CMA complexes were first reported in 2017 and quickly attracted attention as TADF emitters because they combine easy synthesis, tunable emission colors, and high internal and external quantum efficiencies with unusually fast radiative decay rates. TADF emitters work by harvesting both singlet and triplet excitons: triplet excitons undergo reverse intersystem crossing (rISC) back to the emissive singlet state, allowing nearly all electrically generated excitons to produce light without relying on scarce and expensive iridium or platinum phosphors. Yet despite their promise, OLED devices built from d10 CMA emitters with genuinely long operational lifetimes have remained extremely rare, and previous work by the same team showed that steric engineering of N-heterocyclic carbene ligands could push device lifetimes to LT95 values of up to 2082 hours for gold emitters and 3582 hours for copper emitters at a luminance of 1000 cd m−2. What was still missing was a rigorous, quantitative explanation of how the metal atom itself governs the emission kinetics.

The new study tackled this question with an unusually deep computational arsenal. Because conventional density functional theory (DFT) calculations are too approximate to capture the subtle energetics of these systems, the researchers combined time-dependent DFT with high-level ab initio methods, including similarity-transformed equation-of-motion domain-based local pair natural orbital coupled-cluster theory (STEOM-DLPNO-CCSD) and the combined DFT/multireference configuration interaction (DFT/MRCI) approach. They supplemented these with a battery of bonding analysis techniques — natural adaptive orbital (NAdO) analysis, extended transition-state natural orbitals for chemical valence (ETS-NOCV) decomposition, and charge decomposition analysis — to dissect, atom by atom and electron by electron, how the metal binds to its carbene and carbazole ligands in both ground and excited states.

The central discovery concerns the metal’s (n+1)p orbitals, which sit just above the filled d shell. In a naive picture, a d10 metal ion has no low-lying empty d orbitals to engage in conventional bonding, so metal–ligand π-interactions with strongly donating ligands such as carbazolide would seem improbable. But the calculations show that mixing between the (n+1)p orbital and the nd orbital polarizes the d orbital’s spatial distribution, enlarging its bonding lobe in one direction and suppressing its antibonding lobe in the other. This (n+1)p–nd hybridization creates genuine metal–ligand π-interactions that persist even in the excited state. Crucially, the degree of hybridization differs systematically across the coinage metals: copper displays the largest (n+1)p orbital population (roughly 0.10–0.14 electrons), gold an intermediate amount, and silver the smallest. This single electronic difference cascades through the entire photophysics of the emitter.

The practical consequence involves molecular rotation. In the excited singlet state, an electron is promoted from the highest occupied molecular orbital — largely localized on the carbazole ligand — to the lowest unoccupied orbital, producing a ligand-to-ligand charge-transfer (LLCT) transition. Because the HOMO is half-occupied in this state, the π-interaction between the metal and the carbazolyl nitrogen is actually enhanced relative to the ground state, and the nitrogen-side π-interaction is weaker than the corresponding metal–carbene carbon π-interaction. As a result, the carbazole and carbene ligands can rotate about the metal–nitrogen bond during the excited-state lifetime, swinging between semi-coplanar and orthogonal conformations. NAdO and ETS-NOCV analyses quantified these interactions, showing, for example, that the Au–N π-interaction stabilizes the excited state of a representative gold emitter by about 10.85 kcal/mol, compared with roughly 26.94 kcal/mol for the Au–C π-interactions — confirming that rotation occurs preferentially at the metal–nitrogen bond.

This rotational flexibility turns out to be the key to solving a longstanding trade-off in TADF design. Emitters need a small singlet–triplet energy gap (∆ES1–T1) to enable efficient reverse intersystem crossing, but the usual strategy for shrinking that gap — spatially separating the donor and acceptor orbitals — simultaneously reduces the oscillator strength of the S1→S0 transition, slowing prompt fluorescence. The study shows that flexible excited-state rotation offers an escape route: semi-coplanar rotamers provide large orbital overlap and a fast radiative rate, while orthogonal rotamers provide a small ∆ES1–T1, and rapid thermal interconversion between them allows the molecule to exploit both regimes simultaneously. Silver, with the weakest M–N π-interaction and the greatest rotational freedom, achieves the highest TADF radiative decay rates; copper, with the strongest π-interaction and stiffest rotation, the lowest. The predicted ordering — Ag > Au > Cu — matches experimental measurements across multiple emitter series and solvents, resolving a question that had persisted since the first CMA emitters were reported.

The team then translated this mechanistic insight into practical ligand design rules. Calculations of a series of gold emitters bearing electron-withdrawing groups (cyano, trifluoromethyl) or electron-donating groups (methoxy, tert-butyl) on the carbazole ligand revealed a clean trend: electron-withdrawing substituents reduce the metal’s (n+1)p–nd hybridization, weaken the excited-state M–N π-interaction, increase rotational flexibility, and thereby accelerate TADF. Electron-donating substituents do the opposite. Extending the π-system of the carbazole achieves a similar effect through a different route — the hole migrates onto the extended aromatic system, delocalizing the M–N π-electrons and loosening the nitrogen-side bond. The researchers also identified an experimentally accessible descriptor: the emitter’s oxidation potential, measured by cyclic voltammetry, correlates linearly with the calculated strength of the excited-state π(M…N) interaction, giving chemists a simple electrochemical handle for screening new candidates.

But the story comes with a crucial caveat. Weakening the excited-state M–N π-interaction also weakens the bond itself, making the metal–ligand framework more vulnerable to dissociation under the harsh conditions of electroluminescence. To test this, the team fabricated OLEDs based on four gold emitters with matched device structures: Au-1 and three derivatives carrying electron-withdrawing groups. While all emitted light in nearly the same spectral region — 467 to 469 nm — their operational lifetimes differed dramatically. The parent Au-1 device achieved an LT90 of 1709 hours, whereas devices based on Au-1CN, Au-12CF3, and Au-12CN lasted only 131, 1.01, and 0.37 hours, respectively, at an initial luminance of 1000 cd m−2. Because the substituents sit far from the metal center and exert negligible steric protection, the correlation points squarely at the strength of the excited-state M–N π-interaction as a determinant of device durability.

The implication is a genuine design dilemma: the same electronic tuning that maximizes radiative decay rate and efficiency may simultaneously shorten the operational lifetime of the device. The authors emphasize that many factors govern OLED stability — emission energy, steric shielding, molecular packing, and excited-state bond dissociation among them — but their data establish, for the first time, a clear and quantifiable correlation between excited-state metal–nitrogen π-interaction strength and device longevity in this emitter class. Navigating this trade-off, they suggest, will be essential for realizing the full industrial potential of d10 CMA emitters, particularly for blue and green devices where high-energy emission already stresses molecular stability.

Beyond its immediate practical guidance, the study provides a conceptual reframing of TADF mechanism in metal complexes. Rather than treating spin–orbit coupling and the singlet–triplet gap as the sole arbiters of delayed fluorescence kinetics, the work demonstrates that excited-state conformational dynamics — governed by orbitals that textbook bonding pictures would ignore — can dominate the radiative decay rate. The spin–orbit coupling matrix elements, which follow the order Cu > Au > Ag due to differences in d-orbital participation and heavy-atom effects, turn out to be of secondary importance compared with rotational flexibility. For a field racing to replace precious-metal phosphors with earth-abundant copper and silver emitters in displays and lighting, that insight reframes the design problem entirely: the fastest emitters may be those whose metal–ligand bonds are engineered to be just strong enough to survive, yet just weak enough to let the molecule dance.