— shown
The model on this page places the bonds by letting them repel on a sphere until they stop moving — VSEPR as an algorithm, with nothing about the individual atoms in it. Against the measured angle, that turns out to be — of the — molecules here to within a degree. The interesting ones are the misses, and they miss in a pattern.
—
A bond is not either covalent or ionic; it is somewhere on a slope. The difference in electronegativity between the two atoms says where. Below about 0.4 the pair is shared evenly enough to call the bond nonpolar; past about 1.7 one atom has effectively taken the electron and you have an ion. Everything in between — which is most of chemistry, and nearly every bond on this page — is polar covalent: shared, but not fairly.
—
The periodic table on this site says what the elements are. Almost nothing you have ever touched is an element — it is a molecule, and what decides how it behaves is not which atoms are in it but the shape they take.
Water and carbon dioxide are both a small atom with two oxygens on it. One is bent and is the most important solvent on the planet; the other is straight and is a gas you breathe out. The whole difference is 104.5 degrees against 180. A bent molecule with polar bonds has a net pull to one side; a straight one has the two pulls cancelling exactly.
Not one coordinate on this page was typed. Each molecule is stored as a list of which atom is bonded to which, and the geometry is relaxed out of it: every bond and every lone pair is put on a sphere around its atom and allowed to push the others away until nothing moves. That is VSEPR — “electron pairs get as far apart as they can” — stated as an algorithm instead of a table. Four things around a carbon settle at 109.47°; three around a boron settle at 120°; six around a sulfur settle at 90°. None of those numbers appears in the code.
Which makes the failures worth as much as the successes. A lone pair is held by one nucleus rather than shared between two, so it spreads wider and pushes harder. Give it a 20% larger push and ammonia comes out at 107.2° against a real 107.8, and water at 105.3° against a real 104.5. No single value fits both: tuned to ammonia alone it predicts water almost two degrees out, and tuned to water it misses ammonia. The story is the right direction and an approximation, not a law — and the panel tells you the size of the miss for every molecule that has one.
Molar mass and every bond polarity are computed from periodic-data.js — the same generated file the periodic table is drawn from, not a copy. Glucose comes out at 180.16 g/mol because the six carbons, twelve hydrogens and six oxygens were added up from that dataset, and the twelve hydrogens themselves were placed by filling unsatisfied valences rather than typed. If an atomic weight is ever corrected, both tools move together.
The bond types are read the same way: the difference in electronegativity, straight from the dataset, against the usual cutoffs. Sodium and chlorine differ by 2.23 and the bond is ionic; two carbons differ by nothing and it is not polar at all. And the cutoff itself is a rule of thumb that fails here in public: hydrogen fluoride differs by 1.78, which the 1.7 rule files as ionic — and HF is the textbook polar covalent molecule. It is 0.08 over a line somebody drew. The page reports what the rule says and leaves the disagreement visible.
Both are a small atom with two oxygens on it; the difference is 104.5 degrees against 180. Water’s oxygen carries two lone pairs that push the bonds together, and a bent molecule with polar bonds has a net pull to one side. Carbon dioxide’s two pulls cancel exactly — which is the whole difference between the planet’s solvent and a gas you breathe out.
Not one coordinate. Each molecule is stored as a list of which atom is bonded to which, and the geometry is relaxed out of it: every bond and lone pair is put on a sphere around its atom and allowed to push the others away until nothing moves. That is VSEPR stated as an algorithm instead of a table, and its misses are shown as honestly as its hits.
Because they are not molecules. Salt and diamond are lattices that repeat until the crystal ends; NaCl is a ratio, not a count, so there is no molecule to weigh.
Because the 1.7 rule says so, and the rule is wrong there. HF differs by 1.78 in electronegativity, which the cutoff files as ionic — and HF is the textbook polar covalent bond. The rule of thumb fails in public, and the page lets it.
CO₂ and SO₂ look almost the same written down. One is straight and one is bent at 119°, and that is why one is a greenhouse gas you exhale and the other dissolves into acid rain.
Benzene and ozone both have to be drawn with alternating single and double bonds, because a line has to be one or the other. In reality every bond in benzene is the same length. The electrons are spread out; the picture cannot be.
Salt and diamond have no molecule to point at: they are lattices that repeat until the crystal ends. NaCl is a ratio, not a count, which is why the panel refuses to give them a molar mass.
A real ethanol molecule is a population of shapes rotating about its bonds millions of times a second. What you can turn here is one of them, frozen. For small rigid molecules that is honest; for anything with a chain it is a portrait, not a photograph.
This page has changed in 160 archived releases, but each of those was a site-wide pass, so none is attributable to this tool on its own and none is listed here. That is not a claim that the tool never changed — a release that reworked many pages at once may well have altered this one too. The changelog has them.