Bend a good acceptor into a ring
Ring-shaped conjugated molecules are stiff, cannot fold up, and keep their π-electrons on a closed loop. Such rings have been made for decades, but most are built from one kind of unit, so their colour is hard to tune and they absorb only a narrow slice of sunlight. Solar cells made with donor–acceptor rings had reached only 3.6%.
The authors start from a molecule that already works well. BTP-C3 is an A-DA′D-A acceptor: electron-poor end groups (A), electron-rich fused units (D) and an electron-poor core (A′). Two BTP-C3 molecules and two thiophenes (a small donor unit) are joined end to end into one ring, the "ring-like conjugated molecule" RCM.
Schematic, after Fig. 1. The block drawing shows the sequence of units, not the real shape.
A template that holds the halves together
Closing a large ring is usually the hard step. The two ends must meet at the right angle, and the open chains tend to react with other molecules instead, so yields are often poor.
Here the two BTP-C3 halves are first tied together at their cores by a flexible alkyl chain. This dimer, DBTP-C3, is the template. A palladium-catalysed Stille coupling then attaches two thiophenes and closes the ring in one pot, with a yield above 50%. DBTP-C3 is also kept as a reference: it has both halves but no ring.
Schematic, after Fig. 1; yield from the text (Fig. S2).
A narrower gap, and a new blue band
In chloroform all three molecules absorb strongly between 600 and 800 nm, peaking at 727 nm (RCM), 718 nm (DBTP-C3) and 712 nm (BTP-C3). In films the peaks shift further red, to 795, 763 and 753 nm. The absorption onset of the RCM film gives an optical gap of 1.33 eV, against 1.51 eV for the two open molecules.
RCM also has a band between 400 and 550 nm that neither open molecule has. It is very broad, about 0.8 eV, and the authors assign it to higher-energy states delocalised around the ring. The red shift from solution to film, and absorption that changes with the solvent, point to strong aggregation.
Spectra schematic, after Fig. 2a; peak positions from the text, gaps from Table 1.
Rings that stack into crystals
X-ray scattering at grazing incidence (GIWAXS) shows how the molecules pack. Films of BTP-C3 and DBTP-C3 give one π–π stacking peak at qz = 1.65 Å−1 out of plane, typical of face-on stacking.
RCM gives two π–π peaks, at 1.30 and 1.63 Å−1, and a strong (111) reflection near 0.53 Å−1. In plane there are arcs near 0.65 and 1.3 Å−1. Together with electron diffraction (SAED), the authors infer a crystalline film with layered, herringbone-like packing and an extra T-shaped arrangement in plane.
Line cuts schematic, after Fig. 2e and Figs. S28–S29; peak positions from the text. Packing sketch illustrative.
No melting below 350 °C
Thermogravimetry heats a sample and weighs it. BTP-C3 and DBTP-C3 lose 5% of their weight at 330.3 °C; RCM holds out to 393.8 °C. In calorimetry the open molecules melt, at 232.4 °C (BTP-C3) and 286.2 °C (DBTP-C3), while RCM shows no melting or crystallisation between 30 and 350 °C.
The difference carries over to devices. Inverted solar cells kept at 80 °C for 570 hours retain 92% of their starting efficiency with RCM, 76% with DBTP-C3 and 61% with BTP-C3.
Weight-loss curves schematic, after Fig. 2d; temperatures and retention values from the text (Fig. 3c).
A stiff ring wastes less energy as heat
An excited molecule can give its energy back as light or lose it as vibrations. The photoluminescence quantum yield ΦPL measures the fraction that comes out as light. In solution it is 11.1% for RCM, against 3.8% and 3.9% for the open molecules, and the RCM emission lives close to 1 ns. In films it is 4.1% against 1.2% and 1.7%.
Cooling a film usually makes it brighter, because heat-activated loss channels close. All three films brighten at low temperature, but RCM much less. Since its end groups are tied into the ring and neat RCM does not split excitons, the authors attribute the weak temperature dependence to the rigid ring. (A later macrocycle study takes this idea further.)
ΦPL from Table 1. Temperature curves schematic, after Fig. 3e (shape only, no values).
More current at almost the same voltage
Each acceptor is blended with the donor polymer PM6. All three cells give an open-circuit voltage near 1 V. The RCM cell gives much more current, 23.5 mA cm−2 against 17.4 and 17.6 mA cm−2, and reaches 14.2% (13.7 ± 0.2% on average) against 12.2% and 11.5%. Its fill factor is the lowest of the three.
The external quantum efficiency (EQE) shows where the extra current comes from. The RCM cell converts 70–80% of photons over a broad range from 400 to 800 nm. The open molecules stay around 65% at most, between 500 and 780 nm. The electron mobility in the RCM blend, 2.17 × 10−4 cm2 V−1 s−1, is also the highest (1.20 × 10−4 and 7.13 × 10−5 for the others).
J–V curves drawn from a simple model matched to the reported VOC, JSC and FF, not the measured curves. EQE schematic, after Fig. 3b.
Ten times brighter, 60 mV less lost
The voltage loss is the optical gap of the cell (taken from its EQE) minus qVOC. It is 0.50 V for RCM and 0.56 V for the other two. Radiative recombination below the gap costs all three about 0.03 V.
The non-radiative part follows from how well the cell works as a light-emitting diode:
ΔVnr = (kBT/q) · ln(1/EQEEL)
The RCM cell has EQEEL = 4 × 10−4, an order of magnitude above the others (3–4 × 10−5), so ΔVnr drops from 0.26 to 0.20 V. The electroluminescence of the blend overlaps that of neat RCM, which suggests that the charge-transfer state mixes strongly with the RCM excited state. A brighter acceptor then makes a brighter, lower-loss cell. (More on this trade-off in the voltage-loss explainer.)
Values from Table 2. The grey "rest" is the voltage loss minus the radiative and non-radiative parts, computed here.
Lower loss, without slower charge transfer
A worry with low-loss blends is that charge generation suffers. Transient absorption says it does not here. Right after excitation at 700 nm, mostly absorbed by the acceptor, the PM6:RCM spectrum looks like excited RCM; after 1–6 ns it shows features of both PM6 and RCM: the hole has moved to PM6 and the electron stays on RCM.
Fitting the rise of the charge signal gives a separation time of 2.1 ± 0.1 ps with RCM, 2.7 ± 0.1 ps with BTP-C3 and 5.3 ± 0.2 ps with DBTP-C3. Meanwhile the excited state of neat RCM lives longer (PL lifetime 410 ps, against 300 and 200 ps), which leaves more time to reach an interface.
In the blend RCM stacks more tightly, 3.65 Å against 3.73 and 3.72 Å. The PM6 lamellar peak at 0.35 Å−1 disappears with RCM, a sign that donor and acceptor mix well, and microscopy shows a fibril network. The authors conclude that combining a donor–acceptor sequence with a closed ring is a route to stable, low-loss acceptors.
Rise curves drawn as single exponentials with the fitted times, after Fig. 3f (decay omitted). Lifetimes and spacings from the text (Figs. S42, 4, Table S3).