Organic solar cells have spent a decade climbing toward a number that kept receding. Built from carbon-based semiconductors that can be printed like ink, they rose from single-digit efficiencies into the high teens on the strength of non-fullerene acceptors. Yet one loss mechanism has stubbornly capped the climb: non-radiative recombination, in which excited charges dissipate their energy as heat instead of contributing to the voltage a cell can deliver. It is the classic organic-photovoltaic tax.
A team led by Yingze Zhang and Sai Ho Pun of the Hong Kong University of Science and Technology, together with Gang Li of the Hong Kong Polytechnic University and colleagues in Shanghai, Beijing, Shenzhen, and Hangzhou, now reports a way to cut that tax — not by redesigning the blend’s morphology or re-tuning its energy levels, the two conventional levers of the field, but by adding a molecule that glows. Their paper, “Solid-state emissive acceptors enable thermally accessible radiative pathways for efficient organic solar cells,” appeared in Nature Photonics, and the headline result is a single-junction device at 21.2% power conversion efficiency in the laboratory, independently certified at 20.84%, with an open-circuit voltage of 0.93 volts and a non-radiative voltage loss as low as 0.142 electronvolts.
The certified figure is a laboratory cell result, not a module specification, and it comes from one host system. But the mechanism behind it is what makes the paper worth more than its record claim: it turns luminescence — normally a loss channel in a solar cell — into a design tool.
The tax: why charge-transfer states go dark
The problem lives at the boundary between the two light-absorbing materials inside an organic photovoltaic cell. When a photon is absorbed, it creates a tightly bound electron-hole pair, an exciton, which must travel to the interface between the electron-donating polymer and the electron-accepting molecule to be split into free charges. In the process, an interfacial charge-transfer state forms, with the electron and hole sitting on adjacent molecules, still attracted to each other but no longer bound as a single exciton.
These charge-transfer states are the workhorses of charge generation, and they are also a liability. Because the electron and hole are spatially separated, their wavefunctions overlap only weakly, which means they couple poorly to light. A charge-transfer exciton cannot easily emit a photon. Instead it tends to decay non-radiatively, shedding its energy as heat. Every charge-transfer exciton lost this way erodes the open-circuit voltage — the maximum voltage the cell can produce — and therefore the overall efficiency.
This is the same voltage-loss story this site covered in its explainer on excitonic solar cells: the energy offset that drives exciton splitting is energy that does not end up as output voltage. Modern non-fullerene acceptors have narrowed the offset needed for efficient splitting, preserving more voltage, but the non-radiative decay of charge-transfer states has remained. In organic photovoltaics, non-radiative voltage losses of 0.2 to 0.3 electronvolts are common. The new paper’s figure of 0.142 electronvolts is, by that standard, ultralow — and the route to it is photophysical rather than morphological.
The trick: a bright state within thermal reach
The team’s idea is to introduce a small fraction of a solid-state emissive guest acceptor — a molecule built around a tetraphenylethylene, or TPE, core — into a state-of-the-art host system built around Y-series non-fullerene acceptors. The guest is chosen so that its localized excited state, the excited state confined to a single molecule, sits energetically close to the charge-transfer manifold at the interface.
That energetic resonance is the key to the whole scheme. When an interfacial charge-transfer exciton forms, it normally faces a choice between radiative decay, which is slow and improbable, and non-radiative decay, which is fast and dominant. But if a localized excited state of the host acceptor lies just slightly above the charge-transfer state, thermal energy at room temperature is enough to push the exciton back up onto that localized state. From there, the excitation can hop onto the emissive guest molecule through resonance energy transfer, the same near-field mechanism that transfers energy between closely spaced fluorescent dyes. The guest, engineered to luminesce strongly even in the solid state, then has a real chance of emitting a photon.
The result is an indirect but effective radiative pathway: charge-transfer excitons that would otherwise be lost as heat are thermally recycled into bright, emissive states. Note the deliberate inversion. In a conventional design, luminescence is something you minimize — every photon emitted is a charge that did not become current. Here, the radiative pathway rescues excitations that were already doomed: an exciton that decays non-radiatively is gone either way, and the electroluminescence measurements show the balance between radiative and non-radiative decay shifting in the cell’s favor. That shift is what shows up as recovered voltage.
