Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “MD”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 415 records · Page 23

Investigation of fracture in porous materials: a phase-field fracture study informed by ReaxFF

Microscopic features (e.g, pore shapes, sizes, and distribution) in porous material substantially affect the overall mechanical properties such as stiffness and strength. In turn, these material properties determine the macroscopic behaviors of fracture in the porous material. In certain cases, macroscopic properties can be derived from the porous skeleton and void ratio (i.e., porosity), but in many other cases, such derivation is a challenging task. This paper presents a numerical investigation of microporosity and micropore shapes effect on the macrofracture behavior in porous amorphous silica. For this study, we extend the recently-proposed combined molecular dynamic (MD) and phase-field (PF) fracture modeling approach by including different pore shapes in the atomistic domain. In the MD simulations, we adopt ReaxFF to evaluate the material properties, where four different micropore cases are considered. Based on the material properties derived from MD simulations, the macrofracture propagation of porous media is studied using hybrid PF simulation. In the characterization of the pore structure, the concept of pore ligament is proposed to relate the pore shape and the critical energy release rate. Two classical fracture problems were used to evaluate the effect of pore shape on the macrofracture behavior. The results of the case study show that although the micropore shapes change the macrofracture behaviors, these effects vary with the geometry and loading conditions of macroscopic boundary value problems. Furthermore, the case study also shows that the influence of micropore structure can be captured at the macroscopic level through the material properties derived from the MD simulations.

42 ENGINEERING↗

Atomistic simulations of dislocation pileup: Grain boundaries interaction

Here, using molecular dynamics (MD) simulations, we studied the dislocation pileup–grain boundary (GB) interactions. Two Σ11 asymmetrical tilt grain boundaries in Al are studied to explore the influence of orientation relationship and interface structure on dislocation activities at grain boundaries. To mimic the reality of a dislocation pileup in a coarse-grained polycrystalline, we optimized the dislocation population in MD simulations and developed a predict-correct method to create a dislocation pileup in MD simulations. MD simulations explored several kinetic processes of dislocations–GB reactions: grain boundary sliding, grain boundary migration, slip transmission, dislocation reflection, reconstruction of grain boundary, and the correlation of these kinetic processes with the available slip systems across the GB and atomic structures of the GB.

36 MATERIALS SCIENCE↗

Dislocation transmission across Σ3{112} incoherent twin boundary: a combined atomistic and phase-field study

Grain boundaries (GBs) in polycrystalline materials act as impediments to dislocation motion and result in strengthening. Understanding slip transmission through GBs, specifically twin boundaries, is essential to understand the plastic deformation behavior of polycrystalline fcc materials. Here the interaction between a glide dislocation and Σ3{112} incoherent twin boundary (ITB) in copper is investigated using a combined atomistic and mesoscale approach. The material parameters and structure of the GB in the mesoscale phase field dislocation dynamics (PFDD) model are informed from Molecular Statics (MS) simulations. The structural unit of the ITB consists of an array of three partial dislocations. The interaction between a glide dislocation impinging on each of the GB partial dislocations is investigated using both PFDD and Molecular Dynamics (MD) with two boundary conditions. Transmission planes predicted by both PFDD and MD (NVT) are in agreement, and show that not all transmission events are direct. Critical transmission stresses predicted by PFDD are in the range of 276 MPa to 1380 MPa, while MD predictions are in the range from 100 MPa to 700 MPa. The PFDD and MD predictions of slip transmission are explained using dislocation theory based on isotropic linear elasticity.

36 MATERIALS SCIENCE↗

The role of deviatoric stress and dislocations on the α to ω phase transformation in Ti

