Liquid-liquid phase-separated commensal membrane gates mass transport in inorganic nanocells

Liquid-liquid phase-separated commensal membrane gates mass transport in inorganic nanocells | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Liquid-liquid phase-separated commensal membrane gates mass transport in inorganic nanocells Bin Chen, Yong Lu, Zhexuan Song, Zetan Cao, Linfeng Xu, Haoran Liu, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6660582/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Liquid-liquid phase separation (LLPS), often forming membraneless compartments, is prevalent in proteins and polymers, allowing the functionalities for biological activities and soft material engineering. Yet, direct visualization and manipulation of the full membrane-bound LLPS evolutions with high spatiotemporal resolution remains challenging and undefined. Inspired by dynamic modulation from biological membranes, we in-situ design inorganic nanocells from exciting nanoscale cinnabar with simultaneously forming flexible liquid-like membranes and dense Hg nanodroplets by the electron-responsive LLPS strategy. A full LLPS picture from birth to disappearance, including membrane-associated gating of mass transport either in single nanocell or across multiple nanocells is vividly revealed. Periodic reversible cross-feeding occurs among nanodroplets confined in the single nanocell, in contrast to the conventional Ostwald ripening or coalescence behavior. However, once the ionic balance of the membranes is disturbed by nanobubbles or electrolytes, the nanodroplets collapse. The released less dense species proceed cell-to-cell transport over long distances through nanochannels and are irreversibly crystallized into Hg(I/II) compounds. Ab initio molecular dynamics simulations suggest that the nanodroplet-membrane interface undergoes dynamic charge fluctuations, recognizing the unique membrane-bound LLPS in inorganic systems. The flexible membrane is stabilized through the balance between Hg atoms and ions, which can be destroyed by nanobubbles. Physical sciences/Materials science Physical sciences/Nanoscience and technology Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Photon and electron are the most critical environmental stimuli for substance development, serving as the energy source for regulating many aspects of the photon/electron-responsive processes, e.g., liquid-liquid phase separation (LLPS) behavior. LLPS is an intriguing phenomenon in which a mixture separates into two distinct liquid phases with differing component concentrations, usually forming membrane-free compartments (or photobodies) through photon exposure in soft biological and polymeric systems 1,2,3 . Despite the rare report of such behavior in inorganic materials, the in-situ atomic level observation of full LLPS evolution has been also lacking because of the low contrast as well as the beam sensitivity for the organic system under high-resolution microscopy characterization. On the other side, the membrane, when available, forms a functionally distinct configuration that separates two compartments with specific selectivity, enabling spatially regulated functions and offering unique applications in electrochemical processes 4 , catalysis 5,6 , filtration and purification 7 , sensors 8 and biological activities 9,10,11 . For example, membranes with typical thicknesses from several nanometers to microns have been shown to selectively regulate ion and mass exchange in electrochemical processes, thereby enhancing performance metrics such as battery capacity, electrocatalytic property, and cycling behavior 12,13 . Notably, the membrane, with its dynamical adjustment of configurations, mediates the inclusion or exclusion of specific ions for physiological activities 14 . Thus, a fresh perspective on complex membrane-bound LLPS processes, especially when confined in the nano-scaled domain, is essential for comprehensively understanding interfacial dynamics and tailoring the membrane-associated functionalities at the featured level of basic structural unit under environmental stimuli in nature. Here, we propose fabricating the dynamical membrane-nanodroplet system within inorganic nanocells through the electron-responsive reversible/irreversible LLPS strategy. As such, the gating behavior on mass transfer through such flexible membranes is simulated with the consideration of the nanoscale confinement effect. Our method of choice is in-situ liquid cell transmission electron microscopy (LCTEM), which provides a capability for directly observing and simultaneously manipulating dynamic LLPS evolution with atomic-scale resolution. Indeed, it has demonstrated significant applications in liquid-associated dynamics, including solution-based crystal growth 15,16,17 , etching behavior 18,19,20 , and electrochemical activities 21,22 in recent years. Using LCTEM, we choose HgS nanostructures embedded in a tris • HCl buffer solution as the studied system due to the facts that (i) the electron-beam-induced species in this buffer are well-known and also relatively gentle for etching the nanoparticles (NPs) 23 ; (ii) the offer of HgS as a matrix allows the realization of nanoscale confined cells through etching; and (iii) the full membrane-bound LLPS evolution picture – from the birth of liquid Hg nanodroplets with the associated membranes, and subsequent gating for mass transport, to the later disappearance – could be thereby achieved. The formation of Hg nanodroplets originates from exciting HgS, proceeding through a two-step nucleation process. The membranes gate the entrance and exit of the Hg species, mediating the reversible formation-to-dissipation of the liquid condensate for single nanodroplet and cross-feeding of these species among different nanodroplets within individual nanocells. Furthermore, long-range mass transport is achieved through the nanochannels connecting different nanocells, leading to the nano-confined irreversible crystallization of Hg-related compounds. The Hg nanodroplets either instantly dissipate or exhibit typical Ostwald-dominated coalescence behavior when the membrane is absent or disrupted by external factors such as electrolytes or nanobubbles. The above phenomena are further validated through cryo-electron microscopy (cryo-EM) and elemental characterizations, and supported by ab initio molecular dynamics simulations. Experimental implementation for in-situ visualization The HgS nanostructures were synthesized using a seed-mediated epitaxial growth technique 24,25 , with the structures of the seeds and resulting products detailed in Supplementary Figs. 1-2. Fig. 1a shows the in-situ visualization of the LLPS evolution from HgS matrix within an aqueous Tris•HCl solution sandwiched between two ultrathin carbon films. Further details can be found in Methods and Supplementary Fig. 3. Upon the electron beam (e-beam) excitation, free electrons are generated within the semiconducting HgS NPs, while radiolysis-induced redox active species, such as solvated electrons, hydroxyl radicals, and tris • + , are simultaneously formed in the aqueous solution 23 . Those reductive and oxidative species trigger distinct reactions in the HgS semiconductor NPs, enabling the direct observation of the membrane-bound LLPS evolution dynamics via LCTEM. The HgS in Fig. 1b displays a spiral arrangement of Hg/S atoms along the c axis 26 . The typical TEM image presents the bipyramidal shape of the as-synthesized HgS NPs (Fig.1c). The measured d-spacing of 3.18 Å corresponds to the (003) lattice plane of HgS viewed along the [ ] zone axis according to the diffraction pattern (Fig.1d). Fig. 1e shows a typical image of the as-prepared HgS nanobipyramids in LCTEM, with some nanobubbles as marked by the arrows. The energy dispersive X-ray (EDX) characterization indicates that the Hg and S elements are homogenously distributed in the nanobipyramid present in the high-angle annular dark-field (HAADF) image (Fig. 1f). The height profile of the HgS nanobipyramids is detected by atomic force microscopy, showing the thickness of about 20 nm (Fig. 1g), which is in the same order of magnitude of typical liquid-layer thickness (tens of nanometers) reported in the carbon film/graphene liquid cells 23,27,28 . When the liquid layer is significantly thicker (≈100 nm; more production of etching species upon the e-beam excitation) in the SiN liquid cells, it is noteworthy that almost only etching prevails for HgS (Supplementary Fig. 4). Competing to pure etching, on the other hand, reductive electrons are simultaneously generated in HgS semiconductor after exposure to the e-beam, leading to the reduction of Hg 2+ to Hg. This effect becomes more obvious in the carbon liquid cells with thinner liquid layer. Indeed, the liquid Hg nanodroplet was emerged during the LLPS process, which was evidenced from the elemental Hg/S mapping in Fig. 1h. This behavior contrasts with the pure etching- or growth-dominated processes observed in metallic nanostructures, as reported in the literature 16,18,19,20,29 . Reversible/irreversible LLPS in inorganic nanocell With the above experimental implementation in the carbon liquid cell, the typical LLPS evolution from a HgS nanobipyramid with the formation of membrane-bound Hg nanodroplets is shown in Fig. 2a. At the dose rate of ≈1040 e − Å 2 s -1 , the initial nanobipyramid (0 s) gradually changed into a hollow structure with the remaining outer shell (110 s). The detailed evolution process was illustrated in Supplementary Fig. 5. It was noticed that two distinct phases were formed: one with a less dense phase (nanocomplexes, less dark contrast) moved randomly inside the nanocell and was merged into a denser phase that took the shape of liquid nanodroplet (condensates, darker contrast) with an outer membrane at 115 s (marked by the solid and dotted circles, respectively; see Supplementary Movie 1 for vivid illustration). With the contribution from such nanocomplexes (indicated by the dotted arrows), the nanodroplet grew bigger to a diameter of ≈7 nm at 120 s and further to ≈11 nm at 138 s, and then gradually shrunk as the e-beam irradiation continued (e.g., the state at 338 s). Both the elemental identification (Fig. 1h) and the cryo-EM results (Supplementary Fig. 6 and Supplementary Table 1 verified that the as-formed nanodroplet was Hg. When the electron dose rate was increased to ≈2390 e − Å 2 s -1 , the nanodroplet expanded again in a short time (347 s) followed by the slow reduction in size, as detailed in Supplementary Fig. 7. Interestingly, the membrane flexibly regulated its shape but with constant thickness during the growth and shrinking processes of the nanodroplet. Such nanodroplet finally disappeared at 432 s. Note that the formation of the Hg nanodroplets and their evolution were induced by the e-beam since they kept invariant with the absence of the e-beam (Supplementary Fig. 8). The e-beam-induced temperature rise was normally in the range of several kelvins in the liquid cells 30,31 , which was not responsible for the above phenomena, as also confirmed by the in-situ heating experiment (Supplementary Fig. 9). Such general phenomena were observed in the graphene liquid cells as well (Supplementary Fig. 10). Since carbon liquid cells were relatively more accessible, they were used for the comprehensive demonstration in the following section. The time-dependent images with high resolution (Supplementary Movie 2) were further captured to unveil the formation of the hollow nanocells, as illustrated in Fig. 2b. As a demonstration, the fast Fourier transform (FFT) pattern in the inset of each image corresponds to the diffraction from the marked square area, simultaneously revealing the order-disorder transition during the structural evolution. Upon the e-beam illumination, the initial HgS nanobipyramid (0 s) with intact structure and diffraction pattern gradually turned into the hollow one with the increment of structural disordering. For example, several voids (marked by the dotted line) were formed at 20 s. These small ones were coalesced into the large one at 24 s, followed by the subsequent growth and spread (e.g., 108 s). Finally, a hollow nanocell was created with the amorphous nature evidenced by the ring diffraction pattern at 246 s. It has been reported that Hg species have the strong tendency to be adsorbed on HgS matrix 32,33 . Thus, the as-formed nanocell provides a confined domain to develop the evolution processes of liquid Hg species, offering a fundamental understanding of membrane-gated mass transport behavior with the input of nanoscale confinement. The nucleation and associated processes of the nanodroplet inside the nanocell are detailed in Fig. 2c (Supplementary Movie 3). For convenience, the moment of interest for investigation was regarded as the initial state (0 s) in the following section. A single Hg nanodroplet (≈9.2 nm) with the outer membrane was initially formed, which presented slight change during the first ≈4 s. In contrast, the swarming ‘transient’ nanocomplexes with the amorphous liquid-like feature were aggregated into the big one (marked by the solid arrow) after extended e-beam exposure for 0.9 s. Through the condensation of such liquid-like species, a small nucleus of Hg nanodroplet (darker contrast) nucleated inside the nanocomplex at the subsequent stage (1.0 s), which shared the similarity with the two-step nucleation mechanism 34,35,36 . The nucleus gradually grew bigger, from ≈2.2 nm in diameter at 1.2 s to that of ≈5.7 nm at 1.5 s. Without feeding additional species, the as-formed smaller Hg nanodroplet then dissipated into the nanocomplex at 3.4 s (marked by the dotted arrow), and eventually vanished at 3.8 s. As the time proceeded, the preexisting large nanodroplet became degraded, followed by the instantaneous release of Hg-associated species into the nanocell gated through the membrane, leading to the formation of new nanocomplexes (e.g., at 16.9 s indicated by the solid arrow). As such, a new Hg nanodroplet began to nucleate in the center of the nanocomplex again at 17.1 s, and grew to ≈6.5 nm in diameter at 17.5 s. Subsequently, the two nanodroplets were contracted and broken into the nanocomplexes at 20.1 s, which then disappeared at 20.5 s. Such reversible processes could be repeated, e.g., a new circle (nucleation-growth-dissipation-disappear) occurred again from 41.2 to 43.0 s. Overall, such assembly and disassembly of condensates (nanodroplet) caused by the LLPS process potentially act as electron sensors (tentatively termed ‘electrobody’ here), periodically forming and dissipating in response to the exposure of e-beam, which is analogous to the LLPS-induced reversible photobody observed in biological systems 37,38 . Fig. 2d shows the subsequent crystallization (irreversible LLPS) process after the rupture of Hg nanodroplet (Supplementary Movie 4). When the initial nanodroplet (0 s) touched the cell wall, we noticed that the other half fraction (contact region) of the membrane was broken and the core condensate began to shrink rapidly (e.g., 29.4 s). Without the membrane-related regulation, the core and the neighboring nanocomplexes were then coalesced (29.5 s, marked by the dotted arrows), and immediately became condensed to crystallize inside the nanocell (29.7 s). The crystallization phase was obviously observed at 32 s, which was recognized from the high-resolution TEM (HRTEM) image and the associated FFT analysis. The d-spacings of 3.02, 3.64 and 2.99 Å measured in HRTEM corresponded to the ( ), ( ) and ( ) lattice planes of the HgSO 4 · H 2 O crystal viewed along the [100] zone axis according to the FFT (see the detailed analysis in Supplementary Fig. 11 and Supplementary Table 2). On the other hand, the Hg(I) compound, e.g., the Hg 2 SO 4 , could also be crystallized in some cases (Supplementary Fig. 12 and Supplementary Table 3). It should be pointed out that such crystallized phases would not be reversed into the less dense nanocomplexes again, which shared the similarity with irreversible LLPS process in biology 39,40 . Upon the e-beam irradiation, it has been known that the oxidative tris • + radical was created in the tris • HCl solution, leading to the oxidation of selenium in the PbSe system 23 . Thus, the occurrence of high valence state of sulfur in this work was considered to be oxidized by tris • + . During the evolution, the typical relationship between the diameter of the core condensate and the thickness of the membrane is shown in Fig. 2e. Analogous to the photon-responsive photobodies, the liquid inorganic condensates display the sensitivity to the intensity of electrons. At a fixed dose rate (e.g., the 1040 e − Å 2 s -1 here), the core presented the real-time variation in size because of the exposure to electrons. Increasing the dose rate (≈2390 e − Å 2 s -1 at 338 s) increased the diameter of the core temporally. Such core region was finally vanished upon the continuous e-beam irradiation. However, the thickness of the membrane layer maintained constant (≈3 nm) regardless of the dissipation or expansion process of the core condensate. Fig. 2f shows the schematic diagram of the overall LLPS evolution from the HgS nanobipyramid in the solution upon the e-beam excitation. The intact nanobipyramid is gradually evolved into a hollow structure through the formation and coalescence of the voids, which is accompanied with the prenucleation of the less dense nanocomplexes randomly swarming inside the nanocell. The condensation of the nanoclusters results in the nucleation and subsequent growth of dense Hg nanodroplet enclosed by a layer of membrane. In response to the electrons, the membrane dynamically gates the entrance and exit of the Hg species, resulting in the periodic reversible cycling of the nucleation-growth-dissipation-disappearance processes of Hg nanodroplet. However, such nanodroplet is broken once touching the cell wall, resulting in the irreversible formation of new nanocrystals inside the nanocell. Regulation of mass transport by membranes Distinct from the LLPS-induced membrane-free photobody in biological system 41,42,43 , the membranes bounded in the inorganic