Machine learning for layer-by-layer nanofiltration membrane performance prediction and polymer candidate exploration

In this study, machine learning-based models were established for layer-by-layer (LBL) nanofiltration (NF) membrane performance prediction and polymer candidate exploration. Four different models, i.e., linear, random forest (RF), boosted tree (BT), and eXtreme Gradient Boosting (XGBoost), were formed, and membrane performance prediction was determined in terms of membrane permeability and selectivity. The XGBoost exhibited optimal prediction accuracy for membrane permeability (coefficient of determination (R2): 0.99) and membrane selectivity (R2: 0.80). The Shapley Additive exPlanation (SHAP) method was utilized to evaluate the effects of different LBL NF membrane fabrication conditions on membrane performances. The SHAP method was also used to identify the relationships between polymer structure and membrane performance. Polymers were represented by Morgan fingerprint, which is an effective description approach for developing modeling. Based on the SHAP value results, two reference Morgan fingerprints were constructed containing atomic groups with positive contributions to membrane permeability and selectivity. According to the reference Morgan fingerprint, 204 potential polymers were explored from the largest polymer database (PoLyInfo). By calculating the similarities between each potential polymer and both reference Morgan fingerprints, 23 polymer candidates were selected and could be further used for LBL NF membrane fabrication with the potential for providing good membrane performance. Overall, this work provided new ways both for LBL NF membrane performance prediction and high-performance polymer candidate exploration. The source code for the models and algorithms used in this study is publicly available to facilitate replication and further research. https://github.com/wangliwfsd/LLNMPP/.

Preparation of Chemically Resistant Cellulose Benzoate Hollow Fiber Membrane via Thermally Induced Phase Separation Method

For the first time, we have successfully fabricated microfiltration (MF) hollow fiber membranes by the thermally induced phase separation (TIPS) and non-solvent induced phase separation (NIPS) methods using cellulose acetate benzoate (CBzOH), which is a cellulose derivative with considerable chemical resistance. To obtain an appropriate CBzOH TIPS membrane, a comprehensive solvent screening was performed to choose the appropriate solvent to obtain a membrane with a porous structure. In parallel, the CBzOH membrane was prepared by the NIPS method to compare and evaluate the effect of membrane structure using the same polymer material. Prepared CBzOH membrane by TIPS method showed high porosity, pore size around 100 nm or larger and high pure water permeability (PWP) with slightly low rection performance compared to that by NIPS. On the contrary, CBzOH membranes prepared with the NIPS method showed three times lower PWP with higher rejection. The chemical resistance of the prepared CBzOH membranes was compared with that of cellulose triacetate (CTA) hollow fiber membrane, which is a typical cellulose derivative as a control membrane, using a 2000 ppm sodium hypochlorite (NaClO) solution. CBzOH membranes prepared with TIPS and NIPS methods showed considerable resistance against the NaClO solution regardless of the membrane structure, porosity and pore size. On the other hand, when the CTA membrane, as the control membrane, was subjected to the NaClO solution, membrane mechanical strength sharply decreased over the exposure time to NaClO. It is interesting that although the CBzOH TIPS membrane showed three times higher pure water permeability than other membranes with slightly lower rejection and considerably higher NaClO resistance, the mechanical strength of this membrane is more than two times higher than other membranes. While CBzOH samples showed no change in chemical structure and contact angle, CTA showed considerable change in chemical structure and a sharp decrease in contact angle after treatment with NaClO. Thus, CBzOH TIPS hollow fiber membrane is noticeably interesting considering membrane performance in terms of filtration performance, mechanical strength and chemical resistance on the cost of slightly losing rejection performance.

Preparation of Microfiltration Hollow Fiber Membranes from Cellulose Triacetate by Thermally Induced Phase Separation

For the first time, self-standing microfiltration (MF) hollow fiber membranes were prepared from cellulose triacetate (CTA) via the thermally induced phase separation (TIPS) method. The resultant membranes were compared with counterparts prepared from cellulose diacetate (CDA) and cellulose acetate propionate (CAP). Extensive solvent screening by considering the Hansen solubility parameters of the polymer and solvent, the polymer's solubility at high temperature, solidification of the polymer solution at low temperature, viscosity, and processability of the polymeric solution, is the most challenging issue for cellulose membrane preparation. Different phase separation mechanisms were identified for CTA, CDA, and CAP polymer solutions prepared using the screened solvents for membrane preparation. CTA solutions in binary organic solvents possessed the appropriate properties for membrane preparation via liquid-liquid phase separation, followed by a solid-liquid phase separation (polymer crystallization) mechanism. For the prepared CTA hollow fiber membranes, the maximum stress was 3-5 times higher than those of the CDA and CAP membranes. The temperature gap between the cloud point and crystallization onset in the polymer solution plays a crucial role in membrane formation. All of the CTA, CDA, and CAP membranes had a very porous bulk structure with a pore size of ∼100 nm or larger, as well as pores several hundred nanometers in size at the inner surface. Using an air gap distance of 0 mm, the appropriate organic solvents mixed in an optimized ratio, and a solvent for cellulose derivatives as the quench bath media, it was possible to obtain a CTA MF hollow fiber membrane with high pure water permeance and notably high rejection of 100 nm silica nanoparticles. It is expected that these membranes can play a great role in pharmaceutical separation.

