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:** - :math:`\phi_I`: Interphase volume fraction (from multi-:math:`\phi` SAOS) - :math:`\delta_m`: Mobile interphase thickness (from :math:`\phi_I` vs NP geometry) - :math:`\beta_I`: Reinforcement ratio (surface chemistry / confinement strength) **Dual Activation Energies:** - :math:`E_a^{mat}`: Matrix activation energy (from multi-T relaxation) - :math:`E_a^{int}`: Interfacial activation energy (from multi-T relaxation) - :math:`\Delta E_a^{surf} = E_a^{int} - E_a^{mat}`: Surface confinement penalty **Strain Amplification:** - :math:`X(\phi)` from modulus vs :math:`\phi` calibration - Deviation from Guth-Gold suggests non-spherical NPs or aggregation **Two Topological Freezing Temperatures:** - :math:`T_v^{mat}`: Matrix vitrimer freezing (BER arrest) - :math:`T_v^{int} > T_v^{mat}`: Interfacial freezing (higher barrier) **Payne Onset Strain:** - :math:`\gamma_c^{NC} = \gamma_c / X_I`: Reduced critical strain from amplification Parameter-to-Physics Map ------------------------- .. list-table:: :widths: 25 35 40 :header-rows: 1 * - Parameters - Derived Quantity - Physical Meaning * - :math:`\phi, R_{NP}, \delta_m` - :math:`\phi_I` - Interphase fraction → NP dispersion quality * - :math:`\beta_I` - :math:`G_{I,eff}` - Surface chemistry / confinement strength * - :math:`E_a^{int} - E_a^{mat}` - :math:`\Delta E_a^{surf}` - Surface confinement penalty * - :math:`V_{act}^{int} / V_{act}^{mat}` - Ratio - Interfacial mechanochemical coupling * - :math:`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 :math:`G_P`, :math:`G_E`, :math:`G_{I,eff}`, mode timescales 2. **Multi-phi SAOS**: Extract :math:`\beta_I`, :math:`\delta_m` (interphase geometry) 3. **Relaxation**: Confirm 4-mode spectrum (:math:`\tau_D, \tau_E, \tau_I, \infty`) 4. **Multi-T relaxation**: Extract :math:`E_a^{mat}`, :math:`E_a^{int}` (dual Arrhenius) 5. **Startup**: Identify :math:`V_{act}^{mat}`, :math:`V_{act}^{int}` (TST coupling) 6. **LAOS amplitude sweep**: Confirm Payne onset at :math:`\gamma_c / X_I` Common Pitfalls ---------------- **Dual factor-of-2 confusion:** Both the E-network and I-network exhibit the :ref:`factor-of-2 `: :math:`\tau_{E,eff} = 1/(2k_{BER,0}^{mat})` and :math:`\tau_{I,eff} = 1/(2k_{BER,0}^{int})`. A naive Maxwell fit to SAOS data will yield :math:`\tau_{fit} = \tau_{eff}`, not the true bond exchange time. When converting to BER rates, multiply the fitted time constant by 2: :math:`k_{BER,0} = 1/(2\tau_{fit})`. See also the HVM derivation (:ref:`hvm-factor-of-2`). **Guth-Gold deviations:** The Guth-Gold formula :math:`X(\phi) = 1 + 2.5\phi + 14.1\phi^2` is accurate for well-dispersed spherical NPs at low to moderate loading (:math:`\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 :math:`G'/G'_{unfilled}` vs :math:`\phi` and compare with Guth-Gold to diagnose. **Parameter identifiability with** :math:`\phi`: From single-:math:`\phi` SAOS data alone, :math:`\beta_I` and :math:`\phi_I` are correlated — only the product :math:`G_{I,eff} = \beta_I G_E \phi_I` is identifiable. Multi-phi SAOS data separates them because :math:`\phi_I` varies with :math:`\phi` (via NP geometry) while :math:`\beta_I` does not. **Frozen interphase at low** :math:`T`: The interphase typically has higher activation energy than the matrix (:math:`E_a^{int} > E_a^{mat}`), so it freezes at a higher temperature. Below :math:`T_v^{int}`, the I-network behaves as an elastic spring and its contribution becomes indistinguishable