condensates demonstrate the regulation of mass transport in the nanocells. Multiple nanodroplets, whether contained within a single nanocell or multiple nanocells, show a novel gated transport behavior (Fig. 3). When two nanodroplets with the outer membrane layers located in the single nanocell (Fig. 3a), the repeated behavior of mass transfer (cross-feeding) could be realized through the gating of the membranes (Supplementary Movie 5). By this, the big droplet (D1) presented the reduction in size while the initial small one (D2) became larger (e.g., at 12.4 s). Afterwards, the changes were reversed, namely, the D1 grew bigger, followed by the decrease of D2 at 37.8 s. Such intracellular interchange process between two compartments was repeated back and forth for many times (e.g., from 49.4 to 105.0 s), which was analogous to the breathing (in/out causing rhythmical expansion/shrinkage of the core condensate) behavior in biology. From the time-dependent volume curve (Fig. 3b), it is clear that the two Hg droplets showed the opposite trend in volume changes, and their total volumes were gradually reduced upon the continuous e-beam irradiation. Owing to the gating-induced regulation effect from the membranes, however, the above interchange behavior was markedly distinct from the typical Ostwald ripening (growth of large droplets at the expense of shrinking and annihilating the smaller ones) despite the close contact between two droplets. Surprisingly, we also noticed that the total volume would stay no obvious changes when the droplets located within a spatially enclosed nanocell (Fig. 3c and Supplementary Movie 6). D3 and D4 droplets were in close proximity to each other, which were several nanometers away from D5. The oscillation in volume (repeated switching between the large and small size) also occurred for the D3 and D4 droplets in most of the period (e.g., the states shown in the in-situ images). A crucial observation revealed that both of the two droplets could sometimes share the shrinking moments simultaneously, with the involvement of another D5 droplet (e.g., at 14.2 s). In this case, the D5 one became larger for a short period. Such behavior was highlighted in the time-dependent volume changes of these three droplets (Fig. 3d). It is clear that the volume reduction of the D3 and D4 droplets at certain time points (marked by the arrows) was almost compensated by the increment of the D5 one (see Supplementary Fig. 13 for detailed images), maintaining their total volumes nearly at a constant value. This observation demonstrates that the transport of species would be mediated by the membranes, occurring among the droplets regardless of their relative distances inside the nanocell. Besides the transfer among nanodroplets confined within one nanocell, the transport behavior could also occur across different nanocells through nanochannels. In order to directly visualize such nanoscale process within the camera field of view, we further conducted the in-situ experiments by using different morphological HgS nanocrystals (e.g., polyhedral and ellipsoidal shapes) with smaller sizes. Despite the distinct shapes, they also displayed the evolutions in each single cell (Supplementary Fig. 14) as that observed in the ones with the bipyramidal morphology. Favorably, the intercellular mass transfer would be conveniently monitored due to the small size. For the polyhedral sample (Fig. 3e), the transport was initiated from the cell 1 to the cell 2 at 5.7 s (guided by the dotted arrow; see Supplementary Movie 7 for vivid illustration). Afterwards, the nanocomplexes inside the cell 2 transferred to the adjacent cell 3 (10.3 s). Those nanocomplexes spread and broke the ionic balance of the membrane after meeting the preexisting Hg nanodroplet inside the cell 3, causing the dissipation and eventual disappearance of the nanodroplet (35.1 s). The species could further enter the neighboring cell 4 and again touch the membrane so that the corresponding nanodroplet also vanished rapidly (102.9 s). When the cell 2 and 5 were bridged through the formation of the nanochannel (see the enlarged view in Supplementary Fig. 15), the existing nanocomplexes in the cell 2 started to diffuse into the cell 5 (104.1 s). Akin to the “relay race”, the species in the cell 5 continued moving to the adjacent cells, e.g., to the cells 6 and 7 at 111.3 s, and then to the cell 8 at 143.0 s. The intercellular mass transfer process was also observed in the ellipsoidal sample, as displayed in Fig. 3f (Supplementary Movie 8). Through the nanochannels (Supplementary Fig. 16) created among the adjacent cells, the nanocomplexes transferred from the initial cell 1 to the next cell one by one, until they eventually stopped in the cell 6. Confined in the nanocells with smaller sizes, the swarming nanocomplexes possessed high probability to collide with the outer walls of the cells during the transport, which resulted in the crystallization process similar as that presented in Fig. 2d. The crystalline products were determined to be HgSO 4 and Hg 2 SO 4 (Supplementary Fig. 17, Supplementary Table 4 and 5). It should be pointed out that the nanochannel could not be formed when the distance between the cells was far away (e.g., larger than ~5 nm), leading to the termination of the mass transfer shown at the stage of 141.7 s. Fig. 3g schematically summarizes the mass transfer and associated crystallization processes under the confinement from the nanocells. In the isolated single cell, the quantity of the Hg-related complexes is limited, and gradually dissipates to the surrounding regions upon the e-beam irradiation. Such liquid-like complexes are composed of Hg atoms and ions, as revealed by the HRTEM results (crystalline Hg(I)/Hg(II) compounds shown above), the X-ray photoelectron spectroscopy identification (Supplementary Fig. 18) and also the simulations in the following section. Mediated by the membranes through the transition reaction of , the Hg-associated species can enter into the dense Hg nanodroplets (the equilibrium of the reaction is towards the left) or be released from those ones (the equilibrium is towards the right), resulting in the periodic reversible big-to-small transition of the nanodroplets. Under such equilibrium of the reaction, the thickness of the membranes is maintained constant during the evolution. The key characteristic of condensates undergoing reversible LLPS is that they are dynamic, i.e., they can rapidly exchange with the surrounding cellular nanocomplexes through the gating effect from the membranes. On the other hand, the nanocells would be connected with each other under the e-beam interaction when the nanocells are close enough. Subsequently, the connected bridge, acting as a transfer nanochannel of materials, facilitates the long-range transport of the species from nanocell to nanocell, until the termination if the distance between the nanocells is large so that the nanochannel cannot be formed. Once the membrane is destroyed, the Hg nanodroplets and released nanocomplexes touch the walls of the nanocells, which become irreversibly crystallized into the HgSO 4 or Hg 2 SO 4 nanocrystals depending on the relative quantity of the Hg + and Hg 2+ species. Evolution dynamics after disturbance of the membrane We also investigated the system with higher concentration of the tris • HCl buffer, and found similar evolution processes of the Hg nanodroplets (Supplementary Fig. 19). In contrast, other factors like the addition of electrolytes or the existence of nanobubbles were found to break the ionic balance of the membrane, thereby leading to the distinct irreversible processes observed below. To offer an insight into the LLPS field of biological activity as well, the common NaCl was chosen as the electrolyte. Fig. 4a shows the typical evolutions of the Hg nanodroplets with the addition of NaCl aqueous solution (Supplementary Movie 9). Upon the e-beam excitation, a number of Hg nanodroplets were also formed (e.g., at 7.1 s). Surprisingly, no membranes were observed in the outer surfaces of those nanodroplets. Without the modulation from the membranes, the classical Ostwald ripening or coalescence was mainly responsible for the growth of the nanodroplets (e.g., the period during 22.3-29.5 s, indicated by the dotted arrows; see also Supplementary Fig. 20), which was different from the reversible LLPS observed in Figs. 2 and 3 (the membrane acted as a “armor” to hinder the swallowing of small nanodroplets by the large ones). Meanwhile, some rectangular features (marked by the dotted rectangles) were seen in the surroundings at 29.5 s, and became more obvious at 88.5 s. Such features expanded rapidly in the field of view (e.g., 131.0 s), as presented vividly in Supplementary Movie 9. The HRTEM and associated FFT pattern identified that the rectangular nanocrystals were HgCl 2 , with the d-spacing of 2.93 Å corresponding to its (002) lattice plane viewed along the [ 0] zone axis (Supplementary Fig. 21 and Supplementary Table 6). Note that the framework of HgS was etched off, resulting in the crystallization extending to the surrounding regions (upon comparison to the confinement in the nanocells shown in Figs. 2 and 3). When the nanobubbles met the Hg nanodroplets in the solutions，the asymmetric rupture of the nanodroplets would occur, as shown in Fig. 4b (Supplementary Movie 10). Initially, a nanobubble (marked by the solid arrow) was approaching the spherical Hg nanodroplet. As the nanobubble touched the membrane, the ionic equilibrium inside the membrane was suddenly disrupted so that the Hg nanodroplet was asymmetrically contracted (0.2 s). Followed by such continued interaction, the nanodroplet gradually shrunk (0.7 s), and further became smaller with the emergence of an additional nanobubble (marked by the dotted arrow, 2.1 s). Later on, another nanobubble (marked by the triangle) gradually expanded and attacked other part of the droplet, accompanied by the continuous shrinkage of the nanodroplet (2.5~5.0 s). This process proceeded until the nanodroplet completely disappeared at 6.0 s. Further, we prepared the Hg-HgS heterostructures (Supplementary Fig. 22 and Supplementary Table 7) based on the synthesis procedure reported previously 44 , featuring the droplet behavior without the primitive membrane. Fig. 4c shows the evolution of such heterostructures in water liquid cell (Supplementary Movie 11). As seen, no membrane existed on the surface of the Hg droplet. Upon the e-beam excitation, the Hg droplet started to dissipate (refer to the contrast change from 6 to 25 s). In contrast to that observed in Figs. 2 and 3, the Hg droplet only presented the dissipation (e.g., 30-36 s) and eventually vanished, remaining the HgS nanorod section at 37 s. With the disturbance or absence of the membrane, the above results thereby verified that no gating of the mass transfer repeatedly occurred for the Hg-related species. Simulation of membrane-associated behavior To understand the microstructure and dynamic evolution of the Hg droplets system, we performed ab initio molecular dynamics (AIMD) simulations (see Supplementary Materials for details). For the primary components of the quasi-liquid membrane, the HAADF intensity profiles across the membrane were analyzed to evaluate the ratio between the H 2 O molecules and the Hg-related species 45 . Such ratio was estimated to be ~1.2:1 in the membrane (Supplementary Fig. 23 and Supplementary Note). Accordingly, a model (denoted as system A) containing 72 H 2 O molecules and 60 neutral Hg atoms was constructed. Fig. 4d shows the equilibrium state of the system, where the Hg atoms rapidly aggregate into clusters (refer to Supplementary Fig. 24 for detailed evolution). Based on the first coordination shell in the Hg-Hg radial distribution function (Supplementary Fig. 25), the Hg-Hg distance smaller than 3.3 Å was chosen as forming a cluster (indicated by the orange connecting lines in Fig. 4d). The time-averaged charge distribution of Hg atoms varies from -0.06 to 0.05 |e|, with an average charge close to zero per atom. To compare the aggregation behavior and stability, we then removed one hydrogen atom from each H 2 O molecule, resulting in a system (denoted as system B) containing 72 OH⁻ ions and 60 positively charged Hg ions. Simulations in Fig. 4e and Supplementary Fig 24 show that the Hg ions are almost uniformly dispersed (the time-averaged charge distribution ranges from 0.05 to 0.65 |e|). This distinct dynamic behavior is caused by the electrostatic repulsion between the positively charged mercury ions, indicating that the presence of Hg ions prevents the aggregation and further enhances the relative stability of the quasi-liquid membrane. The Hg core-membrane complex structure of the nanodroplets was further simulated with the construction of a large system C that contained 200 H 2 O molecules, 100 OH − ions, and 150 Hg-associated particles (either Hg atoms or Hg cations) in the aqueous solution. After 10 ps of simulation (Supplementary Fig. 26 for other snapshots and Supplementary Movie 12), the system evolved into a core-membrane structure, with neutral Hg atoms forming the core and Hg ions occupying the outer membrane (Fig. 4f). The charges of Hg species range from -0.08 to 0.84 |e| (Fig. 4g). Note that the Hg species with a charge value no greater than 0.1 |e| are classified as Hg atoms (labeled Hg a ), otherwise they are classified Hg ions (labeled Hg i ). As such, the plots of the two-dimensional density distributions (Fig. 4h) show that neutral Hg atoms (≈38%) locate in the core, while positively charged Hg ions (≈62%) occupy the outer membrane. Together, all Hg species exhibit a gradient from the core to the outside, transitioning from a dense to a more diffuse structure. The simulations agree well with the core-membrane structure observed in the HRTEM experiments (Fig. 2), affirming that the balance between Hg ions and atoms maintains the stability of the membrane. The dynamic equilibrium maintained in the core-membrane structure is realized through the charge transfer. As shown in Supplementary Fig. 27 and Supplementary Movie 13, the core and outer Hg species maintain atomic state and positively charged one, respectively. However, the interfacial species undergo the charge fluctuations between 0.0 and 0.3 |e|, implying they dynamically transit between the atomic and charged states at the interface. The radial distribution function g(r) was further analyzed to gain insights into the associated complex structure. The g(r) in Fig. 4i exhibits two sharp peaks. The peak at 3.3 Å is mainly attributed to the Hg a -Hg a pairs (partly from the Hg i -Hg a pairs), supporting the formation of dense clusters at the core of the nanodroplet. It is noteworthy that another peak at 2.7 Å originates from the Hg i -Hg i pairs, indicating the formation of Hg 2 2+ , in consistent with the previous reported distance of 2.52 Å in Hg 2 2+ . 46 Therefore, the simulations verify that the dynamic reaction of regulates the entrance or exit of Hg-associated species to the Hg nanodroplets, leading to the experimental observation of periodic big-to-small transition of the nanodroplets. To investigate the disturbance effect of nanobubbles on the membrane, we constructed another system D by randomly adding 20 H 2 molecules to the above system C. Simulations in Fig. 4j revealed that H 2 molecules diffused into the interior of the Hg droplet and disrupted its core-membrane structure (see Supplementary Fig. 28 and Supplementary Movie 14 for details), supporting the experimental observation that the nanobubbles caused the fragmentation of Hg nanodroplets (Fig. 4b). In contrast, when the system was composed of neutral Hg atoms and H 2 O molecules (system E), the presence of H 2 did not cause significant structural change of the nanodroplet (Supplementary Fig. 29 and Supplementary Movie 15). We proposed that the introduction of H 2 molecules could alter the charge distribution within the core-membrane structure through interactions with the surrounding ions (Supplementary Fig. 30), leading to the disruption of ionic balance that ultimately caused the disintegration of the nanodroplet. Discussion With the in-situ visualization under the atomic level resolution, we revealed a full LLPS evolution picture regarding the formation, growth/shrinkage, and collapse of the Hg nanodroplets, and the comparison of gating effect on the mass transport with/without the dynamical membranes (Fig. 4 k). For the membrane-bound nanodroplets within single nanocell, the repeated reversible cross-feeding among different nanodroplets occurs. In contrast, the long-range propagation across multiple nanocells through nanochannels is observed once the membranes are disturbed. The less dense nanocomplexes released from the membranes transit into the crystallized NPs (accompanied by the Ostwald ripening or coalescence behavior) during the transport, which become the irreversible process. The same strategy could be extended to other buffer systems for realizing the LLPS behavior, e.g., the glycine-HCl buffer solution, and the similar phenomena have been also observed (Supplementary Fig. 31). Although this study focuses on inorganic nanocells, the involvement of buffer solutions, NaCl and nanobubbles with the atomic-scale resolution could also provide valuable insights for biological system. As is well known, Hg was commonly used as an ingredient in immortal pills in ancient alchemy. Based on the direct visualization presented in this work, it is suggested that oxidative species and radicals in the human body, e.g., reactive oxygen species 47 , 48 , 49 , may facilitate the formation of a gating membrane encapsulating Hg, thereby retarding the further oxidation into Hg + or Hg 2+ species. This process could delay the onset of toxic reactions so that the immortal pills were not considered to be harmful in ancient years. Besides the well-known binding of those species with the sulfhydryl-containing proteins affecting cell functionalities 50 , our results further identify other distinct pathways dependent on local microenvironment. The existence of local high concentration of