Skip to content

A Coarse-Grained Molecular Dynamics Model for Analysis of Mesoscale Carbon Nanothread Structures

FieldValue
JournalPolymers
Published2026-08-31
AuthorsJiaJia Cheng, Junshan Si, Nan Wu, Jun Liu, Su Ju
DOI10.3390/polym18172114

A coarse-grained molecular dynamics model predicted a glass transition temperature of 1485 K for a crystalline diamond nanothread aggregate. At a 6 Å coarse-graining distance, the model preserved the cohesive energy and axial sliding response used to select its scale. The mesoscale calculation used 30,000 beads representing 3,000 nanothread chains, a system size that makes direct all-atomistic treatment impractical.

Model construction and atomistic calibration

Section titled “Model construction and atomistic calibration”

Straight zigzag DNT-I and tubular DNT-II models used the same number of carbon atoms. DNT-I was 75.29 Å long because its zigzag skeleton contained Stone-Wales defects, while DNT-II was 64.93 Å long. AIREBO described covalent C-C, C-H, and dispersive interactions, with a 2.0 Å cutoff for covalent interactions. All calibration simulations ran at 1 K to suppress thermal fluctuations.

Each structure was energy-minimized with a conjugate gradient algorithm before equilibration. Tension tests fixed one end and displaced the other at 0.005 Å/ps after 10 ns of NVT equilibration. The integration step was 1 fs, and a Nosé-Hoover thermostat controlled the 1 K ensemble. Stress came from the virial tensor, with an assumed outer radius of 2.5 Å. Young’s modulus was fitted within strains of 3% or less.

Pure bending used a virtual spherical Lennard-Jones 9/3 wall, followed by 10 ns of relaxation and 5 ns at the imposed curvature. Radial adhesion tests displaced one of two parallel threads through an NPT simulation at 1 K. The radial test used axial periodicity to remove end-edge effects. The energy curve gave radial equilibrium separation and cohesive energy per unit length.

Calibrated quantityDNT-I, zigzagDNT-II, tubular
Equilibrium length75.29 Å64.93 Å
Young’s modulus at strain no greater than 3%404.5 GPa1104.4 GPa
Axial stiffness, EA1142.6 kcal/mol/Å3118.4 kcal/mol/Å
Bending stiffness, EI1289.6 kcal/mol4241.0 kcal/mol
Radial equilibrium separation6.25 Å6.55 Å
Cohesive energy per unit length-0.8331 kcal/mol/Å-0.9874 kcal/mol/Å

The coarse-grained model replaced each segment with one bead at spacing r0. Harmonic stretching used 1/2 kb(r-r0)², and harmonic bending used 1/2 kθ(θ-θ0)². Nonbonded interactions used a 12-6 Lennard-Jones potential. Parallel chains were staggered, and interactions were retained through the sixth neighbor bead. The higher-neighbor correction factor was 1.0988.

Bonded coefficients were mapped from continuum beam and rod energies. A radial adhesion curve supplied the equilibrium separation and cohesive energy. These values set the Lennard-Jones well depth and zero-energy distance. The selected mesoscale scale was r0 = 6 Å for both topologies.

Parameter at r0 = 6 ÅDNT-IDNT-II
Equilibrium bead distance6 Å6 Å
Tensile stiffness, kb190.43 kcal/mol/Ų519.73 kcal/mol/Ų
Bending stiffness, kθ214.93 kcal/mol/rad²706.83 kcal/mol/rad²
Lennard-Jones distance parameter, σ6.18 Å6.42 Å
Lennard-Jones well depth, ε2.2746 kcal/mol2.6959 kcal/mol

The equilibrium bead angle was 180 degrees.

The authors varied r0 from 2 Å to 8 Å to test spatial resolution and interfacial response. Every coarse-grained model returned the radial equilibrium distances of its all-atomistic reference. For DNT-II at r0 = 6 Å, the cohesive energy agreed with the all-atomistic result. However, the model underestimated the peak separation force, and the best force match occurred at r0 = 3 Å. This force-energy discrepancy arises because grouping multiple atoms into one bead smooths local force gradients.

Axial sliding produced regular sine-wave oscillations in both all-atomistic and coarse-grained models. The DNT-II coarse-grained amplitude reached 0.313 nN/nm at r0 = 6 Å, close to the all-atomistic value of 0.32 nN/nm. For DNT-I, the all-atomistic amplitude was about 1.3 nN/nm, and every coarse-grained model remained below it.

