Skip to content
LumenaryAcademy
Inorganic 70 min

Coordination chemistry

Metal ions, the molecules that crowd round them, and why the result is so often coloured, magnetic or alive.

Code tested · 5AI-draftedAwaiting review by a chemist

Your guides: Chloe · The Chemist, Margaret · The Analyst, Gideon · The Sceptic

MotivationH

One ion, two gemstones

A ruby is red and an emerald is green, yet the colour of both comes from the same thing: a small fraction of chromium(III) ions, Cr³⁺, sitting in a crystal that would otherwise be colourless. In ruby the chromium replaces some aluminium in aluminium oxide; in emerald it replaces some aluminium in beryl, a beryllium aluminium silicate. In both crystals each chromium ion is surrounded by six oxygen atoms. The difference in colour comes down to how hard those six neighbours push on the chromium's electrons.

The same idea, a metal ion and the atoms that surround it, explains why blood is red and turns brighter when it picks up oxygen, why copper sulfate solution is blue, why some iron compounds are pulled into a magnetic field and others are not, and why one arrangement of platinum, chlorine and ammonia is at the heart of a treatment that cures most testicular cancers while another arrangement of the same atoms is far less active. A metal ion with molecules or ions bound round it is a coordination complex, and the chemistry of these complexes is coordination chemistry.

Unlike most bonds in Chemical bonding, these bonds take both their electrons from the molecule attached to the metal, and the ideas of Molecular orbitals and Spectroscopy explain what they do to the metal's d electrons.

DefinitionsH

Complexes, ligands and coordination number

Definition— coordination complex

A coordination complex is a central metal atom or ion bonded to a set of surrounding molecules or ions called ligands. Each ligand bonds through one or more donor atoms, which supply an electron pair to the metal; the metal accepts it. In G. N. Lewis's language of acids and bases, the ligand is a Lewis base and the metal a Lewis acid. The bond is called a coordinate (or dative) bond.

Definition— coordination number

The coordination number is the number of donor atoms bonded to the metal. Six (octahedral) is the most common, then four (tetrahedral or square planar); numbers from two to nine occur, the largest mainly for big ions such as the lanthanoids.

Definition— denticity and chelation

A ligand that bonds through one donor atom is monodentate (water, ammonia, chloride, cyanide). One that bonds through two or more is polydentate: ethane-1,2-diamine (abbreviated en, two nitrogen donors) is bidentate, and the ethylenediaminetetraacetate ion (EDTA⁴⁻, two nitrogens and four oxygens) is hexadentate. A polydentate ligand wraps the metal in a ring, a chelate (from the Greek for a crab's claw).

The formula of a complex is written in square brackets, with its overall charge outside: [Co(NH₃)₆]³⁺, [Fe(CN)₆]⁴⁻, [CuCl₄]²⁻. A salt of a complex ion keeps its counter-ions outside the brackets: [Co(NH₃)₆]Cl₃ contains the cation [Co(NH₃)₆]³⁺ and three chloride ions. To find the metal's oxidation state, subtract the ligands' charges from the overall charge. In [Fe(CN)₆]⁴⁻, six cyanide ions contribute −6, so iron is +2, written iron(II). Transition-metal ions lose their 4s electrons first (see Atomic structure and the periodic table), so Fe²⁺, from [Ar] 3d⁶ 4s², is a d⁶ ion. The d-electron count is the number the rest of this lesson runs on.

IUPAC's names (from the Red Book) list the ligands alphabetically, with prefixes for how many, then the metal with its oxidation state in Roman numerals; an anionic complex ends in -ate. So [Co(NH₃)₆]Cl₃ is hexaamminecobalt(III) chloride (ammine, with two m's, is ammonia as a ligand), [CoCl(NH₃)₅]Cl₂ is pentaamminechloridocobalt(III) chloride, and K₄[Fe(CN)₆] is potassium hexacyanidoferrate(II).

Note

The chelate effect. Chelating ligands bind far more tightly than the same donor atoms supplied by separate molecules. Nickel(II) with six ammonia molecules has an overall formation constant of about 10910^{9}; with three en molecules, which provide the same six nitrogen donors, it is about 101810^{18}. The bond strengths are similar; the difference is mostly entropy, since replacing six ammonia molecules by three en frees three more molecules into solution. This is why EDTA is used to measure metal ions in water (one of the titrations in Analytical chemistry) and, in hospitals, to remove lead from the body.

