HVNM Knowledge Extraction Guide

This guide explains how to extract physical knowledge from HVNM model parameters and fitting results.

What Knowledge Can Be Extracted

Interphase Characterization:

  • \(\phi_I\): Interphase volume fraction (from multi-\(\phi\) SAOS)

  • \(\delta_m\): Mobile interphase thickness (from \(\phi_I\) vs NP geometry)

  • \(\beta_I\): Reinforcement ratio (surface chemistry / confinement strength)

Dual Activation Energies:

  • \(E_a^{mat}\): Matrix activation energy (from multi-T relaxation)

  • \(E_a^{int}\): Interfacial activation energy (from multi-T relaxation)

  • \(\Delta E_a^{surf} = E_a^{int} - E_a^{mat}\): Surface confinement penalty

Strain Amplification:

  • \(X(\phi)\) from modulus vs \(\phi\) calibration

  • Deviation from Guth-Gold suggests non-spherical NPs or aggregation

Two Topological Freezing Temperatures:

  • \(T_v^{mat}\): Matrix vitrimer freezing (BER arrest)

  • \(T_v^{int} > T_v^{mat}\): Interfacial freezing (higher barrier)

Payne Onset Strain:

  • \(\gamma_c^{NC} = \gamma_c / X_I\): Reduced critical strain from amplification

Parameter-to-Physics Map

Parameters

Derived Quantity

Physical Meaning

\(\phi, R_{NP}, \delta_m\)

\(\phi_I\)

Interphase fraction → NP dispersion quality

\(\beta_I\)

\(G_{I,eff}\)

Surface chemistry / confinement strength

\(E_a^{int} - E_a^{mat}\)

\(\Delta E_a^{surf}\)

Surface confinement penalty

\(V_{act}^{int} / V_{act}^{mat}\)

Ratio

Interfacial mechanochemical coupling

\(G_P \cdot X(\phi)\)

Effective plateau

Actual permanent modulus with amplification

Diagnostic Decision Tree

  1. Single relaxation mode in :math:`G’’` → use HVM (no interphase needed)

  2. Two relaxation modes + phi dependence → use HVNM

  3. Third slow mode in :math:`G’’` → include_diffusion=True

  4. Stress softening in cyclic tests → include_interfacial_damage=True

  5. Monotonic :math:`G’(phi)` matching Guth-Gold → standard HVNM

  6. :math:`G’(phi)` deviates from Guth-Gold → investigate NP aggregation

Multi-Protocol Fitting Strategy

  1. SAOS first: Identify \(G_P\), \(G_E\), \(G_{I,eff}\), mode timescales

  2. Multi-phi SAOS: Extract \(\beta_I\), \(\delta_m\) (interphase geometry)

  3. Relaxation: Confirm 4-mode spectrum (\(\tau_D, \tau_E, \tau_I, \infty\))

  4. Multi-T relaxation: Extract \(E_a^{mat}\), \(E_a^{int}\) (dual Arrhenius)

  5. Startup: Identify \(V_{act}^{mat}\), \(V_{act}^{int}\) (TST coupling)

  6. LAOS amplitude sweep: Confirm Payne onset at \(\gamma_c / X_I\)

Common Pitfalls

Dual factor-of-2 confusion:

Both the E-network and I-network exhibit the factor-of-2: \(\tau_{E,eff} = 1/(2k_{BER,0}^{mat})\) and \(\tau_{I,eff} = 1/(2k_{BER,0}^{int})\). A naive Maxwell fit to SAOS data will yield \(\tau_{fit} = \tau_{eff}\), not the true bond exchange time. When converting to BER rates, multiply the fitted time constant by 2: \(k_{BER,0} = 1/(2\tau_{fit})\). See also the HVM derivation (Factor-of-2 in Relaxation).

Guth-Gold deviations:

The Guth-Gold formula \(X(\phi) = 1 + 2.5\phi + 14.1\phi^2\) is accurate for well-dispersed spherical NPs at low to moderate loading (\(\phi < 0.3\)). Deviations can indicate:

  • Higher-than-predicted modulus: NP aggregation (effective larger particles)

  • Lower-than-predicted modulus: poor NP-matrix adhesion

  • Phi-dependent exponent: non-spherical NPs (rods, platelets)

Plot \(G'/G'_{unfilled}\) vs \(\phi\) and compare with Guth-Gold to diagnose.

