MXene: A two-dimensional material in selective water separation via pervaporation

2.1 MXene fabrication, properties, transport and application in membranes

Many studies report the synthesis of multilayered MXene sheets (Anasori et al., 2017), including wet etching with HF (Ren et al., 2016), HCl-LiF (Chen et al., 2018) and HCl-NaF (F. Liu et al., 2017). In general. there are three fabrication strategies for MXene-based membranes: (i) MXenes are used as skeleton materials to prepare a lamellar-structure; (ii), distinct additives or other nanomaterials are incorporated to fabricate mixed matrix membranes with MXenes; and (iii) MXenes are applied as coating materials to modify a membrane support layer. To some extent, the preparation protocol of MXene sheets dictates the final structural properties of the material and thus the separation performance of the resulting membranes (Ahmed et al., 2021; L. Ding et al., 2017). For instance, Fig. 2 graphically illustrates a typical preparation protocol of an MXene membrane (a), together with the schematic representation of d-spacing and interlayer spacing between the two adjacent MXene nanosheets (b) and the hypothetic water permeation mechanism across the material (c). As mentioned previously, MXene surfaces present abundant oxygen-containing groups (e.g., O, OH,) and/or F groups that confer a hydrophilic nature to the material (Fig. 2b) (Naguib et al., 2011). In the PV process, such hydrophilic surface, together with the fast and selective water transport channels given by the MXene laminates, contribute to the high-water transport following the solution-diffusion model, as observed in Fig. 2c.

Close

Fig. 2

Traditional fabrication of a MXene membrane (a) (L. Ding et al., 2017), illustration of d-spacing and interlayer spacing in MXene sheets (b) (Y. Sun et al., 2019), and hypothetic water permeation mechanism (c) (G. Liu et al., 2018).

Export to PPT

It is worth mentioning that such interlayer spacing of MXene can be fine-tuned by creating pillared structures based on the spontaneous intercalation of surfactants. Of course, the nature of the surfactant used and the alkyl chain lengths (together with the temperature of the solution) play an important role in controlling and tuning the interlayer spacing between layers (Bhat et al., 2021; Simon, 2017). For instance, Luo et al. (2017) successfully fabricated pillared Ti₃C₂ MXenes with 10 nm Sn particles between the MXene layers. Fig. 3 illustrates such a preparation method of Ti₃C₂ MXene containing Sn(+IV)-complexed ions, labeled as CTAB-Sn(+IV) Ti₃C₂, in which the synthesis was reached in two distinct steps. Initially, the authors synthesized a prepillared Ti₃C₂ MXene by intercalating cationic surfactants with various lengths of hydrophobic alkyl chains. Once Ti₃C₂ was immersed in the cationic surfactant solution, the cationic surfactant was self-assembled and intercalated between the negatively charged layers of Ti₃C₂ by electrostatic interactions, provoking an interlayer spacing increase between Ti₃C₂ layers. Even if the interlayer spacing did not vary linearly with the alkyl chain length and the temperature of the surfactant solution, they successfully tailored Ti₃C₂ with controllable interlayer spacings between 1 and 2.708 nm, which were lower than the original interlayer spacing for the pristine Ti₃C₂ MXene (approximately 0.977 nm) (Luo et al., 2017). After this, the authors intercalated Sn(+IV) ions between the Ti₃C₂ layers prepillared with cetyltrimethylammonium bromide (CTAB@Ti₃C₂) showing a spacing of 2.23 nm.

Close

Fig. 3

Graphical depiction of preparation of CTAB − Sn(IV)@Ti₃C₂ via HF etching, CTAB prepillaring, and Sn⁴⁺ pillaring methods (Luo et al., 2017).

