Roll-To-Roll Fabricating MXene Membranes with Ordered Interlayer Distances For Molecule And Ion Separation

Application-oriented assembly of two-dimensional nanosheets with uniform nanochannels is critical for fabricating sophisticated, high-performance membranes for water treatment. However, fabricating the desired membranes by a simple, fast, and eﬀective method is a challenge as most of the previous methods are based on batch processes rather than a continuous roll-to-roll process. Here, a simple Meyer rod-coating approach to continuously fabricate large-size and ﬂat MXene membranes at a scale up to 5 m is introduced. This study demonstrated that a high MXene concentration, above 10 mg mL − 1 , is critical in processability due to the desired viscosity, surface tension, and viscoelastic properties. The as-made MXene membranes show that shearing and solutal-Marangoni ﬂow can considerably improve the ordering of the stacked MXene nanoﬂakes. Thus, the rod-coated MXene membranes demonstrate a smaller surface roughness and interlayer distance compared to the MXene membranes fabricated by the most commonly vacuum-assisted ﬁltration. The rod-costed MXene membranes show superior performance in dye and mono/divalent cation separation. The proposed roll-to-roll Meyer rod-coating method can also be used to fabricate MXene-based composites, such as MXene/carbon nanotubes and MXene/polymer, using the inks containing high concentration MXene and other desired compositions. This roll-to-roll method will promote an industry-level fabrication and application of MXene-based membranes.


Introduction
Separation and purification are critical for protecting the environment and obtaining valuable resources, such as drinking water and desired metal salts. Plenty of effort has been put into identifying promising materials for molecule and ion separation. [1] Two-dimensional (2D) materials have shown promising potential in membrane separations, which is a superior process to other separation techniques, [2] due to its modularization and high efficiency. [3] One of the most striking properties of 2D materials is their ability to form laminar structures with sub-nanosize interlayer distances for molecule and ion separation. This feature has been extensively used in fabricating novel membranes for advanced membrane separation. [4,5] The applications of 2D materials are heavily dependent on processing and assembly methods. [6,7] It has been shown that 2D building blocks must be assembled into membranes to translate their intrinsic properties into a useful separation process. [4] 2D materials can be made into various membranes, depending on the rational design and corresponding microstructures. For ion and molecule separation or selection, 2D materials must be assembled into an appropriate and controllable porous structure. The tunable structure combined with the material properties determines the overall performance of the separation process.
MXene, a new family of 2D transition metal carbide/nitride, has attracted expanding interest worldwide, particularly in molecule and ion separation, due to its richness in compositions, tunable surface chemistry, and interlayer structures. [8][9][10][11] However, there are several hurdles, particularly large-scale fabrication, which need to be overcome before MXene can be practically applied in membrane separation and other applications where continuous fabrication and controlled interlayer distance play critical roles. 2D materials, including MXene, lend themselves ideally to fabricating continuous membranes with ordered interlayer distance because of the high aspect ratio and strong interflake interaction of the 2D building blocks. [12] Vacuum-assisted filtration (VAF) is the most commonly used method to assemble MXene nanoflakes into supported and free-standing membranes. [13][14][15][16][17] www.advancedsciencenews.com www.advmatinterfaces.de Although impressive results have been achieved for the VAF membranes, several limitations and disadvantages, such as high energy and time consumption, and limited membrane areas, intrinsically limit the scalability of the VAF method. Various alternative methods, such as spray coating, drop-casting, and slot-die coating, have been proposed to fabricate large-area MXene membranes. [18,19] Spray coating uses a spray gun to deposit nanoflakes, featuring fast drying and easy scale-up. However, the spray method is hard to fabricate well-ordered laminar structures. [20] The drop-casting method has impressive performance in producing free-standing MXene membranes with micro-patterns and outstanding electromagnetic interference shielding efficiency. [12] The membranes made by drop-casting underwent natural sedimentation, producing thick membranes in several tens micrometers with no mirror effect due to the lack of shear force and less uniform deposit. [12,19] Slot-die coating is superior in the production of smooth and thin MXene membranes, showing promising potential in the large-scale fabrication of highly conductive membranes and nanofiltration. [18,19,21] However, most of the existing methods for fabricating MXene membranes are batch processes rather than continuous ones. The batch processes could limit the production rate if the fabrication systems are scaled up to industrial production levels. Although slot-die coating and drop-casting can produce largearea MXene membranes, their continuous fabrication is yet to be demonstrated. No continuous roll-to-roll fabrication of MXene membranes has been demonstrated up to now.
