LearnChem · Part II — Why do atoms stick together?

Why do molecules have shapes?

Nobody sculpts a molecule. Its electron pairs shove each other as far apart as they can get, and the shape is where the shoving stops.

SpineQ2 — why do atoms stick together at all?
Timeabout 40 minutes
Benchesthree
NeedsΔχ, the outer-floor headcount

Have you ever wondered…

“Why does soap taste horrible and sugar taste sweet — how does a tongue tell molecules apart?”

Where this came from

Last time, two atoms on a spring fell into a valley, and the valley priced the bond: how deep, how long. The Δχ dial added a third number: the lean. But that was two atoms and one bond. Water has two bonds; methane has four. A price list for each bond says nothing about which way the bonds point — and pointing is the difference between a molecule and a shopping list. Something must decide the angles. This module watches it decide, then hands the answer to your tongue.

§1

The four balloons

Start with methane, CH4 — the gas that CNG buses and autos run on, and the gas a gobar-gas plant brews — one carbon holding four hydrogens. Module 05's bookkeeping puts four electron pairs on carbon's outer floor, one pair per bond. Every schoolbook draws it the same way: a carbon in the middle of the page, four H's at the corners of a flat cross, ninety degrees apart.

Now think like the pairs. Each pair is a lump of negative charge, and module 01's one borrowed rule has not gone anywhere: like charges push apart, and pushing them together costs energy. Four mutually repelling lumps, all tethered to one atom, will shove each other apart until no shove gains anything more — the same settling into lowest energy that dug module 06's valley. The only question is what arrangement the shoving ends in.

Four electron pairs around one carbon, each repelling each of the others. Commit: when they settle as far apart as they can get, what angle sits between neighbouring pairs?

Commit before you look. First answers are counted anonymously, never named.

The notebook drawing is wrong, and it is wrong because paper is flat. Pairs live in space. Lift one pair off the page and the other three can back away from it — every pair gains room the flat cross never offered. Run the shoving to the end and four pairs settle pointing at the corners of a tetrahedron: a triangular pyramid with the carbon at its centre, every pair 109.5° from every other. Squash four pairs onto a flat page and the best they can manage is 90° between neighbours; give them space and every pair gains. (A flat three-way does reach 120° — but only for three. Four can never share a flat circle that generously.) No committee chose this. Watch the shoving find it:

Bench 1 · Let the pairs push apart and watch a shape appearThis is a simulation — charged lumps on a tether, repelling and settling. The angles are measured from wherever the lumps end up, never set in advance.

Scramble as often as you like: start the pairs anywhere, and the same angles come back — 180° for two pairs, 120° for three, 109.5° for four. The shape is not stored anywhere in the molecule. It is rediscovered, every time, by charge running downhill — which is why every methane molecule on Earth settles around the same shape, its average angle exactly the tetrahedral one. Chemists call this valence shell electron pair repulsion — VSEPR, in your textbook. You have just watched the whole of it happen.

Slow down — the four balloons

  1. Blow up four balloons and tie them together at the necks. Press them flat and they refuse; let go and they spring into exactly this pyramid, each balloon shoving the other three aside. Electron pairs are the balloons. The carbon is the knot.
  2. Where it breaks, twice — and both breaks teach. Balloons come all one size; the next section meets pairs that are not, and the angles move. And balloons splay because squashed rubber pushes back, while pairs repel through empty space, by charge alone. The picture is scaffolding; the charge is the law.

Quick check — why do the electron pairs push apart in the first place?

§2

Turn the molecule over

Methane's four pairs all end in a hydrogen. But nitrogen brings five outer electrons to ammonia, NH3: three pairs bond to hydrogens, and the leftover two electrons sit as a pair on the nitrogen itself — a lone pair, present at the shoving, invisible in the formula. Oxygen brings six to water: two bonding pairs, two lone pairs. Same four lumps of charge as methane, same rough tetrahedral splay — but now some corners hold no atom, and the visible shape is what remains: ammonia a low pyramid, water a bent line.

And the measurements catch the lone pairs at work. These are measured geometries, not drawings — turn each molecule and read its angle:

Bench 2 · Turn the molecule overMeasured shapes chemists have collected — each angle is the experimental value, cited below. Drag the molecule to turn it; the ghostly lobes are lone pairs, drawn as bookkeeping, not photographs.
Data: experimental geometries — NIST CCCBDB, Release 22 (May 2022), SRD 101, DOI 10.18434/T47C7Z; H2O angle and bond length from Hoy & Bunker, J. Mol. Spectrosc. 74, 1 (1979); CH4 r(CH) 1.087 Å and CO2 r(CO) 1.162 Å from the same database. NIST data is US-government work, public domain.