Under extreme conditions, α-Ti becomes unstable and transforms either into β-Ti at high temperature or into ω-Ti at high pressure. In what concerns the α to ω phase transformation (PT), there has been a wide range of experimentally reported transition pressures from approximately 2 to 15 GPa at room temperature. Deviatoric stresses and internal defects are often assumed to be the root cause of this variation. Here, in this study, these postulates are revisited using both continuum mechanics and molecular dynamics (MD) simulations. First, a simple continuum model, assuming linear elasticity and isotropic plasticity, is developed to describe the effects of applied stress and dislocations on the stability of an ω nucleus in an infinite α domain. Second, a new MD simulation method is developed to generate an ω nucleus in the α domain utilizing the displacement field identified from the topological analysis. Results from MD simulations show that despite the fact that phase diagrams typically delineate the limits between two phases in terms of only P and T, deviatoric stress promotes the α to ω phase transformation by reducing the critical radius above which an ω nucleus is stable. Furthermore, the required deviatoric stress to nucleate and stabilize a nanoscale ω nucleus is likely emanating from the internal stress of defects such as dislocations. The MD-informed micromechanics models are used to identify favorable configurations where dislocations help favor the α to ω transformation. These configurations show that the interaction with a basal or prismatic dislocation reduces the critical radius of a ω nucleus by about 10 or 16 %, respectively. In addition, prismatic edge dislocations are found to promote the growth of ω nucleus when interacting with the ($\bar{1}$$\bar{1}$20) α //(0001) ω interface. Importantly, a simple model of the arrival of dislocation at an ω nucleus suggests that PT does not necessarily require a pile-up to be present but could alternatively be mediated by a constant rapid flow of dislocations.

36 MATERIALS SCIENCE↗

Discovery of multi-functional polyimides through high-throughput screening using explainable machine learning

Polyimides have been widely used in modern industries because of their excellent mechanical and thermal properties, e.g., high-temperature fuel cells, displays, and aerospace composites. However, it usually takes decades of experimental efforts to develop a successful product. Aiming to expedite the discovery of high-performance polyimides, we utilize computational methods of machine learning (ML) and molecular dynamics (MD) simulations. Our study provides compelling evidence for the effectiveness of a data-driven approach in discovering novel polyimides. We first build a comprehensive library of more than 8 million hypothetical polyimides based on the polycondensation of existing dianhydride and diamine/diisocyanate molecules. Then we establish multiple ML models for the thermal and mechanical properties of polyimides based on their experimentally reported values, including glass transition temperature, Young’s modulus, and tensile yield strength. The obtained ML models demonstrate excellent predictive performance in identifying the key chemical substructures influencing the thermal and mechanical properties of polyimides. The use of explainable machine learning describes the effect of chemical substructures on individual properties, from which human experts can understand the cause of the ML model decision. Applying the well-trained ML models, we obtain property predictions of the 8 million hypothetical polyimides. Then, we screen the whole hypothetical dataset and identify three (3) best-performing novel polyimides that have better-combined properties than existing ones through Pareto frontier analysis. For an easy query of the discovered high-performing polyimides, we also create an online platform https://polyimide-explorer.herokuapp.com/ that embeds the developed ML model with interactive visualization. Furthermore, we validate the ML predictions through all-atom MD simulations and examine their synthesizability. The MD simulations are in good agreement with the ML predictions and the three novel polyimides are predicted to be easy to synthesize via Schuffenhauer’s synthetic accessibility score. Following the proposed ML guidance, we successfully synthesized a novel polyimide and the experimentally obtained high glass transition/thermal decomposition temperature demonstrated its excellent thermal stability. Here our study demonstrates an efficient way to expedite the discovery of novel polymers using ML prediction and MD validation. The high-throughput screening of a large computational dataset can serve as a general approach for new material discovery in other polymeric material exploration problems, such as organic photovoltaics, polymer membranes, and dielectrics.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Comparative analysis of various pretreatments to mitigate fouling and scaling in membrane distillation

Membrane distillation (MD) has shown strong promise for treating hypersaline produced water (PW) streams. Scaling and fouling on the membrane surface, however, have been identified as major challenges for commercialization and widespread application. The focus of this study was to first identify the nature of foulants and scalants on the membrane surface and then investigate the pretreatment strategies that mitigate the extent of fouling and scaling on the membrane surface. Vacuum MD (VMD) was selected for this study because of its higher flux at more moderate feed temperatures that eventually leads better assessment of fouling and scaling on the membrane surface. VMD flux tests were carried out with three different real PW samples at total dissolved solids concentrations in the range of 120-160 g/L. It was found that an efficient pretreatment prior to MD is required to mitigate the fouling potential of the organic matter, mostly suspended constituents, in PW samples. Suspended organics fouled and quickly wetted the MD membrane. Major conventional treatments, including filtration, oxidation, coagulation, air floatation, and aeration were investigated to pretreat the PW before VMD tests. A comprehensive comparative analysis is presented showing that the ultrafiltration and coagulation pretreatments displayed the best performance in mitigating the fouling and scaling, while oxidative pretreatment was found to be effective in reducing the iron concentration to less than 5 ppm. The surface of scaled membranes was carefully characterized to further evaluate the performance of each pretreatment. The components that had the highest tendency to precipitate on the membrane surface were identified as strontium sulfate, calcium carbonate, sodium chloride, iron oxide, and silica. In closing, these findings are supported by thermodynamic estimation of the saturation index.

