Hello everyone,

I am encountering a persistent convergence failure during the geometry optimization step when building an amorphous model for a rubber composite (EPDM/carbon black/vulcanizing agent/antioxidant) using Materials Studio. Despite several attempts, the issue remains unresolved. I would appreciate any insights or suggestions.

1. Model Overview

• Matrix:​ 12 EPDM rubber chains, each with a polymerization degree of 100. Each individual chain has been successfully geometry optimized and converged to a reasonable structure.

• Filler:​ Multi-layer graphene sheets used as an equivalent for carbon black.

• Additives:​ Vulcanizing agent DCP, antioxidants RD and MB.

• Target Density:​ Approximately 0.89 g/cm³ (based on experimental value).

2. Modeling Procedure

1. Used the Amorphous Cell​ module to pack all components (12 EPDM chains, graphene sheets, DCP, RD, MB) into a single cubic cell.

2. Set the initial density to 0.3 g/cm³​ (a low value to provide relaxation space), with the goal of compressing to the target density in subsequent NPT dynamics.

3. In the Amorphous Cell options:

   • Checked "Check close contacts"​ and "Check energies".

   • Checked "Restrain ring"​ (to preserve ring geometries in components like ENB, if present).

   • Unchecked "Optimize geometry"​ (to prevent internal optimization by AC that might interfere).

   • Set the Quality to "Fine".

4. The AC construction completed, producing an initial cell.

3. The Problem: Geometry Optimization Failure

Upon importing the AC cell into the Forcite module for geometry optimization (in preparation for subsequent NPT equilibration), the optimization task stops immediately after calculating the initial energy, failing to proceed with iterations.

• Regardless of using the "Smart"​ or "Steepest descent"​ algorithm, the task terminates after the first step. The output contains only one frame (Frame 1), marked as "Converged: No".

• The energy decomposition shows severe abnormalities: Bondand Angleenergies are extremely high (hundreds to thousands of kcal/mol), and the van der Waalsenergy is even positive​ (indicating severe atomic overlaps/non-physical repulsion).

• Example Output (Excerpt):

  Frame  Converged   Total (kcal/mol)  Bond      Angle     vdW
    1       No        -12757.272       909.208  4133.526  234.236

4. Attempted Solutions

• Confirmed that single EPDM chains are independently optimized and converged.

• Rebuilt the AC cell multiple times, trying even lower initial densities (e.g., 0.25 g/cm³). The problem persists.

• In Forcite, attempted using the Steepest descent​ algorithm with significantly relaxed convergence criteria (e.g., Force tolerance = 0.5 or 1.0 kcal/mol/Å). The optimization still fails to take the first iterative step.

• Visual inspection of the initial AC configuration confirms severe atomic penetrations, especially between the graphene sheets and polymer chains.

5. Specific Questions

1. Build Order for Multi-Component Systems:​ For such a complex system containing long polymers, sheet fillers, and small molecules, is there a recommended build order? For example, should one first build a cell with polymer+filler, equilibrate it, and then insert the small molecules?

2. Filler Handling:​ During the build or initial optimization, is it necessary to Fix​ the graphene sheet atoms or apply constraints to prevent unreasonable penetration with polymer chains? If so, at what stage should these constraints be removed?

3. Initial Density:​ For systems with sheet-like fillers, is there an empirical range for the initial AC build density? Is 0.3 g/cm³ still too high?

4. Optimization Strategy:​ When facing extremely high initial energy and severe atomic overlap, are there alternative "forceful relaxation" methods besides Steepest descent with relaxed criteria? (e.g., running a short high-temperature NVT dynamics with a very small timestep to push atoms apart initially?).

5. Force Field Considerations:​ Using the COMPASSIII force field, are any special adjustments needed for non-bonded treatment (cutoff, correction methods) for the interactions between graphene and the polymer matrix?

6. Environment

• Software: Materials Studio 2023

• Modules: Forcite, Amorphous Cell

• Forcefield: COMPASSIII

• OS: Linux

Any advice, shared experience, or workflow suggestions would be immensely helpful for progressing with this project. Thank you for your time and assistance!