Transition Metals

The transition metals are the widest family in the table and the only one whose membership depends on which definition you accept. IUPAC's wording is precise: a transition element is one whose atom has an incomplete d subshell, or which can give rise to a cation with an incomplete d subshell. That second clause is doing real work. It admits copper, whose atoms have a full 3d¹⁰ but whose Cu²⁺ ion is 3d⁹, and it excludes zinc, cadmium and mercury, which are d¹⁰ as atoms and d¹⁰ in every ordinary compound they form.

Most periodic tables, including this one, colour zinc, cadmium and mercury with the d-block anyway, because they occupy d-block columns and because separating them out confuses more readers than it informs. Nothing chemical is being asserted by that choice; it is a decision about what a table is for. The honest summary is that group 12 sits inside the block but outside the definition, and mercury has been dragged partway back in since 2007, when mercury(IV) fluoride was made at low temperature — briefly, and under conditions nobody would call ordinary.

Everything follows from a shell that is half-open

An s or p valence shell gives an element one obvious thing to do. A partly filled d subshell gives it several, all of comparable energy, and four familiar consequences follow.

Variable oxidation state. Because the 3d and 4s levels are close, an atom can surrender different numbers of electrons at similar cost. Manganese runs from +2 to +7 in stable, isolable compounds. Vanadium's four common states each have a different colour in solution, which is why vanadium chemistry is a standard teaching demonstration of the concept rather than of vanadium.

Colour. In a complex, the five d orbitals no longer have the same energy: the ligands' electron density raises some more than others, splitting them by an amount that happens to correspond to visible-light photons. An electron promoted across that gap absorbs one wavelength, and the compound shows the complement. Change the ligand and you change the gap and therefore the colour, which is why the same metal ion is one colour in water and another in ammonia. The two ends of the block — Sc³⁺ with no d electrons, Zn²⁺ with ten — are colourless, and that is the strongest possible evidence for the mechanism.

Catalysis. A metal that can hold two oxidation states within reach can accept electrons from one reactant and pass them to another, and a metal with vacant d orbitals can bind a small molecule and weaken its bonds. That combination is why iron fixes nitrogen industrially, why palladium and nickel make carbon–carbon bonds, why vanadium(V) oxide oxidises sulfur dioxide, and why the three-way catalytic converter is a platinum-group problem rather than a main-group one.

Magnetism. Unpaired d electrons give paramagnetism generally, and in iron, cobalt and nickel the atomic moments align across whole domains to give ferromagnetism. Only a handful of elements do this at room temperature and three of them are consecutive in this block.

Underneath all four sits the ability to form complexes at all. Vacant d orbitals of the right size and energy let a small cation gather six ligands around it in a fixed geometry, and working out that this was what was happening took Alfred Werner most of the 1890s against furious opposition — the prevailing chain theory had no room for a metal with a coordination number distinct from its valency. Werner's proof rested on counting isomers: his structure predicted exactly two forms of a particular cobalt complex where the rival predicted three, and the missing third was never found. He took the 1913 Nobel Prize in Chemistry for it, the first awarded to an inorganic chemist, and the following year resolved a purely inorganic complex into optical isomers, which removed the last argument that chirality required carbon.

The trend runs across, not down, and it is remarkably flat

Descending a main-group column changes things dramatically. Crossing a transition series barely changes them at all, and the flatness is the interesting part.

Each step to the right adds a proton and a d electron. A d electron is poor at shielding the nucleus from the outer s electrons — its orbital shape puts it partly inside them — but it does shield reasonably well within its own subshell. The net effect is that effective nuclear charge creeps up slowly, so atomic radii from titanium to copper vary by only a few picometres, densities climb steadily, and melting points stay high across the whole middle of each row. Metals that are similar in size and electronic character alloy readily with one another, which is why steel accepts chromium, nickel, vanadium, molybdenum and manganese in almost any combination and still behaves like steel.