History & global context

Werner's puzzle: counting isomers to see shapes

In the nineteenth century chemists knew a family of cobalt compounds with the formulas CoCl₃·6NH₃ (yellow), CoCl₃·5NH₃ (purple) and CoCl₃·4NH₃, which came in two forms, one green and one violet. How could cobalt, with a valence of three, hold six ammonia molecules? The leading answer, championed by the Danish chemist Sophus Mads Jørgensen, strung the ammonias together in chains, like carbon chains.

In 1893 Alfred Werner, aged 26 and working in Zürich, proposed something simpler: the cobalt ion holds six groups directly, at the corners of an octahedron, and any chloride not among them is a free ion, so the compounds are [Co(NH₃)₆]Cl₃, [CoCl(NH₃)₅]Cl₂ and [CoCl₂(NH₃)₄]Cl. Silver nitrate, which precipitates only free chloride, gave three, two and one mole of silver chloride per mole of cobalt, as predicted, and the conductivities of the solutions, measured with Arturo Miolati, matched salts that break into four, three and two ions.

The geometry came from counting isomers. If six ligands sit at the corners of an octahedron, a complex with four of one kind and two of another, MA₄B₂, has exactly two arrangements: the two B ligands next to each other (cis) or opposite (trans). A flat hexagon or a trigonal prism would give three. Werner found only two forms of [CoCl₂(NH₃)₄]⁺, the green and the violet. The program below counts by brute force: it lists every placement of the ligands on the six sites and treats two placements as one isomer when a rotation of the shape turns one into the other.

python
import itertools
import numpy as np

def rotation(axis, degrees):
    """Rotation matrix about an axis (Rodrigues' formula)."""
    k = np.array(axis, float) / np.linalg.norm(axis)
    K = np.array([[0, -k[2], k[1]], [k[2], 0, -k[0]], [-k[1], k[0], 0]])
    t = np.radians(degrees)
    return np.eye(3) + np.sin(t) * K + (1 - np.cos(t)) * K @ K

def symmetry_group(sites, generators):
    """Every permutation of the sites produced by combining the generating operations."""
    def as_permutation(R):
        return tuple(int(np.argmin([np.linalg.norm(R @ v - w) for w in sites])) for v in sites)
    gens = [as_permutation(R) for R in generators]
    identity = tuple(range(len(sites)))
    group, frontier = {identity}, [identity]
    while frontier:
        products = [tuple(g[i] for i in h) for g in frontier for h in gens]
        frontier = [p for p in set(products) if p not in group]
        group.update(frontier)
    return group

def count_isomers(ligands, group):
    """Placements of the ligands that no symmetry operation turns into one another."""
    classes = set()
    for placement in set(itertools.permutations(ligands)):
        classes.add(min(tuple(placement[i] for i in g) for g in group))
    return len(classes)

ring = [np.array([np.cos(a), np.sin(a), 0.0]) for a in np.radians(range(0, 360, 60))]
prism = [np.array([np.cos(a), np.sin(a), z]) for z in (1.0, -1.0) for a in np.radians([0, 120, 240])]
octahedron = list(np.vstack([np.eye(3), -np.eye(3)]))
mirror = np.diag([1.0, 1.0, -1.0])       # reflection through the horizontal plane

shapes = {
    "planar hexagon": (ring, [rotation((0, 0, 1), 60), rotation((1, 0, 0), 180)]),
    "trigonal prism": (prism, [rotation((0, 0, 1), 120), rotation((1, 0, 0), 180)]),
    "octahedron":     (octahedron, [rotation((1, 0, 0), 90), rotation((0, 1, 0), 90)]),
}
print("isomers, counting a mirror-image pair once (and, in brackets, twice)")
print(f"{'shape':16} {'MA4B2':>8} {'MA3B3':>8} {'MA2B2C2':>9}")
for name, (sites, rotations) in shapes.items():
    proper = symmetry_group(sites, rotations)               # rotations only
    full = symmetry_group(sites, rotations + [mirror])      # rotations and reflections
    cells = [f"{count_isomers(f, full)} ({count_isomers(f, proper)})"
             for f in ("AAAABB", "AAABBB", "AABBCC")]
    print(f"{name:16} {cells[0]:>8} {cells[1]:>8} {cells[2]:>9}")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

