When does Avogadro2 "draw" a bond

JGrantHill · June 17, 2026, 4:28pm

Potentially a silly question, but one that a colleague has asked me. When does Avogadro (when reading an XYZ file) draw a covalent bond between two atoms? I’d presume it was when the two atoms are within the sum of their covalent radii + some factor.

Is that true?

If so, what’s the additional factor/length on top of the sum of the covalent radii?

TIA!

ghutchis · June 17, 2026, 5:28pm

The default is the sum of the covalent radii + 0.45Å. In the UI, you can change this tolerance.

github.com/OpenChemistry/avogadrolibs

avogadro/core/molecule.h

b0cb2de6e


      
          Array<Vector3> vibrationLx(int mode) const;
          void setVibrationLx(const Array<Array<Vector3>>& lx);
          
          /**
           * Perceives bonds in the molecule based on the 3D coordinates of the atoms.
           *  atoms are considered bonded if within the sum of radii
           *  plus a small @p tolerance.
           * @param tolerance The calculation tolerance.
           * @param minDistance = atoms closer than the square of this are ignored
           */
          void perceiveBondsSimple(const double tolerance = 0.45,
                                   const double minDistance = 0.32);
          
          /**
           * Perceives bonds in the molecule based on preset residue data.
           *
           * Use this if you have residue data available (e.g., reading PDB or MMTF
           * files) Otherwise consider @sa perceiveBondsSimple and @sa
           * perceiveBondOrders
           */
          void perceiveBondsFromResidueData();

ghutchis · June 17, 2026, 6:08pm

BTW, If there’s feedback (e.g., specific use cases, etc.) I’m open to it. There are a few special cases (e.g., never create bonds to He, Ne, etc.).

JGrantHill · June 18, 2026, 8:20am

Thanks. This came from a somewhat philosophical discussion of “when does an intermolecular interaction become a covalent bond?”, so it’s good to be able to put a number on the Avogadro settings.

ghutchis · June 18, 2026, 2:18pm

To my mind, there are three regimes:

two atoms separated by approx. the sum of the covalent radii ⇒ clear bond
two atoms separated by a bit more than the sum of their VdW radii ⇒ non-bonded interaction
the fuzzy middle: less than the sum of the VdW radii but farther apart than the sum of the covalent radii

(Note, I’m ignoring “ionic radii” here because they’re stubbornly hard to pin down.)

JGrantHill · June 18, 2026, 4:13pm

We actually go on to this topic via the “fuzzy middle” part. An interesting way to express this is the penetration index proposed by Alvarez (I’m not a fan of the name).

Here’s a link to the paper, if you’ve not seen it. The borderless world of chemical bonding across the van der Waals crust and the valence region - Chemical Science (RSC Publishing)