Developing continuous and scalable technologies for fabricating uniform and tunable MXene membranes using nanoflakes, which can leverage the full potential of MXene, can facilitate the wide and practical utilization of MXene. The rod-coating method has displayed good compatibility with roll-to-roll deposit processes. [22,23] A meniscus is formed and deposited across the substrate by viscous forces induced by the rod-coating head and the substrate on the dispersion, guiding the membrane coating using the nanoflakes in the dispersion. The formed meniscus can confer nanoflake alignment, producing continuous and wellorganized laminar membranes. [24] The contact portion of the rod with the deposited substrate is smooth, which can reduce and eliminate the sharp edge of the blade-coating method, producing defect-free membranes for molecule and ion separation. Meyer rod-coating has been employed as a simple, scalable, and cost-effective method to fabricate MXene-on-paper electrodes. [25] However, the MXene membranes made by Meyer rod-coating have a thickness of 125 μm with a rough surface because the MXene powders without exfoliation were used as the precursors for making slurries to coat. The rod-coating method involves various associated processing parameters which would control the stack structure of nanoflakes and alter the performance of the deposited membranes. Up to now, the exact rod-coating processes and corresponding membrane performances are still not completely understood.
In this study, we introduce a new and facile Meyer rod-coating method using a gap-controllable wire rod and high-concentration MXene dispersion with desired concentration-determined rheological properties and surface tension. The correlation between rod-coating processes and membrane integrality was studied. The MXene dispersion, coating parameters, and drying procedures were optimized to enhance the controllability and repro- ducibility of the rod-coated MXene membranes. The dryingdependent interlayer nanochannels of rod-coated MXene membranes were studied and compared with the VAF ones. The proposed versatile method can also be utilized for fabricating MXene/carbon nanotubes and MXene/polymer composites using highly concentrated MXene dispersions as the matrix. The asmade MXene membranes showed outstanding performance in removing dyes and sieving mono-/divalent cations due to their narrow and uniform nanochannels produced by shear force in rod-coating. Figure 1 schematically shows the roll-to-roll method for continuous large-scale production of MXene membranes. Shear force on MXene dispersion is produced when there is relative movement between the Meyer rod and the substrate. The dispersion can be uniformly coated on the porous polyethersulfone (PES) substrate under the constant shear force. The water in the dispersion evaporates along the meniscus line formed between the rod-substrate gap, resulting in the dispersed MXene nanoflakes aggregation and precipitation. The water evaporation rate at the meniscus increases gradually along the tip region, and the uneven water evaporation changes the MXene concentration, resulting in the surface tension gradient driving the fluid to form a Marangoni flow along the meniscus. Both the shear force and capillary effect improve the directional packing uniformity of the MXene nanoflakes, forming highly ordered MXene membranes.

Results and Discussion
The continuous and scalable fabrication of MXene membranes was demonstrated in Movie S1 (Supporting Information) by the roll-to-roll method. A roll of MXene membrane with a length over 5 m was demonstrated in Figure 2a. The as-made membrane shows metallic luster, indicating its smooth surface. The thicknesses of the as-made MXene membranes can be easily turned by controlling the coating cycle numbers. Three typical membranes with thicknesses of ≈300, 650, and 1400 nm were produced and shown in Figure 2c,d. These membranes confirm that this roll-to-roll continuous coating method can readily control the as-made MXene membranes from several hundred nanometers to the micrometer scale. The cross-sections of the membranes indicate the dense stack of nanoflakes due to the shear force resulting in the highly oriented stacking of MXene nanoflakes. [21] As the thickness increases, a few voids appear in the membranes (see Figure 2c,d), typical for thicker membranes made by coating. [21] It is difficult to maintain the ideal nanoflake alignment for thicker membranes during drying. The interlocked MXene nanoflakes, as shown in Figure 2e-g, formed the gleamy surfaces of as-made MXene membranes. The homogeneity of the membranes was confirmed by the electrical conductivity tests. Five randomly selected spots on a piece of MXene membrane (with a thickness of 300 nm) show almost identical electrical conductivity of ≈ 4410 ± 172 S cm −1 (Table S1, Supporting Information), confirming the impressive homogeneity of the MXene membranes made by this roll-to-roll method.
