pH-dependent water permeability switches in MoS2 membranes and their memory
文章出处：C. Y. Hu, A. Achari, P. Rowe, H. Xiao, S. Suran, Z. Li, K. Huang, C. Chi, C. T. Cherian, V. Sreepal, P. D. Bentley, A. Pratt, N. Zhang, K. S. Novoselov, A. Michaelides, R. R. Nair. pH-dependent water permeability switching and its memory in MoS2 membranes. Nature 2023, 616, 719-723.
Intelligent transport of molecular species across different barriers is essential for a variety of biological functions and is achieved through the unique properties of biofilms. The two essential characteristics of intelligent transportation are: (1) the ability to adapt to different external and internal conditions; (2) Memorize previous states. In biological systems, the most common manifestation of this intelligence is retardation. Although many advances have been made in smart membranes over the past few decades, making a synthetic membrane with stable molecular transport lag behavior remains a challenge. Here, the authors demonstrate memory effects and stimulus-regulated molecular transport that respond to external pH through an intelligent, phase-transition MoS2 membrane. The authors show that water and ions passing through the 1T' MoS2 membrane follow a pH-dependent lag, and its permeation rate can switch by orders of magnitude. The authors determined that this phenomenon is specific to the 1T' phase of MoS2 due to the presence of surface charge and surface exchange ions. The authors further demonstrate the potential application of this phenomenon in autonomic wound infection monitoring and pH-dependent nanofiltration. The authors' work deepens the understanding of nanoscale water transport mechanisms, opening a pathway for the development of smart membranes.
Membranes with the penetration of lagging molecules in response to external stimuli such as pH can find use in many applications, such as neuromorphology, biomedical, and blast research, where pH-based memory can be adapted to mimic the function of nerve cells. Smart membranes can exist in multiple states (permeable and non-permeable) under the same external stimulus (e.g. pH), favoring the filtration of certain biomolecules, such as proteins or DNA with narrow pH stability. One possible direction in the manufacture of such membranes is the use of phase change materials, which can change their structure depending on external conditions.
Two-dimensional transition metal disulfides (MoS2) are a strong candidate because they can exist in several structural phases (and can switch controllably between several structural phases) (Figure 1a-1c). The most stable polytype of bulk phase MoS2 is the hexagonal 2H phase. MoS2 also exists in a variety of metastable triangle forms such as 1T, 1T', and 1T", characterized by a series of Peierls distortions to form different superstructures. In these polytypes, 1T′ is the energy-preferential metastable phase with an excess negative charge where Mo atoms aggregate into jagged chains. Recently, the potential of transition metal disulfide membranes in water filtration applications has been studied, and rapid transport of water through primitive and functionalized MoS2 membranes has been observed. However, the water permeability reported in the literature has changed by orders of magnitude, inconsistent with the expected water permeability of laminar flow membranes. This may be due to the poor layered structure of the membrane, as seen by cross-sectional electron microscopy, leading to the formation of frame defects. These defects can bypass water transport between MoS2 layers, shortening the molecular transport length, resulting in higher permeability. In order to take advantage of the special surface properties of phase change MoS2 to fabricate smart membranes, it is necessary to ensure the transport of molecules between the MoS2 layers by fabricating a membrane with a high-quality laminar flow structure. Here, the authors report a pH-dependent lag of water passing through the layered MoS2 membrane.
The author's MoS2 membrane (Figure 1d) was prepared by Lilification of bulk phase MoS2 powder (2H phase), followed by stripping and vacuum filtration of the previously reported dispersion. To obtain a membrane with a high laminar flow, the authors ensure that the multilayer MoS2 is removed from the dispersion and that the pH of the dispersion is neutral during membrane fabrication. Atomic force microscopy (AFM) studies have shown that the dispersion consists mainly of a monolayer of MoS2 sheets with a thickness of approximately 1 nm (Figure 1d). The authors know that Liification leads to the stripping of 2H MoS2 resulting in a large number of electrons doped into MoS2, resulting in a phase transition from 2H to 1T'. Thus, X-ray photoelectron spectroscopy (XPS) of the MoS2 membrane prepared by the authors indicates that the primary phase is 1T' (56%), consistent with previous reports. Since 1T′ is the main phase for the preparation of the membrane, the authors refer to it as the 1T′ MoS2 membrane. These membranes can be converted back to the 2H phase after roasting at 300 oC in an inert atmosphere.
