Water hyacinth can act as “nature’s kidney” by accumulating metal ions from polluted water, thus preserving the Earth’s valuable water resources(Zheng et al., 2016). Water hyacinth can effectively expel heavy metal ions, such as lead, copper, and zinc ions(Singh, Kumar & Kumar, 2022). The root system of water hyacinth absorbs toxic compounds present in wastewater. The accumulation processes include heavy metal uptake by the root cell wall, metal transport from the cell wall to the inner cell, cellular compartmentation, and sequestration (Fig. 1a). Almost 99% of metals in water can be removed and enriched in plant cells(Huynh, Chen & Tran, 2021). However, water hyacinth has been rated by the International Union for Conservation of Nature (IUCN) as one of the 100 most aggressive invasive species in the world and one of the top 10 weeds in the world(Thamaga & Dube, 2019). This has prompted attempts to investigate green substitute systems(Wu et al., 2022), in which cellulose and lignin are most attractive for the construction of a cellular-like separation system. As shown in Fig. 1b, this system consists of a nanocellulose/lignin-based microdevice (NLMD) with a wrapped oil membrane layer. The NLMD has a porous structure for the accommodation of metal ions, while the oil layer works similar to the cell wall membrane in water hyacinth root. Beside the structure, the ion transportation mechanism was also inspired by water hyacinth: Metal Uptake Protein and Metal Efflux Protein in plant cell are corresponding to the carrier (D2EHPA) and stripping agent (H2SO4) in our biomimetic membrane system (BMS). Both of the metal transportations were due to the ion exchange at the double interfaces. Detailed, the D2EHPA in the oil membrane (Fig. 1d) simulate the metal efflux protein in the plant cell (Fig. 1c), which form complex interaction with metal ions and make the metal ions passable to the NLMD. And at the interfaces of membrane, the metal ions are able to be locked in the NLMD (Fig. 1d) by the ion exchange, like whatever metal uptake protein strip the metal ions in the plant cell (Fig. 1c) (Duan et al., 2019; Wu, Saleem, He & He, 2021). Thus, metal ions can be selectively separated by the oil membrane and locked in the NLMD. This biomimetic membrane system was expected to be an efficient biomimetic system for the extraction and concentration of metal ions from water for the high efficiency liquid membrane and high-pour NLMD. In addition, we envisioned that the BMS could work as a versatile separation system for metal ions, by just altering the carrier and stripping agent in the oil membrane and NLMD, respectively.
As shown in Fig. 2a, the NLMD is prepared by using an ultrasonic atomizer, whereby the mixed aqueous suspension (nanocellulose + lignin-PAE, Fig. 2b) was atomized to aerosol and frozen in liquid nitrogen. The lignin-PAE nanoaggregates was prepared by mixing prepared by mixing equal vlolume of the 2 wt%lignin and 2 wt%PAE, and the mixture with 2 wt% NCF suspension is without any precipitates (Fig. S1). After the atomization, the ice template was removed by freeze-drying to avoid nanocellulose skeleton collapse, thus enabling the highly porous structural NLMD. This technique has been maturely used for the preparation of many porous cellulose structures(Cai et al., 2014). After the structure shaping, the NLMD was further heated at 100 ℃ for 30 min to allow the crosslinking between PAE-LS and the nanocellulose, and the mechanism is discussed in the next section. The morphologies of the NLMD are shown in Fig. 2c–e with three different magnifications, which reveal the particle size and the microstructure of the NLMD. It is noteworthy that the particle size is almost linearly increased with the ultrasonic output, that is, a higher power yields a smaller particle size (Fig. 2f). In this study, 50 KHz was implemented to produce an NLMD with a mean size of 12 µm, because a smaller particle size is conductive to water in oil stability. From the calculation from SEM, the particle size is shown within 1–25 µm, and most of the NLMDs are distributed in the range of 5–20 µm at pH = 5 (Fig. 2h). With a higher magnification as shown in Fig. 2d, e, the NLMD is shown to be a very highly porous structure with a porosity of ~ 95%. Lignin-PAE nanoaggregates (average size ~ 80 nm, Fig. 2f) were utilized as a strengthening agent to crosslink the NLMD. This strategy has been previously reported to prepare high mechanical performance cellulose paper and foams(Yin et al., 2020). The particle size and charged properties were tested for analyzing the stability of the nanoaggregates as shown in Fig. S2, they show similar diameters but the decreased zeta potential from pH = 1 to 11. Moreover, for the higher pH value condition (e. g. pH ༞12), the lignin-PAE aggregate was turning to precipitate because the electric charge of PAE was screened(Zhang, Lan, Peng, Hu & Zhao, 2020). Nevertheless, the lignin-PAE adhered and bonded with the nanocellulose skeletons without covering the pores (Fig. 2e).
