The upper bound of thin-film composite (TFC) polyamide membranes for desalination

Highlights

• TFC polyamide membranes are constrained by permeance-selectivity tradeoff.

• An upper bound is established based on TFC membranes from >300 papers.

• The upper bound is given by A/B = 16000A^−3.2 (A/B in bar⁻¹ and A in Lm⁻²h⁻¹bar⁻¹).

• The effects of various parameters on membrane performance are summarized.

Abstract

Membrane separation properties are constrained by a tradeoff relationship between permeability and selectivity. This tradeoff relationship has been well established for gas separation membranes in the form of the Robeson's upper bound. In contrast, the upper bound relationship is much less established for thin-film composite (TFC) polyamide membranes used for desalination. In this work, we analyzed the tradeoff between the water permeance and the water/NaCl selectivity for TFC membranes gathered from more than 300 published papers. A clear upper bound behavior relationship is established, and the various effects of membrane synthesis conditions and modifications are reviewed in relation to this permeance-selectivity tradeoff. Our work provides a critical tool for the evaluation and benchmarking of future membrane development works in the context of desalination and water reuse.

Introduction

Membrane-based desalination and water reuse have gained increasing popularity in arid regions to cope with water scarcity [1,2]. These applications typically use thin-film composite (TFC) reverse osmosis (RO) and nanofiltration (NF) membranes, where a thin polyamide rejection layer is synthesized on top of a porous substrate by an interfacial polymerization (IP) reaction [3]. TFC membranes with greater water permeability can significantly reduce the specific energy consumption, whereas increasing their salt rejection is beneficial to improve the product water quality [4]. Nevertheless, there exists a strong tradeoff between membrane water permeability and selectivity: increasing water permeability generally leads to reduced salt rejection [[5], [6], [7]].

Historically, the tradeoff between membrane permeability and selectively was first introduced in the context of gas separation. In 1991, Robeson [8] published his classical work on the “upper bound” for the separation factor and permeability for two-gas systems (e.g., O₂/N₂, CO₂/CH₄, etc.), which quickly became a standard benchmark for gas separation membranes. Owing to the huge success of this seminal work, Robeson [9] published a follow-up paper in 2008 to update the upper bound by including several newly developed membrane materials (e.g., ladder-type and perfluorinated polymers). To date, the 2008 Robeson's upper bound (Fig. 1a) is regarded as the golden ruler to gauge nearly all new membrane development works in the gas separation field [[10], [11], [12], [13], [14], [15], [16], [17]].

Compared to the huge success of the Robeson's upper bound in gas separation, the upper bound is much less established for desalination membranes. For example, Tang and coworkers [7] reported the tradeoff relationship between water permeability coefficient and NaCl rejection based on 11 commercial TFC polyamide RO and NF membranes. In 2011, Geise et al. [5] formalized the theoretical framework for the upper bound of desalination membranes on the basis of classical solution-diffusion theory [18] for the first time. These authors then provided a log-log upper bound plot of the intrinsic water/NaCl permeability selectivity (P_w/P_s) vs. the intrinsic water permeability P_w for a set of 26 membranes of various chemistries (Fig. 1b). In their approach, the calculation of P_w and the intrinsic NaCl permeability P_s requires the determination of the exact thickness of rejection layers, which is often challenging for TFC polyamide membrane due to the nanosized voids contained in their “ridge-and-valley” surface roughness structures [[19], [20], [21], [22]]. In a more recent review paper, Werber et al. [6] reported the tradeoff relationship in the form of water-salt permselectivity A/B vs. the water permeability coefficient A (Fig. 1c), which provides a simpler way to evaluate polyamide RO membranes. Nevertheless, all the existing attempts relied on relative small-sized data sets. A more comprehensive survey of the literature is yet to be performed to establish the state-of-the-art upper bound for desalination membranes.

In this study, we analyzed the separation performance of TFC RO and NF membranes, both commercial ones and those prepared in-house, using a dataset collected from more than three hundred of papers published in the last three decades. On this basis, we formulated their upper bound relationship, which could serve as a standard reference in the field of desalination membranes much like the Robeson's upper bound for gas separation membranes. We further examined the effect of various membrane synthesis conditions on the resulting separation performance using the upper bound as a reference framework. Our study provides a new critical tool for the evaluation of membrane development works.