Machine Learning in Membrane Design: From Property Prediction to AI-Guided Optimization

Abstract

Porous membranes, either polymeric or two-dimensional materials, have been extensively studied because of their outstanding performance in many applications such as water filtration. Recently, inspired by the significant success of machine learning (ML) in many areas of scientific discovery, researchers have started to tackle the problem in the field of membrane design using data-driven ML tools. In this Mini Review, we summarize research efforts on three types of applications of machine learning in membrane design, including (1) membrane property prediction using ML, (2) gaining physical insight and drawing quantitative relationships between membrane properties and performance using explainable artificial intelligence, and (3) ML-guided design, optimization, or virtual screening of membranes. On top of the review of previous research, we discuss the challenges associated with applying ML for membrane design and potential future directions.

Fast Water Desalination with a Graphene–MoS2 Nanoporous Heterostructure

“…In recent years, severe droughts have ravaged all the world, leaving behind significant socioeconomic, environmental, and ecological consequences. − Providing fresh water for the world population has become more challenging due to global warming and deprecation of water resources. Water desalination through reverse osmosis (RO) can be a viable solution to provide fresh water to people since 71% of the earth’s surface is covered with saline seawater. − However, this method still has the drawback of high energy consumption due to the low water permeation rates of traditional polymeric or zeolite membranes. , The ultrathin nature of 2D materials facilitates the accelerated permeation of water molecules through nanopores, resulting in a more energy-efficient RO water desalination process. − Nanoporous graphene (NPG) membranes exhibit water flux rates that are 2–3 orders of magnitude higher than conventional RO membranes, while maintaining a high percentage of salt rejections. − Furthermore, the exceptional mechanical strength exhibited by 2D materials enables them to withstand the pressures encountered during RO desalination operations. , Recently, some studies presented a comprehensive analysis of how the presence of multiple layers, the distance between layers, and the alignment of pores affect the desalination performance of NPG membranes. − The ion rejection rate and water permeation in nanopores used for water desalination can also be influenced by geometrical and material factors of nanopores. , The shape and size of the pores play a critical role in determining the efficiency and effectiveness of the desalination process; − however, the challenge in optimizing the membrane is the trade-off between ion rejection and permeation rate. Maximizing the permeation rate sacrifices the ion rejection efficiency.…”

“… 21 − 24 The ion rejection rate and water permeation in nanopores used for water desalination can also be influenced by geometrical and material factors of nanopores. 25 , 26 The shape and size of the pores play a critical role in determining the efficiency and effectiveness of the desalination process; 27 − 35 however, the challenge in optimizing the membrane is the trade-off between ion rejection and permeation rate. Maximizing the permeation rate sacrifices the ion rejection efficiency.…”

A practical machine learning approach for predicting the quality of 3D (bio)printed scaffolds

“…In the contemporary landscape of biomedical research, AI has emerged as a transformative force across various domains of regenerative medicine, encompassing drug discovery, disease modeling, predictive modeling, personalized medicine, tissue engineering, cell therapy, clinical trial design, patient monitoring, patient education, and regulatory compliance [16][17][18][19][20]. Our examination specifically narrows to the integration of AI within tissue engineering.…”

Combined Machine Learning and High-Throughput Calculations Predict Heyd–Scuseria–Ernzerhof Band Gap of 2D Materials and Potential MoSi2N4 Heterostructures

“…The advancement of novel 2D systems is being propelled by traditional analytical methods such as density functional theory (DFT) and molecular dynamics (MD), which have been utilized to examine the microstructures of materials and predicted future investigations. Nonetheless, the increasing complexity of data and constrained computational resources have made these methods increasingly time-intensive . To address those challenges, researchers are now leveraging the capabilities of data-driven machine learning (ML) techniques to overcome these barriers.…”