from an enhanced :math:`G_P`. Check :math:`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: .. list-table:: :widths: 25 25 50 :header-rows: 1 * - Parameter - Physical Meaning - How to Determine * - :math:`\beta_I` - Surface binding strength - Multi-:math:`\phi` SAOS: fit :math:`G_{I,eff}` vs :math:`\phi_I` * - :math:`E_a^{int}` - Surface exchange barrier - Multi-T relaxation of slow mode * - :math:`\delta_m` - Interphase thickness - :math:`\phi_I = \phi[(R_{NP}+\delta_m)^3/R_{NP}^3 - 1]`; fit from multi-:math:`\phi` * - :math:`V_{act}^{int}` - Interfacial mechanochemistry - Stress overshoot in startup at high :math:`\phi` **Interpreting** :math:`\Delta E_a^{surf} = E_a^{int} - E_a^{mat}`: - :math:`\Delta E_a \approx 0`: interphase exchanges as fast as matrix (weak NP-polymer interaction) - :math:`\Delta E_a \sim 20` - 50 kJ/mol: moderate surface confinement (typical for silica in epoxy vitrimers) - :math:`\Delta E_a > 80` kJ/mol: strong confinement (chemically grafted NPs) Payne Effect Interpretation ----------------------------- The Payne effect — the decrease of :math:`G'` with increasing strain amplitude in filled rubbers — is naturally captured by HVNM: 1. At small :math:`\gamma_0`: all networks respond linearly, :math:`G' = G_P X + G_E + G_D + G_{I,eff} X_I` (full modulus) 2. As :math:`\gamma_0` increases: the I-network natural state begins tracking the deformation via BER, reducing :math:`\sigma_I` 3. At large :math:`\gamma_0`: :math:`\sigma_I \to 0` at steady state, and :math:`G'` drops to the unfilled level The **onset strain** is reduced by strain amplification: :math:`\gamma_c^{NC} = \gamma_c / X_I`, where :math:`\gamma_c` is the onset for the unfilled material. Higher :math:`\phi` lowers the onset strain. Worked Example: Identifying :math:`\phi_I` from Multi-:math:`\phi` SAOS ------------------------------------------------------------------------- **Procedure:** 1. Prepare samples at :math:`\phi =` 0, 0.05, 0.10, 0.15, 0.20 2. Fit SAOS with HVNM at each :math:`\phi`, extracting :math:`G_{I,eff}(\phi)` 3. Compute theoretical :math:`\phi_I(\phi)` from NP geometry: :math:`\phi_I = \phi[(R_{NP}+\delta_m)^3/R_{NP}^3 - 1]` 4. Plot :math:`G_{I,eff}` vs :math:`\phi_I` — slope gives :math:`\beta_I G_E` 5. With :math:`G_E` known from the :math:`\phi=0` fit, extract :math:`\beta_I` 6. From the :math:`\phi_I(\phi)` relationship, extract :math:`\delta_m` **Validation:** The unfilled (:math:`\phi=0`) fit should match HVM exactly, and :math:`\beta_I` should be independent of :math:`\phi`. When to Use HVNM vs HVM ------------------------- .. list-table:: :widths: 50 50 :header-rows: 1 * - Use HVM if - Use HVNM if * - Unfilled vitrimer - NP-filled vitrimer * - Single relaxation mode - Multi-timescale relaxation * - No :math:`\phi` dependence - Modulus depends on :math:`\phi` * - No Payne effect - Payne effect observed * - :math:`T_v^{mat}` sufficient - Two freezing temperatures needed .. _hvnm-cross-protocol-validation: Troubleshooting ---------------- **Modulus doesn't match Guth-Gold scaling:** Guth-Gold :math:`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 :math:`G'/G'_{\phi=0}` vs :math:`\phi` and compare with the quadratic. **Interphase appears frozen at experimental temperature:** The interphase activation energy :math:`E_a^{int}` is typically higher than :math:`E_a^{mat}`. If :math:`k_{BER,0}^{int} < 10^{-6}` s\ :sup:`-1`, the interphase is effectively elastic on experimental timescales. To model this, use a high :math:`E_a^{int}` (up