From Crystal Field to Ligand Field Theory

What the point-charge model gets right — and what it cannot

Through the last four guides you have leaned hard on [[crystal-field-theory|crystal field theory]], and it has paid you back generously. By pretending each ligand is nothing more than a negative point charge, it predicted that the metal's five d orbitals split in an octahedron — the two pointing straight at the ligands (eg) shoved up, the three pointing into the gaps (t2g) sunk down — and from that single split came the colour of a complex, its high-spin or low-spin magnetism, and even why d8 ions go square planar. That is a magnificent return on one crude assumption. But a model this useful tempts you to forget it is a model, so this guide is where we cash in the honesty.

Here is the trouble. If ligands were truly just point charges, then the size of the splitting, delta-o, should track one thing only: how much negative charge each ligand carries and how close it sits. A doubly charged oxide ion, O2-, ought to split a metal's d orbitals far harder than neutral, uncharged carbon monoxide. Yet the experimental ordering of ligands by the delta-o they produce — the [[spectrochemical-series|spectrochemical series]] — says the opposite. Roughly: I- < Br- < Cl- < F- < OH- < H2O < NH3 < en < NO2- < CN- < CO. Neutral CO and CN- sit at the strong-field top; the small, highly charged fluoride and hydroxide languish in the weak-field middle; bare halide anions are weakest of all. Pure electrostatics gets the order almost exactly backwards.

Rebuilding the splitting from molecular orbitals

The fix is to stop treating the bond as electrostatic pushing and start treating it the way the molecular-orbital rung taught you to treat every bond: as orbitals overlapping. This is [[ligand-field-theory|ligand field theory]]. The recipe is exactly the MO diagram machinery you already know, applied to a whole octahedron at once. Each of the six ligands offers a lone pair pointing straight at the metal — a sigma-donor orbital. Combine the six into symmetry-matched groups, line them up against the metal's nine valence orbitals (the five 3d, the one 4s, and the three 4p), and let orbitals of the same symmetry mix.

Now watch the d orbitals sort themselves. The two eg orbitals (dz2 and dx2-y2) point straight at the ligand lone pairs, so they overlap well and mix — this is [[sigma-donation|sigma donation]], the ligand handing its lone pair toward the metal. Mixing produces a bonding combination that sinks low (it ends up mostly ligand in character, filled by the donated electrons) and an antibonding partner that rises high (mostly metal in character). That high antibonding partner *is* the eg level of crystal field theory. Meanwhile the three t2g orbitals (dxy, dxz, dyz) point into the gaps between ligands; in a pure sigma picture they find nothing of the right symmetry to overlap with, so they stay put as nonbonding orbitals at the metal's original energy.

Stand back and look at what just happened. The gap between the nonbonding t2g and the antibonding eg* is delta-o — the very same splitting crystal field theory drew, but now with a completely different cause. It is no longer the cost of an electron snuggling up to a point charge; it is the cost of putting an electron into an *antibonding* orbital. A stronger sigma donor overlaps better, pushes its eg* partner higher, and so makes delta-o bigger. That already rescues part of the spectrochemical series: ammonia is a better sigma donor than water through nitrogen's lone pair, so NH3 sits above H2O, exactly as observed. Same diagram, deeper truth.

Octahedral MO picture (sigma only), where the d-block lives:

   metal s,p + ligands  ->  a1g*, t1u*  (very high, empty)

        eg*  (dz2, dx2-y2)   <- ANTIBONDING: metal-d vs ligand sigma
        ___ ___
         |
         |  delta_o  = energy to occupy an antibonding orbital
         |
        ___ ___ ___          t2g  <- NONBONDING (sigma-only): pure metal d
        (dxy, dxz, dyz)          this is where the d electrons go

        ===  ===  === ...    bonding MOs (mostly ligand), full of
                             the 12 donated sigma electrons

Stronger sigma donor -> better overlap -> eg* pushed higher -> bigger delta_o

Pi changes everything: donors shrink delta, acceptors swell it

The sigma-only story still puts neutral CO below charged fluoride, which is wrong. The missing ingredient is the one thing crystal field theory could never have: pi interactions involving those formerly idle t2g orbitals. Recall from the diatomics rung that an atom or small molecule can carry pi-type orbitals as well as sigma lone pairs. The metal's t2g set (dxy, dxz, dyz) has exactly the right shape to overlap sideways with ligand pi orbitals — and whether that helps or hurts depends entirely on whether the ligand's pi orbitals are *full* or *empty*.

