orca-step 2026.7.16

A SEAMM plug-in for ORCA

SEAMM ORCA Plug-in

A SEAMM plug-in for ORCA

• Free software: BSD-3-Clause

• Documentation: https://molssi-seamm.github.io/orca_step/index.html

• Code: https://github.com/molssi-seamm/orca_step

Features

A SEAMM plug-in for ORCA, a general-purpose quantum-chemistry program, with an emphasis on accurate molecular calculations such as DLPNO-CCSD(T).

Like the MOPAC and Gaussian steps, the ORCA step is a sub-flowchart: you add an ORCA node to your flowchart and then build a small sub-flowchart of ORCA capabilities inside it. Initially the available capabilities are:

• Energy – a single-point energy.

• Optimization – a geometry optimization.

Further capabilities (frequencies, properties, …) will be added.

Methods are described by the step’s metadata and can be set either explicitly in the ORCA dialog (similar to the Gaussian step) or, by default, taken from a preceding Model Chemistry step. Basis sets default to ORCA’s built-in families (Pople, Dunning cc, and Karlsruhe def2), with the Basis Set Exchange available as a planned opt-in source.

Acknowledgements

This package was created with the molssi-seamm/cookiecutter-seamm-plugin tool, which is based on the excellent Cookiecutter.

Developed by the Molecular Sciences Software Institute (MolSSI), which receives funding from the National Science Foundation under award CHE-2136142.

History

2026.7.16 – Structure handling for Optimization/Frequencies, and unit fixes

• Bugfix: the optimized geometry from an Optimization sub-step is now correctly carried forward, so a following sub-step (e.g. Frequencies) runs at the optimized structure. Previously ORCA’s orca.xyz was discarded before it could be read, so the frequencies were computed at the original, un-optimized geometry.

• The Optimization and Frequencies sub-steps now have the standard Structure handling options – overwrite the current configuration (default), create a new configuration, create a new system and configuration, or discard the structure – to control where the resulting structure and its properties are stored.

• Units: the zero-point energy, enthalpy, and Gibbs free energy are now reported in kJ/mol (SEAMM’s SI-based default), and the HOMO/LUMO orbital energies in eV (the conventional unit for orbital energies) throughout the ORCA step.

• The Frequencies step now also reports the largest of the 5 or 6 nominally-zero translation/rotation frequencies – a gauge of the numerical accuracy of the Hessian – and writes the frequencies (and IR intensities) to frequencies.csv in the step directory. Because ORCA projects the translations/rotations to exactly 0.00 in its printed frequencies, this residual is computed from the raw, un-projected mass-weighted Hessian (orca.hess), so it shows the true numerical value rather than zero.

• The Frequencies output now includes a table of the frequencies and IR intensities, and the step writes an ``IR_spectrum.graph`` file with the IR spectrum as a stick trace plus a Lorentzian-broadened trace that mimics an experimental spectrum.

• Each ORCA sub-step’s output is now followed by a blank line, so the sub-steps are visually separated in the output.

• When a sub-step runs, its output now names the actual level of theory – the resolved model-chemistry level spec including the resolved basis (e.g. ORCA:DFT@B3LYP/def2-SVP; the basis is appended when the spec itself omits it and the step fills in its own), or the explicit method/basis – instead of the generic “the model chemistry”.

2026.7.15 – Frequencies sub-step and an MDI Hessian command

• New Frequencies sub-step: the Hessian and harmonic vibrational frequencies, IR intensities, and thermochemistry (zero-point energy, enthalpy, entropy, Gibbs free energy) via ORCA’s analytic (AnFreq) or numerical (NumFreq) second derivatives, at a chosen temperature. Imaginary frequencies are reported and flagged.

• The ORCA MDI engine now answers a custom ``<HESSIAN`` command, returning the analytic Cartesian Hessian (via AnFreq). It advertises <HESSIAN only when ORCA has an analytic Hessian for the method (HF, MP2, ordinary DFT) – not double hybrids or (DLPNO-)CCSD(T) – so a driver’s capability check is truthful. A driver (e.g. the Normal Mode Sampling step) pulls the analytic Hessian over a warm MDI connection when offered, or finite-differences the forces otherwise.

2026.7.13.1 – BSSE: optionally write the wavefunction for DDEC6 charges

• New Write the wavefunction (wfx) file option on the BSSE sub-step (default off). When on, the dimer’s density is retained and converted to an orca.wfx (via orca_2aim), so a following Atomic Charges step can compute DDEC6 charges on the counterpoise complex – just as after an Energy step.

2026.7.13 – Bugfix: correct the canonical CCSD(T)-F12 keyword

• The canonical F12 method keyword was CCSD(T)-F12D, which ORCA rejects; it is now CCSD(T)-F12D/RI (canonical F12 uses the RI approximation for the F12 integrals). DLPNO-CCSD(T)-F12D is unchanged (DLPNO implies RI, and ORCA rejects a /RI on it).