Functionalization of PEG-AgNPs Hybrid Material to Alleviate Biofouling Tendency of Polyethersulfone Membrane

Membrane-based processes are a promising technology in water and wastewater treatments, to supply clean and secure water. However, during membrane filtration, biofouling phenomena severely hamper the performance, leading to permanent detrimental impacts. Moreover, regular chemical cleaning is ineffective in the long-run for overcoming biofouling, because it weakens the membrane structure. Therefore, the development of a membrane material with superior anti-biofouling performance is seen as an attractive option. Hydrophilic-anti-bacterial precursor polyethylene glycol-silver nanoparticles (PEG-AgNPs) were synthesized in this study, using a sol-gel method, to mitigate biofouling on the polyethersulfone (PES) membrane surface. The functionalization of the PEG-AgNP hybrid material on a PES membrane was achieved through a simple blending technique. The PES/PEG-AgNP membrane was manufactured via the non-solvent induced phase separation method. The anti-biofouling performance was experimentally measured as the flux recovery ratio (FRR) of the prepared membrane, before and after incubation in E. coli culture for 48 h. Nanomaterial characterization confirmed that the PEG-AgNPs had hydrophilic-anti-bacterial properties. The substantial improvements in membrane performance after adding PEG-AgNPs were evaluated in terms of the water flux and FRR after the membranes experienced biofouling. The results showed that the PEG-AgNPs significantly increased the water flux of the PES membrane, from 2.87 L·m⁻²·h⁻¹ to 172.84 L·m⁻²·h⁻¹. The anti-biofouling performance of the PES pristine membrane used as a benchmark showed only 1% FRR, due to severe biofouling. In contrast, the incorporation of PEG-AgNPs in the PES membrane decreased live bacteria by 98%. It enhanced the FRR of anti-biofouling up to 79%, higher than the PES/PEG and PES/Ag membranes.

Evaluation of process condition impact on copper and lead ions removal from water using goethite incorporated nanocomposite ultrafiltration adsorptive membranes

Polyacrylonitrile (PAN) adsorptive membrane incorporated with nanosize-goethite (α-FeO(OH)) hydrous metal oxide particles (GNPs), prepared with optimal flux and Cu(II) removal in the previous study, was used to evaluate the process parameter on the Cu(II) removal. Box-Behnken Design (BBD) based on the Response Surface Methodology (RSM) was employed to evaluate the impact of Cu(II) feed solution characteristics such as pH, initial concentration of metal ion, and transmembrane pressure (TMP) on copper removal efficiency. The outcomes indicated that the RSM optimization technique could be utilized as an applicable method to find the optimum condition for the maximum Cu(II) removal with slight variance compared with the experimentally measured data. The effect of each process parameter and the coupling effect of parameters on the Cu(II) removal was assessed. Finally, the optimum condition of pH, Cu(II) concentration, and transmembrane pressure (TMP) to obtain high copper removal efficiency was decided. In the optimum condition of the Cu(II) removal, the removal of lead (Pb(II)) metal ion was evaluated by the same membrane.

Ultrafiltration of α-Lactalbumin Protein: Acquaintance of the Filtration Performance by Membrane Structure and Surface Alteration

α-Lactalbumin is an essential protein with multiple roles in physiological and the nutritional functionalities, such as diabetic prevention, blood pressure stabilization, and cancer cell inhibition. In the present work, polyethersulfone (PES)-based membranes were developed by incorporating Pluronic F127 and carbon nanotubes with single- and multi-walled dimensions (Sw-Cnts and Mw-Cnts) as additives. The resulting membranes were evaluated for use in the filtration of α-lactalbumin protein solution. Four series of membranes, including PES pristine membrane, were fabricated via the phase inversion process. The characteristics of the membrane samples were analyzed in terms of morphology, membrane surface hydrophilicity and roughness, and surface chemistry. The characterization results show that the incorporation of additive increased the surface wettability by reducing the surface water contact angle from 80.4° to 64.1° by adding F127 and Mw-Cnt additives. The highest pure water permeability of 135 L/(m2·h·bar) was also exhibited by the PES/F127/Mw-Cnt membrane. The performance of the modified membranes was clearly better than the pristine PSF for α-lactalbumin solution filtration. The permeability of α-lactalbumin solution increased from 9.0 L/(m2·h·bar) for the pristine PES membrane to 10.5, 11.0 and 11.5 L/(m2·h·bar) for membranes loaded with Pluronic F127, Sw-Cnts, and Mw-Cnts, respectively. Those increments corresponded to 17, 22, and 28%. Such increments could be achieved without altering the α-lactalbumin rejections of 80%. Remarkably, the rejection for the membrane loaded with Sw-Cnts even increased to 89%.