Take a [[pi-donor-ligand|pi-donor ligand]] first — a halide like Cl-, or an oxide, with *filled* p lone pairs left over after sigma bonding. These full ligand pi orbitals lie below the metal t2g, so when they mix, the bonding combination drops onto the ligands and the antibonding combination — now the metal's t2g — is pushed *up*. Raising t2g shrinks the t2g-to-eg* gap. So pi donors make delta-o *smaller*. This is why fluoride, oxide, and water (a weak pi donor through its second lone pair) sit at the weak-field end: they are not feeble at sigma donation, they are actively closing the gap from below by donating pi as well.

Now the reverse, and the punchline of the whole rung. A [[pi-acceptor-ligand|pi-acceptor ligand]] like carbon monoxide has *empty* pi* antibonding orbitals lying just *above* the metal t2g. When metal t2g mixes with these empty ligand pi* orbitals, the bonding combination — heavily metal t2g in character — drops *down*. Lowering t2g widens the t2g-to-eg* gap, so pi acceptors make delta-o *much bigger*. The electrons flow the other way too: the metal donates from its filled t2g into the ligand's empty pi*, the famous [[pi-back-donation|pi back-donation]] you will recognize from the Dewar-Chatt-Duncanson picture of metal-alkene and metal carbonyl bonding. This synergy — ligand gives sigma, metal gives pi back — is exactly why CO and CN- crown the strong-field top of the series, and why a neutral molecule can outsplit a doubly charged anion. Point charges never stood a chance.

Reading the spectrochemical series in one glance

With sigma and pi both in hand, the whole series finally lines up as a single readable story. Walk through it as a three-step recipe applied to any ligand.

1. Ask how good a sigma donor it is. A better sigma donor pushes eg* higher and enlarges delta-o. This alone explains the mid-series climb H2O < NH3 < en, all sigma-only-ish ligands ranked by donor strength.
2. Ask whether it has filled pi orbitals to donate. If yes — halides, oxide, water's spare lone pairs — it is a pi donor: it raises t2g, shrinks delta-o, and drops to the weak-field bottom (I- < Br- < Cl- < F-, the weaker-bound, more diffuse halide donating pi more easily, hence weakest field).
3. Ask whether it has empty pi* orbitals to accept. If yes — CO, CN-, NO, bipyridine, the tertiary phosphines — it is a pi acceptor: metal-to-ligand back-donation lowers t2g, swells delta-o, and rockets the ligand to the strong-field top.

And the nephelauxetic effect drops out for free. Because metal and ligand orbitals genuinely mix, a metal d electron is no longer confined to the metal; it is partly spread onto the ligands, in a bigger, more diffuse orbital. Spread the electrons out and they repel each other less — the cloud has expanded, exactly the word the effect carries. The same ligands that are most covalent (the heavy, soft, polarizable donors and the pi acceptors) cause the biggest cloud expansion, giving the parallel nephelauxetic series. Crystal field theory, with its rigid metal-only d orbitals, has no way even to phrase this; ligand field theory makes it inevitable.

Honest limits, and what to carry up the ladder

Be clear about what ligand field theory does and does not claim. It is not a rival to crystal field theory so much as its honest upgrade: it keeps the priceless t2g/eg* picture and everything you built on it — colour from d-d transitions, the high-spin/low-spin choice from delta-o versus pairing energy — but it explains *why* delta-o has the size it does. The everyday name 'ligand field theory' usually means this MO-flavoured account used at the level we have drawn it; the full quantitative machinery, with its Racah parameters and Tanabe-Sugano diagrams, is a further refinement you can meet later. Even the labels are a controlled fib: in a real complex the t2g and eg* orbitals are not pure metal d, but we keep calling them by their crystal-field names because they behave, for counting electrons, just as before.

One last caution, since this rung sits in inorganic chemistry where words mislead. 'Covalent' here does not mean the bond is fully shared like the one in H2; it means *partly* shared — every metal-ligand bond is a blend of ionic and covalent character, and the spectrochemical and nephelauxetic series are really maps of how covalent each ligand makes its bond. And remember that delta-o is an energy gap we read out of spectra and magnetism, not a charge or a length you could measure with a ruler. Keep the t2g/eg* picture, keep the sigma-donor and pi-donor-versus-pi-acceptor reflexes, and you carry up the ladder the single most powerful way of thinking in the chemistry of the d-block.