The water vapor permeability of 1T' and 2H MoS2 membranes was studied by weight loss measurement of water-filled metal containers sealed with MoS2 membranes of different thicknesses. The authors observed significant weight loss in the 1T' MoS2 membrane, but not in the 2H MoS2 membrane (Figure 1e). In addition, increasing the proportion of the 2H phase in the 1T' MoS2 membrane by controlled roasting results in a monotonic decrease in water permeability: when the 2H ratio exceeds 90%, the water permeability drops below the detection limit of the author's experiment (Figure 1e). To rule out the effect of membrane drying on phase change permeability properties, the authors performed additional experiments to examine the water permeability of vacuum-dried MoS2 membranes. The authors found that although vacuum drying removed the intercalation water, the membrane quickly returned to a hydrated state after exposure to environmental conditions, allowing water to penetrate. In order to further explore the permeability of water through the MoS2 laminate membrane, the authors performed water vapor permeation experiments after treating the MoS2 membrane in acidic and alkaline aqueous solutions at different pHs. To do this, the membrane is first soaked in an acidic (HCl) or alkaline (LiOH) solution at the desired pH (1.0-12.1) for 10 hours, then washed and dried with deionized water. Unexpectedly, the authors found that the 1T' MoS2 membrane treated at pH = 1.0 (HCl at 0.1 M) prevented the permeation of water vapor (Figures 1e and 2a), while for the same membrane treated at pH = 12.1 (0.1 M LiOH), the water vapor permeability returned to the level of the prepared 1T' MoS2 membrane.
Figure 1
Figure 2a shows the change in water vapor permeability of 1T' and 2H MoS2 membranes with pH over a complete pH change cycle (from acidic to alkaline and back to acidic). Interestingly, when the pH of the immersion solution increased from 1.0 to 12.1, the membrane remained impermeable until the solution was treated with a pH of about 11.0, after which it quickly transitioned from an impermeable state to a permeable state. Similarly, when the pH of the immersion solution decreases from 12.1 to 1.0, the membrane remains water-permeable until it is treated with a solution with a pH of approximately 4.0, at which point the membrane rapidly switches from a permeable state to an impermeable state (Figure 2a). This lagging behavior of water vapor permeation conversion is unexpected. In contrast, 2H MoS2 membranes remain impermeable across the entire pH range. The water permeation of this pH-controlled 1T' MoS2 membrane is reversible and robust. Even though the pH alternately changed 10 times between 1.0 and 12.1, the membrane performance did not degrade (Figure 2a). In addition, in order to confirm the stability of the membrane properties, the authors performed water vapor penetration after 48 hours of immersion in the acid-treated and alkali-treated membranes in deionized water, and observed no change in their performance. The pH response behavior of 1T' MoS2 membranes is also confirmed when the membrane comes into contact with solutions of different pH, so changes in pH of the feed solution spontaneously alter the permeation rate. To understand the kinetics of pH response, the authors studied permeation as a function of pH exposure time and found that membrane response time can be adjusted by varying the thickness of the membrane. For example, a response time of 2 min is required for a film with a thickness of 500 nm, while a membrane with a thickness of 2 μm takes 30-60 min.