• F12 methods are no longer advertised as generic model chemistries – they need a specific F12 orbital basis, not the generic advertised bases.

2026.7.10.1 – F12 methods, automatic grid for high-L bases, and GUI fixes

• Added the explicitly-correlated CCSD(T)-F12D and DLPNO-CCSD(T)-F12D methods to the Method pull-down, and the cc-pVDZ-F12 / cc-pVTZ-F12 / cc-pVQZ-F12 orbital bases to the basis list.

• An F12 method now adds its complementary auxiliary basis (CABS) automatically – <basis>-CABS derived from the chosen F12 basis – unless one is already in the extra keywords.

• Selecting an F12 method narrows the basis-set list to the F12 bases and hides the Basis-set source control (forced to ORCA-internal, as the Basis Set Exchange has no CABS), so only a valid basis can be chosen.

• When the integration grid is left on default, it is automatically set to ``DEFGRID3`` for high-angular-momentum basis sets (h functions or above, e.g. cc-pV5Z), determined from the Basis Set Exchange.

• Bugfix: sub-steps can now be deleted – right-clicking a sub-step in the ORCA sub-flowchart now shows the popup menu (Edit / Delete); the menu code was missing.

• Documented that F12 largely removes basis-set superposition error, so the counterpoise (BSSE) correction is unnecessary for F12 methods.

2026.7.10 – BSSE: energy-only mode (enables CCSD(T))

• New Compute the gradient control on the BSSE sub-step. With it set to no (energy only), the counterpoise correction runs without a gradient, which is cheaper and works for methods that have no analytic gradient in ORCA – notably CCSD(T) / DLPNO-CCSD(T) – for gold-standard counterpoise interaction energies. The default (yes) is unchanged and still produces the corrected energy and gradient for MLFF training.

2026.7.9.2 – BSSE (counterpoise) sub-step

• New BSSE sub-step: the counterpoise-corrected (Boys–Bernardi) energy and gradient of a two-fragment complex, in a single ORCA run, for BSSE-free machine-learned-force-field training data. It drives ORCA’s Compound facility (the BSSEGradient script by D. G. Liakos & F. Neese).

• Reports the BSSE-corrected energy, the uncorrected (raw) energy, and the correction (in E_h and kcal/mol), plus the corrected gradient; each can be saved from the Results tab.

• Fragments are found automatically from the two molecules in the structure (so it works directly on a Dimer Builder dimer) or specified by atom; an option relaxes the free monomers before the correction.

• First version: a neutral, closed-shell complex of exactly two fragments with an ORCA-internal basis set. Any analytic-gradient method works, including dispersion-corrected and double-hybrid DFT and MP2.

2026.7.9.1 – SCF SThresh control on the Energy step

• New SCF SThresh control: set ORCA’s SCF convergence threshold (the %scf SThresh value, in E_h). Leave it at default to let the SCF convergence preset (or ORCA’s own default) govern SThresh, or give an explicit value – e.g. ORCA’s nominal 1.0e-07 – to write it out and override whatever the preset would otherwise set. Lower it for a tighter SCF (smoother energies and forces), raise it to converge more loosely. Available on the Energy and Optimization sub-steps.

2026.7.9 – MDI engine, integration-grid and SCF controls, and config fixes

• ORCA can now be driven as a persistent MDI engine, so steps that set up a model chemistry and evaluate it at many geometries (for example the Dimer Builder’s energy-based contact search) can use ORCA – for methods with an analytic gradient. Set this up with a Model Chemistry step; the ORCA step does not need configuring for it.

• New Integration grid control: choose ORCA’s grid preset (DEFGRID1, DEFGRID2, or DEFGRID3), or leave ORCA’s default.

• New SCF convergence control: choose the convergence-tolerance preset (SLOPPYSCF … EXTREMESCF), or leave ORCA’s default. It defaults to TIGHTSCF for smooth energies and forces (previously this was applied via the extra-keywords default, which is now empty).

• Bugfix: a $variable typed into the basis-set field (e.g. to vary the basis in a Loop) is now expanded to its value instead of being passed to ORCA literally.

• Bugfix: the pre-run description of a step showed the basis as a raw dictionary; it now shows the basis name.

• How to find and launch ORCA – its executable path and, for parallel runs, the OpenMPI library directory – now lives in ~/SEAMM/orca.ini (a template is created on install). The [orca-step] section of the main SEAMM configuration keeps only the user run options (ncores, memory). NOTE: if you previously set library-path (or a path) in the [orca-step] section, move it into ~/SEAMM/orca.ini.

2026.7.8 – Ordered basis-set list and complete-basis-set (CBS) extrapolation

• The basis-set list is now ordered by family and, within a family, into valence / polarization / diffuse ladders that each rise DZ -> TZ -> QZ -> 5Z, so a sensible progression is a single ladder read top to bottom.

