Phthalonitrile melting point prediction enabled by multi-fidelity learning

Abstract

Phthalonitrile (PN) resins have been widely used in various fields for their excellent thermal stability and mechanical properties, but they suffer from poor processability due to their high melting point. Data-driven machine learning (ML) can assist in screening PNs with low melting points but is limited by the lack of experimental data. Using error correction and multi-fidelity co-training methods, we established two multi-fidelity models for predicting PN melting points. This work demonstrates that through multi-fidelity learning, limited experimental data can be effectively utilized with the assistance of all-atom molecular dynamics simulation data to establish ML-based property prediction models. A comparison between these two multi-fidelity prediction models was made, and the contribution of chemical units to the PN melting point was analyzed based on one of the models. Our work offers feasible ML tools for future designing PNs with good processability.

Keywords

Phthalonitrile / melting point / machine learning / molecular dynamics / multi-fidelity

References

[1] Jia Y,Dong J.Catalytic polymerization of phthalonitrile resins by carborane with enhanced thermal oxidation resistance: experimental and molecular simulation.Polymers2022;14:219 PMCID:PMC8747195

[2] Zhang H,Yang Z,Peng D.Synthesis, curing and thermal properties of the low melting point phthalonitrile resins containing glycidyl groups.Polym Bull2023;80:725-38

[3] Mouritz A,Burchill P.Review of advanced composite structures for naval ships and submarines.Compos Struct2001;53:21-42

[4] Bai S,Chen X,Zhang Q.Synthesis and properties of a thioether bonded phthalonitrile resin.Mater Today Commun2020;24:101352

[5] Liu C,Sun M.Novel low-melting bisphthalonitrile monomers: Synthesis and their excellent adhesive performance.Eur Polym J2021;153:110511

[6] Wang H,Guo H.A novel high temperature vinylpyridine-based phthalonitrile polymer with a low melting point and good mechanical properties.Polym Chem2018;9:976-83

[7] Peng W,Hu J.Renewable protein-based monomer for thermosets: a case study on phthalonitrile resin.Green Chem2018;20:5158-68

[8] Gu H,Du A.An overview of high-performance phthalonitrile resins: fabrication and electronic applications.J Mater Chem C2022;10:2925-37

[9] Wu Z,Zong L,Wang J.Novel phthalonitrile-based composites with excellent processing, thermal, and mechanical properties.High Perform Polym2018;30:720-30

[10] Nechausov S,Morozov O,Babkin A.Low-melting phthalonitrile monomers containing maleimide group: synthesis, dual-curing behavior, thermal and mechanical properties.React Funct Polym2021;164:104932

[11] Hu Y,Qi Y.Self-curing triphenol A-based phthalonitrile resin precursor acts as a flexibilizer and curing agent for phthalonitrile resin.RSC Adv2018;8:32899-908 PMCID:PMC9086328

[12] Dominguez DD.Properties of phthalonitrile monomer blends and thermosetting phthalonitrile copolymers.Polymer2007;48:91-7

[13] Babkin AV,Bulgakov BA,Avdeev VV.Low-melting siloxane-bridged phthalonitriles for heat-resistant matrices.Eur Polym J2015;66:452-7

[14] Zhang J,Yuan H,Zhang X.Thermal properties of C17H36/MCM-41 composite phase change materials.Comput Mater Sci2015;109:300-7

[15] Qiao Z,Zhou J.Molecular dynamics simulations on the melting of gold nanoparticles.Phase Transit2014;87:59-70

[16] Li JF,Liu J.Molecular dynamics simulations on melting of aluminum.Appl Mech Mater2013;423-6:935-8

[17] Ganz E,Yang LM.The initial stages of melting of graphene between 4000 K and 6000 K.Phys Chem Chem Phys2017;19:3756-62 PMCID:PMC5319407

[18] Liu Y,Yu T,Guo W.Melting point prediction of energetic materials via continuous heating simulation on solid-to-liquid phase transition.ACS Omega2019;4:4320-4

[19] Du S,Wang L,Du L.Polymer genome approach: a new method for research and development of polymers.Acta Polym Sin2022;53:592-607

[20] Yang Z,Liu L,Xia W.Applications of machine learning methods in the studies of polymer glass formation.Acta Polym Sin2023;54:432-50

[21] Gong X.Advances and challenges of machine learning in polymer material genomes.Acta Polym Sin2022;53:1287-300

[22] Li Y,Wang L.Data and machine learning in polymer science.Chin J Polym Sci2023;41:1371-6

[23] Habibi-Yangjeh A,Danandeh-Jenagharad M.Prediction of melting point for drug-like compounds using principal component-genetic algorithm-artificial neural network.Bull Korean Chem Soc2008;29:833-41

[24] Qu N,Liao M.Ultra-high temperature ceramics melting temperature prediction via machine learning.Ceram Int2019;45:18551-5

[25] Hu Y,Wang L,Du L.Machine-learning-assisted design of highly tough thermosetting polymers.ACS Appl Mater Interfaces2022;14:55004-16

[26] Xu X,Hu Y.Discovery of thermosetting polymers with low hygroscopicity, low thermal expansivity, and high modulus by machine learning.J Mater Chem A2023;11:12918-27

[27] Chen C,Zuo Y,Ong SP.Graph networks as a universal machine learning framework for molecules and crystals.Chem Mater2019;31:3564-72

[28] Zeng K.Phthalonitrile matrix resins and composites. In: Wiley Encyclopedia of Composites. Wiley; 2011. pp. 1-14.

[29] Laskoski M,Keller TM.Synthesis and properties of a bisphenol A based phthalonitrile resin.J Polym Sci A Polym Chem2005;43:4136-43

[30] Shi X,Ji P,Yu X.A phthalonitrile resin with a low melting point and high storage modulus containing high‐density aromatic ether bonds.ChemistrySelect2020;5:12213-7

[31] Liu K,Tao L,Xiao R.Synthesis and characterization of inherently flame retardant polyamide 6 based on a phosphine oxide derivative.Polym Degrad Stabil2019;163:151-60

[32] Łaszkiewicz B.Preparation of organophosphorus polyurethane block copolymers.J Appl Polym Sci1967;11:2295-301

[33] Jain P,Varma IK.Effect of phosphorus content on thermal behaviour of diglycidyl ether of bisphenol-A/phosphorus containing amines.J Therm Anal Calorim2022;67:761-72

[34] Sastri SB.Phthalonitrile polymers: cure behavior and properties.J Polym Sci A Polym Chem1999;37:2105-11