The solid-state part of the design matters as much as the energetics. Most organic molecules that glow brightly in dilute solution go dark in the solid state: packed tightly together, their excited states are quenched by the very π-π stacking that makes them good semiconductors. The twisted TPE core frustrates tight stacking, so the guest molecules retain strong emission when blended into the host film — confirmed structurally by single crystals grown to resolve their molecular packing, and by photoluminescence quantum yield measurements inside the blend.
The evidence: spectroscopy and simulation, not just a device curve
A record efficiency alone would not establish that this mechanism is really operating. The paper’s case rests on spectroscopy and simulation, and each addresses a different way the story could be wrong.
Transient spectroscopy — including time-resolved photoluminescence and electroluminescence measurements performed with collaborators at Zhejiang University — traced the flow of excitation from the charge-transfer states to the localized states and finally to the emissive guest. That staged hand-off is the experimental signature of the proposed pathway: if the guest were merely improving morphology or passivating defects, the excitation would not show it. Molecular dynamics simulations, run with the GROMACS package, showed that the energetic resonance is thermally accessible across realistic molecular configurations — not just at one idealized geometry, but across the disorder of a real blend.
The device-level consequences follow from the same physics. Cells containing the emissive guest showed markedly enhanced electroluminescence quantum efficiency, the fraction of injected charges that emerge as photons when the device is driven like a light-emitting diode. A device that emits more efficiently when run in reverse is a device that loses less to non-radiative decay when run as a solar cell. The paper reports that the guest did not compromise the other pillars of performance — charge generation remained efficient, charge transport was preserved, and devices still delivered around 17% in the binary host system.
The full payoff came in the ternary blend, in which the donor polymer, the host acceptor, and the emissive guest acceptor coexist in a single active layer: 21.2% in the laboratory, 20.84% under independent certification, at 0.93 volts open-circuit. The coverage characterizes the certified figure as matching or exceeding the best values reported for any organic photovoltaic architecture to date. Treat that carefully: certified records are a moving target, different architectures are certified under different protocols, and a single certified cell does not settle a “best ever” claim. What is solid is the certified number itself and the direction of the mechanism behind it.
Why this is a third lever, not just a better molecule
For years, the field has pursued two main routes to higher efficiency: controlling the nanoscale morphology of the donor-acceptor blend so that charges separate cleanly, and engineering the frontier orbital energies of the molecules to minimize the energetic offset at the interface, preserving voltage without sacrificing splitting efficiency.
The new work adds a third lever, distinctly photophysical. By deliberately engineering a thermally accessible bridge between dark charge-transfer states and bright localized states, and by supplying a luminescent sink that gives those excitons somewhere to go, the researchers effectively convert a loss channel into a radiative one. The same principle of charge-transfer to localized-state coupling has been discussed theoretically, notably in work on hybridized exciton states, but demonstrating it with a practical guest-acceptor additive in a high-efficiency device is a significant step from theory to technology.
The distinction matters for how the field should read the result. A better molecule is a one-off gain: it improves one system, and the next system starts over. A new lever is, in principle, complementary to the existing ones. If the emissive-guest strategy can be combined with the best morphologies and the best energetics, the gain from reducing a fundamental loss channel should stack with gains from absorbing more light. The improvement comes from loss reduction, not from absorbing more light — and organic photovoltaics need their losses engineered away more than they need another absorber.
What this is not: one system, unverified stability, and a matching problem
The honest limits of the paper are stated in its own coverage, and they should travel with the headline number wherever it goes.
First, the guest acceptor strategy was demonstrated in one particular host system. Extending it to other donor-acceptor pairs will require careful matching of the localized excited state energy to each interface: the localized state has to sit just above the charge-transfer state, close enough for room-temperature thermal energy to bridge the gap, but not so far above that the exciton never climbs it. A laboratory that wants to use this approach cannot simply stir the guest into any blend; it must re-tune the energetics for its own donor and acceptor.
Second, long-term operational stability — always a central concern for organic photovoltaics — will need to be verified for the ternary blends. A three-component active layer adds a new morphology to stabilize: the guest must stay where it is needed, at the right concentration and distribution, under illumination, heat, and time. No stability claim should be attached to this result until operational data exists, and none is reported in the coverage of the paper.