The octahedron gives two isomers of MA₄B₂ and two of MA₃B₃, where the hexagon and the prism give three of each. The bracketed numbers count a pair of mirror images twice: for MA₂B₂C₂ the octahedron allows five arrangements, one of which exists as a non-superimposable mirror-image pair, so six stereoisomers in all.

Werner received the 1913 Nobel Prize in Chemistry (Nobel Prize Outreach) for this work.

ExplanationH

Isomerism in complexes

Complexes show every kind of isomerism you met in Stereochemistry, plus a few of their own.

Structural isomers have different bonds.

  • Ionisation isomers swap a ligand with a counter-ion: [CoBr(NH₃)₅]SO₄ (violet) releases sulfate ions in solution, while [Co(SO₄)(NH₃)₅]Br (red) releases bromide ions.
  • Hydrate isomers do the same with water. Solid "chromium(III) chloride hexahydrate" exists as violet [Cr(H₂O)₆]Cl₃ and as green forms with one or two chlorides bound to the metal.
  • Linkage isomers use a different donor atom of the same ligand. The nitrite ion can bind through nitrogen (nitrito-κN, historically "nitro") or through oxygen (nitrito-κO); thiocyanate can bind through sulfur or nitrogen.

Stereoisomers have the same bonds arranged differently in space.

  • Geometric isomers: cis and trans for MA₄B₂ and square-planar MA₂B₂, and fac (three identical ligands on one face of the octahedron) and mer (three in a line round a meridian) for MA₃B₃.
  • Optical isomers: a complex with no mirror plane and no inversion centre is chiral. [Co(en)₃]³⁺ is the classic example: its three chelate rings make a propeller that turns either left (Λ) or right (Δ).
ClClNH₃NH₃NH₃NH₃CoClNH₃ClNH₃NH₃NH₃CoClNH₃NH₃ClPtClNH₃ClNH₃Pttrans (green)cis (violet)cisplatintransplatin
Geometric isomers. Left: trans- and cis-[CoCl₂(NH₃)₄]⁺, Werner's green and violet compounds; the shaded square is the octahedron's equatorial plane, drawn in perspective. Right: the square-planar platinum(II) complexes cisplatin and transplatin, which have the same formula, PtCl₂(NH₃)₂, and very different effects in the body.

Isomers can differ in colour, solubility, reactivity and, as the section on medicine shows, in whether a drug works.

ExplanationH

Crystal field theory: splitting the d orbitals

Why are so many complexes coloured, and some magnetic? Crystal field theory, begun by the physicist Hans Bethe in 1929 and applied to chemistry in the 1950s, treats each ligand as a point of negative charge (or the negative end of a dipole) and asks how those charges affect the metal's five d orbitals.

In a free ion the five d orbitals have the same energy. Put six negative charges on the ±x\pm x, ±y\pm y and ±z\pm z axes. Two of the d orbitals, dz2d_{z^2} and dx2−y2d_{x^2-y^2}, point their lobes straight at the ligands, so an electron in them is repelled strongly and its energy rises. The other three, dxyd_{xy}, dxzd_{xz} and dyzd_{yz}, point their lobes between the axes, so an electron there is repelled less. The five orbitals split into two sets, called ege_g (the upper pair) and t2gt_{2g} (the lower three), separated by the octahedral splitting Δo\Delta_o.

eg: dz², dx²−y²t2g: dxy, dxz, dyz+0.6 Δo−0.4 ΔoΔofive d orbitalsspherical fieldoctahedral field
Crystal field splitting in an octahedral complex. The five d orbitals of the ion in a spherical field (left) split into a lower set of three, t₂g, and an upper set of two, e_g (right). The splitting is Δo. The average energy (dashed) is unchanged, so t₂g lies 0.4 Δo below it and e_g lies 0.6 Δo above it.