NaCl and/-or nanobubbles destroys the outer membrane of the elemental Hg droplets, resulting in the sudden asymmetrical collapse of the droplets and subsequent formation of ionic Hg species that rapidly propagate across cells through nanochannels and crystallize into Hg-related compounds (e.g., the toxic HgCl 2 that significantly affects human health). Moreover, the crystallized nanocrystals preferentially grow on the cell walls, indicating that they would accumulate on the blood vessels or other organs over months and years and finally lead to the chronic poisoning. This investigation offers a thorough understanding of the mass transport behavior in both nanoscale confined and unconfined regions mediated by the flexible membranes while their distinctive properties pave the path for developing multifunctional capabilities in other systems. Declarations Acknowledgements We thank Prof. Chang Yan for helpful discussion. This work is supported by the National Natural Science Foundation of China (Nos. 92061116, 22273004, and 224B2302) and the National Key Research and Development Program of China (2024YFA1509602). We also acknowledge the financial support from the Science and Technology Commission of Shanghai Municipality (22ZR1428400). Author contributions B. C. and Y. L. conceived the research. Y. L. prepared the samples and performed the in-situ experiments. Z. S. and J. X. conducted the ab initio molecular dynamics simulations. Z. C., L. X., and H. L. assisted in the synthesis of HgS nanostructures. Y. L., Z. S., J. X., and B. C. wrote the manuscript. All authors discussed the results and commented on the manuscript. Supplementary Information is available in the online version of the paper. References Feng Z et al (2024) Liquid–liquid phase separation of TZP promotes PPK-mediated phosphorylation of the phytochrome A photoreceptor. Nat Plants 10:798–814 Wang X, Lin C (2025) The two action mechanisms of plant cryptochromes. Trends Plant Sci Jiang B et al (2023) Light-induced LLPS of the CRY2/SPA1/FIO1 complex regulating mRNA methylation and chlorophyll homeostasis in Arabidopsis. Nat Plants 9:2042–2058 Tao L et al (2023) Reversible Switch in Charge Storage Enabled by Selective Ion Transport in Solid Electrolyte Interphase. J Am Chem Soc 145:16538–16547 Zeng T et al (2022) Hybrid bilayer membranes as platforms for biomimicry and catalysis. Nat Rev Chem 6:862–880 Tang BY, Bisbey RP, Lodaya KM, Toh WL, Surendranath Y (2023) Reaction environment impacts charge transfer but not chemical reaction steps in hydrogen evolution catalysis. Nat Catal 6:339–350 Cheng XQ et al (2023) Biodegradable electrospinning superhydrophilic nanofiber membranes for ultrafast oil-water separation. Sci Adv 9:eadh8195 Yang X et al (2024) Ultralow-pressure-driven polarization switching in ferroelectric membranes. Nat Commun 15:9281 Hannesschlaeger C, Horner A, Pohl P (2019) Intrinsic Membrane Permeability to Small Molecules. Chem Rev 119:5922–5953 Kozlov MM, Taraska JW (2022) Generation of nanoscopic membrane curvature for membrane trafficking. Nat Rev Mol Cell Biol 24:63–78 Simunovic M (2017) Biology and physics rendezvous at the membrane. Science 358:1265–1265 Fang W et al (2024) Durable CO2 conversion in the proton-exchange membrane system. Nature 626:86–91 Lagadec MF, Zahn R, Wood V (2018) Characterization and performance evaluation of lithium-ion battery separators. Nat Energy 4:16–25 Cao L et al (2017) Channel-facilitated molecule and ion transport across polymer composite membranes. Chem Soc Rev 46:6725–6745 Zheng HM et al (2009) Observation of Single Colloidal Platinum Nanocrystal Growth Trajectories. Science 324:1309–1312 Yuk JM et al (2012) High-Resolution EM of Colloidal Nanocrystal Growth Using Graphene Liquid Cells. Science 336:61–64 Ross FM (2015) Opportunities and challenges in liquid cell electron microscopy. Science 350:aaa9886 Ye X et al (2016) Single-particle mapping of nonequilibrium nanocrystal transformations. Science 354:874–877 Moreno-Hernandez IA, Crook MF, Ondry JC, Alivisatos AP (2021) Redox Mediated Control of Electrochemical Potential in Liquid Cell Electron Microscopy. J Am Chem Soc 143:12082–12089 Vigil SA, Moreno-Hernandez IA (2024) Dissolution Heterogeneity Observed in Anisotropic Ruthenium Dioxide Nanocrystals via Liquid-Phase Transmission Electron Microscopy. J Am Chem Soc 146:11133–11140 Zhang Q et al (2024) Atomic dynamics of electrified solid–liquid interfaces in liquid-cell TEM. Nature 630:643–647 Yoon A et al (2025) Revealing catalyst restructuring and composition during nitrate electroreduction through correlated operando microscopy and spectroscopy. Nat Mater Yan C et al (2022) Facet-selective etching trajectories of individual semiconductor nanocrystals. Sci Adv 8:eabq1700 Cao Z, He J, Liu Z, Zhang H, Chen B (2021) Chirality Affecting Reaction Dynamics of HgS Nanostructures Simultaneously Visualized in Real and Reciprocal Space. ACS Nano 15:16255–16265 Wang P-p, Yu S-J, Govorov AO, Ouyang M (2017) Cooperative expression of atomic chirality in inorganic nanostructures. Nat Commun 8:14312 Momma K, Izumi F (2011) VESTA 3for three-dimensional visualization of crystal, volumetric and morphology data. J Appl Crystallogr 44:1272–1276 Pedrazo-Tardajos A et al (2024) Direct visualization of ligands on gold nanoparticles in a liquid environment. Nat Chem 16:1278–1285 Crook MF et al (2023) EELS Studies of Cerium Electrolyte Reveal Substantial Solute Concentration Effects in Graphene Liquid Cells. J Am Chem Soc 145:6648–6657 Zheng H et al (2009) Observation of Single Colloidal Platinum Nanocrystal Growth Trajectories. Science 324:1309–1312 Liao HG et al (2014) Facet development during platinum nanocube growth. Science 345:916–919 Yang J et al (2019) Formation of two-dimensional transition metal oxide nanosheets with nanoparticles as intermediates. Nat Mater 18:970–976 Burkstaller JE, McCarty PL, Parks GA (1975) Oxidation of cinnabar by iron(III) in acid mine waters. Environ Sci Technol 9:676–678 Hsieh YH, Tokunaga S, Huang CP (1991) Some chemical reactions at the HgS(s)—water interface as affected by photoirradiation. Colloids Surf 53:257–274 Yang J et al (2019) Amorphous-Phase-Mediated Crystallization of Ni Nanocrystals Revealed by High-Resolution Liquid-Phase Electron Microscopy. J Am Chem Soc 141:763–768 Kim HJ et al (2022) Heterointerface Effect on Two-Step Nucleation Mechanism of Bi Particles. Nano Lett 22:3252–3259 Tan P, Xu N, Xu L (2014) Visualizing kinetic pathways of homogeneous nucleation in colloidal crystallization. Nat Phys 10:73–79 Ianiro A et al (2019) Liquid–liquid phase separation during amphiphilic self-assembly. Nat Chem 11:320–328 Harris R, Berman N, Lampel A (2025) Coacervates as enzymatic microreactors. Chem Soc Rev Kitahara R et al (2021) Pressure-Jump Kinetics of Liquid–Liquid Phase Separation: Comparison of Two Different Condensed Phases of the RNA-Binding Protein, Fused in Sarcoma. J Am Chem Soc 143:19697–19702 Shen Y et al (2020) Biomolecular condensates undergo a generic shear-mediated liquid-to-solid transition. Nat Nanotechnol 15:841–847 Lafontaine DLJ, Riback JA, Bascetin R, Brangwynne CP (2021) The nucleolus as a multiphase liquid condensate. Nat Rev Mol Cell Biol 22:165–182 Ruan K, Lin Y, Yang P, Zhou W (2023) Condensation matters. Nat Chem Biol 19:1180–1182 Sahin C, Leppert A, Landreh M (2023) Advances in mass spectrometry to unravel the structure and function of protein condensates. Nat Protoc 18:3653–3661 Wu L, Quan B, Liu Y, Song R, Tang Z (2011) One-Pot Synthesis of Liquid Hg/Solid β-HgS Metal – Semiconductor Heterostructures with Unique Electrical Properties. ACS Nano 5:2224–2230 Loh ND et al (2017) Multistep nucleation of nanocrystals in aqueous solution. Nat Chem 9:77–82 Rosdahl J, Persson I, Kloo L, Ståhl K (2004) On the solvation of the mercury(I) ion. A structural, vibration spectroscopic and quantum chemical study. Inorg Chim Acta 357:2624–2634 Schneider NM et al (2014) Electron–Water Interactions and Implications for Liquid Cell Electron Microscopy. J Phys Chem C 118:22373–22382 Ye M et al (2023) Revealing Dominant Oxidative Species in Reactive Oxygen Species-Driven Rapid Chemical Etching. Nano Lett 23:7319–7326 Sies H, Jones DP (2020) Reactive oxygen species (ROS) as pleiotropic physiological signalling agents. Nat Rev Mol Cell Biol 21:363–383 Ajsuvakova OP et al (2020) Sulfhydryl groups as targets of mercury toxicity. Coord Chem Rev 417:213343 Hauwiller MR et al (2019) Gold Nanocrystal Etching as a Means of Probing the Dynamic Chemical Environment in Graphene Liquid Cell Electron Microscopy. J Am Chem Soc 141:4428–4437 Kühne TD et al (2020) CP2K: An electronic structure and molecular dynamics software package - Quickstep: Efficient and accurate electronic structure calculations. J Chem Phys 152:194103 Perdew JP, Burke K, Ernzerhof M (1996) Generalized Gradient Approximation Made Simple. Phys Rev Lett 77:3865–3868 VandeVondele J, Hutter J (2007) Gaussian basis sets for accurate calculations on molecular systems in gas and condensed phases. J Chem Phys 127:114105 Grimme S, Ehrlich S, Goerigk L (2011) Effect of the damping function in dispersion corrected density functional theory. J Comput Chem 32:1456–1465 METHODS Synthesis of α-HgS seeds . The synthesis of α- HgS seeds and the associated bipyramids was based on the wet-chemical reaction route reported previously 24,25 . Basically, 64 mg of Hg(NO 3 ) 2 ·H 2 O and 10 mL of deionized water were mixed in a 25 mL glass vial, followed by the addition of 2 mL of D-penicillamine aqueous solution (0.09 M). The as-formed solution was under ultrasonic treatment for 1 min, and then 0.3 mL of NaOH aqueous solution (2 M) was added into the mixture with stirring. After that, 1 mL of thioacetamide solution (0.18 M) was rapidly injected into the above solution. The vial was then sealed and kept into the 38 ℃ water bath for 15 h under stirring. The HgS seeds were collected by adding isopropanol into the orange-colored solution, followed by centrifugation at 7000 rpm for 10 min. Finally, such as-formed seeds were dispersed in 10 mL of deionized water for later usage. Synthesis of α-HgS bipyramids. Before the epitaxial growth of α-HgS with bipyramid morphology, Hg precursor solution was prepared by mixing 62 mg of Hg(NO 3 ) 2 ·H 2 O and 1.8 mL of D-penicillamine aqueous solution (0.09 M) into 6 mL of deionized water, followed by ultrasonic treatment to obtain colorless solution. Meanwhile, S precursor solution was prepared by dissolving 41 mg of thioacetamide in 20 mL of deionized water under ultrasonic treatment. 0.3 mL of the colloidal solution containing α-HgS seeds and 5 mL of deionized water were added into a three-neck round-bottom flask, followed by the addition of 1 mL of D-penicillamine aqueous solution (0.09 M) under stirring. Then, the flask was transferred into the 38 ℃ water bath with the addition of 0.6 mL of NaOH aqueous solution (2 M) into the flask under stirring. Afterwards, the Hg precursor and S precursor were co-injected into the flask through a syringe pump under an injection rate of 1 mL/h for 2.5 h. After the reaction, the as-synthesized bipyramids were collected by adding isopropanol into the pale orange solution, followed by centrifugation at 7000 rpm for 10 min. The precipitated bipyramids were re-dispersed in 5 mL of deionized water for storage. Synthesis of α-HgS ellipsoids. 32 mg of Hg(NO 3 ) 2 ·H 2 O and 5 mL of deionized water were mixed in a 25 mL glass vial, followed by the addition of 1 mL of D-penicillamine aqueous solution (0.09 M). After the ultrasonic treatment for 1 min, 0.15 mL of NaOH aqueous solution (2 M) was added into the vial under stirring. 1 mL of thioacetamide solution (0.09 M) was then injected into above mixture solution quickly. The vial was sealed and located in the 47 ℃ water bath for 18 h under stirring. By adding isopropanol into the reaction solution, the as-formed ellipsoids were collected through the centrifugation at 7000 rpm for 10 min. Finally, the product was dispersed in 5 mL of deionized water. Preparation of α-HgS polyhedrons. 8 mg of Hg(NO 3 ) 2 ·H 2 O and 11.5 mL of deionized water were mixed in a 25 mL glass vial, with the further addition of 0.23 mL of D-penicillamine aqueous solution (0.1 M). The resulting solution was ultrasonically treated for 1 min, followed by the addition of 0.038 mL of NaOH aqueous solution (2 M) under stirring. After that, 0.23 mL of thioacetamide solution (0.1 M) was immediately injected into the above mixture solution. Then, the sealed vial was put into the 38 ℃ water bath under stirring for 15 h. The polyhedrons were collected by adding isopropanol into the reaction solution, followed by centrifugation at 7000 rpm for 10 min. The as-collected precipitates were finally dispersed in 5 mL of deionized water. Synthesis of Hg-HgS Heterostructures. 7.5 mL of Hg(ClO 4 ) 2 ·3H 2 O (0.01 M) was slowly added into 7.5 mL of cysteine (0.01 M) under stirring, followed by carefully adjusting the pH value of the mixture solution to 11.70 with the addition of NaOH (2 M). Afterwards, the sealed glass vial containing the above reaction solution was immediately kept into an oven with 37 ℃ for one week. During the reaction period, the color of the solution was changed from slight white to yellow green and then to black at the end. Finally, the precipitates of Hg-HgS heterostructures were collected after centrifuging and washing with the deionized water for several times. Liquid cell preparation. Three kinds of liquid cells (LCs), namely, the carbon LCs, graphene LCs and SiN x LCs were fabricated in this work. The carbon LCs were prepared by sandwiching the HgS nanoparticle (NP) colloidal solution with the addition of tris·HCl solution (0.01-0.02 M) between two ultrathin carbon films (thickness of ~10 nm). First, ~2 μL of the above solution was pipetted onto the carbon film-coated side of a 400-mesh gold grid, with the formation of a droplet locating nearly at the center of the grid. After that, another gold grid was quickly placed over this droplet with the carbon-coated side facing downward to encapsulate the droplet. The as-fabricated carbon LC was kept under ambient condition for 2 h before loading to the microscope. Liquid pockets were formed between the two ultrathin carbon films due to van der Waals forces, thereby allowing them for subsequent transmission electron microscopy (TEM) characterization. To study the buffer effect, other buffer solutions, e.g., the glycine-HCl was also used to replace tris·HCl for the LC preparation. Standard procedures were utilized to prepare the graphene LCs 51 . Briefly, the ultrathin carbon supported gold grids were placed onto the flat graphene (~3-5 layers) copper foil (~2 cm × 2 cm), followed by pipetting several drops of isopropanol onto the grids. These grids were attached with the graphene through evaporation of the isopropanol (kept in the ambient condition for ~2 h). The copper foil was etched by placing into the solution of sodium persulfate (1 g/10 mL) with the cooper side downward. Finally, these graphene coated grids were extracted from the etched solution, washed with deionized water to remove the etchant traces. Such graphene coated grids were then used to encapsulate the solution of interest for obtaining the graphene LCs (the following procedures were similar to the fabrication of carbon LCs). For the SiN x LCs, the sandwiched solution was the same as above. Electron transparent SiN x windows (~20 nm thick, 250 μm × 250 μm) supported by a 200-μm-thick silicon frame (diameter of ~3 mm) were used for encapsulating the solution (refer to fig. S3 for detailed illustration). One piece of such SiN x chips was deposited with a ~100-nm-thick layer of gold film as the spacer (the observation area of the electron transparent SiN x window was excluded). Before encapsulation, the SiN x chips were cleaned with plasma for ~10 s to effectively remove the surface contamination. After that, ~2 μL of the solution was pipetted onto the SiN x window of one chip with the gold film, followed by placing another chip onto the gold film to confine the thin liquid layer. Finally, the epoxy adhesive was used to seal the junction area between the two chips, and the liquid cell was realized after the adhesive was dried. To investigate the effect from the electrolytes, the typical NaCl (popularly existed in biological systems for offering insights of Hg droplets on human health as well) aqueous solution (0.02 M) was used as the sandwiched solution between two ultrathin carbon films. All the fabrication steps were the same as that shown in the carbon LCs. Height profile characterization. Briefly, ~20 μL of the diluted HgS colloidal solution was pipetted onto the clean monocrystalline silicon substrate (5 mm × 5 mm), which was kept in ambient condition for drying. The height profiles of the HgS NPs were characterized by atomic force microscopy (AFM, FastScan Bio, Bruker) with the maximum scan rate of 120 Hz and resolution in height of 0.03 nm. X-ray photoelectron spectroscopy analysis. About 0.5 μL of liquid Hg and ~1 mL of Tris·HCl solution (10 mM) were mixed together as the sandwiched solution for the preparation of carbon LCs. After encapsulating such solution between two copper grids with ultrathin carbon films, the gap between the grid edges was sealed carefully by using the glue. The as-fabricated LC samples were characterized by X-ray photoelectron spectroscopy (XPS, ESCALAB QXi, Thermo Scientific) to analyze the binding states of Hg. In-situ visualization of α-HgS nanoparticles evolution dynamics. Direct observation of the evolution dynamics of the Hg nanodroplets and associated membranes from the HgS NPs were in-situ performed by transmission electron microscopy (Talos F200X, Thermo Fisher Scientific) with an accelerating voltage of 200 kV. The electron dose rate of the electron beam (e-beam) was kept in a very low level (1~10 e − Å 2 s -1 ) during searching for the appropriate samples. Afterwards, the electron dose rate was increased to a desired level (100~3000 e − Å 2 s -1 in this work) for initiating the dynamical evolution of the samples. The dynamical processes in the LCs were then recorded by a digital camera. To reveal the elemental