3.1. Synthesis of GO-based porous polymers
Like previous works [43, 48], mother nonaqueous gel emulsions were successfully fabricated with DMSO as the continuous phase, paraffin oil as the dispersed phase, and a nonionic surfactant, Pluronic®F-127, as the stabilizer. Herein, GO-based porous polymers (GO-X) were prepared by dispersing GO within the DMSO phase prior to emulsification. The resultant gel emulsions demonstrated excellent stability, with no observed phase separation for over 1 week at ambient temperature or over 24 h at 70°C. Optical micrographs revealed the average dispersed droplet diameters were 33.38, 30.42, and 27.41 µm for gel emulsions with GO content of 4, 8, and 12 wt% in the continuous phase, respectively (Fig. S1). Rheological measurements confirmed the gel-like nature of these gel emulsions, as the storage modulus, G’, was consistently higher than the corresponding loss modulus, G’’ (Fig. S2a). Additionally, all the GO-based gel emulsions exhibited characteristic shear thinning behavior (Fig. S2b).
GO-X samples with high gel contents of over 90 wt% were successfully synthesized through step-growth polymerization. Densities of GO-4, GO-8, and GO-12 were determined to be 0.09, 0.11, and 0.13 g·cm− 3, respectively. FTIR spectra were used to analyze the chemical components of GO-X (Fig. 2a). All GO-X samples exhibited two distinct peaks: the C = C stretching vibration of the aromatic ring of polyMDI at 1595 cm− 1; and the amide-II (NH-C = O) bending absorption at 1540 cm− 1. The stretching absorptions at 1700 and 1660 cm− 1 corresponded to urethane C = O and urea C = O groups, respectively, which confirmed the presence of both polyurethane (PU) and polyurea (PUA) in GO-X [43].
Macroporous structures of GO-X were observed from SEM images (Fig. 2b, 2c, and 2d). Average macropore diameters were statistically analysed using the ImageJ, and the results were multiplied by 2/(31/2) to correct for the random nature of the section [49]. The calculated macropore diameters were 40.68, 37.60, and 33.89 µm for GO-4, GO-8, and GO-12, respectively. These results showed a strong correlation with the droplet diameters of GO-based gel emulsions. Moreover, the macropores were highly interconnected, and the pore throat sizes ranged from 2 to 15 µm. In the magnified SEM images, the rough scaffold was covered with nanosheets, which had a positive effect on subsequent emulsified oil separation.
3.2. Synthesis of PCL-based porous polymers
PCL-based porous polymers (PCL-Y) were templated by the similar nonaqueous gel emulsions, where PCL-triol was dissolved in the DMSO continuous phase. The PCL-based gel emulsions were found to remain stable at ambient temperature for over 1 week or at 70°C for 24 h, confirming the feasibility of solidifying the PCL-triol through step-growth polymerization. Compared with GO-based gel emulsions, dispersed droplets in PCL-based ones were smaller in size, with average diameters of 11.45, 9.41, and 8.10 µm for the PCL-5, PCL-10, and PCL-20 HIPE, respectively (Fig. S3). Rheological studies further confirmed gel-like properties of the emulsions (Fig. S4).
Monolithic PCL-Y samples were obtained through step-growth polymerization, purification and drying. The PCL-Y samples were also highly crosslinked with a gel content exceeding 90 wt%. However, compared to GO-X, corresponding PCL-Y samples exhibited slightly higher densities: 0.13, 0.16, and 0.23 g·cm− 3 for PCL-5, PCL-10, and PCL-20, respectively. The higher density was the result of shrinkage during fabrication. FTIR spectra of the PCL-Y showed two peaks appeared at 3510 and 3350 cm− 1, which corresponded to the free and bonded N-H stretching from PU or PUA (Fig. 3a) [48]. The stretching absorptions of the urethane C = O and urea C = O groups shifted to 1730 and 1710 cm− 1, respectively.