42 ENGINEERING↗

Computational and experimental study of different brines in temperature swing solvent extraction desalination with amine solvents

Rapid global urbanization and high-salinity wastewater disposal from industrial activities have exerted significant pressure on water resources. Over the past few years, temperature swing solvent extraction (TSSE) has been identified as a promising technique to desalinate hypersaline brines. Despite its potential, the TSSE desalination literature has been mainly based on empirical insights, and the limited molecular simulation studies have primarily focused on NaCl brines. Herein, we use molecular dynamics (MD) simulations to study the TSSE desalination of four different brines, namely, KCl, KBr, NaCl, and NaBr using diisopropylamine as the solvent. Based on both bulk and interfacial brine-diisopropylamine MD results, here we investigate the qualitative and quantitative performance of the simulations by benchmarking these results against our experimental evaluations of these same systems. MD results provide satisfactory qualitative agreement with the experimental data of water solubilization in the organic phase and amine solubilization in the aqueous phase for the KBr, KCl, and NaBr brines. Also, the molecular mechanism of solvation of ionic species by water molecules over diisopropylamine suggested by the MD simulations is in agreement with our experimental data. However, larger qualitative and quantitative deviations were observed for the NaCl brines, and this is likely due to polarization and charge transfer effects, as quantified by our quantum chemical calculations.

42 ENGINEERING↗

Superhydrophobic composite asymmetric electrospun membrane for sustainable vacuum assisted air gap membrane distillation

Membrane distillation (MD) relies on superhydrophobic membranes that must achieve the competing goals of high liquid entry pressure (LEP) and high permeability. Here, in this study, we developed the first superhydrophobic composite electrospun asymmetric membrane using fluorinated graphite (FG). We also provided among the first comparisons of the relative benefits of membrane asymmetry and composite coatings via 3D electrospinning and spraying. FG particles increased the roughness and reduced the zeta potential of the membranes. Superhydrophobic composite membranes had a contact angle of 157° (sliding angle of 7°) and an LEP of 327 kPa. Long-term MD testing revealed that FG particles could increase MD's sustainability while decreasing membrane wetting. Vacuum-assisted air gap membrane distillation (V-AGMD) showed that the composite asymmetric membrane outperformed the non-asymmetric version (composite nanofiber) membrane by 21 % in average water flux and 25 % in average thermal efficiency because of the asymmetric structure of the membrane. The numerical model developed for the V-AGMD system also successfully predicted the water flux generated experimentally from the composite nanofiber and composite asymmetric membranes. Our results confirmed that electrospun asymmetric FG membranes with high contact angle, low sliding angle, high permeability, and high gained output ratio (GOR) would be promising candidates for sustainable MD.

36 MATERIALS SCIENCE↗

Fouling characterization and treatment of water reuse concentrate with membrane distillation: Do organics really matter

Membrane distillation (MD) for the treatment of concentrated brines has been limited in part by membrane fouling, resulting in subsequent flux decline and membrane wetting. This study provides new insight into the identification of fouling and scaling mechanisms and pretreatment strategies for mitigating flux decline with MD treatment of water reuse reverse osmosis concentrate (ROC). Bench-scale direct contact MD experiments were performed with untreated and pretreated ROC. Biological activated carbon (BAC), chemical water softening, or fluidized bed crystallization reactor coupled with ion exchange (FBCR-IX) were selected as pretreatment strategies to isolate the effects of organic fouling and calcium scaling. Organic and inorganic compounds were analyzed by high-performance liquid chromatography (HPLC) and inductively coupled plasma mass spectrometry (ICP-MS). Calcium ions were found to be the major contributor to flux decline despite the high organic content in the ROC. Minimal organic fouling is likely because the organic matter in the ROC is hydrophilic, limiting hydrophobic-hydrophobic interactions between the organics and the membrane. Furthermore, the water flux declined by 63 % after removing organic compounds by BAC pretreatment, with 60 % of the calcium mass precipitating from the solution. Whereas, the water flux remained constant after removing multivalent ions with fluidized bed crystallization. Cleaning the membrane by acid washing and temperature reversal recovered 73 % and 12 % of the water flux, respectively. The analyses outlined in this study can assist in selecting appropriate fouling and scaling mitigation strategies for water reuse ROC and a wide range of feed solutions used in MD applications.

