Skip to content

8.3 Design and audit a QM/MM boundary

These are unexecuted teaching inputs and starting models. Original diagrams are schematics, not calculated results. Validate version-specific syntax, licensed or authorized data, numerical convergence and the scientific model before using this workflow.

8.3.1 Model, units and provenance

Keep basis/potential files and executable versions traceable. Grid controls use Ry; common energy/force outputs use hartree and hartree/bohr. Read the unit in each output heading.

Shared inputs, conventions and evidence

Original schematic: Design and audit a QM/MM boundary. No numerical results are claimed.
Original schematic: Design and audit a QM/MM boundary. No numerical results are claimed.

8.3.2 Unexecuted inputs and explicit deltas

Use the accompanying instructions to identify the parent calculation and placement of every delta; a snippet is not automatically a standalone input. Preserve all blank-line and file-provenance requirements.

8.3.2.1 Input block 1

&LINK
  QM_INDEX 11
  MM_INDEX 12
  LINK_TYPE IMOMM
  QM_KIND H
  ALPHA_IMOMM 1.38
&END LINK

8.3.3 Worked investigation

Intuition and prerequisite. QM/MM places an electronically responsive region inside a cheaper molecular-mechanics environment. The boundary is a scientific approximation, not merely an atom selection. A covalent bond cut requires a link treatment; electrostatic embedding also requires careful treatment of nearby MM charges. Cutting a conjugated group or separating a proton donor from its reactive acceptor can invalidate the model even when CP2K runs successfully. First reproduce an independently supplied, fully parameterized MM system and a no-covalent-cut QM/MM example. The CECAM urea-in-water exercise illustrates that topology preparation and MM equilibration precede reactive QM/MM sampling; its semiempirical setup is not automatically a GPW template.

Scope and original boundary delta. This is an auditable boundary-design exercise, deliberately not a complete runnable topology. Suppose a verified topology contains a saturated C–C bond between atoms 11 and 12. Atom 11 is in the QM region and atom 12 remains MM. The following is valid LINK section syntax to insert inside the existing FORCE_EVAL/QMMM section, after the complete QM_KIND selection, DFT settings, MM force field, topology, coupling, and cells have been supplied independently.

See input block 1 above.

The indices are example topology indices, not transferable chemical identities. ALPHA_IMOMM is dimensionless; 1.38 is a starting illustration, not a fitted value. Determine a defensible ratio of equilibrium QM–MM bond length to QM–H capping-bond length for the actual boundary. The cap is part of the link construction, not a new solvent hydrogen to insert arbitrarily into the original coordinate list. Consult the 2025.2 LINK reference for index and projection conventions.

Embedding decisions. Choose the electrostatic coupling compatible with the actual QM method. In the 2025.2 QMMM reference, GAUSS is the Gaussian-expanded coupling route relevant to GPW, whereas the COULOMB option is not available for GPW/GAPW. This is not a request to change a working model blindly. The QM cell, full-system cell, and their electrostatic boundary conditions must be reviewed together. Set charge and spin for the actual QM electronic system, including its capping construction, rather than copying the total solvated-system charge. The QM_KIND mapping reference identifies QM atoms through MM indices.

Procedure. 1. Build an atom-map table: global index, element, residue, MM type/charge, QM/MM assignment, and every cut bond. Visualize it against the exact coordinate file used for execution. 2. Audit all required parameters and total charges. Confirm that each QM atom appears once and that link bonds are chemically sensible, preferably away from the active center and conjugation. 3. Inspect how the chosen link scheme handles boundary charges. Do not zero a nearby charge without assessing charge redistribution and the total model charge. 4. Compute single-point energies and forces on several representative snapshots before dynamics. Check the boundary for implausibly large forces and compare finite-difference energy derivatives where feasible. 5. Enlarge the QM region to include adjacent groups or important solvent molecules. Compare the target observable, such as a reaction energy or local force, using internally consistent reference states. 6. Re-equilibrate after switching Hamiltonians. Test timestep and energy conservation, then proceed to the intended sampling only after boundary and region-size sensitivity are acceptable.

Interpretation, pitfalls, and exercise. Absolute total energies from different partitions need not be directly comparable; compare reaction differences computed consistently within each partition. A fixed-charge MM environment cannot reproduce every polarization or charge-transfer effect. Link atoms can distort density near the boundary, and strong neighboring point charges can overpolarize the QM region. In CP2K, FIST_SCALE_FACTOR changes an MM charge in the classical description and that changed charge also enters the QM/MM potential, whereas QMMM_SCALE_FACTOR changes only its QM/MM potential contribution; they are not interchangeable switches. Exercise: propose two partitions for a solvated alcohol reaction, identify whether the attacking water and proton relay must be QM, and justify every cut. Deliver a boundary map and validation plan rather than fabricated parameters.

8.3.5 Sources and further reading