Export to PPT

MXene sheets can be assembled into films (also known as supports) by means of different strategies due to the strong Van der Waals forces. Unfortunately, MXene sheets can easily be aggregated, or restack, due to the strong interaction, which may result in the loss of surface area and active sites. To face such an issue, additional phases, such as polymers, carbon materials, among other inorganic materials, are incorporated into the MXene layer to increase the layer spacing. Additionally, reducing some fluoride and oxygen groups on the surface of Ti₃C₂T_x sheets could be contributed by annealing. Regarding the membrane preparation methods, several methods have been utilized by the research community, such as vacuum-assisted filtration, spin-coating, interfacial self-assembly, electrostatic self-assembly, atomic layer deposition, and chemical vapor deposition, among others (X. Li et al., 2021). Vacuum-assisted filtration is the most suitable method for fabricating ultrathin and controlled thickness in membranes ranging from 20 to 500 nm (Huang et al., 2021), but electrostatic self-assembly, atomic layer deposition and chemical vapor deposition can also tailor ultrathin membranes (X. Li et al., 2021).

2.2 Solvent dehydration

The primary application of PV deals with solvent dehydration aimed for solvent upgrading; the process is basically devoted to removing traces of water from organic solvents including ethanol, isopropanol, etc. (R. Castro-Muñoz et al., 2020). Experimentally, the nature of a PV membrane is fundamental for the separation of solvent molecules; for example, a hydrophilic membrane facilitates the transport (diffusion) of polar molecules (including water) while hindering the transport of less polar ones (R. Castro-Muñoz, 2019). On the contrary, a hydrophobic membrane transports preferentially less polar (and non-polar) molecules while hindering the more polar ones. The mass transport in PV membranes is given by the solution-diffusion model (Wijmans and Baker, 1995). In water transport, the water molecules are initially absorbed into the selective layer of the membranes based on chemical affinity; subsequently, they diffuse in vapor phase across the membrane, importantly, water vapor molecules are transported as a result of concentration gradient, and finally, the molecules are desorbed at the permeate side (Baker et al., 2010). The application and commercialization of PV depend on the design of highly permeable and selective membranes. To date, polymers have rendered good enough selectivity for such proposes while their low permeation constricts their commercial establishment, while commercial ceramic membranes based on silica and zeolites display high selective and permeable properties. As part of the researcher’s effort, specific inorganic materials with superior physicochemical properties are intentionally merged into polymers to produce the new generation of membranes that may show enhanced separation efficiency and productivity. MXene has been certainly proposed as filling material into a polymer phase, as reported in Table 1. For example, Xu et al. (2018) embedded the 2D material into a biopolymer, like chitosan (CS), to dehydrate ethanol, ethyl acetate and dimethyl carbonate. The MXene was easily incorporated into the MXene/CS dope solution, which was subsequently coated onto a porous PAN substrate by means of the spinning technique. The resulting membranes had a sub-micron thick layer displaying an average membrane thickness of ∼500 nm for both unfilled CS and MXene MMMs, as shown in Fig. 4a. Interestingly, the MMMs found their optimal filler content at 3 wt% MXene showing a total permeation over 1.4 kg m⁻²h⁻¹ and separation factor of 1,421 for ethanol dehydration (see Fig. 4b). This separation performance seemed to be much better than the one offered by the well-explored graphene oxide (GO) used for ethanol dehydration, e.g., GO/CS (Suhas et al., 2015) and cross-linked GO PVA membranes (R. Castro-Muñoz et al., 2019), which exhibited lower permeation rates (lower than 0.14 kg m⁻²h⁻¹) and separation efficiency of 1093 and 263, respectively. The performance was a result of both sorption and diffusion processes; however, even if the contribution of MXene into CS was not clear in terms of hydrophilic enhancement according to similar hydrophilic nature of both materials, MXene conducted to an improved selective diffusion of water compared to ethanol (Cao et al., 2015). Also, it was believed that MXene sheets were particularly aligned into laminates possessing interlayer channels suitable for water transport. The same membrane formulation (i.e., 3 wt% MXene) was further explored for the dehydration of ethyl acetate and dimethyl carbonate, as represented in Fig. 4c and d, respectively. Similarly, the performance of the MMM was better than that of the pristine CS membrane; but a greater separation efficiency was observed when dehydrating ethyl acetate thanks to variable molecular interactions between water and the solvent (Xia et al., 2011). These MMMs displayed a stable performance in terms of permeation (up to 1.4 kg m⁻²h⁻¹) and high water content (as a result of high separation factor = 4,898) in permeate over the continuous ethyl acetate dehydration for 1800 min (Xu et al., 2018). Such outcomes are competitive with the ones from other types of composite membranes, e.g., PVA/PAN (with separation factor of 54 and permeation of 0.34 kg m⁻²h⁻¹) (Yuan et al., 2011) and PBI/PEI dual-layer membranes (with separation factor of 99, 900 and permeation of 0.82 kg m⁻²h⁻¹) (Wang, 2015).

