The fork in the road every dn ion reaches
By now you can draw the picture from the previous guides: in an octahedral complex the two d orbitals aimed straight at the ligands (dz2 and dx2-y2) are shoved up into the eg set, while the three that point into the gaps (dxy, dxz, dyz) sink down into the t2g set, separated by the gap delta-o. For an ion with one, two, or three d electrons there is no drama — they simply drop into the three lower t2g orbitals one apiece, following Hund's rule. The interesting moment arrives with the fourth electron, and it lasts through the seventh. That is where the complex reaches a genuine fork in the road.
The fourth electron has two ways to go, and they cost different amounts. It can crowd into a t2g orbital that already holds an electron, pairing up with it — but two electrons jammed into one orbital repel each other, so this costs the electron-pairing energy P. Or it can climb up and start a fresh, empty eg orbital all to itself — but to get there it must pay the splitting gap delta-o. Nature is a frugal accountant and takes whichever bill is smaller. That single comparison, delta-o against P, decides the spin state of the whole complex.
Spread the electrons out into the upper set to keep as many unpaired as possible and you have a high-spin complex; pack them into the lower set, pairing up, and you have a low-spin one. This choice is real and measurable, not academic: high-spin ions carry many unpaired electrons and are strongly paramagnetic, low-spin ones carry few or none and are weakly paramagnetic or even diamagnetic. The same Fe2+ d6 center is high-spin and magnetic as [Fe(H2O)6]2+ but low-spin and diamagnetic as [Fe(CN)6]4-. The metal did not change; the ligand did.
Delta versus P: the price war
What exactly is the pairing energy you are weighing delta against? P is not one thing but two costs bundled together. The obvious part is plain electrostatic repulsion — two electrons in one small orbital are two like charges pressed close, and they push back. The subtler part is the loss of exchange energy: a set of parallel-spin electrons in separate orbitals enjoys a quantum-mechanical stabilization (this is the deep reason behind Hund's rule), and the moment you pair two of them, you throw that bonus away. Add the two and you get P, the full toll for forcing two electrons to share a room.
Now the rule writes itself. If delta-o is larger than P, climbing to eg is the expensive route, so electrons would rather pay P and pair down low — the complex goes low-spin. If delta-o is smaller than P, climbing is the bargain, so electrons spread upward and stay unpaired — the complex goes high-spin. Crucially, P is roughly fixed for a given metal ion in a given oxidation state (it grows for smaller, more compact ions whose electrons are squeezed tighter together). Since P barely moves, it is the ligand-controlled delta that actually swings the decision. That is precisely why the next idea — ranking ligands by the delta they produce — is so powerful.
The spectrochemical series: ranking the ligands
If the ligand sets delta, then chemists need a list: which ligands open a wide gap and which open only a narrow one? Reading delta straight off the absorption spectra of many complexes gave a remarkably stable ranking called the **spectrochemical series**. From weakest field (small delta) to strongest field (large delta), a common run is: iodide < bromide < chloride < fluoride < hydroxide < water < ammonia < ethylenediamine < bipyridine < cyanide < carbon monoxide. Weak-field ligands like the halides keep delta small and favour high-spin; strong-field ligands like CN- and CO push delta wide and favour low-spin.
Here is where pure crystal field theory quietly fails, and it is worth being honest about. If ligands were just point charges, the most negative ones should split hardest — yet neutral carbon monoxide outsplits charged fluoride, the very opposite of what bare electrostatics predicts. The repair comes from ligand field theory, which admits the bonds are partly covalent. Strong-field ligands like CO and CN- are pi-acceptors: they have empty pi orbitals that drain electron density out of the metal's t2g set, pushing t2g down and so widening delta. Weak-field halides are pi-donors that do the reverse and shrink it. The order is real; the point-charge story for why is simply too crude.
Cobalt(III), a d6 ion, shows the whole story in one element. [CoF6]3- is high-spin and pale because fluoride sits low in the series and gives a delta too small to beat P. Swap in cyanide, near the top of the series, and [Co(CN)6]3- becomes low-spin and exceptionally stable, because its large delta forces all six electrons to pair into t2g. Same Co3+, opposite spin states, decided entirely by where the ligand sits in the ranking. A fair warning: the series is empirical and only roughly transferable — delta also depends on the metal, its oxidation state, and the geometry, so the exact order can shuffle a little between systems. Treat it as a reliable guide, not an exact universal ruler.