Third, the mechanism itself, involving thermal repopulation of higher-energy states, invites scrutiny of whether the small energetic penalty costs anything in photocurrent at different operating temperatures. A solar cell in the field does not operate at a single tidy room temperature; rooftop and facade modules run hot, and the balance between thermal repopulation and charge extraction could shift with temperature in ways a laboratory measurement at standard conditions does not capture. That is a question for follow-up work, not a flaw in the present result — but it is the kind of question that separates a laboratory mechanism from a deployable one.
Finally, the usual gap between laboratory cell and product applies in full. A certified single-junction cell above 20% is a research milestone, not a module, and the path to a large-area printed module involves yield, uniformity, encapsulation, and lifetime questions that this paper does not address. The perovskite tandem stories this site has covered recently — industrial partnerships aimed at gigawatt-scale coated glass — are a different kind of news: manufacturing bets, with manufacturing risks. This is a photophysics result, with photophysics risks.
What to watch
Generalization to other host systems comes first: the first follow-up papers that apply an emissive guest to a different donor-acceptor pair — and report whether the non-radiative loss falls by a comparable amount — will show whether the lever is general or bespoke. A second system with a similar loss reduction would be stronger evidence than any single record.
Independent reproduction of the loss figure matters more than reproduction of the efficiency. The 0.142-electronvolt non-radiative loss is the mechanistic claim; if other groups reproduce it in their own devices, the mechanism is real in a way that no single laboratory’s device curve can establish alone.
Stability data on ternary blends would convert the stability question from an open worry into a measured quantity — any reported operational lifetime under continuous illumination, elevated temperature, and realistic humidity counts. Until then, stability is unverified, full stop. And temperature-dependent behavior would answer the thermal-repopulation question directly: if the gain holds at module operating temperatures, the approach is more deployable than if it fades when the cell runs hot.
The bottom line
A team in Hong Kong and mainland China has shown that a glowing guest molecule, added to an organic solar cell, can open a radiative escape route for charge-transfer excitons that would otherwise decay as heat — cutting non-radiative voltage loss to as low as 0.142 electronvolts and lifting a single-junction device to 21.2% in the laboratory, 20.84% certified, at 0.93 volts open-circuit. The mechanism is supported by spectroscopy and simulation, not just by a device curve, and it adds a third, photophysical lever to a field that has long relied on morphology and energetics alone.
It is also, at this moment, one host system, with unverified long-term stability and an energy-matching problem to solve for every new donor-acceptor pair. The number to watch is not 21.2% or 20.84%. It is the second system: the first donor-acceptor pair beyond the original host where an emissive guest cuts non-radiative loss by a comparable amount. If that result arrives, luminescence will have earned its place as a design tool in organic photovoltaics.
No investment recommendation is made or implied anywhere in this article.
Sources and further reading
- Zhang, Y., et al. (2026). “Solid-state emissive acceptors enable thermally accessible radiative pathways for efficient organic solar cells.” Nature Photonics. DOI: 10.1038/s41566-026-02009-8 — the primary source for the mechanism and the device figures (21.2% laboratory, 20.84% certified, 0.93 V, 0.142 eV non-radiative loss).
- ScienMag (October 12, 2026), “Glowing Guest Molecules Push Organic Solar Cells Past 20% Efficiency” — coverage of the paper, the source for the team composition (Zhang and Pun, Hong Kong University of Science and Technology; Li, Hong Kong Polytechnic University; collaborators in Shanghai, Beijing, Shenzhen, and Hangzhou), the Zhejiang University spectroscopy collaboration, the GROMACS simulation work, and the stated limits (single host system, stability to be verified, energy matching per pair, temperature questions).
- Excitonic.com, “Excitonic Solar: Can Excitons Make Better Photovoltaics?” — background on exciton binding, bulk heterojunctions, charge-transfer states, and the voltage tax in excitonic solar cells.
- Excitonic.com, “INA Solar’s 3 GW Caelux Deal and the Perovskite-on-Glass Bet” — the recent perovskite tandem coverage this article distinguishes itself from: a manufacturing story, not a photophysics result.
Background on charge-transfer states, non-radiative voltage loss, bulk-heterojunction morphology, and Y-series non-fullerene acceptors is drawn from the standard organic photovoltaics literature. No certified figure beyond the paper’s own reported values is quoted, no certifying laboratory is named (none is named in the available coverage), and no host donor polymer, device area, guest loading fraction, or stability figure is stated, because none is given in the sources consulted.