Aggregation and morphology control enables multiple cases of high-efficiency polymer solar cells

Yuhang Liu, Jingbo Zhao, Zhengke Li, Cheng Mu, Wei Ma, Huawei Hu, Kui Jiang, Haoran Lin, Harald Ade, He Yan•View original

OverviewBalancedalloy voice

Here’s the puzzle that had polymer solar cells stuck. They were getting better every year, but the best devices kept relying on the same small cast of donor polymers paired with one favorite fullerene, PC71BM. Step outside that comfort zone, and performance fell off. When trying to make the active layer thicker—which is necessary for real manufacturing—many of those polymers, especially the PTB7 family, would stumble. Holes moved too slowly. Domains got messy. Efficiency sagged once the film crept past a couple hundred nanometers. Liu and colleagues approached that bottleneck from an angle you can almost feel in your fingertips: side chains. Not the flashy part of the polymer, but the handles that determine how chains approach, stick, and let go. They placed a branched two-octyldodecyl group—people shorthand it as 2OD—at the second carbon along a quaterthiophene backbone. That tiny relocation makes a big difference. It causes the polymer’s aggregation to depend smoothly on temperature. Warm the solution and chains stay apart; let it cool during casting and they gather, crystallize, and sort into domains with a kind of unhurried precision. The payoff is thick films, roughly 250 to 300 nanometers, that still deliver high performance. In their best case, the power conversion efficiency hit ten point eight percent with a fill factor at seventy-seven percent. That’s the headline. The subtext is just as important: the result didn’t hinge on one magical acceptor. It traveled. Why does the branching position matter so much? Liu’s team built near twins to show the contrast. Shift the branch to the first carbon—call that 1ON—and the polymer refuses to aggregate at either eighty-five degrees Celsius or room temperature. It’s too sterically hindered to line up and crystallize during film formation. Devices made from it limped in around zero point six percent. Move the branch to the third carbon—3OT—and you get the opposite problem. Interchain attraction is so strong that the solution can gel before you even cast the film. In other words, too loose and nothing organizes; too tight and you can’t process it. The 2OD position hits the “just right” middle ground where aggregation turns on with cooling and can be steered by the casting conditions. You can see that switch flip in the absorption spectrum. At elevated temperature, the polymer in solution looks disaggregated—broad features, nothing dramatic. Drop it to room temperature and a strong so-called zero-zero transition pops up around 700 nanometers, a telltale of ordered stacks forming in solution. What’s striking is that the room-temperature solution spectrum looks almost like the final solid film when things are optimized. It’s like peeking at the blueprint before the concrete sets. Slow down the spin casting or heat the substrate too much, and that peak weakens and shifts to a longer wavelength, signaling disorder. The process knob is real, and you can dial it the wrong way. Zoom in with X-rays and the story gets even clearer. Using grazing-incidence wide-angle X-ray scattering—GIWAXS for short—you’re looking for how far and how coherently the pi-stacking extends. In a PffBT4T-2OD blend with a C60-type fullerene, the coherence length along the pi-stacking direction, the so-called (010) direction, stretches to about 8.5 nanometers. Think of that as roughly two dozen polymer backbones stacked with consistent spacing. A standard PTB7:PC61BM reference only gets you about 2 nanometers. That’s night and day. The 2OD system also prefers a “face-on” orientation at the interface, which is favorable for shuttling charges vertically through the film. When they built hole-only devices—just to isolate how well holes move—they measured mobilities on the order of one point five to three point zero times ten to the minus two square centimeters per volt-second. That’s the kind of transport that supports those meaty fill factors. Now, organization isn’t just about order; it’s also about boundaries. Resonant soft X-ray scattering—R-SoXS—picks up the size and purity of donor and acceptor domains. Across multiple blends, the domains cluster in the 30 to 40 nanometer range. That’s small enough for excitons, the bound electron-hole pairs, to find an interface before they die, but big enough to avoid excessive recombination at too many boundaries. Purity matters, too. The polymer-rich regions are very pure—about ninety percent of the asymptotic purity signal even before annealing. And when they did push the blends with a long anneal, an example with a fullerene called ICMA still had only about three point two percent fullerene left trapped in the polymer domains. You want donors and acceptors mostly in their own neighborhoods, and this system gets you there without sacrificing crystallinity. Break the recipe, and you see how fragile the chain can be if you lose the aggregation control. When they changed the casting so the polymer didn’t pre-organize—slower spins, hotter substrates—the coherence along the pi stacks fell, the absorption signature drifted, and the hole mobility dropped almost an order of magnitude, down to roughly three point one times ten to the minus three. The device performance followed it off a cliff, down to around three point six percent. That’s a sharp reminder: the molecular design opens the door, but the processing walks you through it. Put the structure and morphology together, and the devices make sense. With active layers around a quarter to a third of a micron thick, PffBT4T-2OD paired with a variety of fullerenes consistently hit efficiencies between nine and eleven