Countless rows of solar modules sit shimmering in the desert, a dark blue ocean stretching to the horizon. This is the Tengger Desert Solar Park in central China, by some estimates the world’s largest photovoltaic array, capable of generating up to 1.5 gigawatts of power. No wonder it’s been dubbed “the Great Wall of Solar Energy.”

Arrays like this are a testament to the remarkable progress of solar photovoltaics (PVs). Global installed capacity is expected to top 400 GW this year—at least 40 times as high as a decade ago—and generate almost 2% of our electricity.

About 90% of these PV modules use the semiconductor silicon to absorb light and convert its energy into an electric current. Most are made in China, where manufacturers have leveraged government subsidies and economies of scale to drive down PV costs until they rival fossil-fuel sources. In a world facing dramatic climate change, the technology offers a glimmer of hope for the future.

But some energy road maps predict that in order to meet emissions targets set out in the Paris Agreement, we will need PV arrays to generate at least 20% of global electricity by 2050. Many researchers believe that silicon cells are not a sustainable way to achieve that step change in capacity. Although commercial modules convert about 20% of the solar energy that falls on them into electricity, silicon is actually pretty poor at capturing sunlight. To achieve that efficiency, the cells need bulky crystalline layers more than 100 µm thick, the production of which consumes a lot of materials and energy.

So the hunt is on for a PV technology that can match the performance of silicon cells but do it with much thinner films of cheap, abundant, and nontoxic elements. Those films should also be simple to manufacture with solution-based methods that are amenable to high-throughput roll-to-roll processing. “We have to have a scalable technology,” says David J. Fermin, a solar PV researcher at the University of Bristol.

Many PV researchers have been pinning their hopes on perovskites—typically metal-organic halides—to meet the challenge. Perovskite cells have made huge progress in the lab, with record efficiencies rising from 4% to 22% in less than a decade. But most perovskites contain lead, raising environmental concerns, and questions linger over the materials’ long-term stability.

So some scientists have turned to a lesser-known family of materials called kesterites, which are very stable and based on commonplace elements such as copper, zinc, tin, and sulfur (Cu₂ZnSnS₄, often abbreviated as CZTS). Despite their promise, kesterites have received less attention than perovskites, and the efficiency of the best kesterite-based cells has plateaued at around 13% for the past five years—frustratingly short of the 20% or so that they need to be commercially viable.

Now, multi-million-dollar research programs are gearing up to rescue kesterites from the doldrums. The largest, a European academic-industry consortium called Starcell, aims to solve the problems that have dogged kesterites and push their efficiency up to 18% by 2020. Another, based in Australia, is targeting 20% efficiency by 2022.

These are ambitious projects, and some PV researchers are frankly pessimistic about the chances of success. They’re also a huge gamble. The $7.6 million Starcell consortium, which kicked off last year, includes the leading kesterite research teams in Europe, Japan, and the U.S. If even they cannot deliver a breakthrough, it may spell the end for kesterite research. “This is a risk,” concedes Edgardo Saucedo, Starcell’s project coordinator, who is based at the Catalonia Institute for Energy Research. “If we don’t succeed in the next couple of years, interest in kesterites will decrease.”

Danger: Low voltage

Kesterites are not the only thin-film PV technology, of course. Cells made from cadmium telluride and copper indium gallium selenide (CIGS) have reached about 23% efficiency in the lab, and commercial module manufacturing is growing at a healthy rate. They offer clear environmental benefits because thin-film PV modules can “pay back” the energy needed to make them more quickly than bulky silicon cells can. Lightweight and potentially flexible, thin-film modules can also be integrated into buildings, an application for solar panels that is expected to grow rapidly in the coming years.

But cadmium is toxic, and tellurium and indium are relatively scarce. The manufacturing of CdTe and CIGS cells has geopolitical drawbacks, too: “For Europe to sustain production of these two technologies is quite complicated,” Saucedo says. “Tellurium, indium, and gallium are mostly produced in other countries, sometimes unstable countries.”

And despite the excitement around perovskites, Aron Walsh at Imperial College London, who works on Starcell, argues that perovskites have overshadowed alternative technologies (ACS Energy Lett. 2017, DOI: 10.1021/acsenergylett.7b00131). “Other interesting systems are being neglected,” he says. Perovskites have racked up five times as many publications as kesterites, and Walsh says it’s time to redress the imbalance.

