WS₂ deposition on cross-linked polyacrylonitrile with synergistic transformation to yield organic solvent nanofiltration membranes

Highlights

• Depositing WS₂-NMP dispersion on polymeric supports yields nanofiltration membranes.

• Exposure to NMP densifies cross-linked polyacrylonitrile substrates.

• Impressively high ethanol permeance of 43.35 L m⁻² h⁻¹ bar⁻¹.

• Long-term sustained rejection of Evans Blue dye at 99%.

Abstract

Using liquid phase exfoliation (LPE) methods established in literature, several different tungsten (IV) disulfide (WS₂) dispersions were prepared in solvents ranging from NMP, ethanol/water to pure water via sonication under ambient conditions. These various dispersions were then deposited directly onto cross-linked ultrafiltration-level polyacrylonitrile (PAN) flat sheet membranes through a pressure-assisted filtration process to produce ready-to-use membranes without any additional modifications. To evaluate and compare the performances of the various membranes produced, organic solvent nanofiltration (OSN) was utilized as the testing method while also representing a potential application of these fabricated membranes. After observing and comparing the outcomes of depositing different WS₂ dispersions, deposition of the NMP dispersion was found to yield a promising membrane due to a concurrent synergistic densification effect with positive impacts on membrane performance. Thus, further emphases were placed on studying the use of different initial WS₂ concentrations and longer sonication times in NMP dispersions. After the optimization, the composite membrane consisting of WS₂ on cross-linked PAN substrates shows impressive performance with a high pure ethanol permeance of 43.35 L m⁻² h⁻¹ bar⁻¹, a rejection of 86% towards Remazol Brilliant Blue R (M_w = 626.54 g/mol) and a long-term sustained rejection of 99% towards Evans Blue (M_w = 960.81 g/mol) in ethanol.

Introduction

In the past decade, there has been growing interest and research in transition metal dichalcogenides (TMDs) which constitutes a group of materials with a generalized formula of MX₂ where M is typically a group 4–7 transition metal and X is a chalcogen (i.e. S, Se or Te) [1]. Interest in these materials stems from their unique properties such as a lamellar structure which allows them to be used in separation processes [[2], [3], [4], [5], [6], [7]], favorable electronic and semiconducting properties for use in high-end electronics [[8], [9], [10]] and large surface areas which are particularly useful for applications in catalysis and sensing [[11], [12], [13]]. In particular, TMDs also stand out from certain other 2D materials such as graphene oxide by displaying superior aqueous stability stemming from the lack of hydrophilic functional groups. This negates the need for cross-linking while the absence of functional groups protruding from the material surface may reduce hindrances to liquid flow [3,14]. TMDs also exhibit impressive out-of-plane rigidity or bending modulus due to the three-layered structure [15,16]. Additionally, more than 40 different types of TMDs exist due to the numerous combinations of chalcogens and transition metals possible. This significantly widens the range of properties that can be displayed and utilized [17]. To access these unique and desirable properties, TMDs have to be exfoliated from their bulk materials to yield single to few-layer flakes. Generally speaking, exfoliation involves disrupting the weak van der Waals forces between the adjacent layers to yield the aforementioned flakes. Amongst the exfoliation methods, TMDs may be chemically exfoliated via organolithium intercalation to achieve a significant proportion of monolayer flakes [[18], [19], [20]]. However, the drawbacks of this method include the need for an inert atmosphere as the process is highly sensitive to the environmental conditions, long reaction times of about 2 days and post-processing to remove the excess organolitihum compounds before finally exfoliating the flakes into monolayers.