information during the evolution, the high-angle annular dark-field (HAADF) and energy dispersive X-ray spectroscopy (EDS) techniques in the scanning transmission electron microscopy (STEM) mode were utilized. In-situ heating characterization. A few drops (~2 μL) of the dispersed solution containing the HgS nanobipyramids was pipetted onto the in-situ heating chip (the observation windows were made of thin SiN x layers), followed by drying under the ambient environment. After drying, the chip was loaded onto the TEM sample holder for the in-situ heating characterization. During the experiments, the temperature of the chip was kept at 100 ℃ with the fluctuation of less than 0.1 ℃ through the control system. The morphologies and diffraction patterns of the HgS nanobipyramids before heating and at 100 ℃ were then captured for the comparison. Cryo-EM observation of the Hg nanodroplets. The LCs encapsulating the HgS nanobipyramids and 10 mM tris·HCl aqueous solution were loaded in the cryo-specimen holder. Initially, the Hg nanodroplets were created by the e-beam illumination at room temperature. Afterwards, the temperature of the sample was lowered to -100 ℃ through liquid nitrogen cooling (the melting point of Hg is -38.9 ℃). The observation procedure was the same as the above in-situ TEM characterization, and the emerging crystalline lattice of the freezing nanodroplets was captured for later analysis. Computational model building. In order to investigate the membrane structure and the behavior of Hg nanodroplets under different conditions, we constructed five different systems. First, to compare the aggregation behavior and stability of the membrane, we built the systems A and B. System A contained 72 H 2 O molecules and 60 neutral Hg atoms that were initially uniformly placed in a cubic box with a side length of 16.5 Å, while system B was constructed by removing a hydrogen atom from each H 2 O molecule in the above system, containing 72 OH − ions and 60 positively charged Hg ions. Further, to simulate the core-membrane structure, we built the system C that contained 200 H 2 O molecules, 100 OH − ions, and 150 Hg-associated species (either Hg atoms or cations) in a cubic box with a side length of 22.5 Å. In addition, the neutral Hg system related to system C was also constructed, which contained 300 H 2 O and 150 Hg atoms. Finally, to study the effect of H 2 on the core-membrane structure, we constructed a system D, which was based on the system C with additional insertion of 20 H 2 molecules. For comparison, another system E that contained 300 H 2 O molecules, 150 neutral Hg atoms, and 20 H 2 molecules was built. Computational setup. We performed ab initio molecular dynamics simulations (AIMD) as implemented in CP2K software 52 .The Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional was used with short-range double-ζ basis set DZVP-MOLOPT-SR-GTH 53,54 . Grimme’s dispersion correction with Becke-Johnson damping (D3-BJ) was employed to account for weak dispersion interactions 55 . The self-consistent field (SCF) cycles in the simulations were converged using the orbital transformation method. Simulations were carried out within the canonical (NVT) ensemble with a timestep of 0.5 fs. Each simulation ran for more than 10 ps and the last 5 ps of the trajectory was used for the detailed analysis. The temperature was set to 298.15 K using a velocity rescaling thermostat with a time. Additional Declarations There is NO Competing Interest. Supplementary Files SupplementaryMovie1.mp4 Supplementary Movie 1 SupplementaryMovie2.mp4 Supplementary Movie 2 SupplementaryMovie3.mp4 Supplementary Movie 3 SupplementaryMovie4.mp4 Supplementary Movie 4 SupplementaryMovie5.mp4 Supplementary Movie 5 SupplementaryMovie6.mp4 Supplementary Movie 6 SupplementaryMovie7.mp4 Supplementary Movie 7 SupplementaryMovie8.mp4 Supplementary Movie 8 SupplementaryMovie9.mp4 Supplementary Movie 9 SupplementaryMovie10.mp4 Supplementary Movie 10 SupplementaryMovie11.mp4 Supplementary Movie 11 SupplementaryMovie13.mp4 Supplementary Movie 13 SupplementaryMovie12.mp4 Supplementary Movie 12 SupplementaryMovie14.mp4 Supplementary Movie 14 SupplementaryMovie15.mp4 Supplementary Movie 15 SI.docx Supplementary Information Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6660582","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":467720233,"identity":"fd131bf1-8c42-41ff-852f-81d39d949916","order_by":0,"name":"Bin Chen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAtklEQVRIiWNgGAWjYDACCRBRwcAM5vAQr+UMAzMPaVoY26CqidLCP7v54YeP8+rY7SUSGB+8bWOQNydoyZ1jxpIztx1m5pFIYDac28ZguLOBgBYDiRw2Zt5tB0Ba2KR52xgSDA4Qo+XvnDqQFvbfxGthbGAG28JMlBawX3qOAf1y5mGz5JxzEoYbCGkBh9iPmrpk9vbkgx/elNnIE7QFBpKBsdPAAI1Z4oAd8UpHwSgYBaNgxAEAqUM2Wq8GTHwAAAAASUVORK5CYII=","orcid":"https://orcid.org/0000-0002-6832-906X","institution":"Shanghai Jiao Tong University","correspondingAuthor":true,"prefix":"","firstName":"Bin","middleName":"","lastName":"Chen","suffix":""},{"id":467720234,"identity":"329b5d76-fee4-40ca-962f-79196406de7b","order_by":1,"name":"Yong Lu","email":"","orcid":"","institution":"Shanghai Jiao Tong University","correspondingAuthor":false,"prefix":"","firstName":"Yong","middleName":"","lastName":"Lu","suffix":""},{"id":467720235,"identity":"45448ea6-1919-413f-aeb0-4b4614cd6035","order_by":2,"name":"Zhexuan Song","email":"","orcid":"","institution":"Beijing Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Zhexuan","middleName":"","lastName":"Song","suffix":""},{"id":467720236,"identity":"cc93a9c7-8893-4128-980e-8fcb3632dcca","order_by":3,"name":"Zetan Cao","email":"","orcid":"","institution":"Shanghai Jiao Tong University","correspondingAuthor":false,"prefix":"","firstName":"Zetan","middleName":"","lastName":"Cao","suffix":""},{"id":467720237,"identity":"0eb8eb16-adee-4c8d-888e-99ea47ff0dda","order_by":4,"name":"Linfeng Xu","email":"","orcid":"","institution":"Shanghai Jiao Tong University","correspondingAuthor":false,"prefix":"","firstName":"Linfeng","middleName":"","lastName":"Xu","suffix":""},{"id":467720238,"identity":"1a3bbf32-b30b-4345-85da-f74b6b5b0537","order_by":5,"name":"Haoran Liu","email":"","orcid":"","institution":"Shanghai Jiao Tong University","correspondingAuthor":false,"prefix":"","firstName":"Haoran","middleName":"","lastName":"Liu","suffix":""},{"id":467720239,"identity":"ecaa6701-04cc-4496-837b-fb27c4ab010d","order_by":6,"name":"Jing Xie","email":"","orcid":"","institution":"Beijing Institute of Technology","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Xie","suffix":""}],"badges":[],"createdAt":"2025-05-14 05:35:17","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6660582/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6660582/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":87808795,"identity":"9c4bc812-0efb-4fa6-b792-a06dc6a31d43","added_by":"auto","created_at":"2025-07-29 08:59:22","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":493455,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eStructural and elemental characterization of HgS semiconductor in liquid cell TEM. a\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eSchematic illustration of the evolution of HgS nanocrystals in the aqueous solution with the buffer addition. \u003cstrong\u003eb\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eAtomic arrangement\u003cstrong\u003e \u003c/strong\u003eof the α-HgS structure. \u003cstrong\u003ec\u003c/strong\u003e, Typical TEM image of the α-HgS nanobipyramid. \u003cstrong\u003ed\u003c/strong\u003e, High-resolution TEM image (HRTEM) and corresponding fast Fourier transform (FFT) pattern of HgS. The 3.18 Å spacing corresponds to the (003) crystal plane of α-HgS. \u003cstrong\u003ee\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eTEM image of the HgS nanobipyramids in the liquid cell with surrounding nanobubbles (indicated by the arrows). \u003cstrong\u003ef\u003c/strong\u003e, EDX mapping and spectrum of the HgS nanobipyramid in the liquid cell. The elements of Au and Cu in the spectrum came from the TEM grid. \u003cstrong\u003eg\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eTypical height profile of the HgS nanobipyramid. The height contour was obtained through the line scan indicated in the inset. \u003cstrong\u003eh\u003c/strong\u003e, HAADF and elemental mapping of the Hg nanodroplet that was originated from the exposure of HgS to the e-beam.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/88dd490fc76840ccaee32409.png"},{"id":87808798,"identity":"43045918-53cd-41f1-8176-6ec4ce3b6ea7","added_by":"auto","created_at":"2025-07-29 08:59:22","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":915736,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eIn-situ evolution of Hg nanodroplet with flexible quasi-liquid membrane. a\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eSequential TEM images showing the typical evolution of a single HgS nanobipyramid induced by the e-beam. The nanodroplet and membrane are marked by the solid and dotted circles, respectively. The dotted arrows indicate the dissipation of the nanocomplexes. \u003cstrong\u003eb\u003c/strong\u003e, HRTEM images and associated FFT patterns (inset) detailing the formation of the hollow nanocell.\u003cstrong\u003e \u003c/strong\u003eEach FFT was from the marked square region. The dotted ellipses indicate the locations of the voids.\u003cstrong\u003e c\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eRepeated nucleation and dissolution processes of Hg nanodroplets within a confined nanocell. The solid arrows indicate the coalescence of nanocomplexes just before the nucleation into the Hg nanodroplet, whereas the dotted arrows represent the dissipation of nanocomplexes into the surrounding. \u003cstrong\u003ed\u003c/strong\u003e, The crystallization process after the collision of Hg nanodroplet to the nanocell wall. The HRTEM and associated FFT reveal that the recrystallized NP is HgSO\u003csub\u003e4\u003c/sub\u003e\u003cstrong\u003e·\u003c/strong\u003eH\u003csub\u003e2\u003c/sub\u003eO with the spacings of 3.02, 3.64 and 2.99 Å corresponding to its (), () and () crystal planes. \u003cstrong\u003ee\u003c/strong\u003e, Typical time-dependent characteristics presenting the diameter of the Hg nanodroplet and thickness of the membrane during the evolution. \u003cstrong\u003ef\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eSchematic diagram illustrating the full evolution picture of the nucleation, growth, dissipation, rupture and crystallization in the nanodroplet system.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/983ce56ee03abde6e4e186a5.png"},{"id":87808796,"identity":"45050499-2309-4b26-8521-9766517e6615","added_by":"auto","created_at":"2025-07-29 08:59:22","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":763842,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDynamic mass transport in single and multiple nanocells. a\u003c/strong\u003e, The mass transfer between the nanodroplets D1 and D2 in the single nanocell. \u003cstrong\u003eb\u003c/strong\u003e, Volume of the nanodroplets D1, D2 and their sum as a function of time. \u003cstrong\u003ec\u003c/strong\u003e, The mass transport among the nanodroplets D3, D4 and D5 in the single nanocell. \u003cstrong\u003ed\u003c/strong\u003e, Time-dependent volume change of the nanodroplets D3, D4, D5 and their sum. Additional TEM images at the states marked by the dotted arrows are shown in Supplementary Fig. 13. \u003cstrong\u003ee\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eMass transfer across multiple nanocells created from the HgS polyhedrons. The transfer direction is guided by the numbers and dotted arrows. \u003cstrong\u003ef\u003c/strong\u003e, Mass transport across multiple nanocells that are formed from the HgS ellipsoids. The numbers and dotted arrows indicate the transport direction. No transport occurs when the distance between neighboring nanocells is far (marked by ×). \u003cstrong\u003eg\u003c/strong\u003e,\u003cstrong\u003e \u003c/strong\u003eSchematic illustration of the mass transport process either in the confined single nanocell or across the multiple nanocells.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/d0c2d7585b82e5752eda4411.png"},{"id":87809303,"identity":"ad2b41a4-244c-4e5b-9803-2c6db5465e18","added_by":"auto","created_at":"2025-07-29 09:07:23","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":656809,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDisturbance to the membrane and \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eab initio\u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003e molecular dynamics simulations. a\u003c/strong\u003e, Evolution of HgS nanobipyramid in the 10 mM NaCl aqueous solution. The dotted arrows and rectangles indicate the locations for Ostwald ripening and recrystallization, respectively. The HRTEM and FFT suggest that the as-formed crystal is HgCl\u003csub\u003e2\u003c/sub\u003e. \u003cstrong\u003eb\u003c/strong\u003e, Effect of the nanobubbles on the core-membrane system. The solid, dotted arrows and triangles indicate the nanobubbles attacking the membrane at different stages. \u003cstrong\u003ec\u003c/strong\u003e, Evolution of the Hg-HgS system without the outer membrane in the liquid cell. \u003cstrong\u003ed\u003c/strong\u003e, Snapshot (at the state of 15 ps) and charge evolution curve for the system A containing 72 H\u003csub\u003e2\u003c/sub\u003eO molecules and 60 Hg atoms. \u003cstrong\u003ee\u003c/strong\u003e, Snapshot and charge evolution curve for the system B that is derived from system A by replacing H₂O with OH\u003csup\u003e−\u003c/sup\u003e. \u003cstrong\u003ef\u003c/strong\u003e, Representative snapshot of the core-membrane structure simulated from the system C that contains 200 H\u003csub\u003e2\u003c/sub\u003eO molecules,150 Hg-associated species, and 100 OH\u003csup\u003e−\u003c/sup\u003e. \u003cstrong\u003eg\u003c/strong\u003e, The charge distribution of system C. A charge of 0.1 |e| serves as the threshold distinguishing Hg atoms (Hg\u003csub\u003ea\u003c/sub\u003e) from Hg ions (Hg\u003csub\u003ei\u003c/sub\u003e). \u003cstrong\u003eh\u003c/strong\u003e, Two-dimensional density distributions of Hg (the sum of Hg\u003csub\u003ea\u003c/sub\u003e and Hg\u003csub\u003ei\u003c/sub\u003e), Hg\u003csub\u003ea\u003c/sub\u003e, and Hg\u003csub\u003ei\u003c/sub\u003e in system C. \u003cstrong\u003ei\u003c/strong\u003e, Radial distribution function between Hg, Hg\u003csub\u003ea\u003c/sub\u003e and Hg\u003csub\u003ei\u003c/sub\u003e in system C. \u003cstrong\u003ej\u003c/strong\u003e, Snapshots of system D (derived from system C by adding 20 H\u003csub\u003e2\u003c/sub\u003e molecules), showing the disturbance to the membrane. Color mode: orange, Hg; grey, Hg\u003csub\u003ea\u003c/sub\u003e; green, Hg\u003csub\u003ei\u003c/sub\u003e; red, O; white, H; blue, H\u003csub\u003e2\u003c/sub\u003e.\u003cstrong\u003e k\u003c/strong\u003e, Schematic diagram illustrating the membrane-bound and membrane-free behaviors in the LLPS-induced systems.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/33e24968640f82b00a121d40.png"},{"id":87811379,"identity":"9a0c993c-f7f3-4ff6-a469-57df6c83dbe7","added_by":"auto","created_at":"2025-07-29 09:24:09","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3846870,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/601970cc-cea4-48f1-a914-a208126a4d6b.pdf"},{"id":87808799,"identity":"d73ba9c0-741f-4389-ad9f-1cbe4527b058","added_by":"auto","created_at":"2025-07-29 08:59:22","extension":"mp4","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":4886934,"visible":true,"origin":"","legend":"Supplementary Movie 1","description":"","filename":"SupplementaryMovie1.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/837ccc2ed5b70333f42e522e.mp4"},{"id":87808802,"identity":"4118aad0-2728-4f94-90aa-1c7a537625ca","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":5317487,"visible":true,"origin":"","legend":"Supplementary Movie 2","description":"","filename":"SupplementaryMovie2.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/64953bc3ea3b4e1b94221d5e.mp4"},{"id":87808805,"identity":"c200a785-5ef2-479d-a375-1732254e5e6f","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":4938017,"visible":true,"origin":"","legend":"Supplementary Movie 3","description":"","filename":"SupplementaryMovie3.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/40b82838da00f1372123c328.mp4"},{"id":87808807,"identity":"53c65118-6bbb-4c38-aef1-8f4123fc724e","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":3509888,"visible":true,"origin":"","legend":"Supplementary Movie 4","description":"","filename":"SupplementaryMovie4.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/d41651303be52f67bad51193.mp4"},{"id":87808800,"identity":"472931f5-0805-4d8f-87a3-513ab0507a0b","added_by":"auto","created_at":"2025-07-29 08:59:22","extension":"mp4","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":4735455,"visible":true,"origin":"","legend":"Supplementary Movie 5","description":"","filename":"SupplementaryMovie5.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/cbf0e7add77fa2e3c3542dcc.mp4"},{"id":87808813,"identity":"14bda76b-6113-4531-8949-8dc1340d4fc8","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":2669226,"visible":true,"origin":"","legend":"Supplementary Movie 6","description":"","filename":"SupplementaryMovie6.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/308f02389704d89a6b4d2b41.mp4"},{"id":87810427,"identity":"66667fe6-40f7-457e-aa64-d2a8a1f29a59","added_by":"auto","created_at":"2025-07-29 09:15:23","extension":"mp4","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":3929126,"visible":true,"origin":"","legend":"Supplementary Movie 7","description":"","filename":"SupplementaryMovie7.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/8af9e7ddbf6ea8c08b5d4125.mp4"},{"id":87808823,"identity":"38dc2593-4fef-47af-8b35-66e2fac0d70b","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":3912987,"visible":true,"origin":"","legend":"Supplementary Movie 8","description":"","filename":"SupplementaryMovie8.