The rod-coated Ti 3 C 2 T x MXene membranes with a thickness of 300 nm display high electrical conductivity of 4410 ± 172 S cm −1 , which is larger than most of the nanomaterials but is smaller than the best MXene membranes (15 000 S cm −1 ) fabricated via the blade-coating method. [21] The degraded electrical conductivity could be due to the smaller nanoflake sizes caused by vigorous sonication. [21] The tip sonicator was used to delaminate Ti 3 C 2 T x MXene in this study to get dispersion with high concentration for rod-coating. The conductivity shows a slight decrease as the thickness increased to 1400 nm (3726 ± 212 S cm −1 ), as shown in Table S1, Supporting Information. The voids, defects in the membrane structure, which were formed in the Ti 3 C 2 T x membrane due to imperfect staking of thicker membranes, could also contribute to the degraded electrical conductivity. [21] The concentration of MXene dispersion plays a critical role in controlling the flow behaviors of dispersions. [7,27,28] Figure 3a shows the concentration-dependent surface tension of MXene dispersions measured by the droplet method at room temperature, which dramatically increases with increased Ti 3 C 2 T x MXene concentration from 0.0 to 2.5 mg mL −1 and saturates at ≈75 mNm −1 as the concentration above 16.5 mg mL −1 .
The increased surface tension with MXene concentration makes the spreading of the dispersions on porous polymer substrates difficult, as indicated by the Ti 3 C 2 T x MXene concentrationdependent contact angles, which increase with the dispersion concentration ( Figure 3b).
Viscosity measurement shows that the Ti 3 C 2 T x MXene dispersion at 27.5 mg mL −1 presents gel-like viscosity reaching 1.49 Pa s at 0.01 s −1 . The dispersion viscosity reduced with the increased shear rate and decreased MXene concentration, as shown in Figure 3c. The larger viscosity at a low shear rate and high MXene concentration, and rheological shear-thinning behavior enabled the Ti 3 C 2 T x MXene dispersions with high concentration to have the desired processability. [6] Additionally, the Ti 3 C 2 T x MXene dispersion showed larger elastic (G') and viscous (G'') moduli as the concentration increased. The G'/G'' ratio of MXene dispersions shows concentration-dependency, reaching a range of 1 to 10 from 0.1 to 10 rad s −1 , indicating the membrane morphology will be maintained after the removal of the shear force. [19,21] Besides dispersion fluidics, the processing parameters, such as coating speed and groove depths of the Meyer rods, must be optimized to allow a high degree of flat and integrality control of the coated Ti 3 C 2 T x MXene membranes. Therefore, control experiments were conducted to figure out the impacts of dispersion concentration, Meyer rod groove depths, and coating speeds on the fabrication of MXene membranes. The corresponding results were provided in Figure S2, Figure S3, and Figure S4 in Supporting Information. When the MXene dispersion concentration was lower than 10 mg mL −1 , the rod-coated membranes were not uniform. There are grooves and stains on the polymer substrates ( Figure S2, Supporting Information) due to the reduced viscosity and viscoelasticity. Larger Meyer rod groove depths and faster coating speeds would destroy the homogeneity and integrality of the rod-coated MXene membranes. The coating speed of 2 to 5 mm s −1 and the groove depth of the Meyer rod of 3 μm facilitated the fabrication of uniform membranes when MXene dispersion with a concentration above 10 mg mL −1 was used. In the following part, 27.5 mg mL −1 was selected for used MXene dispersion with a coating speed of 2 mm s −1 and a groove depth of 3 μm. These conditions facilitated the continuous and uniform MXene membranes on porous PES substrates.