Close

Fig. 4

a) Cross-section view of the MXene/CS membrane containing 3 wt% MXene; b) PV performance of the MMMs as a function of MXene loading for ethanol dehydration (10 wt% water, at 50 °C); c) PV performance of the membrane with 3 wt% MXene for dehydration of ethyl acetate (2 wt% water, at 50 °C) and d) dimethyl carbonate (2 wt% water, at 50 °C). Adapted from Xu et al. (2018).

Export to PPT

Li et al. (2020) also utilized multilayered MXene as a filling material into sodium alginate (SA) to generate MMMs. In comparison with MXene/CS membrane (Xu et al., 2018), MXene/SA MMMs are likely to be more selective to water in ethanol dehydration displaying a separation factor of 994; where MXene stands out as an enhancer of the preferential sorption of water in SA while hindering ethanol molecules. Unfortunately, a poor permeation rate of 0.50 kg m⁻²h⁻¹ was noted in the resulting membranes.

Currently, biopolymers, such as CS and SA, are gaining notable attention as a more environmentally friendly way of membrane fabrication (Galiano et al., 2018). Apart from their swelling issues, both CS and SA are excellent candidate materials for water removal due to their high hydrophilicity. However, several chemically synthesized polymers have been widely investigated including polyimides (Kudasheva et al., 2015), polybenzimidazole (PBI) (G. M. Shi et al., 2012), polyacrylonitrile (PAN) (Okumuş et al., 2003) and poly(vinyl alcohol) (PVA) (Amirilargani and Sadatnia, 2014). This latter polymer is the only one to be consolidated and commercialized at industrial level. For example, international supplier, such as DeltaMem AG (http://www.deltamem.ch), currently manufactures and commercializes cross-linked PVA membranes for PV applications. At this point, current research is still focused on enhancing the separation performance of such hydrophilic polymer. For instance, Cai et al. (2020) proposed MMMs based on cross-linked PVA and MXene for ethanol dehydration. The membrane with 3 wt% MXene displayed enhanced separation factor values over 2,500 with permeation fluxes of 0.074 kg m⁻²h⁻¹ when testing for ethanol solution (7 wt% water at 37 °C). Even though the permeation does not look attractive enough compared to other high-performance membranes, the selective properties of the pristine PVA membrane were substantially improved. In general, the total flux of MMMs decreased by increasing MXene loading, which was associated with a resistance increase towards water molecules to pass through the membrane, as a contribution of the cross-linking density increase, along with aggregation of MXene.

As one of the latest works in the field, Yang et al. (2020) also assayed the embedding of MXene into PVA, followed by chemical in situ cross-linking with sulfosuccinic acid (SSA); the preparation protocol is represented in Fig. 5a. The resulting nanocomposite presented an ultrathin nature with thickness up to ≈230 nm, revealing an outperforming pervaporative separation for the removal of water from various solvents including ethanol, methanol, tert-butanol and isopropanol (IPA). As can be seen in Fig. 5b, the alcohols were easily hindered by the membrane while transporting water to permeate with water content over 97% (at 30 °C), which was a result of high separation factors of 968, 4738, 7913 and 23,786 for water/methanol, water/ethanol, water/IPA and water/tert-butanol, respectively. By maintaining the same concentration (4 wt% water in all alcohol mixtures), it is clear to visualize the greater effect of the sizes of the alcohol molecules, such as methanol (0.36 nm), ethanol (0.45 nm), IPA (0.47 nm), tert-butanol (0.46 nm), over water permeation flux. Moreover, Fig. 5c illustrates the stability of the cross-linked PVA/MXene membranes for 50 h operation in ethanol dehydration, basically, the water flux decreased from 1.46 to 1.31 kg m⁻²h⁻¹ due to the reduction of water concentration in the feed tank. At this point, it could be thought that such membranes may find a stable long-term operation in continuous mode maintaining the concentration in the feed side.