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/c9dc16feb167566ef046b6e3.mp4"},{"id":87808824,"identity":"c75d1662-f042-41b1-92e7-787cd572af5e","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":9,"title":"","display":"","copyAsset":false,"role":"supplement","size":3636932,"visible":true,"origin":"","legend":"Supplementary Movie 9","description":"","filename":"SupplementaryMovie9.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/84068053723ef5b33bf563f4.mp4"},{"id":87809306,"identity":"99ed6a23-41f8-4a60-8406-0b6bedbca6af","added_by":"auto","created_at":"2025-07-29 09:07:23","extension":"mp4","order_by":10,"title":"","display":"","copyAsset":false,"role":"supplement","size":1026188,"visible":true,"origin":"","legend":"Supplementary Movie 10","description":"","filename":"SupplementaryMovie10.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/c5db470d711213be23c6dc32.mp4"},{"id":87808817,"identity":"df72b0da-ce89-477e-841d-cb0caeb813c5","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":11,"title":"","display":"","copyAsset":false,"role":"supplement","size":6563739,"visible":true,"origin":"","legend":"Supplementary Movie 11","description":"","filename":"SupplementaryMovie11.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/6eead462aea02d76b7d6f034.mp4"},{"id":87808810,"identity":"3614b1f7-2dbe-4412-b697-17582e8df99a","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":12,"title":"","display":"","copyAsset":false,"role":"supplement","size":3341712,"visible":true,"origin":"","legend":"Supplementary Movie 13","description":"","filename":"SupplementaryMovie13.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/318bf61b2d4201fc4ede4f69.mp4"},{"id":87808821,"identity":"f5cecb3d-ab17-414e-b856-9dc870bbfcab","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":13,"title":"","display":"","copyAsset":false,"role":"supplement","size":15535424,"visible":true,"origin":"","legend":"Supplementary Movie 12","description":"","filename":"SupplementaryMovie12.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/cadbb1672e2a8a8aead1ea1c.mp4"},{"id":87808822,"identity":"70c7fa01-02fe-4eeb-9803-10fcfa1acd0e","added_by":"auto","created_at":"2025-07-29 08:59:23","extension":"mp4","order_by":14,"title":"","display":"","copyAsset":false,"role":"supplement","size":20555244,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Movie 14\u003c/p\u003e","description":"","filename":"SupplementaryMovie14.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/e3b4b50d5258de4a02fa85e8.mp4"},{"id":87808825,"identity":"36a13169-4265-47ad-ae5f-e87c60f7d0a5","added_by":"auto","created_at":"2025-07-29 08:59:24","extension":"mp4","order_by":15,"title":"","display":"","copyAsset":false,"role":"supplement","size":19475421,"visible":true,"origin":"","legend":"\u003cp\u003eSupplementary Movie 15\u003c/p\u003e","description":"","filename":"SupplementaryMovie15.mp4","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/de3c579f2d08798dfc0b1a25.mp4"},{"id":87808826,"identity":"ac019674-89b5-456a-ad9c-36388359c0d4","added_by":"auto","created_at":"2025-07-29 08:59:24","extension":"docx","order_by":16,"title":"","display":"","copyAsset":false,"role":"supplement","size":30302584,"visible":true,"origin":"","legend":"Supplementary Information","description":"","filename":"SI.docx","url":"https://assets-eu.researchsquare.com/files/rs-6660582/v1/aa6042775428a3c57a8eff76.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Liquid-liquid phase-separated commensal membrane gates mass transport in inorganic nanocells","fulltext":[{"header":"Introduction","content":"\u003cp\u003ePhoton and electron are the most critical environmental stimuli for substance development, serving as the energy source for regulating many aspects of the photon/electron-responsive processes, e.g., liquid-liquid phase separation (LLPS) behavior. LLPS is an intriguing phenomenon in which a mixture separates into two distinct liquid phases with differing component concentrations, usually forming membrane-free compartments (or photobodies) through photon exposure in soft biological and polymeric systems\u003csup\u003e1,2,3\u003c/sup\u003e. Despite the rare report of such behavior in inorganic materials, the in-situ atomic level observation of full LLPS evolution has been also lacking because of the low contrast as well as the beam sensitivity for the organic system under high-resolution microscopy characterization. On the other side, the membrane, when available, forms a functionally distinct configuration that separates two compartments with specific selectivity, enabling spatially regulated functions and offering unique applications in electrochemical processes\u003csup\u003e4\u003c/sup\u003e, catalysis\u003csup\u003e5,6\u003c/sup\u003e, filtration and purification\u003csup\u003e7\u003c/sup\u003e, sensors\u003csup\u003e8\u003c/sup\u003e and biological activities\u003csup\u003e9,10,11\u003c/sup\u003e. For example, membranes with typical thicknesses from several nanometers to microns have been shown to selectively regulate ion and mass exchange in electrochemical processes, thereby enhancing performance metrics such as battery capacity, electrocatalytic property, and cycling behavior\u003csup\u003e12,13\u003c/sup\u003e. Notably, the membrane, with its dynamical adjustment of configurations, mediates the inclusion or exclusion of specific ions for physiological activities\u003csup\u003e14\u003c/sup\u003e. Thus, a fresh perspective on complex membrane-bound LLPS processes, especially when confined in the nano-scaled domain, is essential for comprehensively understanding interfacial dynamics and tailoring the membrane-associated functionalities at the featured level of basic structural unit under environmental stimuli in nature.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eHere, we propose fabricating the dynamical membrane-nanodroplet system within inorganic nanocells through the electron-responsive reversible/irreversible LLPS strategy. As such, the gating behavior on mass transfer through such flexible membranes is simulated with the consideration of the nanoscale confinement effect. Our method of choice is in-situ liquid cell transmission electron microscopy (LCTEM), which provides a capability for directly observing and simultaneously manipulating dynamic LLPS evolution with atomic-scale resolution. Indeed, it has demonstrated significant applications in liquid-associated dynamics, including solution-based crystal growth\u003csup\u003e15,16,17\u003c/sup\u003e, etching behavior\u003csup\u003e18,19,20\u003c/sup\u003e, and electrochemical activities\u003csup\u003e21,22\u003c/sup\u003e in recent years. Using LCTEM, we choose HgS nanostructures embedded in a tris\u003cstrong\u003e•\u003c/strong\u003eHCl buffer solution as the studied system due to the facts that (i) the electron-beam-induced species in this buffer are well-known and also relatively gentle for etching the nanoparticles (NPs)\u003csup\u003e23\u003c/sup\u003e; (ii) the offer of HgS as a matrix allows the realization of nanoscale confined cells through etching; and (iii) the full membrane-bound LLPS evolution picture – from the birth of liquid Hg nanodroplets with the associated membranes, and subsequent gating for mass transport, to the later disappearance – could be thereby achieved. The formation of Hg nanodroplets originates from exciting HgS, proceeding through a two-step nucleation process. The membranes gate the entrance and exit of the Hg species, mediating the reversible formation-to-dissipation of the liquid condensate for single nanodroplet and cross-feeding of these species among different nanodroplets within individual nanocells. Furthermore, long-range mass transport is achieved through the nanochannels connecting different nanocells, leading to the nano-confined irreversible crystallization of Hg-related compounds. The Hg nanodroplets either instantly dissipate or exhibit typical Ostwald-dominated coalescence behavior when the membrane is absent or disrupted by external factors such as electrolytes or nanobubbles. The above phenomena are further validated through cryo-electron microscopy (cryo-EM) and elemental characterizations, and supported by ab initio molecular dynamics simulations.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eExperimental implementation for in-situ visualization\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe HgS nanostructures were synthesized using a seed-mediated epitaxial growth technique \u003csup\u003e24,25\u003c/sup\u003e, with the structures of the seeds and resulting products detailed in Supplementary Figs. 1-2. Fig. 1a shows the in-situ visualization of the LLPS evolution from HgS matrix within an aqueous Tris•HCl solution sandwiched between two ultrathin carbon films. Further details can be found in Methods and Supplementary Fig. 3. Upon the electron beam (e-beam) excitation, free electrons are generated within the semiconducting HgS NPs, while radiolysis-induced redox active species, such as solvated electrons, hydroxyl radicals, and tris\u003cstrong\u003e\u003csup\u003e•\u003c/sup\u003e\u003c/strong\u003e\u003csup\u003e+\u003c/sup\u003e, are simultaneously formed in the aqueous solution\u003csup\u003e23\u003c/sup\u003e. Those reductive and oxidative species trigger distinct reactions in the HgS semiconductor NPs, enabling the direct observation of the membrane-bound LLPS evolution dynamics via LCTEM. The HgS in Fig. 1b displays a spiral arrangement of Hg/S atoms along the c axis\u003csup\u003e26\u003c/sup\u003e. The typical TEM image presents the bipyramidal shape of the as-synthesized HgS NPs (Fig.1c). The measured d-spacing of 3.18 Å\u0026nbsp;corresponds to the (003) lattice plane of HgS viewed along the [\u0026nbsp;] zone axis according to the diffraction pattern (Fig.1d).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFig. 1e shows a typical image of the as-prepared HgS nanobipyramids in LCTEM, with some nanobubbles as marked by the arrows. The energy dispersive X-ray (EDX) characterization indicates that the Hg and S elements are homogenously distributed in the nanobipyramid present in the high-angle annular dark-field (HAADF) image (Fig. 1f).\u0026nbsp;The height profile of the HgS nanobipyramids is detected by atomic force microscopy, showing the thickness of about 20 nm (Fig. 1g), which is in the same order of magnitude of typical liquid-layer thickness (tens of nanometers) reported in the carbon film/graphene liquid cells\u003csup\u003e23,27,28\u003c/sup\u003e. When the liquid layer is significantly thicker (≈100 nm; more production of etching species upon the e-beam excitation) in the SiN liquid cells, it is noteworthy that almost only etching prevails for HgS (Supplementary Fig. 4). Competing to pure etching, on the other hand, reductive electrons are simultaneously generated in HgS semiconductor after exposure to the e-beam, leading to the reduction of Hg\u003csup\u003e2+\u003c/sup\u003e to Hg. This effect becomes more obvious in the carbon liquid cells with thinner liquid layer. Indeed, the liquid Hg nanodroplet was emerged during the LLPS process, which was evidenced from the elemental Hg/S mapping in Fig. 1h. This behavior contrasts with the pure etching- or growth-dominated processes observed in metallic nanostructures, as reported in the literature\u003csup\u003e16,18,19,20,29\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eReversible/irreversible LLPS in inorganic nanocell\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWith the above experimental implementation in the carbon liquid cell, the typical LLPS evolution from a HgS nanobipyramid with the formation of membrane-bound Hg nanodroplets is shown in Fig. 2a. At the dose rate of ≈1040 e\u003csup\u003e−\u003c/sup\u003e Å\u003csup\u003e2\u0026nbsp;\u003c/sup\u003es\u003csup\u003e-1\u003c/sup\u003e, the initial nanobipyramid (0 s) gradually changed into a hollow structure with the remaining outer shell (110 s). The detailed evolution process was illustrated in Supplementary Fig. 5. It was noticed that two distinct phases were formed: one with a less dense phase (nanocomplexes, less dark contrast) moved randomly inside the nanocell and was merged into a denser phase that took the shape of liquid nanodroplet (condensates, darker contrast) with an outer membrane at 115 s (marked by the solid and dotted circles, respectively; see Supplementary Movie 1 for vivid illustration). With the contribution from such nanocomplexes (indicated by the dotted arrows), the nanodroplet grew bigger to a diameter of\u0026nbsp;≈7 nm at 120 s and further to ≈11 nm at 138 s, and then gradually shrunk as the e-beam irradiation continued (e.g., the state at 338 s). Both the elemental identification (Fig. 1h) and the cryo-EM results (Supplementary Fig. 6 and Supplementary Table 1 verified that the as-formed nanodroplet was Hg. When the electron dose rate was increased to\u0026nbsp;≈2390 e\u003csup\u003e−\u003c/sup\u003e Å\u003csup\u003e2\u0026nbsp;\u003c/sup\u003es\u003csup\u003e-1\u003c/sup\u003e, the nanodroplet expanded again in a short time (347 s) followed by the slow reduction in size, as detailed in Supplementary Fig. 7. Interestingly, the membrane flexibly regulated its shape but with constant thickness during the growth and shrinking processes of the nanodroplet. Such nanodroplet finally disappeared at 432 s. Note that the formation of the Hg nanodroplets and their evolution were induced by the e-beam since they kept invariant with the absence of the e-beam (Supplementary Fig. 8). The e-beam-induced temperature rise was normally in the range of several kelvins in the liquid cells\u003csup\u003e30,31\u003c/sup\u003e, which was not responsible for the above phenomena, as also confirmed by the in-situ heating experiment (Supplementary Fig. 9). Such general phenomena were observed in the graphene liquid cells as well (Supplementary Fig. 10). Since carbon liquid cells were relatively more accessible, they were used for the comprehensive demonstration in the following section.\u003c/p\u003e\n\u003cp\u003eThe time-dependent images with high resolution (Supplementary Movie 2) were further captured to unveil the formation of the hollow nanocells, as illustrated in Fig. 2b.\u0026nbsp;As a demonstration,\u0026nbsp;the fast Fourier transform (FFT) pattern in the inset of each image corresponds to the diffraction from the marked square area, simultaneously revealing the order-disorder transition during the structural evolution. Upon the e-beam illumination, the initial HgS nanobipyramid (0 s) with intact structure and diffraction pattern gradually turned into\u0026nbsp;the hollow one with the increment of structural disordering. For example, several voids (marked by the dotted line) were formed at 20 s. These small ones were coalesced into the large one at 24 s, followed by the subsequent growth and spread (e.g., 108 s). Finally, a hollow nanocell was created with the amorphous nature evidenced by the ring diffraction pattern at 246 s. It has been reported that Hg species have the strong tendency to be adsorbed on HgS matrix \u003csup\u003e32,33\u003c/sup\u003e. Thus, the as-formed nanocell provides a confined domain to develop the evolution processes of liquid Hg species, offering a fundamental understanding of membrane-gated mass transport behavior with the input of nanoscale confinement.\u003c/p\u003e\n\u003cp\u003eThe nucleation and associated processes of the nanodroplet inside the nanocell are detailed in Fig. 2c (Supplementary Movie 3). For convenience, the moment of interest for investigation was regarded as the initial state (0 s) in the following section. A single Hg nanodroplet (≈9.2 nm) with the outer membrane was initially formed, which presented slight change during the first ≈4 s. In contrast, the swarming ‘transient’ nanocomplexes with the amorphous liquid-like feature were aggregated into the big one (marked by the solid arrow) after extended e-beam exposure for 0.9 s. Through the condensation of such liquid-like species, a small nucleus of Hg nanodroplet (darker contrast) nucleated inside the nanocomplex at the subsequent stage (1.0 s), which shared the similarity with the two-step nucleation mechanism\u003csup\u003e34,35,36\u003c/sup\u003e. The nucleus gradually grew bigger, from ≈2.2 nm in diameter at 1.2 s to that of ≈5.7 nm at 1.5 s. Without feeding additional species, the as-formed smaller Hg nanodroplet then dissipated into the nanocomplex at 3.4 s (marked by the dotted arrow), and eventually vanished at 3.8 s. As the time proceeded, the preexisting large nanodroplet became degraded, followed by the instantaneous release of Hg-associated species into the nanocell gated through the membrane, leading to the formation of new nanocomplexes (e.g., at 16.9 s indicated by the solid arrow). As such, a new Hg nanodroplet began to nucleate in the center of the nanocomplex again at 17.1 s, and grew to ≈6.5 nm in diameter at 17.5 s. Subsequently, the two nanodroplets were contracted and broken into the nanocomplexes at 20.1 s, which then disappeared at 20.5 s. Such reversible processes could be repeated, e.g., a new circle (nucleation-growth-dissipation-disappear) occurred again from 41.2 to 43.0 s. Overall, such assembly and disassembly of condensates (nanodroplet) caused by the LLPS process potentially act as electron sensors (tentatively termed ‘electrobody’ here), periodically forming and dissipating in response to the exposure of e-beam, which is analogous to the LLPS-induced reversible photobody observed in biological systems\u003csup\u003e37,38\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eFig. 2d shows the subsequent crystallization (irreversible LLPS) process after the rupture of Hg nanodroplet (Supplementary Movie 4). When the initial nanodroplet (0 s) touched the cell wall, we noticed that the other half fraction (contact region) of the membrane was broken and the core condensate began to shrink rapidly (e.g., 29.4 s). Without the membrane-related regulation, the core and the neighboring nanocomplexes were then coalesced (29.5 s, marked by the dotted arrows), and immediately became condensed to crystallize inside the nanocell (29.7 s).