Read the three angles in a row: methane 109.47°, ammonia 106.67°, water 104.48°. Same floor, same four lumps — and the angle shrinks by a couple of degrees for each lone pair present. A lone pair answers to no second nucleus: it sits closer to its atom, spreads fatter, and shoves its neighbours harder than a bonding pair can — in balloon terms, a fatter balloon with nothing tied to its end, §1's first break now measured. The bonds give ground; the angle closes. The squeeze — nearly three degrees for the first lone pair, a little over two more for the second — is the lone pairs' fingerprint, and every ammonia and water molecule carries it.

Quick check — methane's four pairs settle at 109.5°, ammonia's bonds squeeze to 106.7°, water's to 104.5°. Same floor, same count of four pairs. What squeezes?

§3

Add up the pulls

“But the drawing on paper is flat”

Every structural formula you will ever write is a flat shadow of a three-dimensional thing — H–O–H on paper looks as straight as O=C=O, and only one of them is. The shadow is a fine tool for counting atoms and bonds; it is silent about angles, and angles are what this module is about. So carry both: draw flat, think in space. When module 23 hands you two molecules with identical flat formulas and opposite behaviours, the difference will live entirely in the part the paper cannot show.

Now bring module 06's dial back out. Each bond in a molecule has its own Δχ — its own lean, its own slightly negative end and slightly positive end. A bond's lean is an arrow: it points from the losing atom towards the winning one, and its size is the mismatch. The molecule's overall lean is what you get when you add the arrows — and adding arrows means directions matter.

So try it on the simplest case there is.

Each C=O bond in carbon dioxide is strongly polar — Δχ 0.89, oxygen's end negative. The molecule is a straight line, O=C=O. Does the molecule as a whole lean — a negative end and a positive end?

The straight shape is the whole verdict. Both arrows point from carbon to an oxygen — exactly opposite directions, exactly equal sizes. The sum is zero, not roughly zero, and a molecule whose arrows cancel has no negative end and no positive end for the world to take hold of, however polar each bond is on its own. Bend the same two bonds, though, and the cancellation fails: the arrows now share a downward component, and a resultant survives. Water is that bent case, and module 08 shows how much rides on it. Add the arrows yourself:

Bench 3 · Add up the pullsWorked out live: each arrow's direction comes from the measured shape, its size from the Δχ table, and the sum is added as vectors on the spot. The measured molecular lean is printed beside it — the verdict, lean or no lean, checked against experiment.
Arrow sizes: Δχ from the Pauling-scale table (modules 05–06). Shapes: NIST CCCBDB measured geometries (see Bench 2). Measured molecular dipoles, for the check column: H2O 1.857 D, NH3 1.476 D, HF 1.827 D, HCl 1.093 D, CO2 0.000 D — NIST CCCBDB, Release 22 (May 2022), DOI 10.18434/T47C7Z, retrieved 21 Sept 2026; CH4 0 by symmetry.

Run the whole list. Methane: four polar-ish bonds, perfect tetrahedral symmetry, sum zero — measured, zero. Carbon dioxide: two strongly polar bonds, straight line, sum zero — measured, zero. Water: two strongly polar bonds, bent, and a fat resultant survives — measured, 1.857 in the chemists' unit for lean, the debye: big for a molecule of three atoms. Chemists call a molecule with a surviving resultant polar, and the resultant itself its dipole (one warning for later: your physics textbook will draw this arrow from − to +, chemistry's habit is + to −; same lean, opposite pen strokes). The rule you have just built: polarity needs polar bonds and a shape that fails to cancel them. Either alone is not enough.

Quick check — H–F and H–Cl: fluorine out-pulls chlorine (χ 3.98 against 3.16). Both molecules are two-atom lines. Which molecule leans harder — is more polar?

§4

What the world does with shape

Two everyday machines read a molecule's two new numbers, one each: a microwave oven reads the lean, and your tongue reads the shape. Start with the oven.

A microwave oven works by grabbing molecules that have a charged end and twisting them back and forth. Dinner contains water (bent); the air above it contains carbon dioxide (straight). Which gets grabbed?

The oven's field flips several billion times a second, and each flip tugs every water molecule's charged ends around — twist, twist, twist, and the twisting is jostling, and the jostling is heat. Carbon dioxide, arrows cancelled, offers the field no handle at all. (The dry plate stays cooler for its own reason: its charged parts are locked into a rigid solid and cannot swing to follow the field — it warms mostly from the food on it.) Your dinner heats because water is bent. If its two bonds had settled straight, the microwave oven would be a humming cupboard.

The second machine is the one that asked this module's opening question. Your tongue and nose are tiled with receptors — protein sockets, each with a cavity of a particular shape and charge pattern. A molecule drifting past either fits a socket or does not; fitting closes a circuit, and the circuit's name is a taste. And a big molecule is this module's angles over and over — every carbon in sugar is methane's corner-splay again, every oxygen water's bend — so the whole molecule ends up with a shape as particular as a key's. Sugar's ring, with its studding of leaning O–H bonds, fits the socket wired to sweet. Soap's molecules — long, greasy-tailed, charged-headed, nothing like a sugar — miss that socket and mostly trip the bitter alarm, the tongue's standing warning for things best spat out (soap stings and irritates besides; not all of its foulness is a receptor's verdict). The tongue never weighs a molecule and never burns one. It reads shape and lean, the two things this module built — and it reads them in milliseconds, every mouthful, no instrument required.

