Understanding interphase formation, load bearing, and failure of epoxy-based adhesive joints via reactive coarse-grained molecular dynamics - dataset

Title: Understanding interphase formation, load bearing, and failure of epoxy-based adhesive joints via reactive coarse-grained molecular dynamics - dataset Abstract:from [1] Adhesive joints are a crucial technology for efficiently bonding dissimilar materials, offering a superior strength-to-weight ratio compared to conventional fastening methods. Despite their widespread industrial use, a comprehensive understanding required to unlock their full potential remains lacking. This gap is primarily due to the complex influence of adherends on the adhesive's curing process, which alters its nanoscale microstructure and mechanical properties. To address these challenges, we introduce a reactive coarse-grained molecular dynamics (MD) model to investigate the interphase formation, load-bearing mechanism, and failure of epoxy-based adhesive joints. To this end, we calibrate coarse-grained potentials for alumina adherends and combine them with a reactive epoxy model from the literature, serving as the adhesive. We derive the missing interactions between the adherend and adhesive by matching the coarse-grained and atomistic free energies obtained via metadynamics. This ensures that the entire adhesive joint model is grounded in all-atom simulations, following a bottom-up approach. We investigate how chemical bonds between adherends and adhesives (grafting), as well as the adhesive's resin-to-hardener ratio, influence interphase formation, and ultimately affect the stiffness, strength, and toughness of the joint. Notably, we find that slightly resin-poor mixtures effectively mitigate interphase formation, optimizing both strength and toughness. This study presents an advanced simulation framework to elucidate the complex structure-property relations of adhesive joints, facilitating the development of next-generation adhesives with improved performance. Contact: Maximilian Ries Institute of Applied Mechanics Friedrich-Alexander-Universität Erlangen-Nürnberg Egerlandstr. 5 91058 Erlangen [e-mail] Context: Data set supplementing the journal paper: [1] M. Ries, "Understanding interphase formation, load bearing, and failure of epoxy-based adhesive joints via reactive coarse-grained molecular dynamics", Engineering Fracture Mechanics 347 (2026) 112641. https://doi.org/10.1016/j.engfracmech.2026.112641 (incl. supplementary material) This data set contains the coarse-grained Al2O3/DGEBA-PACM adhesive joint simulations (sample preparation and uniaxial tensile tests up to failure) of Sections 3.1, 3.2 and Appendix A2 of [1]. The calibration of the coarse-grained Al2O3 (Section 2.2) and the metadynamics simulations (Section 2.3) are not part of this data set. Funding: DFG - 377472739 (Research Training Group GRK 2423 "Fracture across Scales - FRASCAL"). HPC resources: Erlangen National High Performance Computing Center (NHR@FAU), NHR project b275dc "Molecular Modeling of Adhesive Joints". NHR@FAU hardware is partially funded by the DFG - 440719683. Software: All MD simulations were performed with LAMMPS [2,3], version 10 Sep 2025, including the REACTION package [4,5], on the HPC cluster Fritz (NHR@FAU), 144 MPI tasks per simulation. Compiled with GNU C++ 11.2.0 and OpenMPI 4.1.2 (no OpenMP) Packages required by the input scripts: MOLECULE, REACTION (fix bond/react), MC (fix bond/create, fix bond/break), RIGID (fix rigid), EXTRA-PAIR (pair lj/gromacs) Post-processing: Matlab R2023b, visualization: OVITO [7] License: Creative Commons Attribution 4.0 International Simulation model in brief (see [1] for details): - two CG Al2O3 adherends (200 A x 200 A cross-section, 32.5 A thick, bcc lattice), periodic in x and z, non-periodic in y - CG epoxy model of Giuntoli et al. [6]: DGEBA resin + PACM hardener, 154 000 adhesive beads (e.g. 22 000 DGEBA + 11 000 PACM at stoichiometry) - preparation: insertion -> squeeze/hold at 1000 K -> grafting -> curing (stages I-IV) -> quench to 300 K and hold - uniaxial tensile test in y (plane strain, NVT, 300 K), strain rate 5e-7 1/fs, up to 150 % strain, with bond breaking - timestep 4 fs, LAMMPS units "real" - five replicas (six in 02_bondbreak) with different random seeds per parameter set Dataset structure: readme.txt this file | |-- 01_grafting_strength/ grafting bond stiffness k_grft -> [1] Appendix A2.1, Figs. A2-A3 | |-- DGEBA-6_GrftDeg-0.3_GrftStre-<k_grft>/ k_grft = 3, 6, 9, 12, 15, 18, 21 kcal/(mol A^2) | | |-- DGEBA-6_GrftDeg-0.3_GrftStre-<k_grft>-<replica>/ simulation folders (see below) | | |-- Post/ replica-averaged results (see below) | |-- 02_bondbreak/ grafting bond failure stretch lambda_grft -> [1] Appendix A2.2, Figs. A4-A5 | |-- DGEBA-6_GrftDeg-0.3_GrftStre-15_Break-<lambda_grft>/ lambda_grft = 1.2, 1.3, 1.4, 1.45, 1.5, 1.525, 1.6, 1.7 | | |-- DGEBA-6_GrftDeg-0.3_GrftStre-15_Break-<lambda_grft>-<replica>/ (replicas 0-5, i.e. six replicas) | | |-- Post/ | |-- 03_grafting_frac/ grafting fraction Phi_grft -> [1] Section 3.1, Figs. 4-8 | |-- DGEBA-6_GrftDeg-<Phi_grft>_GrftStre-15_Break-1.525/ Phi_grft = 0, 0.05, 0.1, 0.2, 0.3, 0.4, 0.5 | | |-- DGEBA-6_GrftDeg-<Phi_grft>_GrftStre-15_Break-1.525-<replica>/ | | |-- Post/ | |-- 04_mixing_ratio/ resin-to-hardener ratio x DGEBA : 3 PACM -> [1] Section 3.2, Figs. 9-10 |-- DGEBA-<x>_GrftDeg-0.3_GrftStre-15_Break-1.525/ x = 2, 3, 4, 5, 5.75, 6, 7, 8, 9, 10 (x = 6: stoichiometric) | |-- DGEBA-<x>_GrftDeg-0.3_GrftStre-15_Break-1.525-<replica>/ | |-- Post/ Reference configuration: The reference configuration (x = 6, Phi_grft = 0.3, k_grft = 15 kcal/(mol A^2), lambda_grft = 1.525) is part of all four studies. Its simulations are stored only once, in 03_grafting_frac/DGEBA-6_GrftDeg-0.3_GrftStre-15_Break-1.525/ The corresponding folders in 01_grafting_strength (GrftStre-15), 02_bondbreak (Break-1.525) and 04_mixing_ratio (DGEBA-6) contain only the Post/ folder (same results), so that the replica-averaged results are available in every study. Naming convention for simulation folders: DGEBA-<x>_GrftDeg-<Phi_grft>_GrftStre-<k_grft>_Break-<lambda_grft>-<replica> - DGEBA-<x> resin-to-hardener mixing ratio x DGEBA : 3 PACM (6 = stoichiometric) - GrftDeg-<Phi_grft> grafting fraction, i.e. fraction of Al2O3 surface beads bonded to PACM - GrftStre-<k_grft> grafting bond stiffness in kcal/(mol A^2) (harmonic bond, equilibrium length 3.0 A) - Break-<lambda_grft> grafting bond failure stretch; the grafting bond breaks at lambda_grft * 3.0 A (if missing: lambda_grft = 1.525, e.g. in 01_grafting_strength and the reference configuration) - trailing number replica ID (0-4, or 0-5 in 02_bondbreak) The actual parameters of every simulation are documented in its parameters.prm / lammps.prm. Each simulation directory contains: lammps input files (*.in) all scripts executed by LAMMPS, starting with main.in, which includes the others parameters.prm parameter file governing the simulation (bash + LAMMPS parameters) lammps.prm LAMMPS part of parameters.prm, included by main.in meta.info start date of the simulation Standard_Input/ simulation input files including - CGAl2O3-X_40_Y_100_Z_40_S_6.5_nonperiod.data CG Al2O3 substrate (starting configuration) - DGEBA_centered.data, PACM_centered.data DGEBA and PACM molecule files - rxnlin_*, rxnnet_* reaction templates and maps for the REACTION package (linear and network curing reactions) (grafting is performed with fix bond/create, see grafting.in) LAMMPS_Out/ simulation results - data files (LAMMPS data format) documenting the simulation progress (intermediate steps 1_afterInsert to 4_afterCuring omitted to save space; all grafting and curing bonds are contained in 5_afterCooldown.data) 5_afterCooldown.data equilibrated sample at 300 K 6_afterDeformation.data sample after the tensile test - thermo.dat thermodynamic quantities during sample preparation after grafting (curing, quenching and low-temperature hold) - thermoUT.dat thermodynamic quantities during the uniaxial tensile test - bins/Bins<N>.dat number density of atom type N in 500 slices along y, averaged during the low-temperature hold (LAMMPS fix ave/chunk format), N = 3, 5, 8, 9, 10, 11 Each parameter-set directory (containing the replica folders) additionally contains: Post/ post-processing results, averaged over all replicas (suffix _avg: mean, _std: standard deviation) - UT.Dat stress-strain data (columns of thermoUT.dat, see below) - BIN.Dat spatially binned structure (see below) Atom (bead) types: 1, 2 DGEBA backbone beads 3 DGEBA epoxy end, unreacted 8 DGEBA epoxy end, reacted 4 PACM backbone bead 5 PACM amine bead, unreacted 9 PACM amine bead, reacted once (linear link) 10 PACM amine bead, reacted twice (cross-link) 11 PACM amine bead bonded to the substrate (grafted) 12, 13 Al2O3 beads (bcc lattice: corner and body-centered sites) 14 Al2O3 surface bead bonded to PACM (grafted) 15 Al2O3 surface bead, activated for grafting (reset to type 12 once the target grafting fraction is reached) curing: linear reaction 3 + 5 -> 8 + 9, network reaction 3 + 9 -> 8 + 10 (REACTION package) grafting: 15 + 5 -> 14 + 11 (fix