Cashing it in: crystal field stabilization energy
Splitting the d orbitals does more than colour and magnetize a complex — it can genuinely lower its total energy. The logic is simple bookkeeping. Relative to the unsplit, hypothetical free ion (which we call zero), every electron that lands in the lower t2g set sits below the average and is worth -2/5 delta-o, while every electron forced up into eg sits above and costs +3/5 delta-o. These exact fractions come from the barycenter rule you met earlier: the centre of gravity of the five orbitals must not move, so three orbitals drop by 2/5 and two rise by 3/5. Add up the contributions and you have the crystal field stabilization energy, CFSE.
- Find the d-electron count: from the metal's oxidation state, work out how many d electrons the ion has — for example Co3+ is d6, Fe2+ is also d6, Cr3+ is d3.
- Decide high-spin or low-spin: compare delta-o (read off the ligand's place in the spectrochemical series) against the pairing energy P, then fill t2g and eg accordingly.
- Count electrons in each set: note how many ended up in t2g and how many in eg.
- Add up the energies: CFSE = (t2g count) x (-2/5 delta-o) + (eg count) x (+3/5 delta-o); for low-spin cases, add back the extra pairing energy you paid relative to the high-spin arrangement.
Octahedral splitting (barycenter rule): eg (dz2, dx2-y2) ........ +3/5 delta_o (2 orbitals, raised) --- barycenter (free-ion energy = 0) --- t2g (dxy, dxz, dyz) ...... -2/5 delta_o (3 orbitals, lowered) CFSE = n(t2g)*(-2/5 d_o) + n(eg)*(+3/5 d_o) [+ pairing] low-spin d6 e.g. [Co(NH3)6]3+ : t2g^6 eg^0 CFSE = 6*(-2/5 d_o) = -12/5 d_o = -2.4 d_o (before pairing) high-spin d5 e.g. [Mn(H2O)6]2+: t2g^3 eg^2 CFSE = 3*(-2/5) + 2*(+3/5) = -6/5 + 6/5 = 0
Two end-member cases make the pattern click. Low-spin d6 [Co(NH3)6]3+ packs all six electrons into t2g, scoring 6 x (-2/5 delta-o) = -2.4 delta-o of stabilization (before pairing) — a huge bonus that helps explain why this ion is so robust. At the other extreme, high-spin d5 like Mn2+ in [Mn(H2O)6]2+ puts one electron in each of the five orbitals: three at -2/5 and two at +3/5 sum to exactly zero. CFSE is also zero for d0 and d10 — the symmetric configurations gain nothing — which is one reason Mn2+ complexes are pale, comparatively labile, and unremarkable.
Why CFSE is more than bookkeeping
It would be fair to suspect CFSE is just a tidy accounting fiction. The honest test is whether it leaves fingerprints in real data — and it does. Plot a simple property of the first-row M2+ ions, say ionic radii or hydration enthalpies, from Ca2+ to Zn2+, and instead of the smooth line you would expect from a steadily rising nuclear charge, you get a famous double-humped curve. The humps peak near d3 and d8 (where octahedral CFSE is large) and dip back to the smooth baseline at d0, high-spin d5 (Mn2+), and d10 (Zn2+), exactly where CFSE is zero. Subtract the calculated CFSE from each point and the bumps vanish — the data fall back onto the bare electrostatic line.
One last honesty about the whole picture. All of this rests on the crystal field model, whose ligands-as-point-charges premise is a deliberate fiction — real metal-ligand bonds are partly covalent, which is exactly why the spectrochemical series needed pi-bonding to explain. The numbers like -2.4 delta-o are exact within the model, not exact about nature. Yet the model earns its keep: it predicts spin states, colours, magnetism, geometry preferences, and those double-humped thermodynamic trends with a single, almost arithmetical recipe. When it strains, you reach for ligand field theory; but the delta-versus-P contest and CFSE accounting you have learned here remain the everyday workhorses of d-block chemistry.