Kesterite is a greenish-black mineral that was first plucked from the ground in the 1950s by Russian mineralogists surveying the frozen badlands of Siberia. Fortunately, the synthetic kesterites in PV cells are easier to obtain. Researchers combine simple ingredients in solution and then spin coat the liquid onto a base. Heating it to a few hundred degrees Celsius, a process called annealing, creates a thin crystalline film that’s roughly 1 µm thick.

Between 1996 and 2008, the efficiency of kesterite cells rose from less than 1% to about 7%—enough progress to interest David Mitzi, who then worked at IBM’s Thomas J. Watson Research Center and is now at Duke University. By 2013, he had made a kesterite cell containing selenium in addition to the usual copper, zinc, tin, and sulfur (abbreviated CZTSSe) that boasted a record efficiency of 12.6% (Adv. Energy Mater. 2013, DOI: 10.1002/aenm.201301465).

Although rumors abound of a 13.8% cell from a South Korean group, that work remains unpublished, and Mitzi’s record still represents the high-water mark for kesterites. As the efficiency of kesterites has remained stuck around 12–13%, frustration in the community has grown. “The big question is, what is actually limiting the efficiency?” Walsh says.

PV researchers calculate the maximum power output of a cell by multiplying three variables: its peak voltage, peak current, and a “fill factor” that rates how well the cell compromises between these two extremes. Poor performance in any one of these measures can kill a cell’s efficiency, but the biggest problem for kesterites is voltage, which currently stands at a dismal 60% of the theoretical maximum. “There’s a lot of debate about why the voltage is so low,” the University of Bristol’s Fermin says. The prime suspects, though, are defects within the kesterite (Adv. Energy Mater. 2017, DOI: 10.1002/aenm.201602366). With four or five elements in the material being jiggled around by high-temperature annealing, “the main challenge is to make sure that the atoms all go in the right places,” Fermin explains. “That has a profound impact on performance.”

A zoo of defects

Defects can wreak havoc on the processes that generate current and voltage in a PV cell. When a photon with enough energy hits a PV semiconductor, it boosts an electron from the low-energy valence band up to the high-energy conduction band, which leaves behind a positively charged “hole.” The number of charges mobilized in this way determines the current, while the size of the energy jump—the band gap—affects the voltage. Defects in the material’s crystal structure introduce additional energy levels between the two bands. Some sit close to the edges of the bands, which may effectively reduce the gap in between like silt narrowing a river. Others are smack in the middle, providing a way station where electrons and holes can recombine, wasting their energy.

Kesterites have a tetragonal crystal structure, with copper, zinc, and tin ions occupying specific sites in the crystal lattice. But copper and zinc have almost identical ionic radii, and both have tetrahedral coordination, so it’s all too easy for them to swap places, upsetting the electronic properties of the material. These copper-zinc swaps can disrupt up to 30% of the sites in the kesterite lattice. “That’s a huge number of defects,” Mitzi says. Other kinds of defects, while less numerous, could make performance even worse. “There’s likely a zoo of different defects,” Mitzi says, “all of which play a role in lowering performance.”

Mitzi is working with Starcell on a promising way to tackle these defects. For the past couple of years, his team has been making kesterite cells that use barium in place of zinc (Chem. Mater. 2016, DOI: 10.1021/acs.chemmater.6b01832). Barium has twice zinc’s ionic radius and tends to adopt eightfold coordination rather than fourfold, which makes it harder for barium to swap places with copper or tin. Solar cells made from barium copper tin sulfoselenide do indeed feature far fewer of these defects, although cell efficiencies have yet to top 5%. “It’s still pretty low,” Mitzi acknowledges, “but it is at an early stage of development.”

Starcell is also building on discoveries from another kesterite project called PVTEAM, based in the U.K. and led by Fermin. The four-year project, which wraps up in June, has focused on some of the fundamental properties of kesterites. For example, the group used high-resolution transmission electron microscopy to map individual defects in kesterites, finding that they occur in clusters (Nanoscale 2016, DOI: 10.1039/c6nr04185j). “We’ve been able to map these with exquisite atomic resolution,” Fermin says. Such results can feed into computational studies that predict the formation energy of defects, highlighting where problems are likely to appear and offering potential solutions.

Take a hammer to it

Defects are not the only bugbears for kesterites. The cells typically use a very thin layer of cadmium sulfide to collect freed electrons from one face of the kesterite, and they use molybdenum to gather holes from the opposite face. But molybdenum can react with kesterites during annealing to form molybdenum sulfide, volatile tin sulfide, and other unwanted compounds. MoS₂ is an insulator that increases the cell’s resistance and lowers its efficiency, and when tin sulfide evaporates, it throws the kesterite’s composition out of whack, leaving behind physical voids.