Figure 3 shows the interaction mechanism between NCF and lignin-PAE, which is beneficial to the high mechanical stability of the NLMD, as demonstrated by NLMDs that could withstand 12 h of high-speed agitation after being treated in a hydrothermal reactor at 100 ℃ for 24 h (Fig. 3e). For comparison, NCF microparticles without crosslinking were completely damaged after being stirred at 8000 rpm for 100 min (Fig. 3d, left part), and NCF microparticles crosslinked by merely PAE showed only partially damaged particles after being treated at 8000 rpm for 20 min (Fig. 3d, right part). Figure 3a reveals the interaction schematic involving lignin, PAE, and NCF. The lignin-PAE nano-aggregates fill in the cellulose nanofibrils and form networks through a combination of hydrogen bonding and covalent crosslinking. PAE could rapidly form covalent crosslinking at high temperature (100 ℃, 30 min) on the basis of freeze-dried structures, thus enabling highly mechanical stable networks, that is, the polymeric chain coupling emerging between the azetidium group of PAE and OH, and the interchain crosslinking of azetidium groups with RCOO− from lignin and cellulose(Varanasi, Low & Batchelor, 2015). As confirmed by the FT-IR curves (Fig. 3b), new peaks at 1733 and 1260 cm− 1 appear for the NLMD. These bonds confirm the formation of ester bonds due to the PAE crosslinking during the drying process(Liao et al., 2022), and the mechanism was illustrated in Fig. S3. Accordingly, a significant endothermic peak (52–185°C) appears during the first heating of the NMLD (Fig. 3c), while there is no endothermic peak at the second heating of the DSC curve. These could be explained by the crosslinking reactions as discussed. Overall, the multi-crosslinking structures were successfully established, resulting in the NLMD having excellent mechanical stability and bearing many harsh environments during the water treatment processes.
BMS is formed by dispersing the water absorbed NLMD in an oil phase through mechanical agitation, which is shown in Fig. 4a, b. In this respect, the surface wetting property of the NLMD is an important factor for the stability of BMS. The amphiphilic properties of cellulose and lignin have special affinities for oil and water, thus making them strongly capable of adsorbing polar and nonpolar substances, as well as promoting oil/water interfacial stability (Fig. 4c). Benefiting from these, cellulose and lignin nanoaggregates are widely used for the establishment of pickering emulsions(Guo et al., 2021). Nevertheless, cellulose is much more hydrophilic as compared with lignin, due to its intrinsic molecular structure(Bao, He, Song, Guo, Zhou & Liu, 2022). Hence, the component ratio of NCF/lignin in the NLMD may have a significant influence on the stability of BMS. To investigate this, NLMDs with different lignin/cellulose component ratios were prepared by changing the mixture ratio of NCF and lignin-PAE before the atomization. As shown in Fig. 4d, the lignin contents vary from 0 to 50 wt.%, the samples with a higher lignin-PAE content became more denser in their architecture (Fig. S4). From Fig. 4d,e, the samples with higher lignin content show a remarkable decrease in water absorbency and an increase in water contact angle, indicating a more hydrophobic nature of lignin as compared with cellulose in NLMD (Fig. S5). The interfacial tension tests (Fig. 4f) were carried out to measure the oil/water interfacial tension on the surface interface of NLMD, and the values decreased from 55 to 40 mN/m, corresponding to samples from lignin free to 50 wt.