The splitting keeps the average energy (the barycentre) fixed: two orbitals raised by 0.6 Δo0.6\,\Delta_o balance three lowered by 0.4 Δo0.4\,\Delta_o, since 2×0.6=3×0.42 \times 0.6 = 3 \times 0.4.

Definition— crystal field stabilisation energy

The crystal field stabilisation energy (CFSE) of an octahedral dnd^n complex with ntn_t electrons in t2gt_{2g} and nen_e in ege_g is

CFSE=(−0.4 nt+0.6 ne) Δo.\text{CFSE} = (-0.4\,n_t + 0.6\,n_e)\,\Delta_o .

Some books add the cost of any extra electron pairing; we keep that term, PP per pair, separate.

High spin and low spin

For d1d^1 to d3d^3 each electron simply enters its own t2gt_{2g} orbital. The fourth electron faces a choice. It can go up into ege_g, costing Δo\Delta_o, or pair with an electron already in t2gt_{2g}, costing the pairing energy PP: the extra electron–electron repulsion plus the lost exchange stabilisation of parallel spins.

  • If Δo<P\Delta_o \lt P (a weak field), electrons spread out over all five orbitals before pairing: a high-spin complex with many unpaired electrons.
  • If Δo>P\Delta_o \gt P (a strong field), electrons fill t2gt_{2g} completely before entering ege_g: a low-spin complex.

The choice exists only for d4d^4 to d7d^7. Unpaired electrons make a complex paramagnetic, attracted into a magnetic field; the spin-only formula μ=n(n+2) μB\mu = \sqrt{n(n+2)}\,\mu_B, with nn unpaired electrons and μB\mu_B the Bohr magneton, predicts the measured magnetic moment of most first-row complexes well; the exceptions, such as high-spin cobalt(II), are ions whose electrons' orbital motion adds to the moment.

↑↑↑↓↑↑↑↓↑↓↑↓small Δolarge Δohigh spin, 4 unpaired[Fe(H₂O)₆]²⁺low spin, 0 unpaired[Fe(CN)₆]⁴⁻
The two ways of filling the d orbitals of a d⁶ ion such as Fe²⁺. With a small splitting (weak-field ligands such as water) the six electrons spread out: high spin, four unpaired electrons, paramagnetic. With a large splitting (strong-field ligands such as cyanide) they pair up in t₂g: low spin, no unpaired electrons, diamagnetic.

Example— Iron(II) with water and with cyanide

Both [Fe(H₂O)₆]²⁺ and [Fe(CN)₆]⁴⁻ contain the d⁶ ion Fe²⁺, whose pairing energy is about 17 600 cm⁻¹ (1 cm⁻¹ per ion is 0.01196 kJ/mol). Water gives Δo≈10 400\Delta_o \approx 10\,400 cm⁻¹, less than PP, so the aqua ion is high spin: t2g4eg2t_{2g}^4 e_g^2, four unpaired electrons, CFSE =(−0.4×4+0.6×2)Δo=−0.4 Δo= (-0.4 \times 4 + 0.6 \times 2)\Delta_o = -0.4\,\Delta_o. Cyanide gives Δo≈33 000\Delta_o \approx 33\,000 cm⁻¹, far more than PP, so the cyanide complex is low spin: t2g6t_{2g}^6, no unpaired electrons, CFSE =−2.4 Δo= -2.4\,\Delta_o, paid for with two extra pairings. A strong magnet tests the prediction: iron(II) sulfate is attracted, potassium hexacyanidoferrate(II) weakly repelled.