Parameter identifiability with \(\phi\):

From single-\(\phi\) SAOS data alone, \(\beta_I\) and \(\phi_I\) are correlated — only the product \(G_{I,eff} = \beta_I G_E \phi_I\) is identifiable. Multi-phi SAOS data separates them because \(\phi_I\) varies with \(\phi\) (via NP geometry) while \(\beta_I\) does not.

Frozen interphase at low \(T\):

The interphase typically has higher activation energy than the matrix (\(E_a^{int} > E_a^{mat}\)), so it freezes at a higher temperature. Below \(T_v^{int}\), the I-network behaves as an elastic spring and its contribution becomes indistinguishable from an enhanced \(G_P\). Check \(k_{BER,0}^{int}\) at the experimental temperature before attributing a high plateau to permanent crosslinks alone.

NP-Surface Characterization

The HVNM interphase parameters encode NP-surface chemistry:

Parameter

Physical Meaning

How to Determine

\(\beta_I\)

Surface binding strength

Multi-\(\phi\) SAOS: fit \(G_{I,eff}\) vs \(\phi_I\)

\(E_a^{int}\)

Surface exchange barrier

Multi-T relaxation of slow mode

\(\delta_m\)

Interphase thickness

\(\phi_I = \phi[(R_{NP}+\delta_m)^3/R_{NP}^3 - 1]\); fit from multi-\(\phi\)

\(V_{act}^{int}\)

Interfacial mechanochemistry

Stress overshoot in startup at high \(\phi\)

Interpreting \(\Delta E_a^{surf} = E_a^{int} - E_a^{mat}\):

  • \(\Delta E_a \approx 0\): interphase exchanges as fast as matrix (weak NP-polymer interaction)

  • \(\Delta E_a \sim 20\) - 50 kJ/mol: moderate surface confinement (typical for silica in epoxy vitrimers)

  • \(\Delta E_a > 80\) kJ/mol: strong confinement (chemically grafted NPs)

Payne Effect Interpretation

The Payne effect — the decrease of \(G'\) with increasing strain amplitude in filled rubbers — is naturally captured by HVNM:

  1. At small \(\gamma_0\): all networks respond linearly, \(G' = G_P X + G_E + G_D + G_{I,eff} X_I\) (full modulus)

  2. As \(\gamma_0\) increases: the I-network natural state begins tracking the deformation via BER, reducing \(\sigma_I\)

  3. At large \(\gamma_0\): \(\sigma_I \to 0\) at steady state, and \(G'\) drops to the unfilled level

The onset strain is reduced by strain amplification: \(\gamma_c^{NC} = \gamma_c / X_I\), where \(\gamma_c\) is the onset for the unfilled material. Higher \(\phi\) lowers the onset strain.

Worked Example: Identifying \(\phi_I\) from Multi-\(\phi\) SAOS

Procedure:

  1. Prepare samples at \(\phi =\) 0, 0.05, 0.10, 0.15, 0.20

  2. Fit SAOS with HVNM at each \(\phi\), extracting \(G_{I,eff}(\phi)\)

  3. Compute theoretical \(\phi_I(\phi)\) from NP geometry: \(\phi_I = \phi[(R_{NP}+\delta_m)^3/R_{NP}^3 - 1]\)

  4. Plot \(G_{I,eff}\) vs \(\phi_I\) — slope gives \(\beta_I G_E\)

  5. With \(G_E\) known from the \(\phi=0\) fit, extract \(\beta_I\)

  6. From the \(\phi_I(\phi)\) relationship, extract \(\delta_m\)

Validation: The unfilled (\(\phi=0\)) fit should match HVM exactly, and \(\beta_I\) should be independent of \(\phi\).

When to Use HVNM vs HVM

Use HVM if

Use HVNM if

Unfilled vitrimer

NP-filled vitrimer

Single relaxation mode

Multi-timescale relaxation

No \(\phi\) dependence

Modulus depends on \(\phi\)

No Payne effect

Payne effect observed

\(T_v^{mat}\) sufficient

Two freezing temperatures needed

Troubleshooting

Modulus doesn’t match Guth-Gold scaling: Guth-Gold \(X(\phi) = 1 + 2.5\phi + 14.1\phi^2\) assumes well-dispersed spherical NPs. Deviations indicate NP aggregation (higher-than-predicted modulus) or non-spherical particles (different amplification exponents). Plot \(G'/G'_{\phi=0}\) vs \(\phi\) and compare with the quadratic.