A team led by Richard Haight at IBM has come up with an ingenious work-around (Nat. Energy 2017, DOI: 10.1038/s41560-017-0028-5). After forming a CZTSSe film, the researchers replaced some of the selenide ions by baking it with sulfur. Increasing the proportion of sulfide widens the material’s band gap, enabling the cell to produce a higher voltage. Then they took off the back contact—including the molybdenum substrate and any MoS₂ that had formed—using a highly technical process that they describe as “an impulsive force supplied by [a] hammer.” In its place, they laid down a new back contact to produce a cell with an efficiency of 11.89% and an open-circuit voltage of 670 mV, which was higher than Mitzi’s record-breaking cell. “It was a big jump forward,” Haight says. “No other cell had shown that much voltage at that efficiency.”

Xiaojing Hao of the University of New South Wales has a different approach to the problem. She and her team put alumina between molybdenum and kesterite to stop the chemical reaction that generates MoS₂; this alumina then cracks up during annealing, allowing a decent electrical contact between the two layers. The process allowed the researchers to build a relatively large cell of 1 cm²—more than twice the area of Mitzi’s champion cell—with an independently certified efficiency of 7.6%, a record for a CZTS cell of that size (NPG Asia Mater. 2017, DOI: 10.1038/am.2017.103). “The magic of alumina is that it can be sputtered to form a layer only a few nanometers thick,” Hao says. Since alumina sputtering is suitable for production-line manufacturing, “this offers a realistic potential for kesterites to be commercialized in a short time.”

Further refinements have since pushed the cell’s certified efficiency up to 10%, says Hao, who has just won a major grant to turn her work into a commercially viable technology. The four-year, $3.6 million project, which starts in March, aims to produce CZTSSe cells with 20% efficiency. Achieving that, though, may require a root-and-branch redesign of the kesterite cell.

High hopes

Most kesterite cells still rely on processing methods and a device architecture that they inherited from CIGS cells, including those layers of molybdenum and CdS. As part of a broader effort to overhaul the cell’s design, Starcell plans to replace the CdS layer with a more efficient cadmium-free alternative. The project is also creating larger, less flawed crystals of kesterite that can enable charge to flow more freely. In October 2017, Starcell researchers reported that adding a dash of germanium to a CZTSe cell helped keep ions nicely mixed during annealing (Energy Environ. Sci., DOI: 10.1039/c7ee02318a). The upshot was big, beautiful kesterite crystals that were several micrometers wide—and a cell with a respectable 11.8% efficiency.

The challenge will be to combine the best of all these improvements into a single cell with record-breaking performance. There’s no doubt that will be tough, Starcell’s Saucedo says. But he points out that CdTe and CIGS cells were both stuck at the same efficiencies for years before a few crucial breakthroughs broke the deadlock. Two ingredients were vital, Saucedo says: money and people. “They needed that critical mass of researchers to discover the key steps,” he says. Saucedo believes that a focused, international consortium like Starcell could achieve the same for kesterites. “If we deliver one key step in the next two years, we will have succeeded,” he says.

“I wish them luck,” says Susanne Siebentritt at the University of Luxembourg, who is on Starcell’s scientific advisory panel. But she adds that the project’s 18% efficiency target is a tall order. “If they have a good idea, it’s feasible—but it’ll take a really good idea,” she says. “I stopped making kesterite cells a few years ago because I got quite pessimistic about it.”

Haight is more optimistic, but he doesn’t expect the breakthrough to come from his group at IBM. “We’re not doing much solar-cell work at IBM anymore,” he says. “Enthusiasm was dampened by the difficulty in raising the efficiency.”

Indeed, Mitzi says there are no big projects in the U.S. looking at kesterites. “Funding agencies here have been more reluctant to take big steps in creating huge consortia to address this problem,” he says.

That stands in contrast with the European Union, which recently agreed to a legally binding target of generating at least 27% of its total energy demand from renewables by 2030. Because the EU cannot compete with Asia in making vast quantities of silicon ingots, Saucedo says, kesterites might offer a way for the bloc’s PV industry to help meet that target.

So will kesterites prove to be a dead end, or do they have what it takes to save the day? In the next few years, researchers could settle that question, one way or the other. “I can see that there are some very significant challenges,” Fermin says. “But there isn’t any fundamental reason why it can’t happen.”