The authors selected r0 = 6 Å because cohesive energy governs stable packing, density, and thermal properties. Relative axial sliding also controls aggregate deformation because neighboring threads restrict radial motion. This scale gave DNT-I the largest sliding barrier among the tested coarse-grained models. A uniform r0 also supports consistent multi-component simulations and lowers computational cost.

The mesoscale application used DNT-I only. An amorphous system began at 0.2 g/cm³ with 30,000 beads. It represented 3,000 chains, each containing 10 beads and measuring 6 nm.

Annealing cooled the system from 1800 K to 300 K under NPT conditions at 1 atm. Fourteen cooling cycles lasted 7 ns in total. Each cycle reduced temperature by 100 K over 50 ps, followed by a 450 ps hold. The equilibrated structure reached 1.316 g/cm³ and formed locally hexagonal crystalline domains. This packing was consistent with local hexagonal packing reported in electron diffraction and high-resolution electron microscopy.

The authors identified the glass transition from a slope change in the simulated density during cooling. The resulting 1485 K value was more than 1000 K above conventional flexible-chain polymers. The authors linked this prediction to chain bending stiffness, short thread length, strong van der Waals cohesion, and the sp³ backbone. These mechanisms provide their explanation for the high simulated thermal stability.

Tensile response and deformation mechanism

Section titled “Tensile response and deformation mechanism”

Uniaxial tension was applied along the y-direction at a strain rate of 10⁸/s. The simulation used NPT conditions at 300 K and lateral pressures of 1 atm. Young’s modulus was 1.47 GPa from a fit over strains of 0 to 0.03. This modulus was approximately one order of magnitude above crystalline polyethylene. The yield strength was approximately 64 MPa, twice the value cited for crystalline polyethylene.

Beyond the elastic limit, the aggregate yielded and buckled. Snapshots showed cooperative sliding of parallel chains within aligned crystalline domains. No fracture or chain rupture occurred through strain 0.3. Damage instead developed at low-density amorphous interphases and voids.

The authors describe this process as failure diffusion through the interphases between crystalline domains. They argue that this mechanism dissipates energy and avoids sudden brittle cleavage. The strong sp³ backbone keeps individual chains intact while cooperative sliding accommodates plastic deformation.

The potentials were calibrated from pristine, straight DNT structures. The authors state that high-density defects require direct recalibration through all-atomistic molecular dynamics. Complex helical configurations were used only as a reason for choosing straight symmetric structures, and they were not validated in the mesoscale aggregate study.

The interfacial parameters were calibrated at 1 K and then used across temperatures from 300 K to 1800 K. The tensile strain rate was also far above experimental rates because of molecular dynamics time-scale limits. The authors therefore treat the mechanical values as upper-bound estimates for qualitative and relative comparison.

The paper reports no experimental glass transition temperature or tensile benchmark for the modeled aggregate. The computed values should not be transferred directly to defect-rich material, different chain lengths, or other nanothread topologies without further validation.

The calculated single-thread stiffness separates the two topologies, but the aggregate test used only DNT-I. For composite reinforcement, cohesive energy controls packing, while axial sliding controls plastic energy dissipation. Prior work cited in the paper reports nearly an order of magnitude higher interfacial shear strength for DNT bundles than for pristine CNTs. The paper also notes that reinforcement depends strongly on filler dispersion and loading mode.

The model gives engineers a way to screen mesoscale packing, thermal transitions, and deformation without building every candidate at atomistic resolution. It does not establish aggregate behavior for DNT-II or defect-rich threads. A design team can use the reported potentials for comparative screening, then recalibrate the selected topology and defect state before making quantitative material decisions.

View Original Abstract

Diamond nanothreads (DNTs) represent a promising class of one-dimensional carbon nanomaterials for next-generation structural applications. However, exploring their mesoscale collective properties remains computationally prohibitive via all-atomistic molecular dynamics (MD) simulations. Here, we present a physically consistent coarse-grained model tailored for both zigzag (DNT-I) and tubular (DNT-II) nanothreads. By establishing an energy equivalence framework between all-atomistic MD simulations and molecular mechanics, the bonded potentials (stretching and bending) and non-bonded Lennard-Jones parameters were derived. Moreover, a degree of coarse-graining r0 = 6 Å was determined, which well preserves the interfacial cohesive energy and axial sliding behavior of all-atomistic models. Using the established coarse-grained potentials of DNTs, a high glass transition temperature (Tg = 1485 K) was predicted, and a cooperative intermolecular sliding mechanism that governs the plastic deformation of crystalline DNT aggregates under uniaxial tension was revealed.