to 250 kJ/mol) rather than reducing :math:`\nu_0^{int}` (which may violate parameter bounds). **ODE solver diverges at high phi:** High :math:`\phi` amplifies the affine deformation (:math:`X_I \dot{\gamma}`), creating stiff ODEs. Increase ``max_steps`` or reduce the shear rate. See :ref:`hvnm-numerical` for solver details. **phi=0 gives slightly different results from HVM:** This should not happen — HVNM with :math:`\phi = 0` is verified to recover HVM to machine precision. If discrepancy occurs, check that :math:`\phi` is exactly 0.0 (not a small nonzero value). See :ref:`hvnm-phi-zero` for the mathematical proof. **Parameter identifiability with limited data:** With single-:math:`\phi` SAOS data, :math:`\beta_I` and :math:`\phi_I` are correlated (only their product :math:`G_{I,eff}` matters for SAOS). Multi-phi SAOS data is needed to separate these. Similarly, :math:`\nu_0^{int}` and :math:`E_a^{int}` require multi-temperature data for independent estimation. **Interfacial damage makes results irreversible:** If :math:`D_{int}` accumulates but you expect recovery, ensure self-healing is properly configured. Check that :math:`T > T_v^{int}` (healing is Arrhenius-activated). See :ref:`hvnm-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 :math:`k_{diff}`. See :ref:`hvnm-diffusion-mode`. Cross-Protocol Validation -------------------------- Use multiple protocols to validate the HVNM fit: .. list-table:: :widths: 25 25 50 :header-rows: 1 * - Check - Criterion - Failing Suggests * - :math:`G_P X` from SAOS = :math:`G(\infty)` from relaxation - :math:`\lim_{\omega \to 0} G' \approx G(t \to \infty)` - Incorrect :math:`G_P` or :math:`\phi` * - :math:`\tau_{I,eff}` from SAOS = :math:`\tau_I` from relaxation - Loss peak frequency :math:`\approx 1/\tau_{I,eff}` - Interphase TST distorting linear regime * - :math:`G_P X` increases with :math:`\phi` as Guth-Gold - :math:`G'(\omega \to 0)` vs :math:`\phi` follows quadratic - NP aggregation or non-spherical particles * - :math:`\sigma_I \to 0` at steady state - I-network stress vanishes in long startup - :math:`k_{BER}^{int}` too slow at this :math:`T` This is analogous to the HVM cross-protocol workflow (:ref:`hvm-cross-protocol-validation`). Application Examples --------------------- **Multi-phi SAOS workflow:** Prepare samples at :math:`\phi =` 0, 0.05, 0.10, 0.15, 0.20. Fit SAOS at each :math:`\phi` to extract :math:`G_{I,eff}(\phi)`. Plot vs computed :math:`\phi_I` to determine :math:`\beta_I` (slope) and validate the Guth-Gold amplification. The :math:`\phi = 0` fit should exactly match HVM. **Temperature sweep for dual** :math:`E_a`: Fit SAOS at 3+ temperatures. The E-network loss peak shifts as :math:`\tau_E(T) = 1/(2k_{BER,0}^{mat}(T))` — Arrhenius slope gives :math:`E_a^{mat}`. The I-network loss peak shifts independently — its Arrhenius slope gives :math:`E_a^{int}`. Expect :math:`E_a^{int} > E_a^{mat}` for confined interphase. **Cyclic loading analysis:** Perform strain-amplitude sweeps at fixed :math:`\omega`. The Payne onset occurs at :math:`\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 :math:`T` through the self-healing rate :math:`h_{int}(T)`. **Interphase thickness from geometry:** With known :math:`R_{NP}` (TEM) and :math:`\phi_I` (from multi-phi SAOS fit), solve :math:`\phi_I = \phi[(R_{NP}+\delta_m)^3/R_{NP}^3 - 1]` for :math:`\delta_m`. Typical values: :math:`\delta_m \sim 2\text{-}20` nm for polymer-NP interphases. Compare with the Kuhn length for consistency.