The Meyer rod-coating method is versatile in fabricating various membranes using inks with proper concentrations and rheological behaviors. Ti 2 CT x -based MXene membranes were successfully fabricated by the roll-to-roll Meyer rod-coating method ( Figure S5, Supporting Information). The flat and integral membrane can be made using the Ti 2 CT x dispersion with a concentration of 15 mg mL −1 . This is also the first report on the scalable fabrication of Ti 2 CT x -based MXene membranes due to the versatile rod-coating method. Both Ti 3 C 2 T x /multi-walled carbon nanotubes (Ti 3 C 2 T x /MWCNTs) and Ti 3 C 2 T x /polyvinyl alcohol (Ti 3 C 2 T x /PVA) membranes were successfully fabricated by this rod-coating method ( Figure S6, Supporting Information). These results confirm the generic applicability of the proposed rollto-roll method in continuously assembling MXene-based membranes. High-concentration Ti 3 C 2 T x -based MXene plays a dual role in fabricating Ti 3 C 2 T x /MWCNTs composite membranes. MXene nanoflakes can help to disperse MWCNTs without surfactants due to the steric effect preventing MWCNTs aggregation. This method is expected to fabricate other novel composite mem-branes with MXene dispersion as the matrix to hybridize nanomaterials and polymers.
It is well-known shearing force could improve the ordering of stacked nanoflakes. [29] The stack of Ti 3 C 2 T x MXene nanoflakes in the Meyer rod-coated membrane is significantly different from that in the VAF membrane with the same loading of 0.15 mg cm −2 as shown in Figure 4a, where the rod-coated MXene membrane shows a sharp peak (002) at ≈6.58°compared to the widened peak at ≈6.10°of the VAF MXene membrane, indicating that the shearing inducted nanoflake stack being more ordered and compacted. The sharp peaks with smaller full width at half maximums (FWHMs) for the rod-coated membranes keep constant during various drying times, as shown in Figure 4b, while the VAF membranes have much larger and dry state-dependent FWHMs. The FWHMs of rod-coated membranes are ≈50% smaller than those of VAF membranes, further confirming the improved stack order in the coated membranes via shear force. [29] Additionally, the Meyer rod-coated membranes also have smaller interlayer distances compared to the VAF ones, as indicated in Figure 4c. It is worth noting that rod-coating produced membranes with stable interlayer distances; even when the membranes are in a wet state, the distance shows negligible variation during long-time drying. However, the VAF MXene membranes have larger interlayer distances sensitive to their drying state, with shrunken distance over longer drying time and stabilizing after 12 h drying. The above results demonstrate that Meyer rod-coating is superior in fabricating MXene membranes with ordered and smaller interlayer distances; these nanochannels are desired for molecule and ion separation. [4] The mechanism for loose stack lamellar structures of VAF membranes is that MXene nanoflakes are free to deposit on the polymer substrates during filtration; therefore, there is no control on the stack of nanoflakes, as indicated in Figure 4d. However, Meyer rod-coating induces shear force to facilitate the ordered assembly of the MXene nanoflakes producing ordered and tight laminar structures. The AFM images (Figure 4e) confirmed that the VAF membrane with larger roughness showing a R a of 33.6 nm due to the random and loose stack of Ti 3 C 2 T x MXene nanoflakes, while the surface of the rod-coated membrane is more even, having a smaller Ra as indicated in Figure 4f. The shear force induced by rod-coating also helps the MXene nanoflakes tightly attach to the PES substrate surface. The coated membranes showed integrality after soaking in solutions with a wide pH range from 1 to 13 for 48 h ( Figure S7, Supporting Information). The MXene membranes showed no detaching from the substrate over 48 h soaking, regardless of the drying time of the coated membranes. The XRD results also confirmed the stable interlayer distances, particularly in acid solution ( Figure S7, Supporting Information). In addition, we compared the weight change of rod-coated MXene membranes before and after soaking in deionized water for 48 h (Table S2, Supporting Information). Under different drying time scales, the percentage increase of membrane mass before and after soaking is <6 wt%, because the orderly and compact rod-coated membrane can effectively restrain the swelling.