The program below tabulates every dnd^n configuration, then decides the spin state of five complexes from the energy difference between the two arrangements, using approximate literature values of Δo\Delta_o and PP.

python
import numpy as np

CM_TO_KJ = 0.0119627   # 1 cm⁻¹ per ion corresponds to 0.0119627 kJ/mol (h·c·N_A)

def octahedral(n_d, low_spin):
    """Put n_d electrons into t2g (three orbitals at -0.4 Δo) and eg (two at +0.6 Δo)."""
    t2g, eg = [0] * 3, [0] * 2
    if low_spin:   # fill t2g completely before eg
        steps = [(t2g, i) for i in range(3)] * 2 + [(eg, i) for i in range(2)] * 2
    else:          # one electron in each of the five orbitals before any pairing
        steps = ([(t2g, i) for i in range(3)] + [(eg, i) for i in range(2)]) * 2
    for level, i in steps[:n_d]:
        level[i] += 1
    unpaired = (t2g + eg).count(1)
    pairs = (t2g + eg).count(2)
    cfse = round(-0.4 * sum(t2g) + 0.6 * sum(eg), 1) + 0.0   # in units of Δo
    return sum(t2g), sum(eg), unpaired, pairs, cfse

print("d^n   high spin            CFSE/Δo   low spin             CFSE/Δo")
for n in range(1, 11):
    cells = []
    for low in (False, True):
        t, e, u, p, c = octahedral(n, low)
        cells.append(f"t2g^{t} eg^{e} ({u} unp.)  {c:+5.1f}")
    print(f"d{n:<4} " + "    ".join(cells))

# Approximate literature values in cm⁻¹; published tables differ by up to about 10 %.
complexes = [("[Fe(H2O)6]2+", 6, 10400, 17600), ("[Fe(CN)6]4-", 6, 33000, 17600),
             ("[CoF6]3-", 6, 13000, 21000), ("[Co(NH3)6]3+", 6, 23000, 21000),
             ("[Mn(H2O)6]2+", 5, 7800, 25500)]
print(f"\n{'complex':13} {'Δo':>4} {'P':>5} {'E(low)-E(high)':>15}  spin  unpaired  spin-only μ")
print(f"{'':13} {'kJ/mol':>10} {'kJ/mol':>15} {'':>16} {'μB':>10}")
for name, n, delta, P in complexes:
    d_kj, p_kj = delta * CM_TO_KJ, P * CM_TO_KJ
    hs, ls = octahedral(n, False), octahedral(n, True)
    gap = (ls[4] - hs[4]) * d_kj + (ls[3] - hs[3]) * p_kj     # CFSE difference + extra pairing
    t, e, u, p, c = ls if gap < 0 else hs
    print(f"{name:13} {d_kj:4.0f} {p_kj:5.0f} {gap:+15.0f}  {'low ' if gap < 0 else 'high'} {u:6} {np.sqrt(u * (u + 2)):12.2f}")
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

Other shapes

In a tetrahedral complex the pattern inverts: the ee pair lies lowest and the t2t_2 three lie highest, and the splitting is smaller, Δt≈49Δo\Delta_t \approx \tfrac{4}{9}\Delta_o for the same metal, ligands and distance. So tetrahedral complexes are almost always high spin. A square-planar geometry is what an octahedron becomes when the two ligands on the zz axis are pulled right away: the dx2−y2d_{x^2-y^2} orbital, pointing at the four remaining ligands, rises far above the others. A d⁸ ion with a large splitting (Ni²⁺ with strong-field ligands, and almost always Pd²⁺, Pt²⁺ and Au³⁺) leaves that orbital empty and prefers to be square planar. Cisplatin is one.

A related effect is the Jahn–Teller distortion: when an ege_g level is unevenly occupied, as in Cu²⁺ (d⁹) or high-spin Cr²⁺ (d⁴), the octahedron distorts, usually lengthening the two bonds on one axis, which lowers the energy. Almost every copper(II) complex is distorted in this way.

ExplanationH

Colour and the spectrochemical series

A complex absorbs a photon when the photon's energy matches the gap between two of its energy levels. For the d-block metals one such gap is Δo\Delta_o itself, and Δo\Delta_o is typically 100 to 400 kJ/mol, which corresponds to photons of visible or near-infrared light. The colour you see is what remains of white light after the absorbed part is removed, which is roughly, but only roughly, the "complementary" colour of the absorbed light.