• New complete-basis-set (CBS) extrapolation on the Energy step: set ‘Basis-set extrapolation’ to 2/3, 3/4, or 4/5 and pick a family (cc, aug-cc, def2, or ANO). This is a single ORCA job (its Extrapolate keyword) that runs both basis sets and extrapolates the SCF and correlation parts. When it is on, the fixed basis set is ignored, and gradients are not available (ORCA has no gradient for an extrapolated energy).

• The CBS control is hidden for the Optimization step, which needs a gradient that an extrapolated energy does not provide.

2026.7.6.1 – Bugfix: parallel execution and DFT functionals via Model Chemistry

• Bugfix: the library-path (and orca-path) settings were read under the wrong key and so were ignored; they are now applied. The matching OpenMPI ‘mpirun’ (the sibling ‘bin’ of library-path) is put on PATH so ORCA launches its workers with the correct OpenMPI – a mismatched one (e.g. a newer system OpenMPI) causes parallel runs to abort with a BLAS-ERROR. The loader variables are also exported inside the run command so they survive macOS System Integrity Protection. (On macOS ORCA does not pass DYLD_* to its MPI sub-processes, so the OpenMPI libraries must additionally be on the default loader path, e.g. symlinked into /usr/local/lib; see the User Guide.)

• Bugfix: the DFT functionals are again selectable through the Model Chemistry step (each functional is offered as a method); this regressed when the functionals moved out of the method list. Keywords containing ‘/’ (e.g. REVDSD-PBEP86-D4/2021) appear with ‘_’ in the model-chemistry string, since ‘/’ is reserved there, and are translated back to the real keyword when the calculation runs.

2026.7.6 – All ORCA functionals, forces, database properties, and parallel execution

• Density functional theory now offers the complete set of ORCA functionals (117 of them), organized by type: pick a functional type (local, GGA, meta-GGA, hybrid, range-separated hybrid, or double-hybrid) and then the functional itself, including the double hybrids such as REVDSD-PBEP86-D4/2021.

• Gradients (forces) are produced with the correct ORCA method automatically: the analytic gradient where ORCA has one, or the numerical gradient where it does not (for example DLPNO-CCSD(T), and the non-self-consistent wB97M(2) and wB97X-2 functionals). A note is printed when the slower numerical gradient is used.

• Results can now be saved to the property database, including the gradient, the dipole-moment vector, the Mulliken, Löwdin, and Hirshfeld charges, the Mayer valences, and the rotational constants, in addition to the energies and other scalar results.

• The curated basis-set list has been filled out across the Pople, Dunning (correlation-consistent), and Karlsruhe def2 families, including their diffuse and minimally-augmented variants.

• Choosing ‘Basis Set Exchange’ as the basis-set source now opens the picker directly (it remains available from the ‘…’ button as well).

• ORCA runs in parallel by default, using the cores the machine or batch job provides. The number of cores and the memory per process can be set in the [orca-step] section of orca.ini (the ncores and memory options); parallel runs need ORCA’s OpenMPI runtime, whose location can be given with library-path.

• Documentation: a full User Guide covering methods and functionals, basis sets, forces, saving results, and parallel execution.

2026.6.28.1 – A Basis Set Exchange basis-set picker

• The basis set now uses the shared Basis Set Exchange picker: type a name, pick a common one from the list, or press ‘…’ to browse any basis from the Exchange, filtered to the elements you select on a periodic table. A choice from the Exchange is stored as ‘bse:NAME’, and the element selection is remembered so the picker is restored when the flowchart is reopened.

• Bugfix: the basis-set source control no longer appears on its own when the model chemistry is used; it is shown only with an explicit method and basis.

2026.6.28 – Properties, gradients, citations, and wavefunction export

• Reports many properties from a single calculation: HOMO/LUMO (and the next orbitals) and the gap, the dipole moment, rotational constants, <S^2>, the Mulliken, Löwdin, and Hirshfeld atomic charges, the Mayer bond orders and valences, and the optional dipole polarizability.

• The Mayer bond orders and Hirshfeld charges can be written to a CSV file and applied to the structure.

• Energy gradients can be requested and are written to Results.json for use by driver steps such as Thermochemistry and Reaction Path.

• Full citations for each run: the ORCA program, the DFT functional (from the ORCA manual), the basis set (via the Basis Set Exchange), and the supporting integral and exchange-correlation libraries.

• Basis sets can be taken from the Basis Set Exchange, including a ‘bse:NAME’ shorthand that forces a single basis from the Exchange.

• Can write an analytic wavefunction (.wfx, via orca_2aim) for a following Atomic Charges step to partition into DDEC6 charges.

• Fixed: the Results tab in the GUI was empty; it now lists the available results to save to variables, tables, or JSON.

2026.6.27 – Initial release of the ORCA step

• A sub-flowchart ORCA plug-in with Energy and Optimization sub-steps.

• Single-point energies and geometry optimizations, including DLPNO-CCSD(T).

• The method and basis set can be set explicitly, or taken from a preceding Model Chemistry step. Basis sets use ORCA’s built-in families.