42 ENGINEERING↗

Mechanisms and stability of Li dynamics in amorphous Li-Ti-P-S-based mixed ionic–electronic conductors: A machine learning molecular dynamics study

Mixed ionic–electronic conductors (MIECs) exhibit both high ionic and electronic conductivity to improve the battery performance. In this work, we investigate the mechanism and stability of transport channels in our recently developed MIEC material, amorphous Ti-doped lithium phosphorus sulfide (LPS), using molecular dynamics (MD) simulations with a 99% accurate machine-learning force field (MLFF) trained on ab initio MD data. The achieved MLFF helps efficient large-scale MD simulations on LPS with three Ti concentrations (10%, 20%, and 30%) and six temperatures (25°C to 225°C) to calculate ionic conductivity, activation energy, Li-ion transport mechanism, and configurational entropy. Results show that ionic conductivities and activation energies are consistent with our recent experimental values. Moreover, Li-ion transport occurs via free-volume diffusion facilitated by the formation of disordered Li-S polyhedra. The enhanced stability of transport channels at 10% and 20% Ti doping, compared to 0% and 30%, is observed by analyzing the vibrational and configurational entropy of these disordered Li-S polyhedra. Overall, this study highlights the utility of MLFF-based large-scale MD simulations in explaining the transport mechanism and the stability of Li-ion in Ti-doped LPS electrolyte with significant computational efficiency.

And configuration entropy↗

Using high pressure solutions of polyfluoroacrylate and CO 2 to Seal cement cracks for improved wellbore integrity

Polyfluoroacrylate (PFA) is a hydrophobic and oleophobic polymer that is soluble in high pressure carbon dioxide (CO 2 ). In this study, the ability of PFA-CO 2 solutions to greatly reduce the apparent permeability of split or cracked Portland cement cylindrical samples is assessed. The apparent permeability values of confined samples were determined before and after treatment with PFA-CO 2 solutions. In four tests, PFA-CO 2 solutions were continuously displacing pure CO 2 from the cracked cement and the decrease in apparent permeability due to PFA adsorption and wettability alteration was monitored. The lowest apparent permeability cracked cement sample (81 nD) was completely sealed with a very small amount of solution. Samples with initial apparent permeabilities of 89 μD and 29.4 mD exhibited 92% and 99% reductions in permeability, respectively, before the experiments had to be stopped because of the excessively large increase in pressure drop. A 50% reduction in apparent permeability was observed with a 3.80 mD sample. Four other split cement samples (bound together with tape) with an initial apparent permeability in the 9.0–70 mD range were removed from the core holder and immersed in a PFA-CO 2 solution for 24 h to allow for PFA adsorption. Then the PFA-CO 2 solution was depressurized, allowing for the deposition of additional PFA from the solution within the crack as the pressure fell below the cloud point pressure of the PFA-CO 2 solution. These four samples were then confined again in a core holder and apparent permeability reductions of 29–93% were observed. Finally, results from these eight experiments indicates that the more substantial reductions in the nD – mD apparent permeability of the cracked cement correlated to lower initial crack permeability, higher PFA concentration, and slower injection rate of the PFA-CO 2 solution into the crack.

58 GEOSCIENCES↗

The Importance of the Conformality, Heterogeneity, and Gradient Indices in Evaluating Gamma Knife Radiosurgery Treatment Plans for Intracranial Meningiomas

