Date:
Perovskite solar cells have spent fifteen years setting efficiency records that silicon cannot match. They have also spent those fifteen years failing to reach a factory floor, and the reason is not the perovskite. It is the liquid you have to dissolve it in.
A team at the University of Surrey has now built working solar mini-modules from an ink that swaps the industry’s standard toxic solvents for ethanol and a bio-derived alternative, coated at room temperature and dried with a jet of gas. The modules reached power conversion efficiencies above 18%, and the details of how they got there are more interesting than the number itself.
The results were published in Nano Energy as Solvent systems with significantly reduced toxicity for high-efficiency perovskite solar modules, and the study involved researchers from the University of Oxford, the University of Cambridge, the Helmholtz-Zentrum Berlin, the MAX IV Laboratory in Sweden and the National Physical Laboratory.
Table of Contents
Perovskite is genuinely remarkable at absorbing light. It can also be dissolved into an ink and printed onto flexible surfaces, which is the entire commercial promise: solar cells made the way newspapers are printed, at speed, on rolls rather than in batches.
That promise runs straight into a chemistry problem. The solvents that dissolve perovskite precursors well enough to make high-quality films are, almost without exception, hazardous. The Surrey paper names the main offenders: dimethylformamide (DMF), N-methyl-2-pyrrolidone (NMP), and acetonitrile (ACN). Several are classified as reproductive toxins.
On a laboratory bench, a researcher in gloves and a fume hood can handle those chemicals. On a gigawatt-scale coating line, the same chemicals become four separate industrial problems at once:
The scale at which this bites is easy to underestimate. According to the Surrey release, producing a gigawatt of electricity from these perovskite modules can consume more than 90,000 liters of solvent on a single coating line. That is a chemical plant’s worth of hazardous liquid flowing through a process that is supposed to be cheaper than silicon. The University of Surrey’s announcement of the work is the source for that figure.
This is not a new argument, and it is not a niche one. A 2021 study in Solar RRL estimated that the added cost of managing DMF-based inks specifically works out to about 3.7 US cents per watt. That figure is from a separate modelling paper, not from the Surrey work, and it is a useful yardstick for how much of a solar panel’s cost structure can be consumed by a solvent choice.
Silva’s framing of the prize is deliberately modest. “The potential impact of our innovation extends beyond simply replacing a hazardous solvent,” he said. The process, he added, “points to more compact, resource-efficient, and potentially lower-cost manufacturing for emerging solar technologies.” That is a direction of travel, not a price.
Their formulation is called a significantly reduced-toxicity (SRT) solvent system. The main components are ethanol and 2-methyltetrahydrofuran, the latter derived from renewable sources such as agricultural and forestry waste. The published abstract adds a detail the press release only gestures at: a minimal fraction of dimethyl sulfoxide (DMSO) is included, and the release describes it as a small amount of a milder additive to control how the film forms.
Three consequences follow. First, no DMF, NMP, or ACN is required. Second, the ink is coated at room temperature and dried with a jet of gas rather than baked, which cuts both energy input and process time. Third, the antisolvent drip is eliminated entirely, because the SRT ink crystallizes under mild conditions without one.
The Surrey press release quotes the lead researcher on why the process and the chemistry were designed together. “For a technology to move from the laboratory into a real manufacturing environment, we need to consider not only performance, but also the solvents, processing conditions, safety, resource use and overall complexity of production,” said Dr. Ehsan Rezaee, a research fellow at Surrey.
The process was designed as a single continuous route from agricultural and forestry waste to a bio-derived ink, through room-temperature coating, to a laser-scribed module.
The team fabricated monolithically interconnected mini-modules with active areas up to 11.5 square centimeters, using a laser-based P1 to P4 scribing process to carve the interconnection pattern between cells. The published abstract reports power conversion efficiencies exceeding 18%. The university’s release puts the same result at above 18% and reports it as 18.2%.
Both figures describe active area only. That qualifier matters: efficiency measured over a module’s active area excludes the dead space taken up by interconnects, busbars, and edge margins, so it is always flattering next to aperture-area efficiency, which is what a panel datasheet would eventually quote. A single mini-module of this size is also not a panel. It is a test structure, and the gap between a printed mini-module and a commercially packaged panel includes encapsulation, larger-area uniformity, and years of weathering that no laboratory test has yet supplied.
The second thing to hold onto is that this number is not a record, and it is not certified. It is the authors’ own measurement of their own devices.
Placing 18% in context requires care, because the headline perovskite numbers that circulate widely are a different kind of measurement entirely.
Those two certified records describe cells smaller than a fingernail. The Surrey result describes a printed module roughly the size of a postage stamp and a half. Comparing them directly is not fair to either, but the comparison is not meant to be fair. It is meant to show that perovskite’s famous efficiency figures and perovskite’s manufacturability problem are being solved on different tracks, at different scales, by different kinds of measurement. A technology can hold a certification record and still have no route to a coating line.
The SRT inks in this study were formulated for methylammonium lead iodide, or MAPbI3. That is worth noting plainly, because MAPbI3 is not the composition the field has gravitated toward for stability. The perovskite community has largely shifted to formamidinium and caesium-containing mixtures, often abbreviated FA/Cs, which tolerate heat and humidity considerably better than the methylammonium variant. MAPbI3 was the workhorse of early perovskite research and remains easy to crystallize from solution. A later line of research attacked the lead itself by building hybrid organic-inorganic structures that removed it, arguing that the absorber’s composition is as much a commercial constraint as its efficiency.