% lignin content. This indicated that the BMS had become more stable by introducing lignin. In addition, the emulsion fraction (Fig. 4g) correspondingly confirms that the lignin incorporation is beneficial to build a more stable BMS. Meanwhile, BMS made by NMLD-2 can retain at least 24 h and the time is sufficient for the extraction application (Fig. 4h). Generally speaking, tedious modification procedures are usually needed to link hydrophilic and hydrophobic groups to the surfaces of materials for obtaining superamphiphilic materials. Therefore, constructing hydrophilic and oleophilic microdomains on the surface of materials simultaneously is a major challenge(Song, Zhou, Fan, Zhai, Meng & Wang, 2018). The NLMD is composed of three components, while lignin and cellulose are amphiphilic substances, which are instrumental in the oil/water interfacial stability in addition to the enhancement in mechanical stability. In addition, the NMLD with increased lignin content shows a decrease in water absorbency (Fig. 4e), and this is because the vacant space is occupied by lignin-PAE as explicated by the bulk density results. Considering the trade-off between water accommodation capability and oil/water stability, NLMD-2 with a lignin content of 15 wt.% is chosen for the BMS preparation (Fig. 4i).
A biomimetic membrane system (BMS) was employed to extract heavy metal ions from water (Fig. 5), which consisted of a stripping agent (H2SO4) inside the NLMD and a carrier (D2EHPA) dissolved in the organic phase, and the optical microscopic image of BMS for the extraction of metal ions was shown in Fig. S6. In the extraction process, the BMS was poured into the to-be treated water and simultaneously agitated at 300 rpm, and the BMS was divided into small droplets with sizes of several centimeters by visual inspection (Fig. 5a). The BMS droplets were utilized for the extraction of the metal ions (Pb2+/Zn2+/Cu2+); the ions were transported from the outer water to the internal NLMD as illustrated in Fig. 5b. Bis(2-ethylhexyl)phosphoric acid (D2EHPA) is a well-known extractant on the basis of liquid–liquid extraction, and has been proven to be an efficient carrier for the extraction of Pb2+/Zn2+/Cu2+(Li, Zeng, Du, Zhang, Cao & Wu, 2022; Salman & Mohammed, 2019). In the organic phase, D2EHPA presents as a dimer and reacts with Pb2+/Zn2+/Cu2+ to form a complex of metal2+–D2EHPA [Pb2+/Zn2+/Cu2+R(HR)] at the external interface of the membrane phase, and carries Pb2+/Zn2+/Cu2+ through the membrane phase (Fig. S7). As a proof of concept, the D2EHPA dosage in the organic phase plays an important role for the extraction capability, and 4 wt.% D2EHPA is best in the extraction (Fig. 5e). Meanwhile, at the interface of the NLMD, Pb2+/Zn2+/Cu2+ is stripped by the H2SO4 and transformed into a new species that cannot penetrate the NLMD reversibly(Hong, Yimin, Jing, Tao, Nannan & Kui, 2017). The reactions that occur at the double interfaces of the membrane phase with D2EHPA as carrier and H2SO4 as stripping agent are shown in Fig. 5c.
