selectivity, reflecting adequate polymer/particle compatibility. Importantly, under such conditions the permeability enhancement depended only upon the pore volume within the particle, regardless of the particle chemistry or the morphology of the membrane structure. Remarkably, despite a range of membrane chemistries and filler loadings, this permeability enhancement could be readily described with a simple free volume relationship. The results suggest that nanoparticle porosity should be the focus of research into mixed matrix membrane structures. These results are likely to apply in all diffusivity dominated systems where the particle pore size is significantly larger than the penetrant size, as is the case with glassy polymers and many inorganic additives at low pressures.

Introduction

Mixed matrix membranes (MMMs) have been considered since the 1970s as an effective approach to improve the performance of both gas separation and water treatment polymeric membrane systems [1], [2], [3]. In this approach, nano-particulate fillers are added into the polymer phase, generally causing an enhancement in permeability and/or change in penetrant selectivity. A large number of MMMs using zeolites [4], silicas [5], [6], [7], carbons [8], [9], [10], metal–organic frameworks (MOFs) [11], zeolitic imidazolate frameworks (ZIFs) [12], [13], [14] and porous organic polymers (POPs) [15] have been shown to be promising candidates for gas separation applications. However, these MMMs are normally tested in isolation and rarely compared to each other. In the present work, a comparative study of gas separation using MMMs containing a variety of different nanoparticles has enabled important new insights into the factors that impact membrane performance.

The particles need to be well dispersed to guarantee the separation performance. One of the critical issues for MMMs is the presence of interfacial defects around the particles, which are caused by particle aggregation and poor particle–polymer interaction [16]. Such defects in the membrane can significantly reduce the gas selectivity and are also often related to increases in permeability. Therefore, MMM research has focused not only on the improvement of gas separation performance but also the preparation protocols and particle modifications to prevent or control these interfacial defects [17].

A number of workers have used computational approaches to evaluate the effect of nanoparticle properties [18], [19]. Their performance is also often predicted through mathematical models from the relatively simple Maxwell model [20] to others significantly more complex [21]. The Maxwell model [20] is considered appropriate for low particle concentrations

$$P_{eff}=P_c\left[\frac{P_d+2P_c-2{\varphi }_d\left(P_c-P_d\right)}{P_d+2P_c+{\varphi }_d\left(P_c-P_d\right)}\right]$$

where P_eff is the effective permeability of the mixed matrix membrane, ϕ is the volume fraction and P the permeability of the dispersed (d) and continuous (c) phases respectively. The Bruggeman [22] (Eq. 2) and Lewis–Nielsen [23] (Eq. 3) models were introduced to model systems with higher particle loadings

$$\frac{P_{eff}}{P_c}=\frac{1}{{\left(1-{\varphi }_d\right)}^3}{\left[\frac{\left(P_{eff}/P_c\right)-\alpha }{1-\alpha }\right]}^3$$

$$\frac{P_{eff}}{P_c}=\frac{1+2{\varphi }_d\left[\left(\alpha -1\right)/\left(\alpha +2\right)\right]}{1-{\varphi }_d\psi \left[\left(\alpha -1\right)/\left(\alpha +2\right)\right]}$$

where α=P_d/P_c and ψ=1+[(1−ϕ_m)/ϕ_m²]ϕ_d, in which ϕ_m is the maximum achievable volume fraction of the particles, which is affected by factors such as particle shape, size, size distribution, and agglomeration. Mahajan and Koros also reported a modified Maxwell model that accounted for interfacial defects (i) [16]

$$P=P_c\left[\frac{P_{eff}+2P_c-2\left({\varphi }_d+{\varphi }_i\right)\left(P_c-P_{eff}\right)}{P_{eff}+2P_c+\left({\varphi }_d+{\varphi }_i\right)\left(P_c-P_{eff}\right)}\right]$$

However, an experimental evaluation of the impact of particle properties across a wide range of different nanoparticles has never been reported. In this work, we report for the first time the gas permeation properties of a series of MMMs in a glassy polymer substrate to determine the dominant particle properties and explain the permeation using a simple free volume-based model.