Purpose: To investigate the relationship between the conformality index (CIn), heterogeneity index (HIn), and gradient index (GIn) and the development of toxicity in patients treated with Gamma Knife radiosurgery (GKRS) for intracranial meningiomas. Methods and Materials: Treatment records of patients treated from 1997 to 2009 with at least 6 months of follow-up were reviewed. The following parameters were collected: CIn, HIn, GIn (ratio of the volume receiving half the prescription isodose to the volume receiving the full prescription isodose), brainstem (BS) maximum dose (MD), BS volume receiving {>=}12 Gy (V12), optic apparatus (OA) MD, OA V8 Gy, OA V10, number of isocenters, number of isocenters outside target volume, and the occurrence of six toxicities. Univariate and multivariate logistic regression modeling were used for analysis. Results: This study included 145 patients (148 meningiomas) with a median follow-up time of 27 months (range, 6-113.9 months). The majority of meningiomas were located in the skull base (53%). The median prescription dose was 13 Gy (range, 10-24 Gy) to the 51.50% (range, 50-92%) isodose. A lower HIn was correlated with a higher GIn (p = 0.007). CIn was not associated with any toxicity. Higher HIn was associated with the development of dizziness (odds ratio [OR] 1.9; p = 0.02), whereas a lower GIn was associated with motor deficits (OR 0.38; p = 0.04) and auditory changes (OR 0.59; p = 0.04). The OA MD, V8, and V12 were not associated with visual changes, but visual changes were associated with a higher number of isocenters outside the target volume (OR 1.93; p = 0.07). BS V12 was correlated with the development of auditory changes (OR 1.05; p = 0.05), whereas patients with higher BS MD tended to have increased toxicity. Conclusions: Close attention must be paid to all three indices (CIn, HIn, GIn) when optimal treatment plans are determined. We recommend that the target CIn should be {<=}2.0, the HIn {<=}2.0, and the GIn {>=}3.0 for intracranial meningiomas.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Cryo_fit : Democratization of flexible fitting for cryo-EM

Cryo-electron microscopy (cryo-EM) is becoming a method of choice for describing native conformations of biomolecular complexes at high resolution. The rapid growth of cryo-EM in recent years has created a high demand for automated solutions, both in hardware and software. Flexible fitting of atomic models to three-dimensional (3D) cryo-EM reconstructions by molecular dynamics (MD) simulation is a popular technique but often requires technical expertise in computer simulation. This work introduces cryo_fit, a package for the automatic flexible fitting of atomic models in cryo-EM maps using MD simulation. Additionally, the package is integrated with the Phenix software suite. The module was designed to automate the multiple steps of MD simulation in a reproducible manner, as well as facilitate refinement and validation through Phenix. Through the use of cryo_fit, scientists with little experience in MD simulation can produce high quality atomic models automatically and better exploit the potential of cryo-EM.

59 BASIC BIOLOGICAL SCIENCES↗

Synthesis, phase transition and photoluminescence studies on Eu{sup 3+}-substituted double perovskites-A novel orange-red phosphor for solid-state lighting

Photoluminescence studies on Eu{sup 3+}-doped double perovskites with the formula A{sub 2}CaWO{sub 6} (A=Sr, Ba) revealed that the forced electric dipole (ED) transition is present when Eu{sup 3+} is substituted at the non-centrosymmetric Sr-site vis-a-vis substitution at the centrosymmetric Ca-site shows both ED and magnetic dipole (MD) transition. A series of novel orange-red-emitting phosphor compositions Sr{sub 1.9-x}Ba{sub x}Eu{sub 0.05}Li{sub 0.05}CaWO{sub 6} (x=0-1.9) have also been synthesized and characterized by powder X-ray diffraction (XRD), diffuse reflectance spectroscopy (DRS) and photoluminescence. XRD results reveal a phase transition from monoclinic to pseudo-cubic structure for x{<=}0.2. All the compositions show broad charge transfer band and orange-red (MD and ED) emission. However, the relative intensity of the MD and ED depends on the Ba content present in the host lattice. Select compositions in this system of compounds could find potential application as orange-red phosphors for white light generation using blue/near-UV GaN-based light-emitting diodes (LEDs). - Graphical abstract: A series of orange-red-emitting A{sub 2}CaWO{sub 6} (A=Sr, Ba) and Sr{sub 1.9-x}Ba{sub x}Eu{sub 0.05}Li{sub 0.05}MoO{sub 6} (x=0-1.9) phosphor with double perovskite structure have been synthesized by solid-state reaction. All the compositions show broad charge transfer band and orange-red [magnetic dipole (MD) and electric dipole (ED)] emission. The red emission of Sr{sub 1.5}Eu{sub 0.05}Li{sub 0.05}Ba{sub 0.4}CaWO{sub 6} is found to be {approx}4.5 times higher than that of commercial red phosphor (Nichia) under 465 nm excitation and hence this phosphor could be a potential candidate for white LED based on blue/NUV GaN LED.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

