The price of a small offset
Organic solar cells lose more voltage than silicon or GaAs cells: the loss ΔV = Eg/q − VOC is typically 0.6 V or more, against about 0.3–0.4 V. Part of the reason is that light makes excitons, bound electron–hole pairs, and splitting them takes some energy.
In a donor–acceptor blend that energy comes from two offsets. The LUMO offset drives an electron from an excited donor to the acceptor. The HOMO offset drives a hole from an excited acceptor to the donor. Fullerene cells needed a LUMO offset of at least 0.3 eV. Shrinking the HOMO offset reduces voltage loss, but nothing then stops a hole from going back.
Schematic.
A series that tunes the levels
The donor is the polymer PTO2. The acceptors are ITIC and three derivatives with different side groups: ITCC, ITM and IT4F. Their chemistry is similar, but their energy levels shift step by step. Going from ITCC to IT4F, the LUMO offset grows from about 150 to 360 meV and the HOMO offset from about 40 to 130 meV.
The authors warn that these levels come from cyclic voltammetry on neat films (taken from an earlier study). Such numbers vary between groups by up to a few tenths of an eV, and levels shift in a blend. They should be read as a trend, not as exact values.
Levels from Fig. 1a; offsets are differences of those numbers.
Higher voltage, much lower current
The three small-offset blends give the higher voltages: 1.09–1.19 V, against 0.92 V for PTO2:IT4F. But their short-circuit currents are 30–50% lower, and their efficiencies stay between 7 and 8%. PTO2:IT4F reaches 12.66% with 20.95 mA cm−2.
Weaker long-wavelength absorption explains only about 10–15% of the current difference. The rest, the authors argue, comes from less efficient charge generation. The rest of the paper looks for where those charges are lost.
All values from Fig. 1c.
The LUMO offset sets the electron-transfer speed
A 515 nm pulse excites the donor, and a streak camera records its photoluminescence. In the neat PTO2 film the emission fades slowly. In the blends it is quenched, because the electron moves to the acceptor and forms an interfacial charge-transfer (CT) state.
How fast depends on the LUMO offset. In PTO2:IT4F (~360 meV) the emission disappears faster than the 4 ps time resolution. In PTO2:ITCC (~150 meV) it is still there after 100 ps. PTO2:ITM is the exception: its quenching is as fast as in IT4F, although its offset is only ~230 meV. Transient absorption agrees: the acceptor bleach, a sign of CT states, peaks in under a picosecond in IT4F and ITM blends, and only after more than 30 ps in ITIC and ITCC blends.
Curves schematic, after Fig. 2a; time scales from the text.
Two fixed spectra, free kinetics
A transient-absorption map mixes the signals of excited donors (D*) and of CT or separated charges (CT/CS). To count them, the authors use multivariate curve resolution. The map is written as a sum of two spectra, each multiplied by its own time trace.
Usual global analysis forces the time traces to be exponentials. Here the spectra are fixed instead: the D* spectrum from the neat PTO2 film, and the CT/CS spectrum from the blend signal after 1.5 ns, when no D* is left. Only the time traces are fitted, so they can take any non-exponential shape. The ITM blend gives the highest CT/CS yield and is set to 100%; the other blends are scaled to it.
Schematic of the method; details in the SI.
How many charges survive 2 ns
The CT/CS population peaks after about 2–3 ps in the IT4F and ITM blends, after about 40 ps with ITIC and after about 100 ps with ITCC. The peak heights are 89%, 100%, 59% and 38%.
Then recombination sets in, at very different speeds. At 2 ns, PTO2:IT4F keeps 53% and loses the least. PTO2:ITM falls from 100% to 28%. The 2 ns values track the device currents fairly well. These measurements are at zero electric field. In a working cell the field pulls the charges out, so fewer recombine.
Peak values, peak times and 2 ns values from Table 2; curve shapes schematic, after Fig. 3b.
A small HOMO offset opens a return path
Between 0.1 and 2 ns the charges recombine most slowly in PTO2:IT4F, the blend with the largest HOMO offset. The authors explain this with hole back transfer. With heat, a hole on the donor can hop back to the acceptor. This re-forms the excited acceptor A*, which relaxes to the ground state much faster than a CT state. The charge pair is lost.
With ΔHOMO ≈ 40 meV (ITCC), thermal energy is enough for this step. With ≈ 130 meV (IT4F), the barrier mostly blocks it. Again PTO2:ITM does not fit: it recombines fastest, although its HOMO offset is not the smallest.
Schematic, after Fig. 4 (offsets not to scale). The text also describes the same step as reverse electron transfer from the acceptor to the donor's HOMO.
Holes cross a 40 meV step in under a picosecond
Pumping the acceptors at 670–730 nm starts the process from the other side: the hole must move from the excited acceptor to the donor. The donor bleach appears instantly in every blend, even with ITCC, whose HOMO offset is only ~40 meV. This sub-picosecond step does not depend on the offset. The authors suggest that the hole leaves from an excited state that has not yet relaxed.
Only part of the excitations transfer this fast, though: A* is still present at 2 ns. A slower phase follows, probably from excitons that first have to reach an interface. It takes about 30 ps in PTO2:IT4F and several hundred ps in PTO2:ITCC, so its speed follows the HOMO offset. PTO2:ITM shows no slow phase, perhaps because its charges recombine too fast.
Schematic timeline, after Fig. 5; the time scales are from the text.
About 300 meV and 100 meV
Putting it together: fast and efficient CT formation needs a LUMO offset of about 300 meV. A HOMO offset below about 100 meV still lets holes transfer, but it opens fast recombination by back transfer. Only IT4F clears both in this series, and it gives the most current and the best cell.
What about ITM? Its blend absorption is shifted less, and its emission shows a larger Stokes shift than the others (Fig. 6). The authors read this as a sign that ITM, the more polar molecule, mixes less well with PTO2 and that its LUMO sits lower in the blend than the neat-film value. With this correction it ranks next to IT4F. Because the acceptor gaps widen in the blends, the real LUMO offsets may be smaller than the neat-film values, so the 300 meV figure may be an overestimate. A later study measured the offset in the working device instead.
Offsets from Fig. 1a, 2 ns yields from Table 2; thresholds from the abstract and conclusions. The arrow for ITM is qualitative.