2.3 A molecular energy difference with a controlled scan
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.
2.3.1 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
2.3.2 Worked investigation
Intuition and prerequisites. Energy differences answer a specified transformation, not an unspecified comparison of output numbers. A rigid O–H stretch, a relaxed bond scan and a dissociation curve are different experiments. This case uses a simple symmetric water stretch to make reference conventions transparent. Start from A and understand coordinate transformations; use an external script or careful arithmetic to generate structures.
Original geometry recipe. Let rO=(8,8,8) angstrom, v1=(0.98,0,0), v2=(−0.25,0.95,0). For scale s in 0.94, 0.97, 1.00, 1.03, 1.06, set H1=rO+s v1 and H2=rO+s v2. The angle is fixed and the two starting bond lengths remain slightly different. Keep ENERGY_FORCE and a unique PROJECT for each s. This is a complete coordinate-generation rule; it is not a claim that water's equilibrium bond length lies at s=1.
Workflow. 1. Generate all coordinates and independently recompute the two distances and angle. 2. Run single-point calculations using identical electronics and boundaries. 3. Choose a reference member, explicitly s=1, and plot ΔE(s)=E(s)−E(1). 4. Project the forces onto the displacement direction and compare their signs to the slope of the energy scan. 5. Fit a quadratic only near a sampled minimum and inspect residuals; do not fit a broad dissociation region to a harmonic form. 6. Repeat with a larger basis or tighter grid to see whether the curve shape, rather than its arbitrary vertical offset, changes.
Interpretation. A minimum in this one-dimensional constrained path is not necessarily a full molecular stationary point because the angle and asymmetric stretching degree remain fixed. A smooth curve does not prove accuracy: systematic functional errors can be smooth. Forces provide a stronger local consistency check than a visually smooth line. If you later allow geometry relaxation at fixed s, name the result a relaxed scan and state exactly what remains constrained.
Checks and limits. Use the same energy expression for every member; do not combine free energy at finite electronic temperature with zero-temperature total energy. Avoid using unrelated wavefunction restarts that bias one end toward a different state. A long bond-dissociation scan can need unrestricted or multireference treatment; the short closed-shell stretch here is not an endorsement of PBE restricted DFT at dissociation. Coordinate rounding should be smaller than the deliberate displacement scale.
Exercise. Design a second scan that changes the angle while preserving both bond lengths, then predict which force components would project onto that coordinate. Produce a comparison between a local quadratic fit and the raw sampled points, including fit range and residuals; do not invent a force constant before running.
Diagram. A water molecule with two radial arrows leads to a one-dimensional curve. Dashed arrows mark frozen angular motion. Labels: scan coordinate s, fixed angle, relative energy, local fit window. Mark the curve “schematic.”