Radical OLEDs: A Different Route to Highly Efficient Light

How does an organic radical actually reach the excited state from which it emits light? In our recent preprint, we set out to map this microscopic journey.

OLEDs, electrons, and a little quantum mechanics

OLEDs are now part of everyday life. They illuminate the pixels in smartphones, televisions, watches, and many other displays. But behind every glowing pixel lies an interesting quantum mechanical problem: when positive and negative charges meet inside an organic material, which electronic state do they form, and can that state efficiently produce light?

In a conventional closed-shell fluorescent molecule, all electrons are initially paired. When an electron and a hole recombine, spin statistics lead to the formation of both singlet and triplet excitons.

Approximately 25% are expected to be singlets and 75% triplets. Ordinary fluorescence can use the singlet excitons efficiently, but emission directly from the triplet state is spin-forbidden.

Modern OLED technologies have developed clever ways around this problem. Phosphorescent emitters and thermally activated delayed fluorescence (TADF), for example, can harvest triplet excitons.

Organic radicals offer something fundamentally different.

What makes a radical special?

A radical contains an unpaired electron. This means its electronic structure is different from that of a conventional closed-shell molecule.

Instead of describing the relevant ground and excited states as singlets and triplets, organic radicals can have a doublet ground state, D₀, and an emissive doublet excited state, D₁.

D₁ → D₀ + light

Importantly, this doublet-to-doublet transition can be spin-allowed.

Radical OLEDs therefore bypass the conventional singlet–triplet exciton constraint rather than trying to recover triplet excitons after they have formed.

Experiments have demonstrated nearly 100% doublet-exciton formation in radical OLEDs. If competing non-radiative processes can be sufficiently suppressed, this means that radical emitters can, in principle, approach 100% internal quantum efficiency (IQE).

But there was still a surprisingly basic question

We know that the D₁ state can emit light.

But under electrical excitation, how is D₁ actually formed?

An OLED does not simply place a molecule directly into its emissive state. Electrons and holes move through the material, charges can become localised on different molecules, and several charge-transfer and recombination pathways may be possible.

So I like to think of D₁ as the destination.

We know where the light comes from. But what microscopic road takes the radical there?

Mapping the possible routes

In our recent preprint, Timothy J. H. Hele and I investigated this problem using quantum-chemical calculations together with Marcus-type electron-transfer modelling.

We studied three TTM-based radical emitters: TTM-1Cz, TTM-3PCz, and TTM-3NCz, considering them in a minimal CBP-host environment.

Rather than assuming a particular mechanism from the beginning, we compared several competing possibilities:

• anion-mediated pathways,
• cation-mediated pathways,
• charge–charge annihilation pathways,
• and pathways involving electronically excited charged species.

For each route, the important question is not simply whether it can be drawn on paper. The energetics must be favourable and the corresponding electron-transfer process must occur at a competitive rate.

One pathway consistently stands out

Across the radicals we studied, our calculations point toward an anion-mediated charge-recombination mechanism as the most consistently plausible route to forming the emissive D₁ state.

D₀ → Radical Anion → D₁ → Light

In simple terms, the neutral radical first accepts an electron, producing a negatively charged radical species.

Subsequent recombination with a positive charge can then populate the excited D₁ state from which light is emitted.

Our calculations also suggest that excited-anion pathways could contribute indirectly. Higher-lying electronic states can first be populated and subsequently relax toward D₁.

In contrast, the cation-mediated routes we examined were generally less favourable, while competing processes that return charged species directly to the non-emissive ground state were kinetically suppressed in our calculations.

Why does knowing the mechanism matter?

Finding a molecule that emits efficiently is useful. Understanding why it works is much more powerful.

Once we know the pathways responsible for producing D₁, we can begin asking design questions rather than relying only on trial and error.

Can the electron affinity of the radical be tuned to favour productive anion formation?

Can the surrounding host material be chosen to make the desired charge-transfer process faster?

Can molecular substitution increase the probability of reaching D₁ while suppressing pathways that waste the injected electrical energy?