percent. Fill factors sit in the low- to mid-seventies, sometimes nudging into the high-seventies, and the open-circuit voltage lives around three-quarters of a volt. One representative cell—PffBT4T-2OD blended with a C70 derivative called TC71BM—delivered about zero point seventy-seven volts, a short-circuit current near eighteen point eight milliamps per square centimeter, a zero point seventy-five fill factor, and that ten point eight percent efficiency. The same polymer with the classic PC71BM or with PC61-type fullerenes lands in the same neighborhood. The numbers vary a touch with each acceptor, but the shape of the performance stays the same. Here’s the part that breaks the old mold. The morphology that gives you those results doesn’t seem fussy about which fullerene you pick. R-SoXS and atomic force microscopy keep pointing to that same small-domain, high-purity landscape whether the acceptor is PC61BM, PC71BM, ICMA, or others. GIWAXS shows the polymer keeping its crystalline posture. Even when they stress the film with heat, the polymer tends to hold its order while the fullerene migrates into larger pools, leaving the polymer domains even cleaner. That’s valuable because it decouples your donor design from a single “blessed” acceptor and allows you to explore energy levels and spectra without resetting the entire morphology each time. The generality doesn’t stop with one backbone either. Liu and co-workers made two more donors—PBTff4T-2OD and PNT4T-2OD—that share the same 2OD motif. They behave like members of a family. Cast them warm, let aggregation guide the solidification, and you get the familiar recipe: domain sizes in the few tens of nanometers, strong pi-stacking, and thick-film devices that consistently clear nine percent with fill factors crowding the mid- to upper-seventies. The take-home is not “this one polymer is great.” It’s “this side-chain strategy travels.” They also tested the edges of that strategy. Lengthening the side chain to a 2DT variant—longer branches that add bulk but not absorption—and things start to drift. The domains swell to around one hundred nanometers, lamellar stacking weakens, and the film’s absorption drops simply because those bulky side chains don’t help capture light. The purity of the polymer domains dips as well, from around ninety percent in the 2OD systems to roughly eighty-seven percent. That’s still decent, but you’re giving back some of the hard-won order. In short, the sweet spot is not just “branched” but “branched here, and not too long.” It’s worth pausing on what “warm casting” really means. They’re not baking the film after the fact; they’re tuning aggregation in the liquid before and during deposition. Solutions held in roughly the sixty to eighty degrees Celsius window keep the chains separate, and as the solvent thins and the film cools, those chains find one another, stack, and phase-separate in a controlled, time-ordered way. Do it right and you lock in a morphology that handles thickness gracefully. In fact, when they compared thick to thin devices, the thicker ones picked up extra external quantum efficiency—on the order of ten to twenty percent more signal—and the spectrum’s leading edge shifted red by about twenty nanometers. That combination fed a big gain in current, roughly a thirty percent bump in their comparison, without torpedoing the fill factor. That’s exactly the kind of behavior you want when you think about scalable coating methods. If we pull the threads together, a tidy chain emerges. Start with a quaterthiophene backbone and give it second-position two-octyldodecyl side chains. That architecture makes aggregation a gentle function of temperature. Use warm-solution casting to set the initial conditions, and let the film cool so the polymer pre-stacks and crystallizes as the domains form. The resulting morphology shows small, well-separated donor and acceptor regions with high purity, strong face-on pi-stacking, and coherence lengths that suggest long, orderly stacks. Those structures support hole mobilities on the order of ten to the minus two and enable fill factors of up to about seventy-seven percent. Send light in, and even in films a quarter of a micron thick, you still pull out currents in the upper teens of milliamps per square centimeter and efficiencies around ten percent. Swap the fullerene within reason, and you keep most of that goodness. Swap the donor for one that preserves the 2OD trick, and most of it comes along too. A lot of polymer solar cell work over the last decade has felt like walking a tightrope between chemistry and processing. Change the donor slightly and the morphology collapses; change the acceptor and the whole landscape rearranges. What Liu’s study shows is a way to widen that tightrope into a boardwalk. By encoding a temperature-tunable aggregation response into the side chains, they made the film formation itself do more of the morphological thinking for you. The measurements line up—ultraviolet-visible fingerprints in solution, GIWAXS for crystallinity and orientation, R-SoXS for domain size and purity, transport tests for mobility—and the devices close the loop. Could this approach reach beyond fullerenes to newer acceptors? The paper doesn’t push there, so that remains a short, tantalizing question mark. What it does provide is a robust design rule—second-position two-octyldodecyl side chains on a quaterthiophene core, leveraged by warm casting—that already spans three donors and more than ten donor-fullerene pairings. That’s enough of a platform to think seriously about scale: slot-die coaters, blade coaters, processes that prefer thicker, more forgiving films. And it’s enough to break the old habit of designing around one blessed acceptor. In a field that has too often been about perfect couples, this is a morphology that plays well with others.