That makes it a sensible choice for a solvent demonstration, and it means the solvent result is best read as a statement about processing rather than about durability. Whether the same reduced-toxicity approach transfers to more stable absorber compositions is a separate question that this paper does not answer.
The claim about durability in the Surrey release is a single sentence: the modules stayed stable under continuous light during testing. No test duration is given. No irradiance level is given. No temperature or humidity condition is specified, and no standard test protocol is named. Whatever exposure the devices survived, the public material does not quantify it.
That is a real limit on how much the result can be read as a durability story. Perovskite modules are judged on operational lifetime above almost everything else, and a qualitative statement about continuous illumination cannot be compared against the published degradation figures that competing research groups report. Readers should treat the durability point as encouraging and unquantified rather than as evidence of a working lifetime.
It is also worth separating this result from the device-level stability problems that have dominated perovskite research for years, including the mechanical and chemical failure of interfaces between the layers of a cell stack. Work on toughening those internal interfaces has shown that a layer stack can hold far longer simply by stopping the weakest junction from cracking. This paper’s contribution is upstream of all of that: it addresses what goes into the coating line, not what happens to the finished panel after twenty years outdoors.
The architecture is a specific choice rather than a generic cell. The modules use an inverted, or p-i-n, device structure with an interlayer-free design, where the hole transport layer is a self-assembled monolayer applied by bar coating rather than a separately deposited layer.
Removing that interlayer matters for manufacturing simplicity, because every additional coating step means another deposition station, another solvent, another drying stage, and another chance for a defect. The SRT ink sits on top of a bar-coated monolayer and crystallizes under mild thermal conditions after deposition, with the gas jet driving solvent removal. The abstract notes compatibility with both rigid and flexible substrates.
The coating method itself is described as room-temperature, gas-blowing-assisted bar coating, which is one step short of full roll-to-roll production. Bar coating can run continuously and is compatible with roll-to-roll lines, but a laboratory bar coater is not a manufacturing line. The release describes the process as compatible with roll-to-roll manufacturing rather than as having been validated on production equipment, and that distinction is worth preserving.
The honest summary is that this paper does not produce a better solar panel. It produces a plausibly printable one, using chemicals a factory might actually be allowed to use at scale.
That is a bigger deal than it sounds. For years, the perovskite conversation has been dominated by the efficiency leaderboard, where progress is measured in tenths of a percent on millimeter-scale devices. The commercial blocker has sat elsewhere the whole time: in the solvent inventory, the waste contract, the extraction system, and the antisolvent drip that nobody puts on a conference slide.
Surrey’s contribution is to demonstrate that the whole stack can be printed from an ink that avoids DMF, NMP, and ACN at room temperature, without an antisolvent, at a scale where interconnection losses become real. The efficiency is unremarkable next to certified records. The process is the point.
What remains genuinely open is everything that sits between a printed mini-module and a manufactured panel: quantified stability under standard test conditions, performance at larger widths, transfer to more stable absorber compositions, and whether any of this reaches a production line. The Surrey work answers none of those questions, and it does not claim to.
Solvent chemistry was never going to be the thing that made perovskite famous. It may well be the thing that makes it shippable.
A perovskite solar cell uses a crystalline material with a distinctive atomic structure to absorb light and generate electricity. The name describes a crystal structure rather than one specific compound, and the most studied versions for solar applications contain lead, including the methylammonium lead iodide used in the Surrey study.
Not in the sense of a consumer product. Perovskite photovoltaics are moving toward commercial deployment through tandem designs layered onto silicon, but neither the Surrey mini-modules nor the certified laboratory records described above are products. This paper provides no timeline, and none should be inferred from it.
Because they are different measurements. Certified cell records are taken on devices well under a square centimeter, where losses from interconnects and inactive area are minimal. A module adds up many cells connected in series, and every interconnect, busbar, and edge margin takes a bite out of the available area, so module efficiency always trails cell efficiency for the same material. Add the fact that the certified records are independently measured while the module figure is self-reported, and the comparison becomes even less direct.
It is one of the most concrete ones. Large volumes of DMF, NMP, or acetonitrile bring worker safety obligations, hazardous waste handling, and additional air extraction capacity, all of which are costs and regulatory burdens that arrive before commercial production begins. A separate 2021 assessment in Solar RRL estimated an added cost of roughly 3.7 US cents per watt from managing DMF-based inks alone, which gives a sense of why solvent choice shows up in the economics.
Many perovskite recipes force crystallization by dripping a second solvent onto the wet film while it spins or coats. It works well in a laboratory and is awkward on a continuous line because it is hard to apply evenly across a wide web, and the antisolvents involved are often toxic themselves. Removing that step simplifies the process and cuts a hazardous chemical from the recipe.
This study does not say, and no availability date should be attached to it. The paper describes a process demonstration at a mini-module scale. The distance from that to a commercial panel includes quantified durability testing, larger-area manufacturing trials, and the transfer to more stable absorber compositions, none of which appear in the published result.
Share post:
Popular

source