Ultimately, understanding the mechanism provides a bridge between fundamental electronic-structure theory and the rational design of better OLED materials.

A different way of making light

What I find particularly fascinating about radical OLEDs is that they do not merely improve on the usual closed-shell picture. They change the electronic-state landscape altogether.

The unpaired electron gives us access to doublet states, and the emissive D₁ → D₀ transition provides a spin-allowed pathway for light emission.

And we can now begin to connect the electrical operation of the device to that final photon through a microscopic sequence:

Charge injection → Anion formation → Charge recombination → D₁ → Photon

There is still much to understand, particularly when moving from molecular models toward the complexity of a complete working device.

But having a mechanistic picture gives us something extremely valuable: a set of physical principles that can eventually guide the design of the next generation of radical emitters.

Preprint

Mechanistic Origins of D₁ Excited-State Formation in Radical-Based Organic Light-Emitting Diodes

Prashant Kumar and Timothy J. H. Hele

Read the preprint on ChemRxiv →

From Preprint to Publication: Testing a Radical OLED Mechanism

A scientific paper rarely finishes when the first manuscript is written. Peer review asks whether the conclusions survive when the assumptions behind them are examined more closely. Our recent work on the formation of the emissive D₁ state in radical OLEDs has now completed that journey from preprint to publication.

In the first blog post on this site, I discussed the central question behind the work: how does an electrically driven organic radical actually reach the emissive D₁ state?

Our calculations compared several possible charge-transfer and recombination mechanisms and pointed toward an anion-mediated pathway as the most consistently plausible route.

The work is now published in The Journal of Physical Chemistry A.

Publication was not simply the final step. Peer review provided an opportunity to test which parts of the mechanistic picture were robust, which quantities required more careful interpretation, and where the limitations of the computational model needed to be made clearer.

What did peer review challenge us to think about?

Three particularly useful scientific questions emerged during the review process.

They concerned:

• the effect of the surrounding OLED host environment on the calculated reaction energetics,
• how sensitive the calculated electron-transfer rates might be to the particular molecular configuration used to calculate the electronic coupling,
• and how Marcus and Marcus–Levich–Jortner rate calculations should be interpreted when some processes lie very deep in the Marcus inverted regime.

These questions did not simply require textual clarification. In one case they led to additional calculations, while in the others they pushed us to be more explicit about what the numerical results can — and cannot — tell us.

Testing the effect of the host environment

Our systematic mechanistic comparison had been performed using gas-phase cluster calculations.

That provides a consistent framework for comparing many electronic states and pathways, but a real OLED is not a collection of molecules floating in vacuum. The surrounding host material can polarize in response to charged states and therefore alter their relative energies.

To test this, we carried out an additional set of calculations using a CBP-like dielectric environment.

The relevant structures were treated using PBE0/6-31G(d,p) together with the conductor-like polarizable continuum model, CPCM. The host response was represented using:

ε = 3.0    and    ε_∞ = 2.8

We also performed vibrational-frequency calculations for the structures that could be successfully optimized so that the corresponding Gibbs free energies could be compared with the gas-phase results.

The environment did matter.

Several reaction free energies changed appreciably, and charged states were generally stabilized by the dielectric treatment.

But the important result was that the qualitative mechanistic picture remained unchanged. The anion-mediated pathways remained energetically accessible.

There was also an important practical limitation. Some higher-lying and charged electronic states could not be converged reliably in the dielectric model. This meant that we could not construct a complete, internally consistent solvent-phase dataset for every proposed pathway.

We therefore retained the gas-phase calculations for the systematic comparison of all mechanisms and used the additional dielectric calculations as a test of how strongly host polarization could influence the energetics.

What we learned: absolute reaction energies can be sensitive to the molecular environment, but the principal mechanistic conclusion survived this additional test.

A single molecular geometry is not an amorphous OLED film

Marcus-type electron-transfer calculations require an electronic coupling between the donor and acceptor states.