As with water vapor permeation, a lag in the pH response of liquid water permeation was observed in the 1T' MoS2 membrane in the pressure filtration system (Figure 2b). Within the sensitivity range of the author's 0.01ǀm-2·h-1·bar-1, acid-treated samples had no water penetration, while after treating the same membrane with an alkaline solution (pH = 12.1), the permeability increased to 1.1ǀm-2·h-1·bar-1. The resulting water permeability is inversely proportional to the thickness of the film. The pure water flux was also continuously tested for up to 40 hours, and no significant change in water flux was noted, confirming the stability of the membrane. In addition, the authors tested the permeability of 0.1 M LiOH (pH = 12.1) and 0.1 M HCl (pH = 1.0) solutions directly through the MoS2 membrane and found that the permeability can be repeatedly turned on or off depending on the pH of the feed solution.
To further demonstrate the potential of this unique, pH-sensitive 1T' MoS2 membrane, the authors conducted two additional experiments. First, the authors tested the ion permeation properties as a function of the pH of the feed and MoS2 membrane treatment. The authors' findings suggest that, similar to a water permeation switch, the penetration of ions through the 1T' MoS2 membrane can also be switched by the feed solution or the pH of the MoS2 membrane treatment (Figure 2c). Figure 2c further shows that ion permeation also exhibits hysteresis behavior, depending on the direction in which the pH of the feed solution changes. In addition, the authors investigated the pH-dependent nanofiltration performance of MoS2 membranes and demonstrated that their salt retention can be controlled by the pH treated by the membrane. Second, the authors illustrate its application by using a pH-responsive MoS2 membrane as a sensor for local wound infection detection. Wound healing is a complex process that requires constant monitoring. Clinically, if the pH of the wound is persistently above 8.0 and cannot be restored to 4.5-5.0 due to infection caused by harmful pathogens, it is called a chronic wound. By adjusting simulated wound exudate under different pH conditions, the authors simulated a wound scenario and showed that the MoS2 membrane could monitor wound infection by autonomously detecting pH, that is, only if there were signs of natural infection.
Figure 2
It is generally believed that the penetration of molecules through the two-dimensional laminar flow membrane is carried out through the interfaceted space, and the interlayer distance plays an important regulatory role in the penetration of molecules. As expected, the X-ray diffraction (XRD) plot of 1T' MoS2 shows two peaks of 2θ ≈ 7.7o (001 diffraction) and 14.6o (002 diffraction), corresponding to an interlayer spacing of 11.4 Å (effective free space δ approximately 5.2 Å). Conversely, in the case of 2H MoS2, the peak is located at 2θ ≈ 13.8o (002 diffraction), corresponding to a layer spacing of 6.4 Å (Figure 3a) (δ approximately 0.2 Å). The large interlaminar spacing in phase 1T′ is attributed to the presence of water between the two layers (due to moist ambient air), and X-ray analysis under vacuum further confirms the reduction in interlaminar distance. Interestingly, although the XRD profile of the 1T′ MoS2 membrane after acid treatment was similar to that of the 2H MoS2 membrane, the membrane returned to its original spectrum after treatment with an alkaline solution (Figure 3a). When the pH of the immersion solution gradually decreases from 12.0 to 1.0, the (001) diffraction of the 1T′ MoS2 membrane gradually decreases at 7.7o, and finally disappears at pH = 2.0. It is clear from these experiments that water permeation follows variations in layer spacing: 2H with smaller layer spacing and acid-treated 1T' MoS2 membranes (6.4 Å) hinder water penetration, while primitive 1T' MoS2 membranes (11.4 Å) treated with larger layer spacing allow water infiltration.
Based on the small layer spacing, it is not surprising that samples treated by 2H and acid have no water penetration. However, the reversibility of layer spacing and the pH-dependent lag of water through the 1T' MoS2 membrane are interesting. One possible explanation for the observed reversible changes in layer spacing and associated water permeation is the reversible phase transition of the MoS2 membrane from phase 1T′ to phase 2H; However, the authors' XPS analysis of acid-treated and alkali-treated samples ruled out this possibility (Figure 3b). The only minor difference in the Mo3d XPS spectra of acid-treated samples was a slight decrease in the 1T' ratio, from 56% to 46%. However, the authors did find that the S2p peak after acid treatment underwent a significant change (0.5 eV higher binding energy) compared to base-treated 1T' MoS2 membranes, indicating electron transfer from the electron-rich 1T' phase. To match the 1T′ phase ratio fitted by the Mo3d and S2p peaks, the authors had to include another peak of 161.8 eV (Figure 3b), which corresponds to the S-H peak (16%), which indicates that under acidic conditions the proton binds to the sulfur atom of MoS2, as previously predicted. To further explore the effect of pH on the atomic structure of 1T' MoS2, the authors performed X-ray absorption spectroscopy measurements and found that the coordination number of bonds changed significantly after acid treatment, indicating that the atomic structure underwent partial reversible changes after acid treatment.