An alternative method pioneered by Coleman et al. involves the sonication of the bulk TMD material in a variety of common organic solvents under ambient atmospheric conditions to yield mainly few-layer flakes [21,22]. This method is particularly advantageous since it is able to overcome several of the aforementioned drawbacks faced by chemical exfoliation while allowing for easier scaling-up and reducing contamination in the final product. In the simplest method, the bulk TMDs are merely placed in a suitable solvent such as N-methyl-2-pyrrolidone or dimethylformamide and sonicated using a probe sonicator to disrupt the van der Waals interaction between the layers, thus producing the desired thin flakes. To circumvent the use of these non-volatile solvents, Zhou et al. have also utilized a mixture of greener, lower boiling-point solvents such as ethanol/water to exfoliate TMDs [23]. In using water as the only solvent, certain additives such as poly (acrylic acid) [13] and bovine serum albumin [24] may also be added to aid the sonication-induced exfoliation of the TMDs. In these latter few cases, only a simple bath sonicator is required.

As previously mentioned, such layered TMDs are of interest in the field of membrane science and separations. Thus far, many works have focused on the use of TMDs, particularly molybdenum disulfide (MoS₂), in aqueous-based separations [[3], [4], [5],25,26]. For example, Wang et al. deposited chemically exfoliated MoS₂ onto a polyethersulfone substrate and thoroughly studied the stability and performance of this membrane [3]. They have demonstrated the superior aqueous stability and water permeance of such MoS₂ membranes compared to graphene oxide membranes. Similar superior results were obtained by Sun et al. when they deposited chemically exfoliated WS₂ with added copper hydroxide nanostrand templates onto anodic alumina oxide (AAO) substrates [5]. This work by Sun et al. is one of few works involving the use of WS₂ instead of MoS₂ for membranes. Apart from aqueous-based separations, Wang et al. produced similar chemically-exfoliated MoS₂-based and AAO-supported membranes for use in gas separation [27] while Deng et al. demonstrated the molecular sieving property of MoS₂ membranes for organic vapors [14]. Together, these and other works in literature demonstrate the viability and stability of TMDs for use in separation processes.

Given the promising nature of TMDs in membrane applications, we are inspired to study the use of such materials, particularly the less common WS₂ as opposed to MoS₂, in yet another membrane-based application that is organic solvent nanofiltration (OSN), also known as solvent resistant nanofiltration (SRNF). OSN is characterized by membranes operated and used in organic solvent feeds with the ability to reject solutes that are 0.5–2 nm in size [28], possessing a molecular weight cutoff (MWCO) of between 200 and 1000 g/mol [29,30]. Such a process is of particular importance given that the annual global usage of organic solvents are in the megaton range [31] and such solvents can account for up to 90% of the mass in batch chemical operations [32]. Hence, research in materials suitable for OSN may provide potential solutions for the use of membranes in numerous processes where organic solvents are unavoidable. Furthermore, being a membrane-based process, OSN enjoys typical advantages such as mild operating conditions involving ambient temperatures and low pressures [[33], [34], [35], [36]] and relatively lower energy consumption compared to certain conventional separation processes [[37], [38], [39], [40]].

Given our desire to produce a membrane with potential use in OSN, we focused on the use of sonication-assisted liquid phase exfoliation of WS₂ to prepare the TMD dispersions. It is notable that such a method is in opposition to many of the membrane-related literature cited above where chemical exfoliation is preferred and used [3,5,14,27]. However, since the sonication-assisted liquid phase exfoliation method can produce a fairly large amount of exfoliated products under ambient conditions in shorter times using easily available equipment, we believe that it may prove to be more convenient and preferable for membrane production. To this end, we shall be exploring and comparing membranes fabricated via the deposition of WS₂ dispersions produced through sonication in a variety of solvents such as NMP, ethanol/water and pure water to observe the impact of the dispersion used on membrane quality. To further simplify and generalize our membrane fabrication process [41,42], we simply deposited these various dispersions directly onto solvent-resistant cross-linked PAN ultrafiltration membranes [35,43] rather than more costly substrates such as AAO discs. Through these means, we successfully obtained membranes with desirable and even long-termed OSN performances in the benchmark organic solvent ethanol, particularly via the deposition of WS₂-NMP dispersions.