The Meyer rod-coated membranes were expected to show superior performance in molecule separation and ion sieving owing to their ordered and narrow interlayer distance. It is highly desired for the laminar membranes having both large permeabil-ity and high rejection rate. [4,30] The drying time-dependent permeability is shown in Figure 5a, which is similar to the water permeability of d-MXene membranes assembled by VAF. [30] The permeability decreases as the membranes become drier, which is caused by the loss of interlayer water. Intercalated and confined water molecules interact with oxygen-containing functional groups, play a critical role as spacers and facilitate the transfer of water. [31] Therefore, the water permeability of Ti 3 C 2 T x MXene membranes can be controlled by drying pretreatment. As shown in Figure 5a, a rod-coated 300 nm thick MXene membrane dried for 0.5 h has a large permeability of ≈60.0 L m −2 h −1 bar −1 . When the drying time increased to 12 h, the permeability decreased to ≈6.5 L m −2 h −1 bar −1 . It is worth noting that the permeability is much larger than those of traditional nanofiltration membranes. [5] As mimic wastewater containing 5 mg L −1 methylene blue (MB) was used for treatment, the permeability showed a decrease due to blockage of entrance among the stack nanoflakes. [32] The drying-dependent permeability remained for treating dye-containing wastewater. However, permeability ranging from 3.6 to 43.2 L m −2 h −1 bar −1 can still be maintained, indicating the impressive treatment capability of rod-coated Ti 3 C 2 T x MXene membranes for dye-containing wastewater.
The rejection rate of rod-coated MXene membranes showed drying-independent performance, as indicated in Figure 5b. All the tested membranes demonstrated a rejection rate of 99.95% for MB, regardless of the drying conditions, indicating the impressive dye separation performance of the MXene membranes. The rejection rate of rod-coated Ti 3 C 2 T x MXene membranes also showed no dependence at MB concentration ranging from 2.5 to 10.0 mg L −1 , as indicated in Figure 5c. However, the permeability decreased as the dye concentration increased due to more MB molecules blocked at the entrance (For a detailed explanation, see Figure S8, Supporting Information, and its notes).
The steric effect is one of the main separation mechanisms for laminar membranes assembled from 2D nanoflakes. [32,33] The sizes of entrances are smaller than those of pollutant molecules; therefore, the pollutants are retained while water can easily go through the membranes. The tightly stacked nanoflakes formed entrances smaller than the dye molecules, thus, can separate them from wastewater. The rod-coated MXene membranes also demonstrated impressive ability in removing other dye molecules, such as rhodamine B (RhB) and bromocresol green (BG), from mimic wastewater, with a rejection rate of 98.4% and 96.7% (see Figure 5d). The rejection rate can be improved to 100% and 99.8% by lengthening the drying time to 12 h ( Figure S9, Supporting Information). Since MB, RhB, and BG carry positive, neutral, and negative, respectively, all can be removed by the steric effect regardless of their charges, confirming the versatile feature and promise of the rod-coated MXene membranes in removal dyes.
The rod-coated MXene membranes can also be used for ion removal. As shown in Figure 6a, a piece of the coated membrane with a thickness of 650 nm showed a rejection rate of 88.64% for Cu 2+ with a permeability of 3.47 L m −2 h −1 bar −1 . In contrast, a VAF membrane with identical loading demonstrated a higher permeability of 4.90 L m −2 h −1 bar −1 with a degraded rejection rate of 76.41%. The permeability of both rod-coated and VAF membranes can be dramatically increased to ≈25.50 L m −2 h −1 bar −1 by shortening the drying time, preventing the exces-sive removal of interlayer water. However, the ion rejection performance degraded to 55.37% and 48.67% for the rod-coated and the VAF membranes (see Figure S10, Supporting Information), respectively, indicating that some of the channels are too large to limit the transfer of Cu 2+ . Therefore, it is better to use membranes with longer drying times for ion separation at the cost of permeability since selectivity is believed to be more important for membrane separation. [34] As shown in Figure 6b, the rod-coated membranes with a thickness of 650 nm have shown an impressive rejection rate for divalent cations, such as Cu 2+ , Ni 2+ , and Mg 2+ . In contrast, the removal performance for the most common cation, Na + , is just 24.60%, making the as-made membranes ideal for separating cations with high added value from Na + . Based on the ratio of the rejection rates, the selection for Mg 2+ to Na + is 2.79, which is superior to most of the reported nanofiltration membranes (Table S3, Supporting Information). Nanofiltration tests for mixed MB/Na + separation using rod-coated MXene membranes dried for various times were conducted to evaluate the separation performance through the membrane when dyes and ions coexist ( Figure S11, Supporting Information). The as-made membrane, after drying for 0.5 h, demonstrated a rejection rate of 97.52% for MB, and 12.81% for Na + , with a permeability of 53.26 L m −2 h −1 bar −1 . Based on the ratio of the rejection rates, the separation selection for MB to Na + is 7.61.