Interphase appears frozen at experimental temperature: The interphase activation energy \(E_a^{int}\) is typically higher than \(E_a^{mat}\). If \(k_{BER,0}^{int} < 10^{-6}\) s-1, the interphase is effectively elastic on experimental timescales. To model this, use a high \(E_a^{int}\) (up to 250 kJ/mol) rather than reducing \(\nu_0^{int}\) (which may violate parameter bounds).

ODE solver diverges at high phi: High \(\phi\) amplifies the affine deformation (\(X_I \dot{\gamma}\)), creating stiff ODEs. Increase max_steps or reduce the shear rate. See Numerical Implementation for solver details.

phi=0 gives slightly different results from HVM: This should not happen — HVNM with \(\phi = 0\) is verified to recover HVM to machine precision. If discrepancy occurs, check that \(\phi\) is exactly 0.0 (not a small nonzero value). See \phi = 0 Recovery Verification for the mathematical proof.

Parameter identifiability with limited data: With single-\(\phi\) SAOS data, \(\beta_I\) and \(\phi_I\) are correlated (only their product \(G_{I,eff}\) matters for SAOS). Multi-phi SAOS data is needed to separate these. Similarly, \(\nu_0^{int}\) and \(E_a^{int}\) require multi-temperature data for independent estimation.

Interfacial damage makes results irreversible: If \(D_{int}\) accumulates but you expect recovery, ensure self-healing is properly configured. Check that \(T > T_v^{int}\) (healing is Arrhenius-activated). See Enhanced Damage Mechanics for the healing model.

Relaxation has unexplained slow tail: A long-time tail slower than any Maxwell mode may indicate diffusion-limited exchange. Try include_diffusion=True and fit \(k_{diff}\). See Diffusion-Limited Slow Mode.

Cross-Protocol Validation

Use multiple protocols to validate the HVNM fit:

Check

Criterion

Failing Suggests

\(G_P X\) from SAOS = \(G(\infty)\) from relaxation

\(\lim_{\omega \to 0} G' \approx G(t \to \infty)\)

Incorrect \(G_P\) or \(\phi\)

\(\tau_{I,eff}\) from SAOS = \(\tau_I\) from relaxation

Loss peak frequency \(\approx 1/\tau_{I,eff}\)

Interphase TST distorting linear regime

\(G_P X\) increases with \(\phi\) as Guth-Gold

\(G'(\omega \to 0)\) vs \(\phi\) follows quadratic

NP aggregation or non-spherical particles

\(\sigma_I \to 0\) at steady state

I-network stress vanishes in long startup

\(k_{BER}^{int}\) too slow at this \(T\)

This is analogous to the HVM cross-protocol workflow (Troubleshooting).

Application Examples

Multi-phi SAOS workflow: Prepare samples at \(\phi =\) 0, 0.05, 0.10, 0.15, 0.20. Fit SAOS at each \(\phi\) to extract \(G_{I,eff}(\phi)\). Plot vs computed \(\phi_I\) to determine \(\beta_I\) (slope) and validate the Guth-Gold amplification. The \(\phi = 0\) fit should exactly match HVM.

Temperature sweep for dual \(E_a\): Fit SAOS at 3+ temperatures. The E-network loss peak shifts as \(\tau_E(T) = 1/(2k_{BER,0}^{mat}(T))\) — Arrhenius slope gives \(E_a^{mat}\). The I-network loss peak shifts independently — its Arrhenius slope gives \(E_a^{int}\). Expect \(E_a^{int} > E_a^{mat}\) for confined interphase.

Cyclic loading analysis: Perform strain-amplitude sweeps at fixed \(\omega\). The Payne onset occurs at \(\gamma_c^{NC} = \gamma_c / X_I\) — lower than the unfilled material by the strain amplification factor. If include_interfacial_damage=True, the modulus does not fully recover on unloading (Mullins effect). Recovery timescale depends on \(T\) through the self-healing rate \(h_{int}(T)\).

Interphase thickness from geometry: With known \(R_{NP}\) (TEM) and \(\phi_I\) (from multi-phi SAOS fit), solve \(\phi_I = \phi[(R_{NP}+\delta_m)^3/R_{NP}^3 - 1]\) for \(\delta_m\). Typical values: \(\delta_m \sim 2\text{-}20\) nm for polymer-NP interphases. Compare with the Kuhn length for consistency.