PCL-based porous polymers exhibited well-defined emulsion-templated macroporous structures, indicating greater stability of the gel emulsions during polymerization (Fig. 3b, 3c, and 3d). The polymeric scaffolds were visibly corrugated, particularly the PCL-5 and PCL-10, due to unavoidable shrinkage. The correctional average macropore diameters were located at 17.56, 14.50, and 11.47 µm for PCL-5, PCL-10, and PCL-20, respectively. The macropore sizes fell within 1 to 20 µm, and the average size of interconnecting pore throats ranged from 0.2 to 0.5 µm, like conventional emulsion-templated porous polymers [32].
3.3. Surface wettability and liquid uptake of non-layered porous polymers
The wettability of GO-X and PCL-Y samples were studied using water contact angles (WCAs). The instant WCAs for GO-X decreased sharply from 78.9° to around 64.8° as the GO content increased from 4 to 12 wt% (Fig. 4a ~ 4c). Moreover, the GO-X samples could be perfectly wetted by water at approximately 60, 40, and 25 s for GO-4, GO-8, and GO-12, respectively, due to the hydrophilicity of GO. On the other hand, the average WCAs for the PCL-Y slightly increased from 128.4° to 136.9° with increasing PCL content from 5 to 20 wt% (Fig. 4d ~ 4f). However, the WCAs were hardly be raised any further, possibly because the hydrophilic NH2 groups in PU or PUA and PEO blocks of F-127 were present during or after polymerization [48]. In general, the GO-X samples exhibited hydrophilic properties, while the PCL-Y samples were hydrophobic.
The GO-X and PCL-Y samples could preferentially absorb different liquids, making them suitable to produce asymmetric layered porous materials. The GO-X samples, specifically GO-4, GO-8, and GO-12, were hydrophilic and absorbed significant amounts of water, with capacities of approximately 28.4, 26.5, and 24.4 mL·g− 1, respectively (Fig. 4g). The water absorption capabilities of GO-X were directly related to their porosities. Additionally, the GO-X could still absorb a small quantity of organic solvents or oils, such as hexane, toluene, chloroform, soybean oil, peanut oil, and olive oil. Notably, the GO-4 exhibited highest uptake capacities among the GO-X samples, ranging from 2.5 to 12.0 mL·g− 1, with the maximum capacities observed for chloroform. The absorption of organic solvents or oils by GO-X may be attributed to the polyMDI structure in PU or PUA and PPO blocks of F-127. On the other hand, the hydrophobic PCL-Y showed observably higher uptake capacities of organic solvents or oils (Fig. 4h). Liquid uptakes within PCL-Y decreased with increasing PCL content, and PCL-5 exhibited the highest uptakes for water, hexane, toluene, chloroform, soybean oil, peanut oil, and olive oil, with values of 0, 19.8, 23.0, 26.8, 17.5, 16.2, and 13.9 mL·g− 1, respectively. The uptake capacities indicated that the uptake was mainly contributed by the original porosity, which is different from conventional polyHIPEs where uptake is mostly contributed by gel-swelling-driven pore expansion [32]. The low uptake of GO-X and PCL-Y, associated with pore expansion, could be explained by their high crosslinking degree, demonstrating their dimensional stability as separating membranes.