That coupling is strongly influenced by molecular geometry: intermolecular separation, relative orientation, and orbital overlap can all change its magnitude.

The configuration used in our calculation was not selected randomly. It came from a structural search for low-energy CBP–radical configurations and was subsequently optimized before calculating the electronic coupling.

Nevertheless, it still represents one low-energy local arrangement.

A real amorphous OLED contains a distribution of intermolecular configurations, and those different configurations can produce different electronic couplings.

This matters because in the nonadiabatic Marcus expression the rate depends quadratically on the electronic coupling:

k ∝ |V_DA|²

Different configurations can therefore give different absolute rate constants. They can also alter other quantities, including reaction free energies and reorganization energies.

This led us to make an important distinction much more explicit in the published work:

The calculated absolute rate constants should be regarded as configuration- and method-dependent quantities rather than unique predictions for an entire amorphous OLED film.

Our mechanistic conclusions instead rely on comparing the broad energetic and kinetic regimes of the competing pathways using a consistent computational framework.

When a calculated rate should not be treated as a literal rate

This was perhaps the most conceptually important point raised during review.

Some of the electron-transfer processes investigated in the work fall extremely deep into the Marcus inverted region. In such cases, Marcus or MLJ calculations can predict rates that are effectively zero.

It would be tempting to take those numbers literally.

We do not.

When a system lies far outside the regime where the assumptions of the model are expected to be quantitatively reliable, the exact numerical rate — and particularly the fine ordering of several extremely small rates — should be treated with caution.

The role of Marcus-type theory in this study is therefore better viewed as a common mechanistic framework.

It helps distinguish between processes occupying very different kinetic regimes:

• pathways that remain comparatively accessible,
• pathways that encounter substantial activation barriers,
• and processes that are pushed extremely far into the inverted region and are therefore strongly suppressed within the model.

The conclusion is therefore not:

“This reaction literally occurs at the exact calculated rate.”

Rather, the useful mechanistic statement is:

“Within a consistently applied theoretical framework, this pathway occupies a very different energetic and kinetic regime from the competing alternatives.”

That distinction is important whenever a relatively simple kinetic theory is being applied to complex open-shell molecular systems.

So what changed between the preprint and the published paper?

The central mechanistic picture did not need to be replaced.

Instead, peer review made the work more carefully qualified and, I think, scientifically stronger.

The broader lesson: a model is useful partly because of its limits

Computational chemistry is sometimes presented as if the objective were simply to calculate a number with as many decimal places as possible.

Mechanistic modelling is rarely that simple.

A calculated reaction energy depends on the environment we choose to represent. An electron-transfer rate depends on molecular geometry, electronic coupling, reorganization energy, driving force, and the assumptions built into the kinetic theory.

Recognising those dependencies does not make a model useless. It tells us what kind of question the model is actually capable of answering.

In this case, the strongest conclusion is not a single absolute rate constant.

It is the emergence of a consistent mechanistic picture from the combined electronic-structure, energetic, and kinetic analysis: anion-mediated charge recombination remains the most plausible route among the competing mechanisms examined for forming the emissive D₁ state.

Mechanistic hypothesis → computation → peer-review challenge → additional testing → refined interpretation

Why I found this part of the process useful

The most useful review questions are not necessarily those that ask for another calculation simply for the sake of adding more data.

They are the questions that expose an assumption and ask: how much does the conclusion depend on it?

Does the mechanism survive when environmental polarization is included?

Are the calculated rates properties of the material, or of the particular molecular configuration used in the model?

When does a kinetic model provide a meaningful number, and when is its real value instead the qualitative physical picture that it reveals?

Working through those questions made the distinction between prediction and interpretation considerably sharper.

And that is ultimately what I find most useful about computational chemistry: not only calculating what might happen, but understanding why one physical pathway is more plausible than another.

Published Paper

The work discussed in Blog #001 as a ChemRxiv preprint is now published in The Journal of Physical Chemistry A.

DOI: 10.1021/acs.jpca.6c04284

Read the published paper at ACS →

View the earlier ChemRxiv preprint →