Figure 3
The 1T′ phase of MoS2 is known to be relatively unstable, but can be stabilized by inserting alkali metal ions. Since the original 1T' MoS2 membrane under study was prepared by Lilification and the alkali treatment involved LiOH, the authors carefully analyzed the Li content in the studied 1T' MoS2 membrane using inductively coupled plasma mass spectrometry (ICP-MS). The Li content of the original 1T' membrane was as high as 0.3 mmol·g-1 MoS2, but after acid-base treatment, the concentration of Li+ ions changed significantly. Figure 4a shows the Li/Mo molar ratio of the MoS2 membrane as a function of pH treatment, exhibiting a hysteresis response similar to water permeability. The trend of water permeability and Li/Mo ratio with pH value is the same, indicating that there is a correlation between Li content in the membrane and water permeability. Figure 4b shows the relationship between water permeability and Li/Mo ratio, it can be seen that when Li/Mo is greater than 0.1, the water permeability is higher, and when Li/Mo is relatively low, the water permeability decreases sharply. In addition to Li+, the authors also investigated the effects of different intercalated cations on the water permeability of MoS2 membranes and found similar behavior in almost every case. The energy-dispersive X-ray spectra showed that cations were evenly distributed on the membrane surface at different pH values, which ruled out the possibility of permeation changes caused by cation localization.
The study of water vapor adsorption provides further understanding of the water permeation mechanism of MoS2 membrane. Water adsorption follows a typical type 2 adsorption isotherm, indicating that monolayer and multilayer water adsorption between MoS2 layers subsequently undergoes capillary condensation. The absorbency of samples treated at different pH at 99% relative humidity exhibits pH-related hysteresis, similar to water penetration and Li/Mo ratio. Based on the relationship between Li+ ion adsorption, water permeation and water adsorption, the authors propose that the penetration of water through MoS2 laminates is controlled by an adsorption diffusion mechanism, that is, water molecules are adsorbed on cations by hydration and then diffused between MoS2 layers.
To explain the unique pH response lag when water passes through the 1T' MoS2 membrane, the authors remind that 1T' MoS2 tablets have exchangeable cations due to excessive negative charges. During acid treatment of 1T' MoS2, these cations are exchanged with protons. However, this differs from other ion exchange mechanisms because protons strongly interact with S atoms on the surface of 1T' MoS2 to form S-H bonds, as demonstrated by the XPS data shown in Figure 3b. Therefore, the interlayer charge disappears, preventing water adsorption/embedding, resulting in a decrease in interlayer spacing and no water penetration between MoS2 layers. The authors investigated Li+ ion exchange as a function of different cation (K+ and H+) concentrations and found that it mainly occurs above about 0.1 mM ion concentrations. This corresponds to a pH of about 4.0 for the H+ ion and explains why the osmotic switch is at that pH. On the other hand, exposure of acid-treated samples to alkaline solutions due to the weakly acidic nature of S-H bonds results in bond breaking and cation re-adsorption into the MoS2 layer, which is reflected in water permeation. Due to the high pKa of the S-H bond (approximately pH 10.0), osmotic switching was observed only at high pH. This explains why protons in acid-treated samples cannot be exchanged without bases, while other adsorbed cations can be exchanged with any salt solution (e.g., Li+ can be exchanged with NaCl with Na+, but H+ cannot be exchanged with NaCl with Na+, only with NaOH). On this basis, the authors briefly describe the pH response lag mechanism of 1T' MoS2 membrane through two counteracting phenomena: protonation/hydrogen suppression and deprotonation/Li ation. This model explains the observed osmotic hysteresis and supports the change of Li+ ion adsorption hysteresis with pH. To further test this hypothesis, the authors hydrogenated the MoS2 membrane using hydrogen plasma. As expected, the hydrogenated membrane showed a blockage of water vapor. After re-exposing the membrane to the LiOH solution, the permeability returned to its original value.
