Lanthanides

The lanthanides are the answer to a question nobody set out to ask: what happens when you add electrons to an atom without changing its chemistry? From lanthanum to lutetium the nuclear charge climbs by fourteen, and all fourteen of the electrons that balance it are delivered into 4f orbitals. Those orbitals sit underneath the shells that form the atom's outer surface, so the electrons arriving are never the electrons that meet anything. Fourteen of them join, and from the outside almost nothing happens.

The consequence is that a lanthanide's valence behaviour is set by its 5d and 6s electrons, which are the same for all fifteen. Every one of them forms a stable +3 ion. Their ionic radii differ by less than twenty per cent across the whole series. Their salts crystallise in the same structures, their hydroxides precipitate at almost the same pH, and their solubilities in almost any solvent differ by a few per cent rather than a few orders of magnitude. No other family on the table is defined principally by a lack of variation, and everything interesting about them — the history, the industry, the geopolitics — descends from that.

A hundred years of pulling one element out of another

The historical record for these elements is a sequence of discoveries that turned out to be mixtures. It starts in 1794, when Johan Gadolin isolated a new "earth" from a mineral found at Ytterby, a quarry outside Stockholm that would eventually give its name to four elements. That earth, yttria, was not one substance. Neither was ceria, found separately a few years later.

Carl Gustaf Mosander spent the years around 1839 to 1843 splitting both of them, extracting lanthanum and didymium from ceria and erbium and terbium from yttria. Didymium — the name means "twin" — was itself split by Carl Auer von Welsbach in 1885 into praseodymium and neodymium. Erbium was split again. Names were transferred between substances when later chemists decided the original assignment had been backwards, which is why erbium and terbium refer to something close to the reverse of what Mosander meant by them.

The pattern only stopped because the periodic table put a limit on it. Henry Moseley's 1913 measurement of characteristic X-ray frequencies gave, for the first time, a way to count the protons in an element directly, and it settled how many lanthanides could exist. It also showed that exactly one was still missing, at atomic number 61. Several groups claimed it — illinium, florentium — and all the claims failed to replicate, for a reason that was not chemical: element 61 has no stable isotope. It was finally isolated in 1945 from the fission products of a reactor at Oak Ridge by Jacob Marinsky, Lawrence Glendenin and Charles Coryell, and named promethium.

The separations that defeated the nineteenth century were only solved during the Manhattan Project, when Frank Spedding's group at Ames adapted ion-exchange chromatography to the problem — passing the mixture down a resin column with a complexing agent that binds each ion very slightly differently, and multiplying that tiny difference over thousands of theoretical plates. Modern production uses solvent extraction instead, with hundreds of mixer-settler stages in series, for exactly the same reason: no single step separates these elements, so the answer is to run an almost useless step several hundred times.

The contraction, and the two elements that dodge it

Across the series the ionic radius of the +3 ion falls steadily, from about 103 picometres for lanthanum to about 86 for lutetium. That is the lanthanide contraction, and its cause is the shielding failure of the f orbitals: each added 4f electron screens the added proton imperfectly, so the outer shells experience a steadily larger pull and are drawn inward.

Within the series the contraction is what makes separation possible at all. The heavy lanthanides are smaller, bind chelating agents slightly more strongly, and elute in a predictable order — the whole industrial process is a lever built on a difference of one or two picometres per step.

Two members sit off that smooth curve, and where they sit off it is the informative part: the break shows up in the metallic radius and not in the ionic one. What sets a metal's radius is the number of electrons its atoms hand to the lattice; an ion's radius owes nothing to that. Europium and ytterbium surrender two where the other thirteen surrender three, each of them protecting a 4f arrangement worth protecting — half filled in the first case, completely filled in the second. What that reluctance costs and what it buys is europium's own story, and it is the best-documented case in the series. Cerium leans the opposite way and comes out anomalously small, having decided that an empty 4f shell is worth a fourth electron.

Those same two configurations are the only escape routes from the +3 uniformity. Cerium is stable as Ce⁴⁺ and europium as Eu²⁺, and — not coincidentally — cerium and europium were historically the two easiest lanthanides to obtain pure. Change an element's charge and you change its chemistry enormously; oxidise the cerium in a mixture and it drops out on its own. Samarium, thulium and ytterbium can be pushed to +2 under reducing conditions, but nowhere near as readily.