The simplest case is a d¹ ion. [Ti(H₂O)₆]³⁺ has a single d electron, and its one absorption band, a jump from t2gt_{2g} to ege_g, measures Δo\Delta_o directly. A d⁹ ion, with a single hole, behaves similarly. The program models each band as a smooth hump and prints the light that gets through.

python
import numpy as np

CM_TO_KJ = 0.0119627                  # kJ/mol per cm⁻¹

def band(wavenumber, centre, width, peak_absorbance=1.0):
    """A d-d absorption band, modelled as a Gaussian in wavenumber (cm⁻¹)."""
    return peak_absorbance * np.exp(-0.5 * ((wavenumber - centre) / width) ** 2)

# Approximate band maxima of two aqua ions; real bands are broad and lopsided.
ions = {"[Ti(H2O)6]3+": 20300, "[Cu(H2O)6]2+": 12500}
colours = [(450, "violet"), (495, "blue"), (570, "green"), (590, "yellow"), (620, "orange"), (700, "red")]

for name, centre in ions.items():
    print(f"{name}: band maximum {centre} cm⁻¹ = {1e7 / centre:.0f} nm, "
          f"so the splitting is about {centre * CM_TO_KJ:.0f} kJ/mol")
    print("   nm   colour   transmitted light (each # is 5 %)")
    for nm in range(410, 700, 20):
        T = 10 ** -band(1e7 / nm, centre, width=2500)       # Beer–Lambert: T = 10^(-A)
        colour = next(c for limit, c in colours if nm < limit)
        print(f"  {nm}  {colour:7}  {'#' * round(T * 20)}")
    print()
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

The titanium ion absorbs most strongly in the blue-green, near 493 nm, and lets through red and some violet, so it looks purple. The copper ion absorbs red and near-infrared light and looks pale blue; its smaller splitting, about 150 kJ/mol against 243, fits its lower charge. With more d electrons the spectrum has several bands that also depend on electron–electron repulsion; Tanabe and Sugano's diagrams of 1954 (paper) are the standard tool for assigning them.

The spectrochemical series

By measuring Δo\Delta_o for one metal with many ligands, the Japanese chemist Ryutaro Tsuchida found in the 1930s that ligands always fall in nearly the same order. Abbreviated, from weak field to strong field:

I⁻ < Br⁻ < S²⁻ < SCN⁻ (S-bound) < Cl⁻ < F⁻ < OH⁻ < C₂O₄²⁻ < H₂O < NCS⁻ (N-bound) < NH₃ < en < bipy < phen < NO₂⁻ (N-bound) < PPh₃ < CN⁻ < CO

The order does not follow charge: neutral ammonia beats fluoride, and the halide ions are at the weak end, so the point-charge model is too simple. Ligand field theory, the molecular-orbital version of the picture (see Molecular orbitals), explains it: the ege_g orbitals are antibonding combinations with the ligands' σ-donor orbitals, and the t2gt_{2g} set can interact with ligand orbitals of π symmetry.

  • π-donor ligands (halides, hydroxide) have filled p orbitals that push the t2gt_{2g} level up, shrinking Δo\Delta_o.
  • σ-only ligands (ammonia, en) sit in the middle.
  • π-acceptor ligands (CO, cyanide, phosphines, bipyridine) have empty π* orbitals that accept electron density from the metal's t2gt_{2g} orbitals, pulling that level down and enlarging Δo\Delta_o. This back-bonding is also why carbon monoxide binds so firmly to iron.

The metal matters too. For a given ligand, Δo\Delta_o rises with the oxidation state (Co³⁺ > Co²⁺) and rises by roughly 25 to 50 per cent at each step down a group, from 3d to 4d to 5d. So complexes of rhodium, palladium, iridium and platinum are almost always low spin.

Example— Ruby and emerald again

Cr³⁺ is d³: no spin choice, three unpaired electrons, CFSE =−1.2 Δo= -1.2\,\Delta_o. Its main absorption band lies near 18 000 cm⁻¹ (about 555 nm, yellow-green) in ruby and near 16 000 cm⁻¹ (about 625 nm, orange-red) in emerald, because the oxygen neighbours in beryl give chromium a slightly weaker field. Ruby absorbs the yellow-green and transmits red; emerald's band has moved far enough to let the green through. A shift of about a tenth in Δo\Delta_o turns a red stone green.

ExplanationH

Metals in biology and medicine

About a third of all proteins are thought to need a metal ion, and many of the most important are coordination complexes buried in a protein.