\u0026nbsp;The crystallization phase was obviously observed at 32 s, which was recognized from the high-resolution TEM (HRTEM) image and the associated FFT analysis. The d-spacings of 3.02, 3.64 and 2.99 Å measured in HRTEM corresponded to the (\u0026nbsp;), (\u0026nbsp;) and (\u0026nbsp;) lattice planes of the HgSO\u003csub\u003e4\u003c/sub\u003e\u003cstrong\u003e·\u003c/strong\u003eH\u003csub\u003e2\u003c/sub\u003eO crystal viewed along the [100] zone axis according to the FFT (see the detailed analysis in Supplementary Fig. 11 and Supplementary Table 2). On the other hand, the Hg(I) compound, e.g., the Hg\u003csub\u003e2\u003c/sub\u003eSO\u003csub\u003e4\u003c/sub\u003e, could also be crystallized in some cases (Supplementary Fig. 12 and\u0026nbsp;Supplementary Table 3). It should be pointed out that such crystallized phases would not be reversed into the less dense nanocomplexes again, which shared the similarity with irreversible LLPS process in biology\u003csup\u003e39,40\u003c/sup\u003e. Upon the e-beam irradiation, it has been known that the oxidative tris\u003cstrong\u003e\u003csup\u003e•\u003c/sup\u003e\u003c/strong\u003e\u003csup\u003e+\u003c/sup\u003e radical was created in the tris\u003cstrong\u003e•\u003c/strong\u003eHCl solution, leading to the oxidation of selenium in the PbSe system\u003csup\u003e23\u003c/sup\u003e. Thus, the occurrence of high valence state of sulfur in this work was considered to be oxidized by tris\u003cstrong\u003e\u003csup\u003e•\u003c/sup\u003e\u003c/strong\u003e\u003csup\u003e+\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eDuring the evolution, the typical relationship between the diameter of the core condensate and the thickness of the membrane is shown in Fig. 2e. Analogous to the photon-responsive photobodies, the liquid inorganic condensates display the sensitivity to the intensity of electrons. At a fixed dose rate (e.g., the 1040 e\u003csup\u003e−\u003c/sup\u003e Å\u003csup\u003e2\u0026nbsp;\u003c/sup\u003es\u003csup\u003e-1\u003c/sup\u003e here), the core presented the real-time variation in size because of the exposure to electrons. Increasing the dose rate (≈2390 e\u003csup\u003e−\u003c/sup\u003e Å\u003csup\u003e2\u0026nbsp;\u003c/sup\u003es\u003csup\u003e-1\u003c/sup\u003e at 338 s) increased the diameter of the core temporally. Such core region was finally vanished upon the continuous e-beam irradiation. However, the thickness of the membrane layer maintained constant (≈3 nm) regardless of the dissipation or expansion process of the core condensate. Fig. 2f shows the schematic diagram of the overall LLPS evolution from the HgS nanobipyramid in the solution upon the e-beam excitation. The intact nanobipyramid is gradually evolved into a hollow structure through the formation and coalescence of the voids, which is accompanied with the prenucleation of the less dense nanocomplexes randomly swarming inside the nanocell. The condensation of the nanoclusters results in the nucleation and subsequent growth of dense Hg nanodroplet enclosed by a layer of membrane. In response to the electrons, the membrane dynamically gates the entrance and exit of the Hg species, resulting in the periodic reversible cycling of the nucleation-growth-dissipation-disappearance processes of Hg nanodroplet. However, such nanodroplet is broken once touching the cell wall, resulting in the irreversible formation of new nanocrystals inside the nanocell.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRegulation of mass transport by membranes\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDistinct from the LLPS-induced membrane-free photobody in biological system\u003csup\u003e41,42,43\u003c/sup\u003e, the membranes bounded in the inorganic condensates demonstrate the regulation of mass transport in the nanocells. Multiple nanodroplets, whether contained within a single nanocell or multiple\u0026nbsp;nanocells, show a novel gated transport behavior (Fig. 3). When two nanodroplets with the outer membrane layers located in the single nanocell (Fig. 3a), the repeated behavior of mass transfer (cross-feeding) could be realized through the gating of the membranes (Supplementary Movie 5). By this, the big droplet (D1) presented the reduction in size while the initial small one (D2) became larger (e.g., at 12.4 s). Afterwards, the changes were reversed, namely, the D1 grew bigger, followed by the decrease of D2 at 37.8 s. Such intracellular interchange process between two compartments was repeated back and forth for many times (e.g., from 49.4 to 105.0 s), which was analogous to the breathing (in/out causing rhythmical expansion/shrinkage of the core condensate) behavior in biology. From the time-dependent volume curve (Fig. 3b), it is clear that the two Hg droplets showed the opposite trend in volume changes, and their total volumes were gradually reduced upon the continuous e-beam irradiation.\u0026nbsp;Owing to the gating-induced regulation effect from the membranes, however, the above interchange behavior was markedly distinct from the typical Ostwald ripening (growth of large droplets at the expense of shrinking and annihilating the smaller ones) despite the close contact between two droplets.\u003c/p\u003e\n\u003cp\u003eSurprisingly, we also noticed that the total volume would stay no obvious changes when the droplets located within a spatially enclosed nanocell (Fig. 3c and Supplementary Movie 6). D3 and D4 droplets were in close proximity to each other, which were several nanometers away from D5. The oscillation in volume (repeated switching between the large and small size) also occurred for the D3 and D4 droplets in most of the period (e.g., the states shown in the in-situ images). A crucial observation revealed that both of the two droplets could sometimes share the shrinking moments simultaneously, with the involvement of another D5 droplet (e.g., at 14.2 s). In this case, the D5 one became larger for a short period. Such behavior was highlighted in the time-dependent volume changes of these three droplets (Fig. 3d). It is clear that the volume reduction of the D3 and D4 droplets at certain time points (marked by the arrows) was almost compensated by the increment of the D5 one (see Supplementary Fig. 13 for detailed images), maintaining their total volumes nearly at a constant value. This observation demonstrates that the transport of species would be mediated by the membranes, occurring among the droplets regardless of their relative distances inside the nanocell.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eBesides the transfer among nanodroplets confined within one nanocell, the transport behavior could also occur across different nanocells through nanochannels. In order to directly visualize such nanoscale process within the camera field of view, we further conducted the in-situ experiments by using different morphological HgS nanocrystals (e.g., polyhedral and ellipsoidal shapes) with smaller sizes. Despite the distinct shapes, they also displayed the evolutions in each single cell (Supplementary Fig. 14) as that observed in the ones with the bipyramidal morphology. Favorably, the intercellular mass transfer would be conveniently monitored due to the small size. For the polyhedral sample (Fig. 3e), the transport was initiated from the cell 1 to the cell 2 at 5.7 s (guided by the dotted arrow; see Supplementary Movie 7 for vivid illustration). Afterwards, the nanocomplexes inside the cell 2 transferred to the adjacent cell 3 (10.3 s). Those nanocomplexes spread and broke the ionic balance of the membrane after meeting the preexisting Hg nanodroplet inside the cell 3, causing the dissipation and eventual disappearance of the nanodroplet (35.1 s). The species could further enter the neighboring cell 4 and again touch the membrane so that the corresponding nanodroplet also vanished rapidly (102.9 s). When the cell 2 and 5 were bridged through the formation of the nanochannel (see the enlarged view in Supplementary Fig. 15), the existing nanocomplexes in the cell 2 started to diffuse into the cell 5 (104.1 s). Akin to the “relay race”, the species in the cell 5 continued moving to the adjacent cells, e.g., to the cells 6 and 7 at 111.3 s, and then to the cell 8 at 143.0 s.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe intercellular mass transfer process was also observed in the ellipsoidal sample, as displayed in Fig. 3f (Supplementary Movie 8). Through the nanochannels (Supplementary Fig. 16) created among the adjacent cells, the nanocomplexes transferred from the initial cell 1 to the next cell one by one, until they eventually stopped in the cell 6. Confined in the nanocells with smaller sizes, the swarming nanocomplexes possessed high probability to collide with the outer walls of the cells during the transport, which resulted in the crystallization process similar as that presented in Fig. 2d. The crystalline products were determined to be HgSO\u003csub\u003e4\u003c/sub\u003e and Hg\u003csub\u003e2\u003c/sub\u003eSO\u003csub\u003e4\u003c/sub\u003e (Supplementary Fig. 17, Supplementary Table 4 and 5). It should be pointed out that the nanochannel could not be formed when the distance between the cells was far away (e.g., larger than ~5 nm), leading to the termination of the mass transfer shown at the stage of 141.7 s.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFig. 3g schematically summarizes the mass transfer and associated crystallization processes under the confinement from the nanocells. In the isolated single cell, the quantity of the Hg-related complexes is limited, and gradually dissipates to the surrounding regions upon the e-beam irradiation. Such liquid-like complexes are composed of Hg atoms and ions, as revealed by the HRTEM results (crystalline Hg(I)/Hg(II) compounds shown above), the X-ray photoelectron spectroscopy\u0026nbsp;identification (Supplementary Fig. 18) and also the simulations in the following section. Mediated by the membranes through the transition reaction of \u0026nbsp;, the Hg-associated species can enter into the dense Hg nanodroplets (the equilibrium of the reaction is towards the left) or be released from those ones (the equilibrium is towards the right), resulting in the periodic reversible big-to-small transition of the nanodroplets. Under such equilibrium of the reaction, the thickness of the membranes is maintained constant during the evolution. The key characteristic of condensates undergoing reversible LLPS is that they are dynamic, i.e., they can rapidly exchange with the surrounding cellular nanocomplexes through the gating effect from the membranes. On the other hand, the nanocells would be connected with each other under the e-beam interaction when the nanocells are close enough. Subsequently, the connected bridge, acting as a transfer nanochannel of materials, facilitates the long-range transport of the species from nanocell to nanocell, until the termination if the distance between the nanocells is large so that the nanochannel cannot be formed. Once the membrane is destroyed, the Hg nanodroplets and released nanocomplexes touch the walls of the nanocells, which become irreversibly crystallized into the HgSO\u003csub\u003e4\u003c/sub\u003e or Hg\u003csub\u003e2\u003c/sub\u003eSO\u003csub\u003e4\u003c/sub\u003e nanocrystals depending on the relative quantity of the Hg\u003csup\u003e+\u003c/sup\u003e and Hg\u003csup\u003e2+\u003c/sup\u003e species.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEvolution dynamics after disturbance of the membrane\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe also investigated the system with higher concentration of the tris\u003cstrong\u003e•\u003c/strong\u003eHCl buffer, and found similar evolution processes of the Hg nanodroplets (Supplementary Fig. 19). In contrast, other factors like the addition of electrolytes or the existence of nanobubbles were found to break the ionic balance of the membrane, thereby leading to the distinct irreversible processes observed below. To offer an insight into the LLPS field of biological activity as well, the common NaCl was chosen as the electrolyte. Fig. 4a shows the typical evolutions of the Hg nanodroplets with the addition of NaCl aqueous solution (Supplementary Movie 9). Upon the e-beam excitation, a number of Hg nanodroplets were also formed (e.g., at 7.1 s). Surprisingly, no membranes were observed in the outer surfaces of those nanodroplets. Without the modulation from the membranes, the classical Ostwald ripening or coalescence was mainly responsible for the growth of the nanodroplets (e.g., the period during 22.3-29.5 s, indicated by the dotted arrows; see also Supplementary Fig. 20), which was different from the reversible LLPS observed in Figs. 2 and 3 (the membrane acted as a “armor” to hinder the swallowing of small nanodroplets by the large ones). Meanwhile, some rectangular features (marked by the dotted rectangles) were seen in the surroundings at 29.5 s, and became more obvious at 88.5 s. Such features expanded rapidly in the field of view (e.g., 131.0 s), as presented vividly in Supplementary Movie 9. The HRTEM and associated FFT pattern identified that the rectangular nanocrystals were HgCl\u003csub\u003e2\u003c/sub\u003e, with the d-spacing of 2.93 Å corresponding to its (002) lattice plane viewed along the [\u0026nbsp;0] zone axis (Supplementary Fig. 21 and Supplementary Table 6). Note that the framework of HgS was etched off, resulting in the crystallization extending to the surrounding regions (upon comparison to the confinement in the nanocells shown in Figs. 2 and 3).\u003c/p\u003e\n\u003cp\u003eWhen the nanobubbles met the Hg nanodroplets in the solutions，the asymmetric rupture of the nanodroplets would occur, as shown in Fig. 4b (Supplementary Movie 10). Initially, a nanobubble (marked by the solid arrow) was approaching the spherical Hg nanodroplet. As the nanobubble touched the membrane, the ionic equilibrium inside the membrane was suddenly disrupted so that the Hg nanodroplet was asymmetrically contracted (0.2 s). Followed by such continued interaction, the nanodroplet gradually shrunk (0.7 s), and further became smaller with the emergence of an additional nanobubble (marked by the dotted arrow, 2.1 s). Later on, another nanobubble (marked by the triangle) gradually expanded and attacked other part of the droplet, accompanied by the continuous shrinkage of the nanodroplet (2.5~5.0 s). This process proceeded until the nanodroplet completely disappeared at 6.0 s. Further, we prepared the Hg-HgS heterostructures (Supplementary Fig. 22 and Supplementary Table 7) based on the synthesis procedure reported previously\u003csup\u003e44\u003c/sup\u003e, featuring the droplet behavior without the primitive membrane. Fig. 4c shows the evolution of such heterostructures in water liquid cell (Supplementary Movie 11). As seen, no membrane existed on the surface of the Hg droplet. Upon the e-beam excitation, the Hg droplet started to dissipate (refer to the contrast change from 6 to 25 s). In contrast to that observed in Figs. 2 and 3, the Hg droplet only presented the dissipation (e.g., 30-36 s) and eventually vanished, remaining the HgS nanorod section at 37 s. With the disturbance or absence of the membrane, the above results thereby verified that no gating of the mass transfer repeatedly occurred for the Hg-related species.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSimulation of membrane-associated behavior\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo understand the microstructure and dynamic evolution of the Hg droplets system, we performed \u003cem\u003eab initio\u003c/em\u003e molecular dynamics (AIMD) simulations (see Supplementary Materials for details). For the primary components of the quasi-liquid membrane, the HAADF intensity profiles across the membrane were analyzed to evaluate the ratio between the H\u003csub\u003e2\u003c/sub\u003eO molecules and the Hg-related species\u003csup\u003e45\u003c/sup\u003e. Such ratio was estimated to be ~1.2:1 in the membrane (Supplementary Fig. 23 and Supplementary Note). Accordingly, a model (denoted as system A) containing 72 H\u003csub\u003e2\u003c/sub\u003eO molecules and 60 neutral Hg atoms was constructed. Fig. 4d shows the equilibrium state of the system, where the Hg atoms rapidly aggregate into clusters (refer to\u0026nbsp;Supplementary\u0026nbsp;Fig. 24 for detailed evolution). Based on the first coordination shell in the Hg-Hg radial distribution function (Supplementary\u0026nbsp;Fig. 25), the Hg-Hg distance smaller than 3.3 Å was chosen as forming a cluster (indicated by the orange connecting lines in Fig. 4d). The time-averaged charge distribution of Hg atoms varies from -0.06 to 0.05 |e|, with an average charge close to zero per atom. To compare the aggregation behavior and stability, we then removed one hydrogen atom from each H\u003csub\u003e2\u003c/sub\u003eO molecule, resulting in a system (denoted as system B) containing 72 OH⁻ ions and 60 positively charged Hg ions. Simulations in Fig. 4e and\u0026nbsp;Supplementary\u0026nbsp;Fig 24 show that the Hg ions are almost uniformly dispersed (the time-averaged charge distribution ranges from 0.05 to 0.65 |e|). This distinct dynamic behavior is caused by the electrostatic repulsion between the positively charged mercury ions, indicating that the presence of Hg ions prevents the aggregation and further enhances the relative stability of the quasi-liquid membrane.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe Hg core-membrane complex structure of the nanodroplets was further simulated with the construction of a large system C that contained 200 H\u003csub\u003e2\u003c/sub\u003eO molecules, 100 OH\u003csup\u003e−\u003c/sup\u003e ions, and 150 Hg-associated particles (either Hg atoms or Hg cations) in the aqueous solution. After 10 ps of simulation (Supplementary Fig. 26 for other snapshots and\u0026nbsp;Supplementary\u0026nbsp;Movie 12), the system evolved into a core-membrane structure, with neutral Hg atoms forming the core and Hg ions occupying the outer membrane (Fig. 4f). The charges of Hg species range from -0.08 to 0.84 |e| (Fig. 4g). Note that the Hg species with a charge value no greater than 0.1 |e| are classified as Hg atoms (labeled Hg\u003csub\u003ea\u003c/sub\u003e), otherwise they are classified Hg ions (labeled Hg\u003csub\u003ei\u003c/sub\u003e). As such, the plots of the two-dimensional density distributions (Fig. 4h) show that neutral Hg atoms (≈38%) locate in the core, while positively charged Hg ions (≈62%) occupy the outer membrane. Together, all Hg species exhibit a gradient from the core to the outside, transitioning from a dense to a more diffuse structure. The simulations agree well with the core-membrane structure observed in the HRTEM experiments (Fig. 2), affirming that the balance between Hg ions and atoms maintains the stability of the membrane.