HVM Advanced Theory & Numerical Methods¶
This page documents the thermodynamic foundations, kinematics, and numerical methods underlying the HVM. For the constitutive equations, see HVM Model Reference. For protocol derivations, see HVM Protocol Equations & Derivations.
Thermodynamic Framework¶
Helmholtz Free Energy¶
The total Helmholtz free energy density is the sum of contributions from each subnetwork, with damage coupling on the permanent network:
Permanent network (Neo-Hookean, Gaussian chains):
where \(\mathbf{B} = \mathbf{F}\mathbf{F}^T\) and \(G_P = c_P k_B T\).
Exchangeable (vitrimer) network:
The stress vanishes when \(\boldsymbol{\mu}^E = \boldsymbol{\mu}^E_{nat}\), not when \(\boldsymbol{\mu}^E = \mathbf{I}\). This distinction is the hallmark of associative exchange.
Dissociative (physical) network:
Natural state is always \(\mathbf{I}\) (bonds reform stress-free).
Clausius-Duhem Derivation¶
The second law requires non-negative dissipation:
Expanding \(\dot{\Psi}_{tot}\) and collecting terms linear in \(\mathbf{L}\) identifies the Cauchy stress:
The remaining terms yield the dissipation from kinetic processes, each of which must be individually non-negative:
Exchangeable network dissipation:
Guaranteed non-negative because \((\boldsymbol{\mu}^E - \boldsymbol{\mu}^E_{nat})^2\) is positive semi-definite.
Dissociative network dissipation:
Damage dissipation:
Satisfied because \(\Psi_P \geq 0\) and \(\dot{D} \geq 0\) (damage is irreversible).
Upper-Convected Kinematics¶
The evolution equations use the full velocity gradient \(\mathbf{L}\), not the symmetric part \(\mathbf{D}\) alone:
The decomposition \(\mathbf{L} = \mathbf{D} + \mathbf{W}\) means:
The vorticity terms \(\mathbf{W}\boldsymbol{\mu} - \boldsymbol{\mu}\mathbf{W}\) provide the Jaumann co-rotational correction for rigid-body rotation.
The upper-convected derivative form is:
Simple shear as a special case: For \(\mathbf{L} = \dot{\gamma} \mathbf{e}_1 \otimes \mathbf{e}_2\) with isotropic initial conditions, \(\mu_{22} = \mu_{33}\), which means \(\boldsymbol{\mu}\) commutes with \(\mathbf{W}\), and the vorticity terms vanish. The \(\mathbf{L}\) and \(\mathbf{D}\) formulations then coincide – this is why all protocol derivations in HVM Protocol Equations & Derivations use scalar ODEs.
TST Kinetics Deep Dive¶
Stress-Coupled vs Stretch-Coupled¶
Option A – Stress-based (von Mises invariant, ``kinetics=”stress”``):
Appropriate when the exchange barrier is reduced by total stress magnitude. Isotropic and simple to evaluate.
Option B – Stretch-based (chain stretch invariant, ``kinetics=”stretch”``):
Measures elastic stretch of exchangeable chains relative to their current natural state. More appropriate when force along the chain backbone directly lowers the barrier (Bell model picture).
In the zero-stress limit, both reduce to the thermal rate \(k_{BER,0}(T) = \nu_0 \exp(-E_a / k_B T)\).
Von Mises Computation (Simple Shear)¶
With \(\sigma^E_{ij} = G_E(\mu^E_{ij} - \mu^{E,nat}_{ij})\), the von Mises equivalent stress is:
where \(\Delta_{ij} = \mu^E_{ij} - \mu^{E,nat}_{ij}\).
Square root singularity: At \(\sigma_{VM} = 0\), the gradient of \(\cosh(\cdot)\) can produce numerical issues. The implementation uses \(\sqrt{\max(x, 0) + 10^{-30}}\) to guard against infinite gradients (see Numerical Implementation below).
Phenomenological Fast Mode¶
For computational efficiency (avoiding ODE stiffness from the exponential stress-coupling), a linearized rate is available:
where \(\alpha = V_{act} G_E / (2 k_B T)\) maps TST parameters to the phenomenological enhancement coefficient. This is a first-order Taylor expansion valid for small deformations.
See VLB Advanced Theory & Numerical Methods for the analogous VLB Bell breakage mechanism.
Temperature & Topological Freezing¶
Arrhenius Structure¶
The TST rate directly produces the topological freezing temperature \(T_v\). Defining \(T_v\) as the temperature where \(\tau_{BER} = 1/k_{BER,0}\) exceeds \(10^3\) s:
Temperature regimes:
Regime |
Condition |
Behavior |
|---|---|---|
Below \(T_v\) |
\(T < T_v\) |
Vitrimer behaves as thermoset (no exchange) |
Above \(T_v\) |
\(T > T_v\) |
Active BER, material flows |
Well above \(T_v\) |
\(T \gg T_v\) |
Fast exchange, approaches viscous liquid |
Arrhenius shift factor for time-temperature superposition:
If \(G_P\) and \(G_E\) have the entropic \(T\)-scaling (\(G \propto T\)), a vertical shift \(b_T = T_{ref}/T\) is also needed.