Close

Fig. 5

a) Preparation strategy for the ultrathin PVA/MXene composite membrane; b) performance in desalination (left side) and organic solvent dehydration (right side); c) long-term operation for ethanol dehydration (Yang et al., 2020).

Export to PPT

Towards IPA purification, Liu et al. (2019) developed ultrathin MXene membranes (approximately 100 nm) with highly ordered 2D nanochannels. In this work, a new strategy of membrane preparation was oriented to use positively charged hyperbranched polyethylenimine (HPEI), which was intercalated between negatively charged MXene nanosheets to fit out regular stacking structures via electrostatic interactions; after this, interfacial polymerization with trimesoyl chloride (TMC) was applied to seal possible non-selective defects. By separating IPA solution (10 wt% water, 50 °C), the resulting membrane (i.e., Ti₂CT_x–HPEI/TMC with d-pacing of 0.44 nm) was able to produce permeate fractions containing water content higher than 99 wt%, as a result of high separation factor (over 1,200). The permeation was also estimated as high as 1.092 kg m⁻²h⁻¹, with stable long-term operation over 120 h. In the same research group, the authors explored a different protocol for the molecular separation of water from IPA using MXene membranes (G. Liu et al., 2020); at this point, they introduced various polyelectrolytes (such as PDDA, PEI, PAH) to control the membrane structure during the PV operation. After preliminary studies, the authors claimed that the poly (diallyldimethylammonium chloride) (PDDA) functionalized MXene membrane revealed the highest separation performance ascribed to the optimized electrostatic effect and hydrophilicity. When compared with Ti₂CT_x–HPEI/TMC membrane (G. Liu et al., 2019), this PDDA functionalized MXene demonstrated higher permeation (∼1.237 kg m⁻²h⁻¹) and efficiency (separation factor of 1,932) (G. Liu et al., 2020). Here, MXene acquired plenty of oxygen groups and improved hydrophilic nature from PDDA, contributing to exceptional IPA dehydration. It is worth mentioning that PDDA functionalized MXene was fabricated in an easier way compared to Ti₂CT_x–HPEI/TMC, which is important from the feasibility and commercialization point of view.

The high permeation of 2D materials deals with their well-defined nanochannels and characteristic d-spacing; however, a specific 2D material, such as GO, has suffered from an interlayer d-spacing enlargement in the presence of water; as an example, the interlayer spacing in GO varied from ≈6.4 to 9.8 Å with relative humidity range from 0 to 100%, respectively (Abraham et al., 2017). This enlarged spacing can definitely affect the selective properties of the materials since water and other molecules can be indifferently transported across the channels. The phenomenon was investigated by Wu et al. (2019), who observed that the MXene channels of 1.37 nm exhibited an increase of 0.23 nm after immersing in water. Also, the authors monitored the d-spacing as a function of time showing an increase of 0.2 nm after immersed in water for 2 h. After immersion for 5 days, such an increase in d-spacing was still ∼0.2 nm in comparison with the dry membrane. This gives an idea that the increase in d-spacing is not infinite, noting enough stability in liquid solution. Here, MXene owns an advantage compared with its analogous GO since this latter material can easily be disintegrated in water due to the strong electrostatic repulsion forces among sheets, while MXene has demonstrated a stronger van der Waals attraction force and weaker electrostatic repulsion force. Thereby, MXene stands out as potential material for water separations and impressive stability in water (Lao et al., 2018), Additionally, MXene membranes also displayed an unchanged d-spacing when immersed in ethanol, suggesting a much better anti-swelling stability in polar organic solvents. When tested for ethanol dehydration (5 wt% water), MXene membranes (with 2 µm thick) eventually showed their best separation performance at room temperature with a separation factor and flux of 135 and 0.263 kg m⁻²h⁻¹, respectively (Y. Wu et al., 2019). Regarding the flux improvement, the authors systematically increased the flux by decreasing the membrane thickness up to 0.5 µm with an average flux of 0.60 kg m⁻²h⁻¹. The low permeation was attributed to tortuous 2D channels that produce a mass transfer resistance, which supports Cai’s statement, where an increase of MXene loading lowered the total flux by increasing the mass transport resistance (Cai et al., 2020).