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe dynamic equilibrium maintained in the core-membrane structure is realized through the charge transfer. As shown in\u0026nbsp;Supplementary Fig. 27\u0026nbsp;and\u0026nbsp;Supplementary\u0026nbsp;Movie 13, the core and outer Hg species maintain atomic state and positively charged one, respectively. However, the interfacial species undergo the charge fluctuations between 0.0 and 0.3 |e|, implying they dynamically transit between the atomic and charged states at the interface. The radial distribution function g(r) was further analyzed to gain insights into the associated complex structure. The g(r) in Fig. 4i exhibits two sharp peaks. The peak at 3.3 Å is mainly attributed to the Hg\u003csub\u003ea\u003c/sub\u003e-Hg\u003csub\u003ea\u003c/sub\u003e pairs (partly from the Hg\u003csub\u003ei\u003c/sub\u003e-Hg\u003csub\u003ea\u003c/sub\u003e pairs), supporting the formation of dense clusters at the core of the nanodroplet. It is noteworthy that another peak at 2.7 Å originates from the Hg\u003csub\u003ei\u003c/sub\u003e-Hg\u003csub\u003ei\u003c/sub\u003e pairs, indicating the formation of Hg\u003csub\u003e2\u003c/sub\u003e\u003csup\u003e2+\u003c/sup\u003e, in consistent with the previous reported distance of 2.52 Å in Hg\u003csub\u003e2\u003c/sub\u003e\u003csup\u003e2+\u003c/sup\u003e.\u003csup\u003e46\u003c/sup\u003e Therefore, the simulations verify that the dynamic\u0026nbsp;reaction of \u0026nbsp; regulates the entrance or exit of Hg-associated species to the Hg nanodroplets, leading to the experimental observation of periodic big-to-small transition of the nanodroplets.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eTo investigate the disturbance effect of nanobubbles on the membrane, we constructed another system D by randomly adding 20 H\u003csub\u003e2\u003c/sub\u003e molecules to the above system C. Simulations in Fig. 4j revealed that H\u003csub\u003e2\u003c/sub\u003e molecules diffused into the interior of the Hg droplet and disrupted its core-membrane structure (see\u0026nbsp;Supplementary Fig. 28\u0026nbsp;and\u0026nbsp;Supplementary\u0026nbsp;Movie 14 for details),\u0026nbsp;supporting the experimental observation that the nanobubbles caused the fragmentation of Hg nanodroplets (Fig. 4b). In contrast, when the system was composed of neutral Hg atoms and H\u003csub\u003e2\u003c/sub\u003eO molecules (system E), the presence of H\u003csub\u003e2\u003c/sub\u003e did not cause significant structural change of the nanodroplet (Supplementary\u0026nbsp;Fig. 29 and\u0026nbsp;Supplementary\u0026nbsp;Movie 15). We proposed that the introduction of H\u003csub\u003e2\u003c/sub\u003e molecules could alter the charge distribution within the core-membrane structure through interactions with the surrounding ions (Supplementary Fig. 30), leading to the disruption of ionic balance that ultimately caused the disintegration of the nanodroplet.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eWith the in-situ visualization under the atomic level resolution, we revealed a full LLPS evolution picture regarding the formation, growth/shrinkage, and collapse of the Hg nanodroplets, and the comparison of gating effect on the mass transport with/without the dynamical membranes (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e4\u003c/span\u003ek). For the membrane-bound nanodroplets within single nanocell, the repeated reversible cross-feeding among different nanodroplets occurs. In contrast, the long-range propagation across multiple nanocells through nanochannels is observed once the membranes are disturbed. The less dense nanocomplexes released from the membranes transit into the crystallized NPs (accompanied by the Ostwald ripening or coalescence behavior) during the transport, which become the irreversible process. The same strategy could be extended to other buffer systems for realizing the LLPS behavior, e.g., the glycine-HCl buffer solution, and the similar phenomena have been also observed (Supplementary Fig.\u0026nbsp;31).\u003c/p\u003e\u003cp\u003eAlthough this study focuses on inorganic nanocells, the involvement of buffer solutions, NaCl and nanobubbles with the atomic-scale resolution could also provide valuable insights for biological system. As is well known, Hg was commonly used as an ingredient in immortal pills in ancient alchemy. Based on the direct visualization presented in this work, it is suggested that oxidative species and radicals in the human body, e.g., reactive oxygen species\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e,\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e,\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e, may facilitate the formation of a gating membrane encapsulating Hg, thereby retarding the further oxidation into Hg\u003csup\u003e+\u003c/sup\u003e or Hg\u003csup\u003e2+\u003c/sup\u003e species. This process could delay the onset of toxic reactions so that the immortal pills were not considered to be harmful in ancient years. Besides the well-known binding of those species with the sulfhydryl-containing proteins affecting cell functionalities\u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e, our results further identify other distinct pathways dependent on local microenvironment. The existence of local high concentration of NaCl and/-or nanobubbles destroys the outer membrane of the elemental Hg droplets, resulting in the sudden asymmetrical collapse of the droplets and subsequent formation of ionic Hg species that rapidly propagate across cells through nanochannels and crystallize into Hg-related compounds (e.g., the toxic HgCl\u003csub\u003e2\u003c/sub\u003e that significantly affects human health). Moreover, the crystallized nanocrystals preferentially grow on the cell walls, indicating that they would accumulate on the blood vessels or other organs over months and years and finally lead to the chronic poisoning. This investigation offers a thorough understanding of the mass transport behavior in both nanoscale confined and unconfined regions mediated by the flexible membranes while their distinctive properties pave the path for developing multifunctional capabilities in other systems.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWe thank Prof. Chang Yan for helpful discussion. This work is supported by the National Natural Science Foundation of China (Nos. 92061116, 22273004, and 224B2302) and the National Key Research and Development Program of China (2024YFA1509602). We also acknowledge the financial support from the Science and Technology Commission of Shanghai Municipality (22ZR1428400).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eB. C. and Y. L. conceived the research. Y. L. prepared the samples and performed the in-situ experiments. Z. S. and J. X. conducted the \u003cem\u003eab initio\u003c/em\u003e molecular dynamics simulations. Z. C., L. X., and H. L. assisted in the synthesis of HgS nanostructures. Y. L., Z. S., J. X., and B. C. wrote the manuscript. All authors discussed the results and commented on the manuscript.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupplementary Information\u003c/strong\u003e is available in the online version of the paper.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eFeng Z et al (2024) Liquid\u0026ndash;liquid phase separation of TZP promotes PPK-mediated phosphorylation of the phytochrome A photoreceptor. Nat Plants 10:798\u0026ndash;814\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang X, Lin C (2025) The two action mechanisms of plant cryptochromes. Trends Plant Sci\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJiang B et al (2023) Light-induced LLPS of the CRY2/SPA1/FIO1 complex regulating mRNA methylation and chlorophyll homeostasis in Arabidopsis. Nat Plants 9:2042\u0026ndash;2058\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTao L et al (2023) Reversible Switch in Charge Storage Enabled by Selective Ion Transport in Solid Electrolyte Interphase. J Am Chem Soc 145:16538\u0026ndash;16547\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZeng T et al (2022) Hybrid bilayer membranes as platforms for biomimicry and catalysis. Nat Rev Chem 6:862\u0026ndash;880\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTang BY, Bisbey RP, Lodaya KM, Toh WL, Surendranath Y (2023) Reaction environment impacts charge transfer but not chemical reaction steps in hydrogen evolution catalysis. Nat Catal 6:339\u0026ndash;350\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCheng XQ et al (2023) Biodegradable electrospinning superhydrophilic nanofiber membranes for ultrafast oil-water separation. Sci Adv 9:eadh8195\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang X et al (2024) Ultralow-pressure-driven polarization switching in ferroelectric membranes. Nat Commun 15:9281\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHannesschlaeger C, Horner A, Pohl P (2019) Intrinsic Membrane Permeability to Small Molecules. Chem Rev 119:5922\u0026ndash;5953\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKozlov MM, Taraska JW (2022) Generation of nanoscopic membrane curvature for membrane trafficking. Nat Rev Mol Cell Biol 24:63\u0026ndash;78\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSimunovic M (2017) Biology and physics rendezvous at the membrane. Science 358:1265\u0026ndash;1265\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFang W et al (2024) Durable CO2 conversion in the proton-exchange membrane system. Nature 626:86\u0026ndash;91\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLagadec MF, Zahn R, Wood V (2018) Characterization and performance evaluation of lithium-ion battery separators. Nat Energy 4:16\u0026ndash;25\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCao L et al (2017) Channel-facilitated molecule and ion transport across polymer composite membranes. Chem Soc Rev 46:6725\u0026ndash;6745\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZheng HM et al (2009) Observation of Single Colloidal Platinum Nanocrystal Growth Trajectories. Science 324:1309\u0026ndash;1312\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYuk JM et al (2012) High-Resolution EM of Colloidal Nanocrystal Growth Using Graphene Liquid Cells. Science 336:61\u0026ndash;64\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRoss FM (2015) Opportunities and challenges in liquid cell electron microscopy. Science 350:aaa9886\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYe X et al (2016) Single-particle mapping of nonequilibrium nanocrystal transformations. Science 354:874\u0026ndash;877\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMoreno-Hernandez IA, Crook MF, Ondry JC, Alivisatos AP (2021) Redox Mediated Control of Electrochemical Potential in Liquid Cell Electron Microscopy. J Am Chem Soc 143:12082\u0026ndash;12089\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVigil SA, Moreno-Hernandez IA (2024) Dissolution Heterogeneity Observed in Anisotropic Ruthenium Dioxide Nanocrystals via Liquid-Phase Transmission Electron Microscopy. J Am Chem Soc 146:11133\u0026ndash;11140\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang Q et al (2024) Atomic dynamics of electrified solid\u0026ndash;liquid interfaces in liquid-cell TEM. Nature 630:643\u0026ndash;647\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYoon A et al (2025) Revealing catalyst restructuring and composition during nitrate electroreduction through correlated operando microscopy and spectroscopy. Nat Mater\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYan C et al (2022) Facet-selective etching trajectories of individual semiconductor nanocrystals. Sci Adv 8:eabq1700\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCao Z, He J, Liu Z, Zhang H, Chen B (2021) Chirality Affecting Reaction Dynamics of HgS Nanostructures Simultaneously Visualized in Real and Reciprocal Space. ACS Nano 15:16255\u0026ndash;16265\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang P-p, Yu S-J, Govorov AO, Ouyang M (2017) Cooperative expression of atomic chirality in inorganic nanostructures. Nat Commun 8:14312\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMomma K, Izumi F (2011) VESTA 3for three-dimensional visualization of crystal, volumetric and morphology data. J Appl Crystallogr 44:1272\u0026ndash;1276\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePedrazo-Tardajos A et al (2024) Direct visualization of ligands on gold nanoparticles in a liquid environment. Nat Chem 16:1278\u0026ndash;1285\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCrook MF et al (2023) EELS Studies of Cerium Electrolyte Reveal Substantial Solute Concentration Effects in Graphene Liquid Cells. J Am Chem Soc 145:6648\u0026ndash;6657\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZheng H et al (2009) Observation of Single Colloidal Platinum Nanocrystal Growth Trajectories. Science 324:1309\u0026ndash;1312\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiao HG et al (2014) Facet development during platinum nanocube growth. Science 345:916\u0026ndash;919\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang J et al (2019) Formation of two-dimensional transition metal oxide nanosheets with nanoparticles as intermediates. Nat Mater 18:970\u0026ndash;976\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBurkstaller JE, McCarty PL, Parks GA (1975) Oxidation of cinnabar by iron(III) in acid mine waters. Environ Sci Technol 9:676\u0026ndash;678\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHsieh YH, Tokunaga S, Huang CP (1991) Some chemical reactions at the HgS(s)\u0026mdash;water interface as affected by photoirradiation. Colloids Surf 53:257\u0026ndash;274\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang J et al (2019) Amorphous-Phase-Mediated Crystallization of Ni Nanocrystals Revealed by High-Resolution Liquid-Phase Electron Microscopy. J Am Chem Soc 141:763\u0026ndash;768\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim HJ et al (2022) Heterointerface Effect on Two-Step Nucleation Mechanism of Bi Particles. Nano Lett 22:3252\u0026ndash;3259\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTan P, Xu N, Xu L (2014) Visualizing kinetic pathways of homogeneous nucleation in colloidal crystallization. Nat Phys 10:73\u0026ndash;79\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eIaniro A et al (2019) Liquid\u0026ndash;liquid phase separation during amphiphilic self-assembly. Nat Chem 11:320\u0026ndash;328\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHarris R, Berman N, Lampel A (2025) Coacervates as enzymatic microreactors. Chem Soc Rev\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKitahara R et al (2021) Pressure-Jump Kinetics of Liquid\u0026ndash;Liquid Phase Separation: Comparison of Two Different Condensed Phases of the RNA-Binding Protein, Fused in Sarcoma. J Am Chem Soc 143:19697\u0026ndash;19702\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eShen Y et al (2020) Biomolecular condensates undergo a generic shear-mediated liquid-to-solid transition. Nat Nanotechnol 15:841\u0026ndash;847\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLafontaine DLJ, Riback JA, Bascetin R, Brangwynne CP (2021) The nucleolus as a multiphase liquid condensate. Nat Rev Mol Cell Biol 22:165\u0026ndash;182\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRuan K, Lin Y, Yang P, Zhou W (2023) Condensation matters. Nat Chem Biol 19:1180\u0026ndash;1182\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSahin C, Leppert A, Landreh M (2023) Advances in mass spectrometry to unravel the structure and function of protein condensates. Nat Protoc 18:3653\u0026ndash;3661\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWu L, Quan B, Liu Y, Song R, Tang Z (2011) One-Pot Synthesis of Liquid Hg/Solid β-HgS Metal\u0026thinsp;\u0026ndash;\u0026thinsp;Semiconductor Heterostructures with Unique Electrical Properties. ACS Nano 5:2224\u0026ndash;2230\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLoh ND et al (2017) Multistep nucleation of nanocrystals in aqueous solution. Nat Chem 9:77\u0026ndash;82\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRosdahl J, Persson I, Kloo L, St\u0026aring;hl K (2004) On the solvation of the mercury(I) ion. A structural, vibration spectroscopic and quantum chemical study. Inorg Chim Acta 357:2624\u0026ndash;2634\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSchneider NM et al (2014) Electron\u0026ndash;Water Interactions and Implications for Liquid Cell Electron Microscopy. J Phys Chem C 118:22373\u0026ndash;22382\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYe M et al (2023) Revealing Dominant Oxidative Species in Reactive Oxygen Species-Driven Rapid Chemical Etching. Nano Lett 23:7319\u0026ndash;7326\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSies H, Jones DP (2020) Reactive oxygen species (ROS) as pleiotropic physiological signalling agents. Nat Rev Mol Cell Biol 21:363\u0026ndash;383\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAjsuvakova OP et al (2020) Sulfhydryl groups as targets of mercury toxicity. Coord Chem Rev 417:213343\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHauwiller MR et al (2019) Gold Nanocrystal Etching as a Means of Probing the Dynamic Chemical Environment in Graphene Liquid Cell Electron Microscopy. J Am Chem Soc 141:4428\u0026ndash;4437\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eK\u0026uuml;hne TD et al (2020) CP2K: An electronic structure and molecular dynamics software package - Quickstep: Efficient and accurate electronic structure calculations. J Chem Phys 152:194103\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePerdew JP, Burke K, Ernzerhof M (1996) Generalized Gradient Approximation Made Simple. Phys Rev Lett 77:3865\u0026ndash;3868\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVandeVondele J, Hutter J (2007) Gaussian basis sets for accurate calculations on molecular systems in gas and condensed phases. J Chem Phys 127:114105\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGrimme S, Ehrlich S, Goerigk L (2011) Effect of the damping function in dispersion corrected density functional theory. J Comput Chem 32:1456\u0026ndash;1465\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"},{"header":"METHODS","content":"\u003cp\u003e\u003cstrong\u003eSynthesis of α-HgS seeds\u003c/strong\u003e\u003cstrong\u003e.