The ion sieving performance was evaluated using our previously reported method. [9] The ion permeation rates of Na + , K + , Li + , Mg 2+ , and Ca 2+ were measured in a homemade U-shaped device (Figure 6c). The number of permeated cations was evaluated by measuring the ionic conductivity of the permeate side, as reported in our previous work. [9] The concentration of permeated cations linearly increases with testing time, as shown in Figure 6d. The permeability of cations decreases in the row of Li + > K + > Na + > Mg 2+ > Ca 2+ , confirming that the rod-coated MXene membrane can sieve cations with similar sizes. The permeation rates through rod-coated MXene membranes show the cation type dependency as indicated in Figure 6e. Monovalent cations, such as Na + and K + , show faster permeation rates than divalent cations (Mg 2+ and Ca 2+ ), suggesting the role of cations' charge density and sizes. [9] Despite a larger hydration radius than those of K + and Na + , Li + demonstrates the largest permeation rate of 4.66 mol h −1 m −2 , which is 2.17 times faster than K + and 2.85 times faster than that of Na + , suggesting that the rod-coated MXene membrane has potential for precise sieving of monovalent cations. The permeation rate of Li + is much larger than those of divalent cations, which is 7.14 times faster than Mg 2+ , and 11.95 times faster than that of Ca 2+ , suggesting the membrane is the potential for extracting Li + from brine with high contents of Mg 2+ and Ca 2+ . [35,36] The Li + selectivity over other mono-/divalent cations was plotted in Figure 6f. The selectivity shows the charge dependence, which is much larger for divalent cations compared to separate monovalent ones. It was confirmed that cation permeation through MXene membranes and other 2D materials lamellar membranes is significantly affected by the cation intercalation and controlled by the size and charge of the cations. [37,38] The intercalation of larger divalent cations is kinetically hindered due to the narrow nanochannels and electrostatic interaction between cations and nanoflakes. [9] This work just describes new MXene membranes fabricated by the scalable rod-coating method, which shows the potential in dye removal and cation sieving. The separation performance can be further enhanced by optimizing membrane compositions and structures and adding electrical fields. [39]

Conclusions
To conclude, we have demonstrated that MXene membranes can be fabricated using high-concentration MXene dispersion in a continuous roll-to-roll way by Meyer rod-coating. MXene dispersion with a concentration as high as 30 mg mL −1 was first produced. Additionally, the concentration-dependent surface tension, viscosity, spreading ability, and rheological behaviors enabled the alignment of MXene nanoflakes under the shear force caused by rod-coating, producing highly ordered nanoflakes along the rod-coated membrane plane. The membrane thickness can be easily tuned from several hundred nanometers to several micrometers by repeating rod-coating. The highly oriented nanoflakes lead to narrower and uniform nanochannels in the rod-coated MXene membranes than in VAF membranes. The nanochannels in rod-coated membranes show dry-state independency and impressive stability. The thin and defect-free rodcoated membranes showed outstanding dye removal and monovalence/divalent cation sieving performance owing to the narrow, www.advancedsciencenews.com www.advmatinterfaces.de uniform, and stable nanochannels. The proposed rod-coating method can also be used for fabricating MXene/MWCNT and MXene/PVA composite membranes in a roll-to-roll way using hybrid inks containing highly concentrated MXene. The approach is poised to play a critical role in making MXene membranes in a continuous and up-scaling way.