bond/create, bond type 16) 6, 7 not used (types kept for compatibility with the original epoxy model [6]) Result quantities in thermo.dat (time averages every 1000 steps): TimeStep timestep c_thermo_temp temperature in K v_Ly joint length in y direction in A v_Density mass density in g/cm^3 v_CurDeg curing degree v_MaxCurDegEpoxy maximum epoxy-limited curing degree v_MaxCurDegAmine maximum amine-limited curing degree v_CrossDeg cross-linking degree v_MaxCrossDeg maximum cross-linking degree v_NormMixing normalized resin/hardener mixing ratio v_NormMixingEff normalized effective resin/hardener mixing ratio (corrected by grafting) v_GraftDeg grafting degree (averaged over upper and lower substrate) v_Etot total energy in kcal/mol v_Epair pair energy in kcal/mol v_Ebond bond energy in kcal/mol v_Eangle angle energy in kcal/mol v_Epot potential energy in kcal/mol v_Ekin kinetic energy in kcal/mol v_Stress_xx_MPa normal stress in x direction in MPa v_Stress_yy_MPa normal stress in y direction in MPa v_Stress_zz_MPa normal stress in z direction in MPa v_Stress_xy_MPa shear stress in xy direction in MPa v_Stress_xz_MPa shear stress in xz direction in MPa v_Stress_yz_MPa shear stress in yz direction in MPa Result quantities in thermoUT.dat (time averages every 1000 steps): TimeStep timestep c_thermo_temp temperature in K v_Ly joint length in y direction in A v_Density mass density in g/cm^3 v_Strain_xx engineering strain in x direction v_Strain_yy engineering strain in y direction (loading direction) v_Strain_zz engineering strain in z direction v_Strain_xy engineering shear strain in xy direction v_Strain_xz engineering shear strain in xz direction v_Strain_yz engineering shear strain in yz direction v_Etot total energy in kcal/mol v_Epair pair energy in kcal/mol v_Ebond bond energy in kcal/mol v_Eangle angle energy in kcal/mol v_Epot potential energy in kcal/mol v_Ekin kinetic energy in kcal/mol v_Stress_xx_MPa normal stress in x direction in MPa v_Stress_yy_MPa normal stress in y direction in MPa v_Stress_zz_MPa normal stress in z direction in MPa v_Stress_xy_MPa shear stress in xy direction in MPa v_Stress_xz_MPa shear stress in xz direction in MPa v_Stress_yz_MPa shear stress in yz direction in MPa v_NumBondsBroken number of broken bonds (cumulative) c_PEcrosslink potential energy of cross-link beads (type 10) in kcal/mol c_PElinlink potential energy of linear-link beads (type 9) in kcal/mol c_PairSubstrate non-bonded adherend-adhesive energy in kcal/mol c_GraftingEnergy bond energy of the grafting bonds in kcal/mol Post/UT.Dat contains the same quantities (Step, Temp, Ly, Dens, StrainXX, ..., GraftingEnergy) with suffixes _avg and _std. PEcrosslink, PElinlink, PairSubstrate and GraftingEnergy are given as changes Delta E with respect to the undeformed state (cf. [1] Section 2.6). Result quantities in Post/BIN.Dat (spatial binning along y, see [1] Section 2.6): the 500 slices are combined pairwise from the upper and lower half of the joint (symmetry), giving 250 rows III, V, VIII, IX, X, XI number density of bead types 3, 5, 8, 9, 10, 11 pos slice position along y in A lambda local resin-to-hardener mixing ratio Dcure local curing degree Psi_cure Dcross local cross-linking degree Psi_cross (each with suffix _avg and _std) References: [1] M. Ries, "Understanding interphase formation, load bearing, and failure of epoxy-based adhesive joints via reactive coarse-grained molecular dynamics", Engineering Fracture Mechanics 347 (2026) 112641. https://doi.org/10.1016/j.engfracmech.2026.112641 [2] S. Plimpton, "Fast parallel algorithms for short-range molecular dynamics," Journal of Computational Physics, 1995, 117, 1-19. [3] A. P. Thompson et al., "LAMMPS - a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales," Computer Physics Communications, vol. 271, p. 108171, 2022. [4] J. R. Gissinger, B. D. Jensen, and K. E. Wise, "Modeling chemical reactions in classical molecular dynamics simulations," Polymer, vol. 128, pp. 211-217, 2017. [5] J. R. Gissinger, B. D. Jensen, and K. E. Wise, "REACTER: A heuristic method for reactive molecular dynamics," Macromolecules, vol. 53, no. 22, pp. 9953-9961, 2020. [6] A. Giuntoli, N. K. Hansoge, A. van Beek, Z. Meng, W. Chen, S. Keten, "Systematic coarse-graining of epoxy resins with machine learning-informed energy renormalization," npj Computational Materials, vol. 7, 2021. [7] A. Stukowski, "Visualization and analysis of atomistic simulation data with OVITO - the open visualization tool," Modelling and Simulation in Materials Science and Engineering, vol. 18, no. 1, p. 015012, 2009.