When trying to make the active layer thicker—which is necessary for real manufacturing—many of those polymers, especially the PTB7 family, would stumble. Holes moved too slowly. Domains got messy. Efficiency sagged once the film crept past a couple hundred nanometers.

Liu and colleagues approached that bottleneck from an angle you can almost feel in your fingertips: side chains. Not the flashy part of the polymer, but the handles that determine how chains approach, stick, and let go. They placed a branched two-octyldodecyl group—people shorthand it as 2OD—at the second carbon along a quaterthiophene backbone.

That tiny relocation makes a big difference. It causes the polymer’s aggregation to depend smoothly on temperature. Warm the solution and chains stay apart; let it cool during casting and they gather, crystallize, and sort into domains with a kind of unhurried precision.

The payoff is thick films, roughly 250 to 300 nanometers, that still deliver high performance. In their best case, the power conversion efficiency hit ten point eight percent with a fill factor at seventy-seven percent. That’s the headline.

The subtext is just as important: the result didn’t hinge on one magical acceptor. It traveled.

Why does the branching position matter so much? Liu’s team built near twins to show the contrast. Shift the branch to the first carbon—call that 1ON—and the polymer refuses to aggregate at either eighty-five degrees Celsius or room temperature.

It’s too sterically hindered to line up and crystallize during film formation. Devices made from it limped in around zero point six percent. Move the branch to the third carbon—3OT—and you get the opposite problem.

Interchain attraction is so strong that the solution can gel before you even cast the film. In other words, too loose and nothing organizes; too tight and you can’t process it. The 2OD position hits the “just right” middle ground where aggregation turns on with cooling and can be steered by the casting conditions.

You can see that switch flip in the absorption spectrum. At elevated temperature, the polymer in solution looks disaggregated—broad features, nothing dramatic. Drop it to room temperature and a strong so-called zero-zero transition pops up around 700 nanometers, a telltale of ordered stacks forming in solution.

What’s striking is that the room-temperature solution spectrum looks almost like the final solid film when things are optimized. It’s like peeking at the blueprint before the concrete sets. Slow down the spin casting or heat the substrate too much, and that peak weakens and shifts to a longer wavelength, signaling disorder. The process knob is real, and you can dial it the wrong way.

Zoom in with X-rays and the story gets even clearer. Using grazing-incidence wide-angle X-ray scattering—GIWAXS for short—you’re looking for how far and how coherently the pi-stacking extends. In a PffBT4T-2OD blend with a C60-type fullerene, the coherence length along the pi-stacking direction, the so-called (010) direction, stretches to about 8.5 nanometers.

Think of that as roughly two dozen polymer backbones stacked with consistent spacing. A standard PTB7:PC61BM reference only gets you about 2 nanometers. That’s night and day.

The 2OD system also prefers a “face-on” orientation at the interface, which is favorable for shuttling charges vertically through the film. When they built hole-only devices—just to isolate how well holes move—they measured mobilities on the order of one point five to three point zero times ten to the minus two square centimeters per volt-second. That’s the kind of transport that supports those meaty fill factors.

Now, organization isn’t just about order; it’s also about boundaries. Resonant soft X-ray scattering—R-SoXS—picks up the size and purity of donor and acceptor domains. Across multiple blends, the domains cluster in the 30 to 40 nanometer range.

That’s small enough for excitons, the bound electron-hole pairs, to find an interface before they die, but big enough to avoid excessive recombination at too many boundaries. Purity matters, too. The polymer-rich regions are very pure—about ninety percent of the asymptotic purity signal even before annealing.

And when they did push the blends with a long anneal, an example with a fullerene called ICMA still had only about three point two percent fullerene left trapped in the polymer domains. You want donors and acceptors mostly in their own neighborhoods, and this system gets you there without sacrificing crystallinity.

Break the recipe, and you see how fragile the chain can be if you lose the aggregation control. When they changed the casting so the polymer didn’t pre-organize—slower spins, hotter substrates—the coherence along the pi stacks fell, the absorption signature drifted, and the hole mobility dropped almost an order of magnitude, down to roughly three point one times ten to the minus three. The device performance followed it off a cliff, down to around three point six percent.

That’s a sharp reminder: the molecular design opens the door, but the processing walks you through it.

Put the structure and morphology together, and the devices make sense. With active layers around a quarter to a third of a micron thick, PffBT4T-2OD paired with a variety of fullerenes consistently hit efficiencies between nine and eleven percent. Fill factors sit in the low- to mid-seventies, sometimes nudging into the high-seventies, and the open-circuit voltage lives around three-quarters of a volt.