To further demonstrate the unique water permeability characteristics of MoS2 membranes, the authors conducted a control experiment with a polar solvent ethanol, which is inserted between the MoS2 layers and has an adsorption effect comparable to water. Surprisingly, the authors found that although ethanol can be adsorbed onto the membrane, the permeation rate of ethanol is two orders of magnitude lower than that of water. This suggests that the observed infiltration of water through MoS2 is due to a unique, more rapid diffusion pathway between the MoS2 layers and the absence of significant microscopic defects in the membrane.
The authors understand the transport mechanism of MoS2 interlayer water by ab initio molecular dynamics (AIMD) simulations, and study the structure of water with low (0.02) and high (0.3) Li/Mo ratios between 2H, 1T' and 1T' MoS2, all at a fixed interlaminar distance of 1.0 nm (Figure 4c-4e). The authors' simulations show that pressurized water forms a well-defined bilayer structure regardless of polymorphism or Li concentration. The only difference between the two structures of 1T' MoS2 is the orientation of the water molecules and the relative position of the Li atoms in interlaminar space, as shown in Figures 4d and 4e. However, the authors found that these changes did not affect the dynamics of water in the interlaminar channels, suggesting that the mechanism of water osmotic through MoS2 is not controlled by Li concentration alone.
Figure 4
Classical MD (CMD) simulations reveal water ingress effects in different MoS2 structures. The authors' simulations found that hydration of the MoS2 bilayer containing Li+ ions allows water to enter the channel by extending the mesolayer to 11.5 Å (swelling), while in the absence of ions (2H and acid-treated MoS2), water molecules do not enter the channel and maintain the initial structure with a layer spacing of 6.4 Å. The authors propose that in the case of Li ions in the channel, this spontaneous swelling is due to osmotic water embedding. This explains why 2H membranes and acid-treated MoS2 membranes are impermeable, while 1T' is permeable.
In addition to MoS2, the authors experimentally tested the phase and pH dependence of water passing through the WS2 membrane. The authors found that, much like MoS2, only 1T polycrystalline WS2 is water-permeable and exhibits a similar hysteresis pH response.
In conclusion, the authors demonstrate that 1T' MoS2 laminate film is permeable, while 2H MoS2 laminate film is impermeable. The presence of charges/ions between the layers of 1T' MoS2 is conducive to the adsorption of the membrane by cation hydration. The absence of any charge between the 2H MoS2 layers explains its impermeability. The charge in the 1T' MoS2 layer is controlled by pH, resulting in a switching of water penetration: at an acidic pH, the S atoms undergo reversible hydrogenation, removing charge from the layer and cations from the interlayered space, so the membrane becomes impermeable; At alkaline pH, the adsorption cation concentration and water permeability are restored. The minimum H+ concentration required for protonation and the minimum OH- concentration required to break the S-H bond explain the hysteresis observed in water and ion osmosis. It was found that the rapid transport of water through the MoS2 membrane was independent of the nature of the MoS2 layer. Instead, this is due to the hydrogen bond network of closed water and the relatively weak water-MoS2 interaction. The authors' work demonstrates a pH-responsive smart membrane based on MoS2 laminates and also deepens the understanding of the permeation mechanism of water through two-dimensional material-based membranes. In multiple applications, from filtration to simulating biological processes, further efforts are needed to take advantage of attractive hysteresis molecular transport properties.