An AI-driven inverse design framework identifies optimal polymer structures for enhanced oil recovery by mapping target viscosity to molecular features, achieving a 12% improvement in viscosity through simulation-guided optimization of polymer topology and functionality.

This review summarizes machine learning strategies tailored for polymers under small data regimes, highlighting key advances in data augmentation, domain-informed modeling, and efficient learning paradigms such as transfer and multitask learning. These approaches enhance predictive performance despite data scarcity.

Machine learning models were developed to predict Young's modulus, tensile strength, and elongation at break, exploring the chemical space of thousands of polyimide candidates and offering a cost-effective approach for designing high-performance films.

Self-consistent field theory (SCFT) enables precise calculation of block copolymer phase diagrams, yet it critically depends on human expertise for selecting candidate structures. This work integrates artificial intelligence with SCFT to achieve automated search of self-assembled structures and autonomous construction of phase diagrams, thereby minimizing reliance on human prior knowledge.

We demonstrate the effectiveness of the Simulated Bifurcation (SB) algorithm, based on the concept of conjugate fields, in polymer sequence optimization. Despite requiring numerical gradients, SB outperforms stochastic methods like simulated annealing and exhibits excellent scalability in high-dimensional design spaces.

OpenPoly offers 3,985 curated polymer–property pairs across 26 properties. A unified benchmark shows XGBoost surpasses deep models in sparse data, and the database enables rapid identification of candidates for high-temperature dielectrics and fuel-cell membranes, advancing data-driven polymer discovery.

The work presents a reactive machine learning force field for crosslinked epoxy, capable of accurately predicting reactive events and thermo-mechanical properties. The force field operates approximately 1200 times faster than ab initio molecular dynamics, offering a promising approach for simulating complex polymer systems with quantum-level precision.

We developed a step growth polymerization of diol/COS(CS₂)/dichloride yielding poly(thiocarbonate)s. By suppressing oxygen-sulfur exchange reactions, structurally controlled poly(monothiocarbonate)s were obtained by polymerization of diol/COS/dichloride, exhibiting mechanical properties comparable to high-density polyethylene, while alcoholysis granted recyclability. CS₂-based polymerizations allowed tunable structures and properties via oxygen-sulfur exchange reaction regulation.v

This work reports the non-covalent functionalization of carbon nanotubes by hyaluronic acid-based azo polymer (HA-MWCNT) and its application in the photothermal elimination of tumor cells. Based on the interaction between hyaluronic acid and CD44 receptors and the photothermal effect of carbon nanotube, HA-MWCNTs manifested great elimination effect on tumor cells.

Fluorescent polyurea-carbon dots (PU-CD) were synthesized via co-pyrolysis of polyurea and carbon dots, showing enhanced fluorescence from crosslink-enhanced emission. As a probe, PU-CD detected doxycycline with a 2.9×10^–7 mol/L detection limit, offering a simple, green, and cost-effective method for analytical chemistry and food safety applications.

Developing high-performance polymer dielectric films from biomass vanillin to advance sustainable green energy storage materials. Under optimal conditions, the dielectric constant is 3.4, the breakdown strength is 670.2 MV/m, the energy storage density is 7.1 J/cm³, and the efficiency is over 90%.

In order to achieve controlled multiple boosts of antigen release, we mixed 4-arm PEG-Se-Se-ONH₂, ODEX and 4-arm PEG-ONH₂ to prepare ultrasound-responsive hydrogel (URH) and successfully realized ultrasonic responsive release. URH is capable of sustaining antigen release under physiological conditions and can further boost antigen release upon ultrasound triggering. Due to the optimization of the release mode, the level of humoral immunity was highly enhanced.

This study presents an innovative approach to synthesize bio-based polydiene polymers from citronellall by converting it into the diene monomer 6,10-dimethyl-1,3,9-undecatriene (DMUT) via Wittig reaction, followed by neodymium-catalyzed polymerization.

The microcellular foams of ultra-high molecular weight poly(vinylidene fluoride) (H-PVDF) were successfully fabricated through a supercritical foaming process. The resulting H-PVDF foam exhibits a record-breaking expansion ratio of 55.6-fold, coupled with a uniformly fine and dense microcellular structure, achieving low thermal conductivity, exceptional hydrophobicity, and outstanding flame-retardancy.

The monomer HFBA was grafted from the ZIF-8@GMA surface by metal-free ATRP. The obtained ZIF-8@GMA-PHFBA hybrid material was cured on the fabric surface and the fabric has good superhydrophobicity properties. The oil-water separation experiment and self-cleaning test confirmed that ZIF-8@GMA-PHFBA modified fabric has excellent oil-water separation effect and self-cleaning performance.

PBC prepared via isothermal annealing and pre-stretching in different orders was assessed for crystallinity, mechanical properties, and microstructure to interpret the relationship between the microstructure and macroscopic properties. Here, shish structure emerged under pre-stretching process, crystallinity and tensile strength after annealing-stretching treatment increased to 24.45% and 104.5 MPa, respectively.

with Zn²⁺-imidazole coordination cross-linked structures and polysulfide bonds displayed reprocessability, broadband sound absorption at low frequency, and quick shape memory capability at moderate temperatures. In addition, the use of DBU effectively suppressed C-S bond formation, thereby preserving the dynamic network reconfigurability.

Thermo-responsive dielectric switching is enabled by modulating the polarization within the polymer composite by exploiting the conductivity and modulus changes of low-melting-point alloys during phase transitions. Via electron tunneling, it achieves bending/compression-responsive insulator-conductor transitions. Ultimately, this integrates multi-level overheating warning and small deformation monitoring.

A green eutectogel exhibiting high transparency, mechanical strength, stretchability, ionic conductivity, leakage-resistance, interfacial adhesion, and cryogenic tolerance has been developed. A flexible sensor based on the eutectogel demonstrates ideal linear sensitivity, a broad response range, reliable stability, and quick responsiveness, enabling precise monitoring of human motion and information transmission.

This work presents bentonite-based hybrid hydrogel strain sensors with outstanding mechanical and strain-sensing performance, which is achieved through highly matched interfaces mediated by supramolecular interactions.

Mechanistic and rheological distinction between dynamic and static flocculation: Dynamic flocculation involves strain-induced rearrangement of filler networks through particle motion, whereas static flocculation arises from polymer-mediated interfacial enhancement through chain reorganization.

In a mixed solvent, the polymer chains form spherical aggregates at low pressures owing to the co-nonsolvency effect. Increasing the pressure caused the polymer chains to transform from an aggregated state to a dispersed state.