Haemoglobin. Each of its four subunits holds a haem group: an iron(II) ion bound by the four nitrogen atoms of a flat porphyrin ring, with a fifth bond to a histidine of the protein below the ring. The sixth position, above the ring, is where O₂ binds. Without oxygen the iron is high-spin d⁶, slightly too large for the ring's hole, and sits a little below the ring. When O₂ binds, the iron becomes low spin and smaller and moves into the plane (the bonding of the bound oxygen is subtler than this, and still discussed), tugging its histidine and the protein with it. That movement, passed between subunits, makes each successive O₂ bind more easily, so haemoglobin loads oxygen in the lungs and unloads it where it is scarce. Oxygen-rich blood is bright red and oxygen-poor blood dark red; it is never blue. Carbon monoxide binds to the same site about two hundred times more strongly than oxygen, helped by back-bonding, which is why homes need working CO alarms near fuel-burning appliances. The PDB's Molecule of the Month shows the structure.

Chlorophyll. Chlorophyll is a magnesium complex of a chlorin, a ring closely related to porphyrin. Mg²⁺ has no d electrons, so chlorophyll's colour is not d–d: it comes from the ring's large π system, which absorbs strongly near 430 nm (blue) and near 660 nm (red) and reflects the green between. The metal holds the ring in shape and tunes its energy levels. Vitamin B₁₂ is a cobalt complex of a related ring, whose structure Dorothy Hodgkin solved by X-ray crystallography in the 1950s.

Cisplatin. In the 1960s the American biophysicist Barnett Rosenberg and his colleagues passed a current through a bacterial culture between platinum electrodes and found that the bacteria stopped dividing (Rosenberg et al., 1965). The cause was not the current but small amounts of platinum complexes formed from the electrodes. A closely related complex, cis-[PtCl₂(NH₃)₂], first made by the Italian chemist Michele Peyrone in 1844, was then tested against tumours in mice and worked; as cisplatin it was approved for cancer treatment in 1978. Combined with other drugs it turned testicular cancer from a usually fatal disease into one most patients survive.

Its mechanism is coordination chemistry throughout. In blood, with about 100 mmol/L of chloride, cisplatin keeps its chloride ligands; inside a cell, where chloride is several times scarcer, one is replaced by water. The reactive aqua complex binds to nitrogen atoms of guanine bases in DNA, most often linking two neighbouring guanines on one strand, which kinks the helix and triggers the cell's self-destruction. Transplatin, with its chlorides opposite each other, cannot make that adjacent cross-link and is far less effective. Later relatives such as carboplatin and oxaliplatin were designed to reduce side effects. Other complexes in medicine include gadolinium(III) chelates for magnetic resonance imaging (the chelate stops toxic free Gd³⁺ from being released) and technetium-99m complexes used in millions of diagnostic scans each year.

Note

These are medicines and toxins handled by trained professionals. Platinum anticancer drugs, gadolinium agents and heavy-metal compounds are hazardous outside the clinic and the laboratory; their safety data in PubChem shows why. Safe colours to explore at home are the ones in food: with an adult, compare the colour of red cabbage water, beetroot juice and green spinach water in clear glasses against a white sheet.

ExplanationH

The professional view

Measuring complexes. A new complex is characterised by its single-crystal X-ray structure (deposited in the Cambridge Structural Database, which holds well over a million structures), its UV–visible spectrum, its magnetic moment (from a SQUID magnetometer, or in solution by the Evans NMR method) and, if paramagnetic, electron paramagnetic resonance. For iron, Mössbauer spectroscopy reads the oxidation and spin state from the nucleus itself.

Beyond crystal field theory. Professionals use ligand field theory and its parametrised form, the angular overlap model, which assigns separate σ and π parameters to each ligand. Quantitative work uses quantum chemistry (see Computational chemistry), but transition-metal complexes are a hard case: the energy gap between high- and low-spin states is notoriously sensitive to the choice of density functional, and errors of tens of kJ/mol are common.