2.3 Seawater desalination

One of the key contributions of MXene in PV membranes falls into seawater desalination, where impressive separation performances have been obtained. For example, Liu et al. (2018) created ultrathin MXene-based membranes (thickness ∼60 nm) onto polyacrylonitrile (PAN) ultrafiltration membranes using vacuum filtration, which reported high water flux (up to 85.4 kg m⁻²h⁻¹) and salt rejection (∼99.5%) when separating a 3.5 wt% NaCl solution (at 65 °C). In this work, MXene conferred high hydrophilicity to the resulting membranes as a result of its plenty of e oxygen-containing groups. Additionally, such membranes were able to completely reject other ions including Mg²⁺, Ca²⁺, K⁺, SO₄ ²⁺, Cl⁻, and Br⁻, presenting a good long-term operation for desalination (over 100 h). However, unprecedented separation in seawater desalination has been reported by Ding et al. (2020), who designated ultrathin membranes (ca. 30 nm layer). Particularly, a maleic acid covalent-bridged MXene (MXMA) membrane supported onto nylon support using facile vacuum-assisted filtration synthesis. Here, adjacent MXene nanosheets were smartly fixed by the covalent bonding formed by the esterification reaction between the carboxyl groups in maleic acid and hydroxyl groups in MXene surface, as illustrated in Fig. 6a. The authors compared the performance of the MXMA membranes with the pristine MXene membranes, exhibiting both comparable salt rejection; however, MXMA membranes showed higher permeation values (ca. 22 kg m⁻²h⁻¹) than MXene membranes (ca. 15 kg m⁻²h⁻¹) (see Fig. 6b). It is important to point out that the performance in terms of water permeate flux (70 kg m⁻²h⁻¹) and salt rejection (99%) was impressively achieved when operating at 65 °C. The study stated that the esterification reaction constructed well-defined nano-channels in MXMA membranes with a size of ca. 0.49 nm, which were larger in comparison with the MXene membrane (∼0.37 nm). Considering channels of 0.49 nm, they were still larger than the kinetic diameter of water molecules (∼0.29 nm) but smaller than both hydrated Cl^– (∼0.66 nm) and Na⁺ (∼0.72 nm)(Joshi et al., 2014; Song et al., 2019), conducting to efficient sieving effect. Here, the high permeation was initially related to the still hydrophilic nature of MXMA that promotes water transportation, however, during the cross-linking protocol, Ding et al. (2020) speculated that the usage of hydrophilic hydroxyl groups and the production of hydrophobic ester groups in MXMA membranes resulted in more frictionless regions, introducing fast water flow within the channels. Fig. 6c also shows that MXMA membranes demonstrated a similar performance when treating real seawater while maintaining exceptional stability over 30 h operation. So far, these membranes are among the thinnest membranes with promising potential to compete with PVA/intercalated graphene oxide membranes (water permeation of 98 kg m⁻²h⁻¹, at 85 °C) as the most efficient ultrathin membranes for seawater desalination, surpassing the performance of most of the ultrathin membranes reported in the literature (see Table 2). Once a selective and defect-free morphology is obtained, the ultrathin nature of membranes is pursued to break the trade-off between selectivity and permeance. Besides, the fabrication of ultrathin membranes needs a small amount to design the selective layer, contributing to material cost savings (Roberto Castro-Muñoz et al., 2020).