3.4. Preparation of GO/PCL Janus porous composites
Layered porous composites were obtained by patterning GO- and PCL-based gel emulsions, followed by polymerization, purification, and drying. The pore morphology, interface, and structural asymmetry were characterized using SEM (Fig. 5). Each layered porous composite exhibited a continuous interface and showed two distinct pore morphologies (Fig. 5a ~ 5c). Both sides exhibit a similar porous structure to the original GO-X and PCL-Y samples, but the pore homogeneity of the layered porous composites was lower. This difference in porous morphology was most noticeable in the GO4/PCL5 layered composite, which consisted of the two formulations with the highest porosity [39]. In general, each layer in the layered composites reflected the formulation used to prepare the GO- and PCL-based gel emulsions. Moreover, the interface became less curved as the content of GO and PCL increased, mainly due to the reduced shrinkage during and after the preparation process. Morphological studies confirmed that the patterning process of layered composites did not disrupt the final porous structures, but the interfaces were affected by the porosity of individual porous polymers. Additionally, the structural asymmetry of the GO4/PCL5 layered composite was examined using SEM images without gold sputtering (Fig. 5d). The SEM image showed that the GO4/PCL5 layered composite appeared smeared, suggesting that the entire composite has lower conductivity [50]. The PCL-Y samples were found to be insulative, leading to a high charge in the SEM image, whereas the conductive GO-X samples could be observed without gold sputtering. In Fig. 5d, the highly charged PCL-based layer and the conductive GO-based layer were well combined with a clear interface. Therefore, the layered composites exhibited strict asymmetry, which could be defined as Janus composites.
3.5. Mechanical analysis of GO/PCL Janus porous composites
Compressive stress-strain curves were presented for the GO-X and PCL-Y, which exhibited typical stress-strain behaviors seen in conventional polyHIPEs: a linear region at low strains, followed by a stress plateau region and finally an abrupt increase in stress at the densification or crushing region (Fig. 6a and 6b) [43, 48]. The Young’s modulus (E) of GO-X ranged from 1.6 to 10.2 MPa, while the E of PCL-Y ranged from 6.5 to 97.0 MPa. It was observed that the Young's modulus increased with the density of the corresponding porous polymers. Surprisingly, no failure was observed in these highly crosslinked GO-X and PCL-Y, even at a high compressive strain of 70%, attributed to the influence of their relatively high porosities on a deformation mechanism. Furthermore, the porous polymers, particularly the PCL-Y samples, exhibited excellent resilient-elasticity recovery (Fig. S5). For instance, the PCL-5 sample could completely recover even after being compressed with a high strain of 70%.
Mechanical properties of GO/PCL Janus porous composites were studied using an example, the GO4/PCL5 composite. This composite was compressed in both the vertical and horizontal directions of the interface (Fig. 6c). In the vertical direction, the GO4/PCL5 composite exhibited a compressive stress-strain curve like that of the homogeneous porous polymers. The Young’s modulus (E) for GO4/PCL5 composite was located at 4.7 MPa, which was approximately the arithmetic mean of the homogeneous GO4 and PCL5. However, the stress-strain curve was quite different when the GO4/PCL5 composite was horizontally compressed. The curve showed two elastic regions, which could be attributed to the destruction of the interface and the composite itself. The modulus was about 1.8 MPa in the first elastic region, while it increased to 4.1 MPa in the second region. Therefore, the strengthening of bonding interfaces was closer to that observed in GO-based porous polymers.
3.6. Asymmetric photothermal conversion of GO/PCL Janus porous composites
The temperature evolution on both sides of the GO4/PCL5 Janus porous composite was studied using an infrared thermal imager under the simulated solar irradiation with a xenon lamp (Fig. 7). The GO4 surface showed an efficient photothermal conversion performance. Specifically, the temperature increased from 22.7°C to 39.7°C (Fig. 7a), from 21.1°C to 79.3°C (Fig. 7b), and from 24.8°C to 115.2°C (Fig. 7c) within 90 s under 0.1, 0.6 and 1.2 kW·m− 2 irradiation, respectively. In contrast, the temperature of the PCL5 surface only increased from 20.4°C to 49.8°C after being irradiated under simulative solar of 1.2 kW·m− 2 for 90 s (Fig. 7d). GO exhibits outstanding photothermal properties due to its two-dimensional layer structure, which consists of stacked hexagonally arranged carbon atoms [51]. Moreover, the dispersion of GO was limited to the GO-based layers, causing an asymmetric photothermal conversion effect in the Janus porous composites. To gain a better understanding of this asymmetric photothermal conversion, the surface temperatures were recorded as a function of irradiation time (Fig. 7e). The temperature on GO4 surface rapidly climbed under strong simulated solar of 1.2 kW·m− 2, while the temperature on the PCL5 surface remained lower. The photothermal conversion effect on PCL5 surface (under 1.2 kW·m− 2 irradiation) was comparable to that of the GO4 surface under an extremely weaker simulated solar irradiation of 0.1 kW·m− 2. Interestingly, regardless of the conditions, the surface temperatures initially jumped to a high level during the early stages of irradiation and finally reached a temperature plateau when the heat transforming from the simulated solar and transferring to the surroundings were balanced.