Slow down — the socket and the plug

  1. A receptor is a socket; a molecule is a plug. Shape and charge pattern are the pin layout, and taste is the circuit closing. No fit, no signal — and no plug at all is no signal either: a steel spoon sheds no loose molecules, so it smells of nothing. (The “metal” smell on your fingers afterwards is your own skin's oils, freshly reacted.)
  2. Where it breaks: a wall socket is rigid, and reads pins alone. A receptor is a folded protein that flexes around its guest, reading charge and greasiness along with shape. The fit is loose, not exact — loose enough that saccharin, no sugar at all, sets off the sweet signal; module 28 asks how chilli and mint pull the same trick.

Quick check — your tongue tells sugar from soap in milliseconds, without burning, weighing or boiling either. What is it reading?

§5

Exercises

These check themselves, and “New numbers” deals a fresh set — there is nothing to memorise. Each stem says how exact to be.

Tier 1 · Quick numbers

1. repelling pairs around one centre, none of them lone pairs. What angle do they settle at, to the nearest degree?

degrees

2. Module 05's outer-floor headcount, revisited: . How many lone pairs sit on the central atom? (A whole number.)

lone pairs

3. The squeeze, read from Bench 2's data: . By how many degrees is the second angle squeezed below the first, to 0.1°?

degrees

4. Module 05's league table, revisited: the arrow entering Bench 3 for bond. Look both pulls up on the χ table (module 05's Bench 3, or module 06's dial) and give the arrow's size, Δχ, to two decimals.

Δχ

Tier 2 · One atom swapped, everything changes

Two molecules, three fluorines each: NF3 and BF3. Nitrogen brings five outer electrons; boron brings three. Work each one through: outer electrons on the centre, bonds, lone pairs left over, shape — and whether the arrows cancel. Count only what each atom actually brings; don't top boron up to four pairs. Then type the one that ends up flat and non-polar (name or formula).

Worked through. NF3: five outer electrons, three in bonds, leaves one lone pair — four lumps, one unseen: a low pyramid, like ammonia. Its three arrows share a downward lean and a resultant survives: polar — though only just: measured, NF3 leans a feeble 0.234 D against ammonia's 1.476, because a lone pair carries lean of its own, and nitrogen's here points against its three arrows. The arrows call the verdict; the lone pairs tune the size. BF3: three outer electrons, all three in bonds, no lone pair — three lumps splay flat at 120°, the Bench 1 answer for three. Three equal arrows at 120° add to exactly zero: BF3 is the flat, non-polar one. One atom swapped; lone-pair count changed; shape changed; polarity changed. That chain — headcount to lumps to shape to lean — is this module in one line. (Boron's six-electron outer floor, two short of eight, is real and allowed: eight was always bookkeeping, as module 06 said, not law.)

Tier 3 · The world with straight water

Suppose water's two bonds had settled straight, like carbon dioxide's, and its arrows cancelled. Leave the microwave aside — you have already done that one. Find two other things in your kitchen that would behave differently, and for each, say which arrow or which shape does the damage. There is no marking here; write your reasoning down before looking.

The discussion. The big casualty is dissolving. Straight water would keep its two polar O–H bonds, but it would lose its charged ends — and the ends are what grip an ion. Salt would dissolve far worse; the sugar in your chai, which clings to water by gentler handholds, would probably fare better (module 08 separates those two kinds of grip properly). Sweetness itself is the sneaky second casualty — not because sugar's own arrows change (they don't), but because sweetness needs sugar dissolved in saliva, and saliva is mostly water. A tongue that cannot dissolve sugar probably cannot taste it. Notice what survives untouched: sugar's shape, the receptor's socket, methane's pyramid — none of them depend on water's bend. One angle, 104.48° instead of 180°, and the kitchen quietly rearranges itself around it.

The sentence you keep

Electron pairs push each other as far apart as they can. That is shape.

Further play

PhET: Molecule ShapesBench 1 grown up — build your own molecule, add lone pairs one at a time, and watch the named shapes appear. (The theory's name, for the curious: VSEPR, after Gillespie and Nyholm.)PhET Interactive Simulations, University of Colorado Boulder PhET: Molecule PolarityBench 3 with knobs on — drag the electronegativities themselves and watch the arrows and the field-grab respond.PhET Interactive Simulations, University of Colorado Boulder PhET: Build a MoleculeAssemble molecules from atoms and flip each one into 3D — Bench 2's turning trick, for a whole kit of molecules.PhET Interactive Simulations, University of Colorado Boulder