Dissociative Bond Temperature Dependence¶
The D-network rate can follow Arrhenius:
For force-dependent dissociation (Bell-Evans model):
Numerical Implementation¶
ODE solver: diffrax Tsit5 (explicit 5th-order Runge-Kutta) with
PIDController adaptive stepping (rtol=1e-8, atol=1e-10).
Stiffness handling: TST stress-coupling can make the ODEs stiff at high
shear rates (large \(k_{BER}\) variations within a timestep). The
explicit Tsit5 solver handles moderate stiffness; for extreme cases, reduce
shear rate or switch to kinetics="stretch" (smoother coupling).
Note
Implicit solvers (e.g., Kvaerno5) were tested but produce
TracerBoolConversionError due to lineax LU transpose checks during JAX
tracing. Tsit5 is the recommended solver.
Square-root guard: The BER rate computation involves \(\sqrt{\text{tr}(\boldsymbol{\mu}^E - \boldsymbol{\mu}^E_{nat})}\), which has infinite gradient at zero. The implementation uses:
safe_stretch = jnp.sqrt(jnp.maximum(stretch_invariant, 0.0) + 1e-30)
Initial conditions: All tensors at identity (\(\mu_{xx} = \mu_{yy} = 1\), \(\mu_{xy} = 0\)), \(\gamma = 0\), \(D = 0\).
References¶
Vernerey, F.J., Long, R. & Brighenti, R. (2017). “A statistically-based continuum theory for polymers with transient networks.” J. Mech. Phys. Solids, 107, 1–20. https://doi.org/10.1016/j.jmps.2017.05.016
Vernerey, F.J. (2018). “Transient response of nonlinear polymer networks: A kinetic theory.” J. Mech. Phys. Solids, 115, 230–247. https://doi.org/10.1016/j.jmps.2018.02.018
PDFVernerey, F.J., Brighenti, R., Long, R. & Shen, T. (2018). “Statistical Damage Mechanics of Polymer Networks.” Macromolecules, 51(17), 6609–6622. https://doi.org/10.1021/acs.macromol.8b01052
Meng, F., Saed, M.O. & Terentjev, E.M. (2019). “Elasticity and Relaxation in Full and Partial Vitrimer Networks.” Macromolecules, 52(19), 7423–7429. https://doi.org/10.1021/acs.macromol.9b01123
Shen, T., Song, Z., Cai, S. & Vernerey, F.J. (2021). “Nonsteady fracture of transient networks: The case of vitrimer.” PNAS, 118(29), e2105974118. https://doi.org/10.1073/pnas.2105974118
Wagner, R.J., Hobbs, E. & Vernerey, F.J. (2021). “A network model of transient polymers: exploring the micromechanics of nonlinear viscoelasticity.” Soft Matter, 17, 8742. https://doi.org/10.1039/D1SM00753J
PDFLamont, S.C., Mulderrig, J., Bouklas, N. & Vernerey, F.J. (2021). “Rate-Dependent Damage Mechanics of Polymer Networks with Reversible Bonds.” Macromolecules, 54(23), 10801–10813. https://doi.org/10.1021/acs.macromol.1c01943
Meng, F., Saed, M.O. & Terentjev, E.M. (2022). “Rheology of vitrimers.” Nature Communications, 13, 5753. https://doi.org/10.1038/s41467-022-33321-w
Vernerey, F.J., Rezaei, B. & Lamont, S.C. (2024). “A kinetic theory for the mechanics and remodeling of transient anisotropic networks.” J. Mech. Phys. Solids, 190, 105713. https://doi.org/10.1016/j.jmps.2024.105713
Wagner, R.J. & Silberstein, M.N. (2025). “A foundational framework for the mesoscale modeling of dynamic elastomers and gels.” J. Mech. Phys. Solids, 194, 105914. https://doi.org/10.1016/j.jmps.2024.105914
Karim, M.R., Vernerey, F. & Sain, T. (2025). “Constitutive Modeling of Vitrimers and Their Nanocomposites Based on Transient Network Theory.” Macromolecules, 58(10), 4899–4912. DOI: 10.1021/acs.macromol.4c02872
PDFAlkhoury, K. & Chester, S.A. (2025). “A chemo-thermo-mechanically coupled theory of photo-reacting polymers: Application to modeling photo-degradation with irradiation-driven heat transfer.” J. Mech. Phys. Solids, 197, 106050. https://doi.org/10.1016/j.jmps.2025.106050
White, Z.T., Smith, A.M. & Vernerey, F.J. (2025). “Mechanical cooperation between time-dependent and covalent bonds in molecular damage of polymer networks.” Communications Physics, 8, 265. DOI: 10.1038/s42005-025-02192-0
PDF