Practically, the BMS extraction capabilities toward outer Cu2+ concentration, the volume ratio of BMS to external water, and pH values of the external water were studied (Fig. S8). On this premise, the metal ion concentration in water was detected to monitor the extraction kinetics of BMS (Fig. 5d). For three Pb2+/Zn2+/Cu2+ curves, they show almost 80% removal efficiency of metal ions around 2 min and approximately 90% removal efficiency around 3 min. The Pb2+ extraction capability is the best followed by Zn2+ and Cu2+, and there is no significant difference in their extraction kinetics. Moreover, there has developed a realistic model for the analysis of the comprehensive account of the theoretical aspects of mass transfer with chemical reactions for the D2EHPA and H2SO4 in the liquid membrane system (Fig. S9) (Abd Khalil, Shah Buddin, Puasa & Ahmad, 2023; Sulaiman, Jusoh, Othman, Noah, Rosly & Rahman, 2019). According to that, zinc ions extraction can be predicted as shown in Fig. S10 and shows almost in accordance with the experimental data (Fig. 5d black line), which was attributing to the high porosity of NLMD that makes no obstacle for the mass transfer in our BMS. Significantly, the liquid/membrane-based extraction of the BMS allowed for high-speed mass transfer; thus, the equilibrium time of the BMS is much higher than those of the reported cellulose/lignin-based absorbent materials (Fig. 5f) (Chen et al., 2022; Gao, Zhang, Liu & Liu, 2022; Godiya, Cheng, Li, Chen & Lu, 2019; Liu, Fan, Wang, Zhang, Li & Wang, 2022; Qiao, Li, Li, Liu & Du, 2020; Sirviö & Visanko, 2020; Tang et al., 2020).
The detachment of BMS could be easily performed using a filtration process as shown in Fig. 6a, additionally, with the fast shape-recovery property of NLMD, which is highly desired for the collection of concentrated metal ions (Fig. 6b) as well as the regeneration for repeat extraction (Fig. 6c). The shape-recovery performance is shown in Fig. 6d & Fig. S11. The original water absorbed NLMD was set at a volume of 10 mL, while the compressed NLMD was dewatering and the volume was sharply reduced to 2.5 mL, and when reabsorbing water, the NLMD quickly recovered its original shape within 1 s. This water-responsive shape recovery greatly exceeded that reported for cellulose foams. A possible reason is given: On the one hand, lignin on a fiber surface imparts the fiber with elasticity and can partially prevent the fiber from swelling; on the other hand, the co-crosslinking between cellulose and the lignin–PAE complex makes the network tough and flexible. The stripping agent (H2SO4) content in the NLMD in the absorbing-dewatering cycles was measured as shown in Fig. 6e, and showed that the water absorbing of NLMD maintained its original capability at least up to 50 cycles, because the highly-crosslinking renders the porous structure strong enough to bear the extraction-regeneration treatments (Fig. S12). As shown in Fig. 6f, the extraction capability of BMS toward Pb2+/Zn2+/Cu2+ shows no attenuation up to 50 cycles (Fig. 6f). Metal ions could be highly concentrated by BMS extraction, and easily collected by squeezing the concentrated metal ions out of the NLMD. As shown in Fig. 6g, the metal ion concentration in NLMD is ~ nine times higher than that of the initial external water (Fig. 6g). In addition, this value is almost equal to the ideal metal ion concentration calculated from the reduced metal ions of the external water, indicating that the interactions between metal ions and NLMD are neglected. Overall, the straightforward detachability of BMS allows for durable re-extraction and concentration of heavy metals other than high-performance extraction, making it a high-efficient process for the extraction applications.
Although the extractions by BMS and the conventional emulsion liquid membrane systems are both based on the liquid-liquid mass transfer, the BMS is superior for its easy demulsificaiton compared with the conventional emulsion liquid membrane systems (Fig. 6f-h). Figure 6f shows the extraction capability of our BMS which is on par with the conventional emulsion liquid membrane system (prepared by the same component without NLMD). However, the conventional demulsification process took more than 20 to 100 minutes for the membrane demulsification, while it took only 1 minute for our BMS through filtration as shown in Fig. 6a. In addition, the demulsification performances (demulsification ratio) by the conventional methods (Fig. 6h) are unsatisfied because the very stable emulsions that is hard to be broken down completely. Overall, except the contributions to the easy preparation of the BMS and the high-efficient extraction (Fig. 5) of the metal ions. The NLMD (cellulose-lignin based materials) also contributes to the easy demulsification of the BMS that are hardly attainable in the conventional emulsion liquid membrane systems.