Where complexes work. Most industrial catalysts that work in solution are metal complexes with ligands chosen to tune the metal (see Catalysis and green chemistry). Iridium complexes are the light emitters in many OLED displays; ruthenium complexes were the first dyes of dye-sensitised solar cells. Joining metal ions with organic linkers into extended porous crystals gives metal–organic frameworks, recognised by the 2025 Nobel Prize in Chemistry (Susumu Kitagawa, Richard Robson and Omar Yaghi) and taken up in New reactions and framework materials. The same d-orbital splitting helps set the magnetism of oxides in Lumenary Materials' magnetic materials.

Common misconceptionsH

Common misconceptions

Misconception

"The colour a complex shows is the colour it absorbs." It is the reverse: we see the light that is not absorbed. A solution that absorbs red light looks blue-green. And because absorption bands are broad, the seen colour is the sum of everything transmitted, not simply the "opposite" of the band's peak.

Misconception

"Ligands with more negative charge always give a bigger splitting." The spectrochemical series contradicts this: neutral CO and NH₃ give larger splittings than F⁻ or Cl⁻. What matters most is σ-donation and π-bonding, which the point-charge model ignores.

Misconception

"Every transition-metal complex is coloured." Complexes of d⁰ and d¹⁰ ions, such as Zn²⁺, have no d–d transitions and are usually colourless (unless a charge-transfer or ligand band reaches the visible, as in permanganate). Zinc sulfate solution looks like water.

Exercises

Exercises

1. Coordination chemistry

Warm-upHuman

Potassium tris(oxalato)ferrate is K₃[Fe(C₂O₄)₃]. Each oxalate ion, C₂O₄²⁻, is a bidentate ligand. What is the oxidation state of iron? (Give the number without a sign.)

2. Coordination chemistry

Warm-upHuman

Werner's purple compound, CoCl₃·5NH₃, is [CoCl(NH₃)₅]Cl₂. A solution containing 1 mol of it is treated with excess silver nitrate, which precipitates free chloride ions as AgCl. How many moles of AgCl precipitate?

3. Coordination chemistry

CoreHuman

The octahedral ion [CoCl₂(en)₂]⁺ has two chelating ethane-1,2-diamine ligands, each spanning two cis positions, and two chloride ligands. How many stereoisomers does it have, counting a pair of mirror images as two?

4. Coordination chemistry

CoreHuman

Cyanide is a strong-field ligand. What does crystal field theory predict for [Fe(CN)₆]³⁻?

Choose one

5. Coordination chemistry

CoreHuman

[Co(NH₃)₆]³⁺ is yellow-orange and diamagnetic. [CoF₆]³⁻ is blue-green and paramagnetic. Both contain Co³⁺ (d⁶). Select every correct statement.

Choose all that apply

6. Coordination chemistry

CoreComputational

Write octahedral_config(n_d, low_spin) for an octahedral ion with n_d d electrons (0 to 10). Return a tuple (n_t2g, n_eg, unpaired, cfse), where cfse is the crystal field stabilisation energy in units of Δo\Delta_o (−0.4-0.4 per t2gt_{2g} electron, +0.6+0.6 per ege_g electron), rounded to one decimal place.

  • High spin: one electron in each of the five orbitals (t2gt_{2g} first) before any pairing.
  • Low spin: fill t2gt_{2g} completely (six electrons) before ege_g.
Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

7. Coordination chemistry

StretchComputational

Write predict_spin(n_d, delta_o, pairing) for an octahedral d^n ion, with the splitting and the pairing energy in the same units (say cm⁻¹). Return (spin, unpaired, mu): spin is "low" if the ion is d⁴–d⁷ and Δo>P\Delta_o \gt P, otherwise "high" (use "high" also when no choice arises); unpaired is the number of unpaired electrons; and mu is the spin-only moment n(n+2)\sqrt{n(n+2)} in Bohr magnetons, rounded to 2 decimal places.

The tests include [Fe(H₂O)₆]²⁺ (Δo\Delta_o = 10 400, PP = 17 600), [Fe(CN)₆]⁴⁻ (33 000, 17 600) and [Mn(H₂O)₆]²⁺ (7800, 25 500).

Runs entirely in your browser, in an isolated worker. Nothing is sent to a server.

8. Coordination chemistry

Warm-upHuman

Give the IUPAC name of [Co(NH₃)₆]Cl₃.

Further reading and resources