Close

Fig. 6

a) Hypothetical reaction between MXene and maleic acid via esterification reaction; b) desalination performance of both MXMA and MXene membranes with similar thickness (ca. 30 nm) (3.5 wt% NaCl solution, at 30 °C); c) and long-term operation of the MXMA membranes at the same operating conditions.

Export to PPT

At almost room temperature (30 °C), Yang et al. (2020) reported that ultrathin cross-linked PVA/MXene membranes (ca. 230 nm) offer extremely high desalination performance (water permeation: 62.2 kg m⁻²h⁻¹, salt rejection:99.8%), as shown previously in Fig. 5b. The authors claimed that such an impressive pervaporative performance was ascribed to the MXene incorporation. MXene composite membrane exhibited high water permeation flux due to the combination of several factors including a more amorphous region, increased free volume pore size, higher free volume, improved hydrophilicity and additional permeating paths through MXene and MXene-polymer interphase. Based on the excellent desalination performance towards NaCl, the authors also evaluated the rejection performance towards other salt ions obtaining high rejection values for KCl (99.81%), Na₂SO₄ (99.91%), MgCl₂ (99.93%), CaCl₂ (99.92%) and MgSO₄ (99.93%). It was clear that this composite membrane owned a defined nanostructured, together with water affinity properties favouring high permeation and water selectivity; however, it is worth mentioning that the membranes also displayed a negative charge surface that is positive for ion rejection governed by the Donnan exclusion effect (Schaetzel et al., 1993), in which divalent ions tend to be more sensitive to charge exclusion than monovalent ions.

3.1 Research gaps and recommendations for the new researchers in the field

To finalize, according to the findings of this review, the following suggestions and research gaps for the new researchers in the field are given:

• One of the bottlenecks of hydrophilic pervaporation deals with the swelling of polymer membranes by polar solvents, which conducts to compromising their selectivity. Even though MXenes have apparently enhanced the solvent resistance in polymer composites (P. Liu et al., 2020), there is still a large number of hydrophilic polymers (such as polyimide, polybenzimidazole, polyether ether ketone, among others) that suffer such a phenomenon. Therefore, precise studies on swelling issues should be analyzed in MXene-based nanocomposite declaring the mechanism or interaction for the successful swelling mitigation.

• Until now, most of the research in pervaporation separation has pointed out the ability of MXene- based membranes for a long-term operation (up to 60 h) for binary solvent mixtures and aqueous salt solutions. To some extent, such evaluation is a starting point that opens the window of testing such membranes in harsh environments (real azeotropes, natural seawater, etc.); however, further research is needed to better understand the impact of MXene nanostructure on separation performance as well as the structural stability of selective layer (Zhu et al., 2018). Here, the evaluation of the membrane performance will give us a more realistic approximation of the feasibility of the membranes not only for real yield but also for their commercialization, where MXenes are still at an early stage of the investigation.

• Molecular diffusion for understanding the water transport mechanism in MXenes has been explored by analyzing computer simulation (P. Liu et al., 2020). Unfortunately, parameters associated with the separation mechanism have been minimally investigated. In this context, simulation models should be studied in detail for a better understanding of water transport and separation (and other solvent molecules), along with new outperforming membrane designs (Karahan et al., 2020).

• When dealing with mixed matrix membranes, the incorporation of inorganic materials (like 2D) contributes to higher fluxes compared with pristine polymer membranes. However, filler loading increase does not always guarantee the highest permeation. In particular, overuse of MXene can decrease the permeation rate due to resistance increase towards permeating molecules (Cai et al., 2020). At this point, researchers should optimize the MXene loading to compensate the distortion of the polymer chain, which can provoke the creation of non-selective gaps.

• As widely investigated in its analogous GO, the influence of water content on the nanostructure of MXene, specially d-spacing, must be further evaluated since there is strong evidence of channel enlargement in 2D materials in the presence of water (Abraham et al., 2017), influencing the separation performance. Even if Wu et al. (2019) declared only ∼0.2 nm increase in respect to the initial d-spacing, researchers should be aware of any change in the structural properties of MXene membranes applied in polar separations.