Experimental Section
Exfoliation of Ti 3 C 2 T x and Ti 2 CT x MXene: In a typical run, LiF (2.88 g) was dissolved in HCl (36 mL, 9 mol L −1 ). Ti 3 AlC 2 powders (1.80 g) were slowly added into the LiF-contained acid, and the mixture was magnetically stirred at 45°C for 48 h to produce multilayered Ti 3 C 2 T x MXene (labeled as Ml-MXene). The Ml-MXene was separated by centrifugation and washed six times using deionized water till the pH reached ≈6.5. Ml-MXene was subjected to bath ultra-sonication for 5 min and tip ultra-sonication for one h under flowing N 2 in the ice bath. The dispersion containing delaminated MXene (labeled as d-MXene) was centrifuged at 3500 rpm for 0.5 h to remove the large unexfoliated Ml-MXene, and the concentrated d-MXene solution was obtained. The concentration of the obtained d-MXene dispersion was measured by weighting the d-MXene content via filter separation. The typical dispersion concentration is ≈30 mg mL −1 ; the d-MXene concentration can be easily controlled by dilution for the desired experiment. Ti 2 CT x -based MXene dispersion was prepared according to our previous work, [26] the typical concentration is ≈15 mg mL −1 .

Fabrication d-MXene Membranes in a Roll-to-Roll Manner by Meyer Rod-Coating:
The details of the automatic coater (TBJ-X3-XB, manufactured by Shandong Zhongyi Instrument Co. Ltd.) were presented in Movie S1 in the Supporting Information. The details of the Meyer rod (D10-OSP-d, d corresponds to the groove depth of different Meyer rods manufactured by Japan OSG Co. Ltd.) were presented in Scheme S1 in the Supporting Information. The as-made high-concentration Ti 3 C 2 T x d-MXene dispersion was diluted to 27.5 mg mL −1 to fabricate MXene membranes. The MXene membranes with nominal loadings of 0.15, 0.3, and 0.6 mg cm −2 were prepared. Take the MXene membrane with a loading of 0.15 mg cm −2 for example, the amount of d-MXene dispersion required was calculated according to the coated areas. PES membranes were used as polymer substrates for MXene membrane depositing. The coating was performed at a Meyer rod moving speed of 2 mm s −1 . The water was immediately evaporated after coating the d-MXene dispersion on PES, leaving a dark and uniform MXene membrane. The prepared membranes were naturally dried at room temperature for different times (0.5 h, 1 h, 1.5 h, 4 h, 6 h, and 12 h) before they were used for desired tests to study the drying-dependent structures and separation performance. To obtain MXene membranes with larger loading, multiple coating cycles were conducted by repeating the procedure mentioned above. During the multiple coating, a drying time of 2 min was used after each coating cycle to avoid the detaching of previously coating MXene layers.
Ti 2 CT x MXene membranes were fabricated using d-Ti 2 CT x dispersion with a concentration of 15 mg mL −1 . The Meyer rod (OSP-03, with a groove depth of 3 μm) was used for coating with a coating speed of 2 mm s −1 and PES membrane as the substrate.
T 3 C 2 T x MXene/MWCNTs and T 3 C 2 T x MXene/PVA composite membranes were produced via coating using the desired inks. For making T 3 C 2 T x MXene/MWCNTs membranes, MWCNTs were dispersed by tipsonication for 10 min in the T 3 C 2 T x based d-MXene dispersion with a concentration of 20 mg mL −1 , the mass ratio of MWCNT/MXene was kept at 5:95. The Meyer rod (OSP-03, with a groove depth of 3 μm) was used for coating the composite membranes with a coating speed of 2 mm s −1 and PES membrane as the substrate. For making T 3 C 2 T x MXene/PVA mem-branes, PVA solution (50 mg mL −1 ) was mixed with d-MXene dispersion with a concentration of 20 mg mL −1 under sonication for 8 min, and the mass ratio of PVA/MXene was 5:95. The Meyer rod (OSP-06, with a groove depth of 6 μm) was used for coating the composite membranes with a coating speed of 5 mm s −1 and PES membrane as the substrate.