The dilemma of parotid gland and pharyngeal constrictor muscles preservation—Is daily online image guidance required? A dosimetric analysis

With margin reduction common in head and neck radiotherapy, it is critical that the dosimetric effects of setup deviations are quantified. With past studies focusing on the quantification of positional and volumetric changes of organs at risk (OARs), this study aimed to measure the dose delivered to these the parotid gland (PG) and pharyngeal constrictor muscles (PCMs) using cone beam computed tomography (CBCT). Furthermore, this investigation sought to establish a potential time trend of change in dose delivered to target volumes secondary to ascertaining the need for daily image guidance (IG) to reduce the dose burden to these important OARs. Intensity modulated radiotherapy (IMRT) plans for 5 locally advanced head and neck patients' plans were created and mapped to weekly CBCTs. Each plan was recalculated without heterogeneity correction allowing for dosimetric comparison. Dosimetric endpoints recorded to assess the effect of positional variation were as per ICRU 83 and included D{sub 95} and D{sub 98} for the target volumes, mean dose (MD) and V{sub 30} {sub Gy} for the PGs, and V{sub 50} {sub Gy} and MD for the PCMs. Results were deemed statistically significant if p < 0.05. No significant time trends were established for these OARs. A significant decrease in V{sub 50} {sub Gy} was observed for all PCMs (p < 0.001) on all CBCTs relative to the original plan. Regarding target volumes, a highly significant decrease in MD (MD = 20 Gy, CI: −20.310 to −19.820) in D{sub 98} of the high-dose planning target volume (PTV [70 Gy]; PTVD{sub 98%} = 70 Gy) for case 3 was found (p ≤ 0.001). A nonpredictable, yet significant dosimetric effect was found. A clinically acceptable balance must be achieved between OAR dosimetry and target coverage as can be achieved by frequent IG.

61 RADIATION PROTECTION AND DOSIMETRY↗

Novel thermal efficiency-based model for determination of thermal conductivity of membrane distillation membranes

Development and selection of membranes for membrane distillation (MD) could be accelerated if all performance-determining characteristics of the membrane could be obtained during MD operation without the need to recur to specialized or cumbersome porosity or thermal conductivity measurement techniques. By redefining the thermal efficiency, the Schofield method could be adapted to describe the flux without prior knowledge of membrane porosity, thickness, or thermal conductivity. A total of 17 commercially available membranes were analyzed in terms of flux and thermal efficiency to assess their suitability for application in MD. The thermal-efficiency based model described the flux with an average %RMSE of 4.5%, which was in the same range as the standard deviation on the measured flux. The redefinition of the thermal efficiency also enabled MD to be used as a novel thermal conductivity measurement device for thin porous hydrophobic films that cannot be measured with the conventional laser flash diffusivity technique.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Machine learning prediction on the fractional free volume of polymer membranes

Fractional free volume (FFV) characterizes the microstructural level features of polymers and affects their properties including thermal, mechanical, and separation performance. Experimental measurements and theoretical analyses have been used to quantify the FFV of polymers, but challenges remain because of their limitations. Experimental measurements are laborious and based on semi empirical equations, while Bondi’s group contribution theory involves ambiguities like the determination of van der Waals volume and the choice of factor values in the theoretical equation. To efficiently evaluate the FFV of polymers, this study utilizes high-throughput molecular dynamics (MD) simulations to build a large dataset regarding polymer’s FFV. Based on this large dataset, we further build machine learning (ML) models to establish the composition-structure relation. Inspired by group contribution theory which correlates polymer’s functional groups to FFV, our ML models correlate polymer’s substructures or physico-chemical indexes to FFV. Here, our study first benchmarks the MD simulation protocol to obtain reliable FFV of polymers and then carries out high-throughput MD simulations for more than 6,500 homopolymers and 1,400 polyamides. Such a large and diverse dataset makes the well-trained ML models more generalizable, compared with the group contribution theory. The efficiency of a feed forward neural network model is further demonstrated by applying it to a hypothetical polyimide dataset of more than 8 million chemical structures. The predicted FFVs of hypothetical polyimides are further validated by MD simulations. The obtained FFVs of the 8 million polymers, plus their previously reported gas separation performances, demonstrate the promising capability of ML virtual screening for the discovery of polymer membranes with exceptional permeability/selectivity.

36 MATERIALS SCIENCE↗