3.7. Oil droplet transportation and emulsified oil separation
The Janus porous composites showed unique mass transportation behaviors due to their dual configuration with opposite wettability characteristics [5, 52–54]. It was well known that oil droplets could move through the composite under capillary forces, which determine the state of oil droplets when they encounter the porous media. In a Janus porous composite, its asymmetric wettability leaded to opposite Young-Laplace capillary pressures (P) acting on the oil droplet. The upward capillary pressure (P1), which was generated by the hydrophilic side, repelled the oil out of the composite, while the hydrophobic side created a downward capillary pressure (P2) that drew the oil into the pores. As a result, the oil droplet was directed across the Janus porous composites, achieving the oil/water separation.
The directional transmembrane phenomenon of an underwater chloroform droplet was experimentally recorded with a camera (Fig. 8a and 8b). The droplet transferred from the hydrophilic side to the hydrophobic one within 4 s (Fig. 8a and Movie S1). Initially, the chloroform droplet maintained a globular shape when it met the water-wetted hydrophilic side due to the underwater superoleophobicity. Then, the droplet shrank instead of spreading on the hydrophilic side, indicating it remained in a nonwettable state and underwent directional movement across the composite. Finally, the hydrophilic surface returned to its original state with no traces, confirming the successful transmembrane process of the oil droplet. In comparison, the transmembrane process under simulated solar irradiation was much faster, taking only 1 s (Fig. 8b and Movie S2). This phenomenon could be attributed to the temperature-drive viscosity reducing of chloroform droplets on the GO-based photothermal surface. These GO/PCL Janus porous composites have shown significant potential in the separation of high-viscous crude oil or/and edible oil which viscosity is primarily influenced by temperature, as described by Glaso’s formula [55, 56].
The separation of the oil-in-water miniemulsions was studied both without and with simulated solar irradiation (Fig. 8c and 8d). Chloroform-in-water miniemulsions were prepared using a phase inversion temperature (PIT) method, with the synergistic stabilization of the nonionic surfactant, Tween80, and the cationic surfactant, CTAB [45]. The as-obtained miniemulsions exhibited a typical Tyndall effect when a light beam passed through the separation column. OM images and DLS results confirmed that nanosized droplets were uniformly dispersed within the miniemulsion, with an average droplet diameter of approximately 887.5 nm. After being left to separate under gravity for over 2 h, the liquid within the column became clear without the Tyndall effect (Fig. 8c). Less visible droplets were found in the OM image, and the average size was jumped to approximately 61.1 nm. The separation mechanism of miniemulsions might involve the synergistic effect between the interlaced transmission channel inside the composites [57] and the negative charge of GO on the channel wall [58]. The separation process was significantly accelerated under simulated solar irradiation, resulting in a clear water phase within 30 min without any visible oil droplets from the OM image (Fig. 8d). The infrared spectrophotometer analysis about the depurated water phase confirmed an extremely low oil concentration of about 0.05%. The enhanced separation efficiency could be attributed to the reduction in oil phase viscosity [59] and the phase separation of the nonionic surfactant (cloud point) [60]. In summary, the Janus porous composites could effectively separate oil-in-water miniemulsions combined effects of multiple factors and mechanisms, which were further enhanced under simulated solar irradiation.