From dots on paper to shapes in space
In the previous guide you learned to draw a Lewis structure — to spread the valence electrons of a molecule into bonds and lone pairs. But a Lewis structure is flat ink on a page, and a real molecule lives in three dimensions. Water is not a straight line of H-O-H; it is bent at about 104.5 degrees. Why? VSEPR — valence-shell electron-pair repulsion — answers this with one delightfully blunt idea: the groups of electrons around a central atom repel one another, so they arrange themselves to get as far apart as possible.
The key word is electron domain (also called an electron group). A domain is any region of electron density that wants its own elbow room: a single bond counts as one domain, a double or triple bond still counts as just one domain (the extra electrons travel between the same two atoms), and a lone pair counts as one domain too. So the recipe is simple: count the domains around the central atom, let them spread out evenly, and read off the shape.
The parent geometries, two through six
If every domain were a bond and no lone pairs intruded, the molecular geometry would simply be the most symmetric way to space the domains apart. These ideal arrangements are the parent geometries, and there are only five worth memorizing. Two domains spread to opposite sides: linear, 180 degrees, as in CO2. Three domains splay into a flat triangle: trigonal planar, 120 degrees, as in BF3. Four domains cannot stay flat — they pop into the third dimension as a tetrahedron, 109.5 degrees, as in CH4.
Five domains give the trigonal bipyramid, the one shape whose corners are not all equivalent: three sit in a flat equatorial belt 120 degrees apart, while two axial sites stick straight up and down. PCl5 is the classic example. Six domains give the beautifully symmetric octahedron — think of a central atom at the middle of a die with a bond poking through each of the six faces, all corners equivalent at 90 degrees, as in SF6. Everything else in VSEPR is one of these five parents with one or more corners quietly replaced by a lone pair.
domains parent geometry ideal angle 2 linear 180 3 trigonal planar 120 4 tetrahedral 109.5 5 trigonal bipyramidal 120 (eq) / 90 (ax) 6 octahedral 90
Lone pairs are pushier than bonds
Here is the subtlety that makes VSEPR feel alive. A bonding pair is held between two nuclei, stretched out and pinned in place. A lone pair belongs to the central atom alone, so it is fatter, closer in, and far more demanding of space. The ranking of repulsions is lone pair–lone pair > lone pair–bond > bond–bond. This is the heart of lone-pair repulsion: a lone pair does not just take up a corner, it elbows the bonding pairs closer together and squeezes the bond angles below their ideal value.
Watch it work on the four-domain family. Methane CH4 has four bonds and no lone pairs: a perfect tetrahedron at 109.5 degrees. Ammonia NH3 has three bonds and one lone pair: the lone pair pushes the three N-H bonds down to about 107 degrees, giving a trigonal-pyramidal shape — a tetrahedron with one corner missing an atom. Water H2O has two bonds and two lone pairs: the two lone pairs press hard, closing the H-O-H angle to about 104.5 degrees and giving the familiar bent shape. Same parent geometry, three different molecular shapes, all explained by how many corners are lone pairs.
The distorted shapes, named and explained
Once you accept that lone pairs occupy corners but hide from view, every named shape falls out by subtraction. From a tetrahedron: remove one corner's atom to get trigonal pyramidal (NH3), remove two to get bent (H2O). From a trigonal bipyramid: one equatorial lone pair gives the see-saw of SF4 (picture the axial pair as the plank and the two equatorial bonds as the legs), two equatorial lone pairs give the T-shape of ClF3, three give linear XeF2. From an octahedron: one lone pair gives a square pyramid (BrF5), and two lone pairs — which sit opposite each other to stay maximally apart — give the flat square-planar shape of XeF4.
Notice that the molecules that need five or six domains — SF4, ClF3, XeF2, BrF5, XeF4, SF6 — all have central atoms from period 3 and below carrying more than eight electrons. This is hypervalency, and it is worth being honest about. The old textbook story said sulfur or xenon expands its octet by using its empty d orbitals. Modern calculations show that explanation is largely wrong: the d orbitals are too high in energy to contribute much. The bonding is better described by highly polar bonds and three-center interactions. The good news for you: VSEPR does not care about the reason. It only counts domains, and it still nails the shape.
A worked recipe, and what shape buys you
Let us pin down a procedure you can run on any small molecule. Take the sulfite ion, SO3 with a 2- charge, as a test case. The trick that makes counting fast is to find the steric number of the central atom: the number of atoms bonded to it plus the number of lone pairs on it. That single number is the count of electron domains, and it picks the parent geometry straight off the table.
- Draw the Lewis structure and pick the central atom (usually the least electronegative, here sulfur). For SO3 2-, sulfur bonds to three oxygens and keeps one lone pair.
- Find the steric number: 3 bonded atoms + 1 lone pair = 4 domains. (Double bonds still count as one domain, so it does not matter how you draw the resonance.)
- Read the parent geometry off the table: 4 domains means tetrahedral electron-domain geometry.
- Subtract the lone pair to get the molecular shape: 3 atoms + 1 lone pair from a tetrahedron is trigonal pyramidal, just like NH3, with O-S-O angles a touch under 109.5 degrees.
Why bother predicting shape at all? Because shape decides whether a molecule is polar. If the individual bond dipoles point in directions that cancel, the molecule has no net dipole moment; if the shape is lopsided, they add up. Linear CO2 is nonpolar because its two C=O dipoles point exactly opposite and cancel — but bent H2O is strongly polar because its two O-H dipoles, splayed at 104.5 degrees, do not. Likewise tetrahedral CCl4 is nonpolar while pyramidal NH3 and see-saw SF4 are polar. Shape is also why XeF4, being flat and symmetric, has no dipole even though it is bristling with polar Xe-F bonds.
Where VSEPR ends and the next idea begins
VSEPR is astonishingly good for what it costs — a model with essentially one rule. But be honest about its limits. It is a model of repulsion between electron clouds, not a theory of bonding, so it explains shape but says nothing about bond strength or why a bond forms. It can stumble: some heavy-element molecules are bent or pyramidal where VSEPR predicts symmetry, and it does not handle the geometries of transition-metal complexes well — those are ruled by the d electrons and crystal field effects you will meet in a later rung.
There is also a frequent misreading worth heading off. VSEPR is sometimes taught hand-in-hand with hybridization — sp for linear, sp2 for trigonal planar, sp3 for tetrahedral — as though the hybrid orbitals cause the shape. They do not. The geometry comes first, set by electron-pair repulsion; hybridization is a second, separate bookkeeping language that describes the same arrangement in terms of orbitals. Keep them distinct in your mind. The next guide takes up hybridization properly and shows how it stitches VSEPR shapes onto the valence-bond picture of bonding.