Period 2

Period 2 fills 2s and then 2p, so it is eight elements long, and it is the row where the periodic table stops being a filing system and starts being physics. Two things happen here that happen nowhere else: the energies of the subshells are exposed directly by experiment, and the atoms are small enough to do a kind of bonding that the rest of the table largely cannot.

The row also behaves badly as a set of group members. Lithium, beryllium, boron, carbon, nitrogen, oxygen and fluorine each differ from the element below them more than that element differs from the one below it. The first row of the p-block is the awkward one throughout the table, and everything below follows a pattern that period 2 declines to join.

Two dips in a line that ought to rise smoothly

Plot first ionisation energy against atomic number across this row and the general direction is upward, as effective nuclear charge climbs and the atoms shrink. But the line is not smooth. It falls twice — from beryllium to boron, and from nitrogen to oxygen — and those two dips are the most direct experimental evidence for subshell structure that a single row provides.

Beryllium's outermost electron comes out of a filled 2s orbital. Boron's comes out of 2p, which lies higher in energy, so despite boron having one more proton its first electron is easier to remove. Without subshells there is no reason for that to happen at all.

The second dip exposes Hund's rule as a measurement rather than a diagram. Nitrogen spreads its three 2p electrons across three separate orbitals with parallel spins, which is as uncrowded as a p subshell can be. Oxygen has one more, and by then there is nowhere uncrowded left for it: it must double up in an orbital that is already occupied, and the two occupants push against each other. Taking that crowded electron away is therefore cheaper than taking one from nitrogen, and oxygen's value lands below its neighbour's despite the extra proton.

Period 3 repeats both dips at aluminum and sulfur, more faintly. Here they are unmistakable, which is why this row is the one every course uses to teach the point.

There is no 2d, and it is a hard ceiling

For n = 2 the available angular momentum values are 0 and 1, so there are s and p orbitals and nothing else. A period-2 atom has four valence orbitals, which means a maximum of eight valence electrons, and it cannot be talked out of it.

Nitrogen therefore forms at most four bonds, as in the ammonium ion, and there is no such thing as NCl₅. Phosphorus, one row down, gives PCl₅ and the PF₆⁻ ion without difficulty. Oxygen never exceeds two ordinary covalent bonds; sulfur gives SF₆. Fluorine forms exactly one; chlorine, bromine and iodine form compounds with three, five and seven bonds to oxygen or fluorine.

The traditional explanation is that period 3 elements "use their empty 3d orbitals". Modern calculations do not support that in any quantitative sense — 3d orbitals are far too high in energy to contribute meaningfully — and the better account is that the heavier atoms are simply large enough to accommodate more neighbours, with the extra bonding described by highly polar three-centre interactions rather than by an expanded octet. The observation is solid; the textbook mechanism for it is not, and it is worth knowing which is which.

Sideways overlap, and why life is not silicon-based

The second-row atoms are small. A carbon atom's covalent radius is about 77 picometres against silicon's 111, and that difference decides which kinds of bonds are worth forming.

A double or triple bond needs two p orbitals to overlap side by side. That overlap falls off quickly with distance, so it is efficient between small atoms and poor between large ones. Period 2 therefore does multiple bonding routinely — C=C, C≡C, C=O, C≡N, N≡N, O=O — and period 3 largely does not.

The consequences are visible in ordinary substances. Carbon dioxide is a gas of small O=C=O molecules; silicon dioxide, denied the double bond, builds a giant covalent network of Si–O single bonds and is quartz. Nitrogen is a gas held by a triple bond; phosphorus, unable to make one worth having, forms P₄ tetrahedra and is a solid. Oxygen is O₂; sulfur is a crown-shaped S₈ ring.

Carbon gets a second advantage on top of that. Its bond to itself is about as strong as its bonds to hydrogen and oxygen, so a long carbon chain has no weak link and no thermodynamic incentive to fall apart into something else. Silicon's bond to itself is much weaker than its bond to oxygen, so silicon chains are unstable with respect to silicates, which is where essentially all of the Earth's silicon has ended up. The recurring speculation about silicon-based life founders on that single comparison of bond energies, and the comparison is a period-2-versus-period-3 fact.

Three elements that stars take apart

Lithium, beryllium and boron are strikingly scarce — much rarer in the crust and in the cosmos than their positions in the table would suggest, and rarer than carbon, nitrogen and oxygen immediately to their right despite being lighter.

The reason is nuclear rather than chemical. These three nuclei are fragile enough to be destroyed by proton bombardment at temperatures well below those needed to make them, so stellar interiors consume them and produce essentially none. Almost all the lithium, beryllium and boron in the universe was made instead by cosmic-ray spallation: high-energy particles striking heavier nuclei in interstellar space and chipping them apart. Three consecutive elements in this row have a different cosmic origin from every element around them.

Their neighbours have the opposite story. Carbon, nitrogen and oxygen serve as catalysts in the CNO cycle, the dominant hydrogen-burning mechanism in stars more massive than the Sun: the three nuclei pass protons around a loop and emerge unchanged, having converted four hydrogen nuclei into a helium nucleus. Period 2 both catalyses the burning of period 1 and is partly destroyed by it.

Diagonal neighbours

Because period-2 elements resemble their group-mates so poorly, a different resemblance becomes useful. Moving one place right increases nuclear charge and shrinks the atom; moving one place down does roughly the reverse. Move both at once and the charge density comes out similar, which is why lithium behaves in several respects like magnesium, beryllium like aluminum, and boron like silicon. Boric and silicic acid are both weak, both of their oxides form glasses, and both elements build extended oxo-anion frameworks — a resemblance across a diagonal that neither shares with the element directly below it.

Neon closes the row at 2s²2p⁶, and in doing so it created a rule that has been over-applied ever since. The octet is a genuine period-2 result: eight electrons fill every available valence orbital for n = 2, and carbon, nitrogen, oxygen and fluorine obey it reliably. It fails at the start of the same row, where boron trifluoride sits happily with six valence electrons, and it fails below period 2 wherever an atom is large enough to take more neighbours. The rule that students are taught first is the rule that describes one row of the table exactly and everything else approximately.