Supporting Information for A Multiscale Simulation Workflow for Solvated Molecular Systems in ChemShell: Application to Orthosilicic Acid Condensation

Simulation data accompanying the article: You Lu, Chin Yong, Xu Zhang, Thomas W. Keal, Alexey A. Sokol, and C. Richard A. Catlow, "A Multiscale Simulation Workflow for Solvated Molecular Systems in ChemShell: Application to Orthosilicic Acid Condensation," Journal of Chemical Theory and Computation (Under review). The work models the first condensation (Si–O–Si bond formation) of orthosilicic acid, Si(OH)4, in an acidic water–methanol solution (pH < 1, 333.15 K). Solution models were built and equilibrated by classical molecular dynamics with DL_FIELD and DL_POLY, and a non-periodic QM/MM cluster was cut from the equilibrated snapshot. The reactive free energy surface was sampled by well-tempered, multiple-walker metadynamics (20 walkers, 200 ps per walker) at the GFN2-xTB/OPLS2005 QM/MM level, evaluated through xtb and OpenMM and driven by PLUMED within ChemShell. The two biased collective variables are the attack distance d(O1898, Si2112) and the coordination number n of the electrophilic silicon Si2112 with its own four oxygen atoms. This deposit holds the structural models, the metadynamics output, the collective-variable time series, the reconstructed free energy surfaces, and the analysis scripts needed to reproduce and reanalyse the reported results. Structural models md_snapshot.xyz — PBC-wrapped classical MD snapshot used as the starting configuration (20,205 atoms). shell_model.xyz and shell_model_optimized.xyz — the 8,026-atom QM/MM cluster before and after QM/MM optimization. qm_region_optimized.xyz — the optimized 119-atom QM region (B3LYP-D3/6-31+G**/OPLS2005). Metadynamics run plumed_metadynamics.dat — the ChemShell-generated PLUMED input for the metadynamics production run. HILLS — accumulated Gaussian hills from the 20-walker run (8,000 entries). HILLS_angstrom.dat — the same hill file with d and its width converted to Å and the heights to kcal/mol; this is the file read by the sum_hills command quoted in the Supporting Information. COLVAR_walker00–COLVAR_walker19 — collective-variable time series per walker (d, n, n2–n5; 200,001 entries each). animation_walker00_qm.xyz–animation_walker19_qm.xyz — QM-region trajectories for the 20 walkers (2,001 frames each, 119 atoms, 100 fs stride). Free energy surfaces fes_2d.dat — the two-dimensional free energy surface in (d, n) from which Figure 6a was produced (Å, dimensionless, kcal/mol; zero at the global minimum). fes_d1_sumhills.dat and fes_c1_sumhills.dat — one-dimensional projections along d and along n obtained with PLUMED sum_hills (Å, kcal/mol). fes_1d_d.dat — the Boltzmann-weighted projection onto d used for Figure 6b (Å, kcal/mol; zero at the product minimum). fes_convergence.dat — the forward barrier ΔG‡ and the reaction free energy ΔG as a function of accumulated sampling, underlying Figure S3c. Analysis scripts (Python 3, NumPy only; each is run in this directory with no arguments) hill2fes.py — reconstructs the two-dimensional surface from HILLS, projects it onto d, reports ΔG‡ and ΔG, and writes fes_1d_d.dat. hill2fes_Figure-S3c.py — the same reconstruction repeated on a finer time grid; writes fes_convergence.dat. verify_Table-S3.py — regenerates Table S3 (siloxane bridge formation, proton release and transfer) and Note S3 (transient five-coordinate silicon) directly from the animation_walker*_qm.xyz trajectories; writes _verify_Table-S3.json and prints the corresponding tables. Conventions Cartesian coordinates in the XYZ files are in Angstroms, and atom labels quoted in the article (O1898, Si2112, and so on) follow the shell model's 0-based sequential numbering. In the raw PLUMED files the column headers d1 and c1–c5 correspond to d and n, n2–n5 respectively. In HILLS the centres and widths of d are in Bohr, n and its width are dimensionless, and the hill heights are in Hartree and already carry the well-tempered scaling γ/(γ−1). In the COLVAR files d is in Bohr and n, n2–n5 are dimensionless. Changes in this version This version accompanies the revised manuscript. It adds the two-dimensional free energy surface and its projections, the convergence data behind Figure S3c, the hill file in Å and kcal/mol, and the three analysis scripts. The one-dimensional projection previously deposited as fes_d1.dat is now named fes_d1_sumhills.dat, and the companion projection along n has been added. Please cite the article above when using these data.