One representative cell—PffBT4T-2OD blended with a C70 derivative called TC71BM—delivered about zero point seventy-seven volts, a short-circuit current near eighteen point eight milliamps per square centimeter, a zero point seventy-five fill factor, and that ten point eight percent efficiency. The same polymer with the classic PC71BM or with PC61-type fullerenes lands in the same neighborhood. The numbers vary a touch with each acceptor, but the shape of the performance stays the same.

Here’s the part that breaks the old mold. The morphology that gives you those results doesn’t seem fussy about which fullerene you pick. R-SoXS and atomic force microscopy keep pointing to that same small-domain, high-purity landscape whether the acceptor is PC61BM, PC71BM, ICMA, or others.

GIWAXS shows the polymer keeping its crystalline posture. Even when they stress the film with heat, the polymer tends to hold its order while the fullerene migrates into larger pools, leaving the polymer domains even cleaner. That’s valuable because it decouples your donor design from a single “blessed” acceptor and allows you to explore energy levels and spectra without resetting the entire morphology each time.

The generality doesn’t stop with one backbone either. Liu and co-workers made two more donors—PBTff4T-2OD and PNT4T-2OD—that share the same 2OD motif. They behave like members of a family.

Cast them warm, let aggregation guide the solidification, and you get the familiar recipe: domain sizes in the few tens of nanometers, strong pi-stacking, and thick-film devices that consistently clear nine percent with fill factors crowding the mid- to upper-seventies. The take-home is not “this one polymer is great.” It’s “this side-chain strategy travels.”

They also tested the edges of that strategy. Lengthening the side chain to a 2DT variant—longer branches that add bulk but not absorption—and things start to drift. The domains swell to around one hundred nanometers, lamellar stacking weakens, and the film’s absorption drops simply because those bulky side chains don’t help capture light.

The purity of the polymer domains dips as well, from around ninety percent in the 2OD systems to roughly eighty-seven percent. That’s still decent, but you’re giving back some of the hard-won order. In short, the sweet spot is not just “branched” but “branched here, and not too long.”

It’s worth pausing on what “warm casting” really means. They’re not baking the film after the fact; they’re tuning aggregation in the liquid before and during deposition. Solutions held in roughly the sixty to eighty degrees Celsius window keep the chains separate, and as the solvent thins and the film cools, those chains find one another, stack, and phase-separate in a controlled, time-ordered way.

Do it right and you lock in a morphology that handles thickness gracefully. In fact, when they compared thick to thin devices, the thicker ones picked up extra external quantum efficiency—on the order of ten to twenty percent more signal—and the spectrum’s leading edge shifted red by about twenty nanometers. That combination fed a big gain in current, roughly a thirty percent bump in their comparison, without torpedoing the fill factor.

That’s exactly the kind of behavior you want when you think about scalable coating methods.

If we pull the threads together, a tidy chain emerges. Start with a quaterthiophene backbone and give it second-position two-octyldodecyl side chains. That architecture makes aggregation a gentle function of temperature.

Use warm-solution casting to set the initial conditions, and let the film cool so the polymer pre-stacks and crystallizes as the domains form. The resulting morphology shows small, well-separated donor and acceptor regions with high purity, strong face-on pi-stacking, and coherence lengths that suggest long, orderly stacks. Those structures support hole mobilities on the order of ten to the minus two and enable fill factors of up to about seventy-seven percent.

Send light in, and even in films a quarter of a micron thick, you still pull out currents in the upper teens of milliamps per square centimeter and efficiencies around ten percent. Swap the fullerene within reason, and you keep most of that goodness. Swap the donor for one that preserves the 2OD trick, and most of it comes along too.

A lot of polymer solar cell work over the last decade has felt like walking a tightrope between chemistry and processing. Change the donor slightly and the morphology collapses; change the acceptor and the whole landscape rearranges. What Liu’s study shows is a way to widen that tightrope into a boardwalk.

By encoding a temperature-tunable aggregation response into the side chains, they made the film formation itself do more of the morphological thinking for you. The measurements line up—ultraviolet-visible fingerprints in solution, GIWAXS for crystallinity and orientation, R-SoXS for domain size and purity, transport tests for mobility—and the devices close the loop.

Could this approach reach beyond fullerenes to newer acceptors? The paper doesn’t push there, so that remains a short, tantalizing question mark. What it does provide is a robust design rule—second-position two-octyldodecyl side chains on a quaterthiophene core, leveraged by warm casting—that already spans three donors and more than ten donor-fullerene pairings.

That’s enough of a platform to think seriously about scale: slot-die coaters, blade coaters, processes that prefer thicker, more forgiving films. And it’s enough to break the old habit of designing around one blessed acceptor. In a field that has too often been about perfect couples, this is a morphology that plays well with others.

Related lectures