\u003c/strong\u003e The synthesis of \u003cem\u003eα-\u003c/em\u003eHgS seeds and the associated bipyramids was based on the wet-chemical reaction route reported previously \u003csup\u003e24,25\u003c/sup\u003e. Basically, 64 mg of Hg(NO\u003csub\u003e3\u003c/sub\u003e)\u003csub\u003e2\u003c/sub\u003e·H\u003csub\u003e2\u003c/sub\u003eO and 10 mL of deionized water were mixed in a 25 mL glass vial, followed by the addition of 2 mL of D-penicillamine aqueous solution (0.09 M). The as-formed solution was under ultrasonic treatment for 1 min, and then 0.3 mL of NaOH aqueous solution (2 M) was added into the mixture with stirring. After that, 1 mL of thioacetamide solution (0.18 M) was rapidly injected into the above solution. The vial was then sealed and kept into the 38 ℃ water bath for 15 h under stirring. The HgS seeds were collected by adding isopropanol into the orange-colored solution, followed by centrifugation at 7000 rpm for 10 min. Finally, such as-formed seeds were dispersed in 10 mL of deionized water for later usage.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSynthesis of α-HgS bipyramids.\u003c/strong\u003e Before the epitaxial growth of α-HgS with bipyramid morphology, Hg precursor solution was prepared by mixing 62 mg of Hg(NO\u003csub\u003e3\u003c/sub\u003e)\u003csub\u003e2\u003c/sub\u003e·H\u003csub\u003e2\u003c/sub\u003eO and 1.8 mL of D-penicillamine aqueous solution (0.09 M) into 6 mL of deionized water, followed by ultrasonic treatment to obtain colorless solution. Meanwhile, S precursor solution was prepared by dissolving 41 mg of thioacetamide in 20 mL of deionized water under ultrasonic treatment. 0.3 mL of the colloidal solution containing α-HgS seeds and 5 mL of deionized water were added into a three-neck round-bottom flask, followed by the addition of 1 mL of D-penicillamine aqueous solution (0.09 M) under stirring. Then, the flask was transferred into the 38 ℃ water bath with the addition of 0.6 mL of NaOH aqueous solution (2 M) into the flask under stirring. Afterwards, the Hg precursor and S precursor were co-injected into the flask through a syringe pump under an injection rate of 1 mL/h for 2.5 h.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAfter the reaction, the as-synthesized bipyramids were collected by adding isopropanol into the pale orange solution, followed by centrifugation at 7000 rpm for 10 min. The precipitated bipyramids were re-dispersed in 5 mL of deionized water for storage.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSynthesis of α-HgS ellipsoids.\u003c/strong\u003e 32 mg of Hg(NO\u003csub\u003e3\u003c/sub\u003e)\u003csub\u003e2\u003c/sub\u003e·H\u003csub\u003e2\u003c/sub\u003eO and 5 mL of deionized water were mixed in a 25 mL glass vial, followed by the addition of 1 mL of D-penicillamine aqueous solution (0.09 M). After the ultrasonic treatment for 1 min, 0.15 mL of NaOH aqueous solution (2 M) was added into the vial under stirring. 1 mL of thioacetamide solution (0.09 M) was then injected into above mixture solution quickly. The vial was sealed and located in the 47 ℃ water bath for 18 h under stirring. By adding isopropanol into the reaction solution, the as-formed ellipsoids were collected through the centrifugation at 7000 rpm for 10 min. Finally, the product was dispersed in 5 mL of deionized water.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePreparation of α-HgS polyhedrons.\u003c/strong\u003e 8 mg of Hg(NO\u003csub\u003e3\u003c/sub\u003e)\u003csub\u003e2\u003c/sub\u003e·H\u003csub\u003e2\u003c/sub\u003eO and 11.5 mL of deionized water were mixed in a 25 mL glass vial, with the further addition of 0.23 mL of D-penicillamine aqueous solution (0.1 M). The resulting solution was ultrasonically treated for 1 min, followed by the addition of 0.038 mL of NaOH aqueous solution (2 M) under stirring. After that, 0.23 mL of thioacetamide solution (0.1 M) was immediately injected into the above mixture solution. Then, the sealed vial was put into the 38 ℃ water bath under stirring for 15 h. The polyhedrons were collected by adding isopropanol into the reaction solution, followed by centrifugation at 7000 rpm for 10 min. The as-collected precipitates were finally dispersed in 5 mL of deionized water.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSynthesis of Hg-HgS Heterostructures.\u003c/strong\u003e 7.5 mL of Hg(ClO\u003csub\u003e4\u003c/sub\u003e)\u003csub\u003e2\u003c/sub\u003e·3H\u003csub\u003e2\u003c/sub\u003eO (0.01 M) was slowly added into 7.5 mL of cysteine (0.01 M) under stirring, followed by carefully adjusting the pH value of the mixture solution to 11.70 with the addition of NaOH (2 M). Afterwards, the sealed glass vial containing the above reaction solution was immediately kept into an oven with 37 ℃ for one week. During the reaction period, the color of the solution was changed from slight white to yellow green and then to black at the end. Finally, the precipitates of Hg-HgS heterostructures were collected after centrifuging and washing with the deionized water for several times.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eLiquid cell preparation.\u0026nbsp;\u003c/strong\u003eThree kinds of liquid cells (LCs), namely, the carbon LCs, graphene LCs and SiN\u003csub\u003ex\u003c/sub\u003e LCs were fabricated in this work. The carbon LCs were prepared by sandwiching the HgS nanoparticle (NP) colloidal solution with the addition of tris·HCl solution (0.01-0.02 M) between two ultrathin carbon films (thickness of ~10 nm). First, ~2 μL of the above solution was pipetted onto the carbon film-coated side of a 400-mesh gold grid, with the formation of a droplet locating nearly at the center of the grid. After that, another gold grid was quickly placed over this droplet with the carbon-coated side facing downward to encapsulate the droplet. The as-fabricated carbon LC was kept under ambient condition for 2 h before loading to the microscope. Liquid pockets were formed between the two ultrathin carbon films due to van der Waals forces, thereby allowing them for subsequent transmission electron microscopy (TEM) characterization. To study the buffer effect, other buffer solutions, e.g., the glycine-HCl was also used to replace tris·HCl for the LC preparation. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eStandard procedures were utilized to prepare the graphene LCs \u003csup\u003e51\u003c/sup\u003e. Briefly, the ultrathin carbon supported gold grids were placed onto the flat graphene (~3-5 layers) copper foil (~2 cm × 2 cm), followed by pipetting several drops of isopropanol onto the grids. These grids were attached with the graphene through evaporation of the isopropanol (kept in the ambient condition for ~2 h). The copper foil was etched by placing into the solution of sodium persulfate (1 g/10 mL) with the cooper side downward. Finally, these graphene coated grids were extracted from the etched solution, washed with deionized water to remove the etchant traces. Such graphene coated grids were then used to encapsulate the solution of interest for obtaining the graphene LCs (the following procedures were similar to the fabrication of carbon LCs).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFor the SiN\u003csub\u003ex\u003c/sub\u003e LCs, the sandwiched solution was the same as above. Electron transparent SiN\u003csub\u003ex\u003c/sub\u003e windows (~20 nm thick, 250 μm × 250 μm) supported by a 200-μm-thick silicon frame (diameter of ~3 mm) were used for encapsulating the solution (refer to fig. S3 for detailed illustration). One piece of such SiN\u003csub\u003ex\u003c/sub\u003e chips was deposited with a ~100-nm-thick layer of gold film as the spacer (the observation area of the electron transparent SiN\u003csub\u003ex\u003c/sub\u003e window was excluded). Before encapsulation, the SiN\u003csub\u003ex\u003c/sub\u003e chips were cleaned with plasma for ~10 s to effectively remove the surface contamination. After that, ~2 μL of the solution was pipetted onto the SiN\u003csub\u003ex\u003c/sub\u003e window of one chip with the gold film, followed by placing another chip onto the gold film to confine the thin liquid layer. Finally, the epoxy adhesive was used to seal the junction area between the two chips, and the liquid cell was realized after the adhesive was dried.\u003c/p\u003e\n\u003cp\u003eTo investigate the effect from the electrolytes, the typical NaCl (popularly existed in biological systems for offering insights of Hg droplets on human health as well) aqueous solution (0.02 M) was used as the sandwiched solution between two ultrathin carbon films. All the fabrication steps were the same as that shown in the carbon LCs.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eHeight profile characterization.\u003c/strong\u003e Briefly, ~20 μL of the diluted HgS colloidal solution was pipetted onto the clean monocrystalline silicon substrate (5 mm × 5 mm), which was kept in ambient condition for drying. The height profiles of the HgS NPs were characterized by atomic force microscopy (AFM, FastScan Bio, Bruker) with the maximum scan rate of 120 Hz and resolution in height of 0.03 nm.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eX-ray photoelectron spectroscopy analysis.\u003c/strong\u003e About 0.5 μL of liquid Hg and ~1 mL of Tris·HCl solution (10 mM) were mixed together as the sandwiched solution for the preparation of carbon LCs. After encapsulating such solution between two copper grids with ultrathin carbon films, the gap between the grid edges was sealed carefully by using the glue. The as-fabricated LC samples were\u0026nbsp;characterized by X-ray photoelectron spectroscopy (XPS, ESCALAB QXi, Thermo Scientific) to analyze the binding states of Hg.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIn-situ visualization of α-HgS nanoparticles evolution dynamics.\u003c/strong\u003e Direct observation of the evolution dynamics of the Hg nanodroplets and associated membranes from the HgS NPs were in-situ performed by transmission electron microscopy (Talos F200X, Thermo Fisher Scientific) with an accelerating voltage of 200 kV. The electron dose rate of the electron beam (e-beam) was kept in a very low level (1~10 e\u003csup\u003e−\u003c/sup\u003e Å\u003csup\u003e2\u0026nbsp;\u003c/sup\u003es\u003csup\u003e-1\u003c/sup\u003e) during searching for the appropriate samples. Afterwards, the electron dose rate was increased to a desired level (100~3000 e\u003csup\u003e−\u003c/sup\u003e Å\u003csup\u003e2\u0026nbsp;\u003c/sup\u003es\u003csup\u003e-1\u003c/sup\u003e in this work) for initiating the dynamical evolution of the samples. The dynamical processes in the LCs were then recorded by a digital camera. To reveal the elemental information during the evolution, the high-angle annular dark-field (HAADF) and energy dispersive X-ray spectroscopy (EDS) techniques in the scanning transmission electron microscopy (STEM) mode were utilized.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIn-situ heating characterization.\u003c/strong\u003e A few drops (~2 μL) of the dispersed solution containing the HgS nanobipyramids was pipetted onto the in-situ heating chip (the observation windows were made of thin SiN\u003csub\u003ex\u003c/sub\u003e layers), followed by drying under the ambient environment. After drying, the chip was loaded onto the TEM sample holder for the in-situ heating characterization. During the experiments, the temperature of the chip was kept at 100 ℃ with the fluctuation of less than 0.1 ℃ through the control system. The morphologies and diffraction patterns of the HgS nanobipyramids before heating and at 100 ℃ were then captured for the comparison.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCryo-EM observation of the Hg nanodroplets.\u003c/strong\u003e The LCs encapsulating the HgS nanobipyramids and 10 mM tris·HCl aqueous solution were loaded in the cryo-specimen holder. Initially, the Hg nanodroplets were created by the e-beam illumination at room temperature. Afterwards, the temperature of the sample was lowered to -100 ℃ through liquid nitrogen cooling (the melting point of Hg is -38.9 ℃). The observation procedure was the same as the above in-situ TEM characterization, and the emerging crystalline lattice of the freezing nanodroplets was captured for later analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eComputational model building.\u003c/strong\u003e In order to investigate the membrane structure and the behavior of Hg nanodroplets under different conditions, we constructed five different systems. First, to compare the aggregation behavior and stability of the membrane, we built the systems A and B. System A contained 72 H\u003csub\u003e2\u003c/sub\u003eO molecules and 60 neutral Hg atoms that were initially uniformly placed in a cubic box with a side length of 16.5 Å, while system B was constructed by removing a hydrogen atom from each H\u003csub\u003e2\u003c/sub\u003eO molecule in the above system, containing 72 OH\u003csup\u003e−\u003c/sup\u003e ions and 60 positively charged Hg ions. Further, to simulate the core-membrane structure, we built the system C that contained 200 H\u003csub\u003e2\u003c/sub\u003eO molecules, 100 OH\u003csup\u003e−\u003c/sup\u003e ions, and 150 Hg-associated species (either Hg atoms or cations) in a cubic box with a side length of 22.5 Å. In addition, the neutral Hg system related to system C was also constructed, which contained 300 H\u003csub\u003e2\u003c/sub\u003eO and 150 Hg atoms. Finally, to study the effect of H\u003csub\u003e2\u003c/sub\u003e on the core-membrane structure, we constructed a system D, which was based on the system C with additional insertion of 20 H\u003csub\u003e2\u003c/sub\u003e molecules. For comparison, another system E that contained 300 H\u003csub\u003e2\u003c/sub\u003eO molecules, 150 neutral Hg atoms, and 20 H\u003csub\u003e2\u003c/sub\u003e molecules was built.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eComputational setup.\u003c/strong\u003e We performed \u003cem\u003eab initio\u003c/em\u003e molecular dynamics simulations (AIMD) as implemented in CP2K software\u003csup\u003e52\u003c/sup\u003e.The Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional was used with short-range double-ζ basis set DZVP-MOLOPT-SR-GTH \u003csup\u003e53,54\u003c/sup\u003e. Grimme’s dispersion correction with Becke-Johnson damping (D3-BJ) was employed to account for weak dispersion interactions \u003csup\u003e55\u003c/sup\u003e. The self-consistent field (SCF) cycles in the simulations were converged using the orbital transformation method. Simulations were carried out within the canonical (NVT) ensemble with a timestep of 0.5 fs. Each simulation ran for more than 10 ps and the last 5 ps of the trajectory was used for the detailed analysis. The temperature was set to 298.15 K using a velocity rescaling thermostat with a time.\u0026nbsp;\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-6660582/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6660582/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eLiquid-liquid phase separation (LLPS), often forming membraneless compartments, is prevalent in proteins and polymers, allowing the functionalities for biological activities and soft material engineering. Yet, direct visualization and manipulation of the full membrane-bound LLPS evolutions with high spatiotemporal resolution remains challenging and undefined. Inspired by dynamic modulation from biological membranes, we in-situ design inorganic nanocells from exciting nanoscale cinnabar with simultaneously forming flexible liquid-like membranes and dense Hg nanodroplets by the electron-responsive LLPS strategy. A full LLPS picture from birth to disappearance, including membrane-associated gating of mass transport either in single nanocell or across multiple nanocells is vividly revealed. Periodic reversible cross-feeding occurs among nanodroplets confined in the single nanocell, in contrast to the conventional Ostwald ripening or coalescence behavior. However, once the ionic balance of the membranes is disturbed by nanobubbles or electrolytes, the nanodroplets collapse. The released less dense species proceed cell-to-cell transport over long distances through nanochannels and are irreversibly crystallized into Hg(I/II) compounds. \u003cem\u003eAb initio\u003c/em\u003e molecular dynamics simulations suggest that the nanodroplet-membrane interface undergoes dynamic charge fluctuations, recognizing the unique membrane-bound LLPS in inorganic systems. The flexible membrane is stabilized through the balance between Hg atoms and ions, which can be destroyed by nanobubbles.\u003c/p\u003e","manuscriptTitle":"Liquid-liquid phase-separated commensal membrane gates mass transport in inorganic nanocells","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-29 08:59:18","doi":"10.21203/rs.3.rs-6660582/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"431585d8-dd97-4a64-b16a-85f434a5e6bb","owner":[],"postedDate":"July 29th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":49665867,"name":"Physical sciences/Materials science"},{"id":49665868,"name":"Physical sciences/Nanoscience and technology"}],"tags":[],"updatedAt":"2025-07-29T08:59:18+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-29 08:59:18","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6660582","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6660582","identity":"rs-6660582","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}