Sharp lines, not broad bands

Screening cuts both ways, and the second thing it does is optical. A lanthanide ion's f levels are so little disturbed by whatever crystal it has been put into that its emission wavelength is effectively a property of the ion alone.

Commercially that one sentence is worth more than everything else on this page. It means a lanthanide phosphor can be specified by which element it contains, transplanted into a different host, and still deliver the same colour at the same temperature stability — an engineering luxury almost nothing else offers. Europium is the famous case, red in a picture tube and again in a fluorescent lamp, with terbium supplying the green beside it. Neodymium is the one that carries the most hardware: a neodymium-doped crystal amplifies at 1064 nanometres with enough gain to be the workhorse solid-state laser, and it turns up in everything from a machine shop to a laser-fusion target chamber. Erbium's is the luckiest coincidence in the series — its transition falls at 1550 nanometres, which happens to be exactly where silica fibre is most transparent, so erbium-doped amplifiers boost an intercontinental signal in place without ever converting it back into electricity.

The magnetic consequences are the other half. Unpaired f electrons in a shielded shell give large, weakly-coupled magnetic moments. Samarium-cobalt magnets arrived in the 1960s, and in 1983 Masato Sagawa at Sumitomo and John Croat at General Motors independently announced neodymium-iron-boron, which remains the strongest permanent magnet made. Adding a little dysprosium raises its coercivity enough to survive the temperatures inside a motor, which is why dysprosium — an obscure element by any other measure — is a bottleneck in electric vehicle manufacturing.

Gadolinium exploits the same shielded shell in a third way. Gd³⁺ has seven unpaired electrons, the most of any ion in the periodic table, and that large moment shortens the relaxation time of nearby water protons very efficiently — which is what a magnetic resonance contrast agent does. The free ion is toxic, so it is administered wrapped in a chelate strong enough to keep it bound while it passes through the body.

Not every application needs the elements separated at all. Auer von Welsbach's incandescent gas mantle, which lit European cities before electricity did, worked on a mixed thoria–ceria coating; and the alloy still sold as a lighter flint is mischmetal, the lanthanides in whatever ratio the ore supplied, with iron added. Two of the industry's oldest products deliberately skip the hardest step.

Common, but not conveniently

Calling these elements rare earths is a double misnomer. They are not earths, and the more abundant of them are not rare: cerium is more common in the crust than copper, and even thulium, the scarcest non-radioactive member, outweighs gold and platinum in the crust by a wide margin.

The difficulty is concentration. Because they are chemically almost identical, geological processes that would sort one element from another do not sort these — they travel together and are deposited together, so a rich single-element deposit essentially never forms. Bastnäsite and monazite are the economically important minerals, and both contain a suite of lanthanides in fixed proportions determined by geology rather than demand. A producer wanting neodymium gets lanthanum and cerium whether it wants them or not, which has periodically crashed the market for the light lanthanides while the heavy ones stayed scarce.

Monazite carries a further complication: it concentrates thorium alongside the lanthanides, so processing it generates a radioactive residue, and the regulatory cost of that residue is a significant part of why extraction migrated away from the United States and Europe from the 1990s onwards. China now accounts for the majority of mined output and a still larger share of the separation capacity, and separation — not mining — is the step that is genuinely hard to replicate.

Where the series is misunderstood

The most persistent error is treating "lanthanide" and "rare earth" as synonyms. The standard definition of the rare earths adds scandium and yttrium, which are not f-block elements at all but behave like heavy lanthanides because yttrium's ion happens to be almost exactly the size of holmium's — so it travels with them through every geological and industrial process.

Almost as common is the assumption that these elements are exotic laboratory curiosities. Two of the lightest are bulk industrial chemicals: cerium compounds polish optical glass and sit inside catalytic converters, and lanthanum works in the cracking catalysts that refineries consume by the tonne.

Finally, the row's boundaries are genuinely disputed. Whether lanthanum or lutetium belongs in group 3 beneath yttrium has been argued for decades, and a 2021 IUPAC report came down in favour of lutetium. Tables printed since then still disagree with one another, and both layouts describe the same fifteen elements.