Fabrication of d-MXene Membranes by VAF: The as-made highconcentration Ti 3 C 2 T x d-MXene dispersion was diluted to 2.5 mg mL −1 , and the diluted dispersion was used to fabricate d-MXene membranes by VAF using the PES substrate. In a typical operation, a certain volume of d-MXene dispersion was filtered and deposited on the porous PES membrane with d-MXene loadings of 0.15, 0.3, and 0.6 mg cm −2 by VAF.
Water Flux Test: The loading and drying condition-dependent water flux was studied for the as-made Ti 3 C 2 T x d-MXene membranes. The MXene membranes with the loading of 0.15, 0.3, and 0.6 mg cm −2 were used for the tests. In a typical run, the fresh MXene membranes were made via rod-coating. There was no visible water soon after the membranes were produced. Then the membranes were naturally dried for different periods. After drying, deionized water (10 mL) was filtered through an effective filtration area of 2.68 cm 2 , the pressure was 0.8 bar, and the completion time was recorded to calculate the water flux. The drying condition-dependent performance was studied for membranes dried for 0.5, 1, 1.5, 4, 6, and 12 hours. The water flux was also tested by the above-mentioned procedure using the MB-contained simulative wastewater (5 mg mL −1 ) as the feeding solution.
Nanofiltration Performance of the MXene Membranes: Nanofiltration tests were performed using a customized vacuum filtration setup ( Figure S1, Supporting Information). The rejection rate of dyes and salts was used to evaluate the separation performance of the rod-coated and VAF MXene membranes. The permeation dye solution (6 mL) and the permeation salt solution (10 mL) were used to analyze the rejection rate and flux through the as-made membranes. The tested dyes include positively charged MB, electrically neutral RhB, and negatively charged BG. Ultraviolet visible absorption spectrometer was used to detect the absorbance of the dye solution before and after filtration for quantitative analysis. The concentration of the above dyes used in the experiment ranges from 2.5 to 10 mg L −1 . The tested salt solutions include CuCl 2 , NiCl 2 , MgCl 2 , and NaCl. The concentration of all salt solutions used in the experiment was 10 mg L −1 . Inductively coupled plasma-optical emission spectrometry was used to detect cation concentrations. The permeance (J, L m −2 h −1 bar −1 ) and rejection rate (R, %) were calculated using the following equations: where Vp (L) is the volume of the permeate solution, t (h) is the filtration time, A (m 2 ) is the effective area of the membrane, and ΔP is the applied pressure (0.8 bar in this work). C f and C p are the concentrations (mg L −1 ) of dyes (or salts) in feed and permeation solutions, respectively. Ion Permeation Performance of the Meyer Rod-Coated MXene Membranes: Ion permeation experiments were performed using a homemade U-shaped device. The tested salt solutions included NaCl, KCl, LiCl, MgCl 2 , and CaCl 2 . The concentration of all the salt solutions used in the ion permeation test is 1 M. Taking the permeation test of NaCl as an example, for the rod-coated MXene membrane, NaCl aqueous solution (100 mL) and deionized water (100 mL) were injected into the feed side and the permeation side, respectively. Both the feed and the permeation sides were magnetically stirred to minimize concentration polarization. The conductivity on the permeation side was measured by a conductivity meter (ET915, eDAQ TECH, Australia) and recorded as a function of permeation time. The measured ionic conductivity can be converted into salt concentration by formula (3): Adv. Mater. Interfaces 2023, 10,2300301 where C is the electrolyte concentration, is the measured conductivity, and Λm is the molar conductivity. Ion permeability (Ji) can be calculated using Equation (4): where V is the effective volume of the solution on the permeation side, C is the calculated concentration on the permeation side, A is the effective membrane area (2.68 cm 2 in this work), and t is the test time.
Stability Test of the MEYER rod-Coated MXene Membranes in Different Solutions: Ti 3 C 2 T x d-MXene membranes with a loading of 0.15 mg cm −2 on PES supports were used for the stability test. Briefly, to study the effect of drying on the liquid phase stability of the membranes, several slices of membranes (with a size of 1 cm × 2 cm) were immersed for 48 h in aqueous solvents including DI water, acidic solution (0.1 M HCI, pH = 1) and alkaline solution (0.1 M NaOH, pH = 13) after drying for different periods. After immersion, all the membranes were dried in air for 3 min before the XRD analysis.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.