Light that never comes out
A solar cell run backwards is a light-emitting diode. Detailed balance says the ideal cell would turn every injected charge pair back into a photon. Every pair that instead decays silently, as heat, lowers the open-circuit voltage:
Each factor of ten lost in emission costs about 60 mV at room temperature. In these devices non-radiative recombination is the largest part of the total voltage loss, and its ceiling is set by how well the low-gap acceptor itself emits: its photoluminescence quantum yield, or PLQY.
A ladder of vibrations
Draw the ground and excited states as two parabolas along a molecular coordinate. Each holds a ladder of vibrational levels spaced by the mean high-frequency vibrational quantum hν̄. The relaxed excited state can drop without emitting by landing on a ground-state level of the same energy; how fast depends on how much the two vibrational wavefunctions overlap.
If the vibrations are softer, hν̄ is smaller, the matching level sits higher up the ladder (j = 4 instead of j = 3 in the paper's sketch), the overlap shrinks, and the heat channel slows. The paper writes this as PLQY rising with exp[(ΔE − λS)/hν̄].
Schematic, after Fig. 3g.
Linear acceptors glow weakly
The building block is BTP-C3, a linear A–DA′D–A acceptor of the Y-series family: a fused electron-rich core flanked by two electron-poor end groups. Low-gap linear acceptors like this typically have PLQYs below 5%.
Strategies to raise PLQY, such as adding rigid units or bulky side chains, usually change how the molecules aggregate in the film, and charge transport suffers. The goal is brightness without that trade-off.
Drawing is schematic; see Fig. 1 of the paper for the chemical structures.
Two backbones, one strained ring
In the ring-like conjugated macrocycle (RCM), two BTP-C3 backbones are joined end to end by thiophene bridges, and their central triazole units are tied together across the cavity by an alkyl strap. The strap is six, five or four carbons long, which gives RCM-C6, RCM-C5 and RCM-C4.
The ring is shape-persistent and strained: the conjugated backbone can no longer relax freely. RCM-C6 is the macrocycle the group introduced in Angewandte Chemie 2023, with an electroluminescence efficiency of 4 × 10−4.
Schematic, after Fig. 1a.
Raman shows softer vibrations
Raman spectra probe exactly the backbone vibrations that matter. In the C=C stretching region (1350–1700 cm−1) the macrocycle spectra are redshifted relative to linear BTP-C3, with more weight in the lower-frequency modes. The average high-frequency vibration is lower, which is the softening sketched in step 2.
Solution spectra agree: RCM-C5 and RCM-C4 show a smaller Stokes shift and narrower emission than RCM-C6, signs of a stiffer structure that relaxes less after excitation.
Spectra are schematic, after Figs. 3c and 1d.
Brighter films, smaller voltage loss
Neat films of RCM-C4 and RCM-C6 reach PLQYs of 14.5% and 13.7%, roughly three times the typical value for linear acceptors. RCM-C5 is lower, at about 10% in Fig. 3b.
In devices with the donor PM6, the non-radiative voltage loss ΔEnr is 0.21 eV (RCM-C6), 0.24 eV (RCM-C5) and 0.22 eV (RCM-C4), against 0.26 eV for a linear BTP-C3 device. The ring itself is the main win; within the series the order is not strictly monotonic.
A shorter strap tames aggregation
RCM-C6 emits well but packs badly. Its two backbones sit at 95.8° to each other, and in neat films it forms large crystalline domains with 10.3 nm surface roughness. In blends this excessive aggregation drives recombination and limits fill factor.
Shortening the strap pulls the triazoles together: the cavity shrinks from 8.3 to 7.5 to 6.3 Å, the backbones become more planar, and a controlled twist in the thiophene bridges keeps aggregation in check. The film absorption peak moves from 804 to 796 to 786 nm, a sign of weaker aggregation. RCM-C4 films form an ordered fibrous network with only 1.42 nm roughness.
Ring geometry and film textures are schematic, after Figs. 1b and 4a.
17.1%, the record for macrocycles
With PM6, efficiency rises from 14.55% (RCM-C6) to 16.51% (RCM-C5) to 17.10% (RCM-C4), driven by fill factor (65.6 → 72.1 → 73.9%) and current, with Voc near 0.97 V throughout. At publication 17.1% was the highest efficiency for a macrocycle-based cell.
RCM-C4 also has the longest excited-state lifetime in films, 0.75 ns against 0.57 ns for RCM-C6, giving excitons more time to reach the donor interface, plus more balanced electron and hole mobilities. Its devices kept about 100% of their efficiency after 285 h at 80 °C, where L8-BO and BTP-C3 devices fell to about 80%.
Stiffness first, then packing
The recipe has two steps. First, lock the conjugated backbone into a ring so its vibrations soften and the excitation has fewer ways to turn into heat; that raises PLQY and lowers the voltage tax. Second, tune the geometry of the lock (here the strap length) so the stiff rings do not over-aggregate and charges can still move.
RCM-C6 shows that the first step alone is not enough. RCM-C4 shows that both can be had in one molecule.