Two configurations refuse to follow the filling order. Chromium is 3d⁵4s¹ rather than 3d⁴4s², and copper is 3d¹⁰4s¹ rather than 3d⁹4s². The usual explanation invokes the stability of a half-filled or filled subshell; the more careful one notes that 3d and 4s are so close in energy that the electron–electron repulsion saved by keeping the two 4s electrons apart is enough to tip the balance. Either way, these are not typographical errors in the table, and the same pattern recurs more often and more messily in the 4d row.

Hafnium hides inside zirconium

The single most consequential quirk of this family sits between the second and third d-series. Fourteen f-block elements intervene between barium and hafnium, and 4f electrons shield the nuclear charge badly, so by the time the table resumes at hafnium the atoms have been squeezed. Hafnium ends up almost exactly the same size as zirconium directly above it, instead of being noticeably larger as every other pair in the block is.

Two things follow. First, the third-row transition metals are far denser and higher-melting than their second-row counterparts, because roughly twice the mass has been packed into the same volume — this is where osmium, iridium, tungsten, rhenium and platinum come from. Second, no ordinary chemical test separates zirconium from hafnium, and hafnium was consequently not discovered until 1923, having sat inside zirconium ores unnoticed for a century.

Separating them is expensive and unavoidable, because their nuclear properties are opposite: zirconium is nearly transparent to thermal neutrons and hafnium absorbs them greedily. Reactor fuel cladding must be hafnium-free zirconium, and reactor control rods can usefully be hafnium. One of the least chemically distinguishable pairs in the table has to be told apart to a few parts per million on grounds that are not chemical at all.

Iron dominates and everything else is a rounding error

World steel production runs around 1.9 billion tonnes a year. Every other transition metal put together is a small fraction of that, and the gap is so large that it distorts how the family is usually described: articles about "the transition metals" tend to be articles about iron with appendices.

The extraction routes divide cleanly by how stable the oxide is. Iron oxide is reducible with carbon at a temperature a blast furnace can reach, which is why the Iron Age happened and why steel is cheap. Titanium's oxide is not — carbon reduction gives a carbide instead — so titanium waited until the Kroll process converted the oxide to a chloride and reduced that with magnesium, a batch process that remains slow and costly and is the reason titanium prices bear no relation to titanium's abundance. Copper, silver and gold sit at the other extreme, stable enough as metals to have been found native, which is why they were worked first.

The platinum group metals share a supply peculiarity: they are concentrated in very few deposits, principally the Bushveld Complex in South Africa and Norilsk in Russia, and are usually recovered together, so demand for one drags the others along whether or not anyone wants them. Technetium is the family's true outlier, having no stable isotope at all and being obtained from spent reactor fuel or from a molybdenum-99 generator rather than from any ore.

The 6d row belongs here, but its story is nuclear

Rutherfordium, dubnium, seaborgium, bohrium and hassium sit in the transition-metal columns as the beginning of a 6d series, and they are counted in this family's membership on that basis. What is actually known about them, though, is not d-block chemistry in any recognisable sense: they exist a few atoms at a time, for seconds or less, and their properties are inferred from a handful of volatility and adsorption experiments conducted one atom at a time. That work — how the atoms are made, how a single-atom experiment can be meaningful, and the long argument about who made them first — is treated at Superheavy elements, which is the right place for it. What follows below applies to the elements that can be held in a hand.

What people get wrong about the d-block

• Assuming all transition metal compounds are coloured. Scandium and zinc compounds are not, because their d subshells are empty and full respectively. That is a feature of the explanation, not an exception to it.
• Reading "transition" as "in the middle of the table". Aluminum, tin and lead are middle-ish metals and are not in the block at all; their valence electrons are s and p.
• Filling 3d before 4s and then removing 3d first. The orbitals swap order once the d subshell is occupied. Fe²⁺ is 3d⁶, not 3d⁴4s², and this catches people out more reliably than any other single question about electron configuration.
• Treating the block as chemically uniform. Scandium behaves like a large, hard, colourless main group cation; gold refuses to react with almost anything and forms a stable −1 anion in cesium auride. They are in the same family only in the loosest sense.