Sitting in a dimly lit bar in Japan, then graduate student Michael Lee was scribbling on a beer coaster as night fell, jotting down a list of chemical ingredients before he forgot them. Earlier that day scientists at Toin University of Yokohama had generously shared their groundbreaking recipe for making solar cells from a new material called perovskite rather than the usual silicon. The cells were only 3.8 percent efficient in converting sunlight to electricity, so the world had not taken notice. But Lee was inspired. After the 2011 fact-finding mission, he returned to Clarendon Laboratory at the University of Oxford, where all three of us worked at the time, and made a series of tweaks to the recipe. The changes yielded the first perovskite cell to surpass 10 percent efficiency. His invention sparked the clean-energy equivalent of an oil rush, as researchers worldwide raced to push perovskite cells even higher.

The latest record, set at 20.1 percent by the Korea Research Institute of Chemical Technology in November 2014, marked a fivefold increase in efficiency in just three years. For comparison, after decades of development state-of-the-art silicon solar cells have plateaued at about 25 percent, a target that perovskite researchers like us have squarely in our sights. We are also anticipating a commercial debut, perhaps through a spin-off company such as Oxford Photovoltaics, which one of us (Snaith) co-founded.

Perovskites are tantalizing for several reasons. The ingredients are abundant, and researchers can combine them easily and inexpensively, at low temperature, into thin films that have a highly crystalline structure similar to that achieved in silicon wafers after costly, high-temperature processing. Rolls of perovskite film that are thin and flexible, instead of thick and rigid like silicon wafers, could one day be rapidly spooled from a special printer to make lightweight, bendable, and even colorful solar sheets and coatings.

Still, to challenge silicon's dominance, perovskite cells will have to overcome some significant hurdles. The prototypes today are only as large as a fingernail; researchers have to find ways to make them much bigger if the technology is to compete with silicon panels. They also have to greatly improve the safety and long-term stability of the cells—an uphill battle.

Winning the efficiency race

Today the best silicon cells are 25.6 percent efficient. Why can't solar cells convert 100 percent of the sun's light energy? And why should perovskites be able to surpass the silicon record?

The answers to these questions are found in the excitable and errant electron. When a solar cell is in the dark, electrons in the material stay bound to their respective atoms. No electricity flows. But when sunlight strikes a cell, it can liberate some of the electrons. Infused with energy, the “excited” electrons careen drunkenly through the crystal lattice of the cell until they either exit one end of the cell—whisked away by an electrode as useful current—or run into an obstacle or a trap, losing their energy in the form of waste heat.

The higher the crystal quality, the fewer defects there are to derail the electron's journey. Silicon cells are typically heated to as much as 900 degrees Celsius to remove defects. Perovskites are largely free of such defects even though they are processed at much lower temperatures, around 100 degrees C. As a result, electrons excited by light are just as successful in exiting perovskite cells, and they are unlikely to lose as much energy along the way when colliding with obstacles. Because the electrical power of a cell is the product of the flow of electrons exiting that cell (the current) and the energy that those electrons carry (the voltage), the efficiency of perovskites can rival silicon, with much less processing effort.

But there is a ceiling to how much of the sunlight's energy a solar cell made of semiconductors such as silicon and perovskites can convert into electrical power. That is primarily because of a property of semiconductors called the bandgap—a minimum level of energy needed to liberate electrons. Sunlight includes all wavelengths of light, but only certain wavelengths exceed the energy bandgap. Other wavelengths will simply pass through the material, doing nothing.

The bandgap is different for different semiconductors, and it sets up a fundamental trade-off: the lower the bandgap, the more of the sun's spectrum a cell can absorb to excite electrons, but the lower the energy each electron will have. Because electrical power depends on both the number and energy of electrons, even a cell with the ideal bandgap can convert only around 33 percent of the sun's energy.

Silicon has a fixed bandgap that is not ideal, but it commands the solar industry because effective ways to manufacture the technology are well understood. When making perovskites, however, researchers can adjust the bandgap at will by tweaking the mix of ingredients, which raises the prospect of exceeding silicon efficiencies. Researchers can also layer different perovskites with different bandgaps on top of one another. Double-decker perovskites should be able to break through the nominal 33 percent ceiling; some projections indicate they could put 46 percent of the sun's energy to work.

Teaching an old material new tricks

Mineralogists have known about the natural forms of perovskite in the earth's crust since the 19th century. The crystals graced a 1988 cover of this magazine when scientists thought they could form high-temperature superconductors (some work continues today). During the past two decades engineers also made experimental electronics with man-made perovskites, but they overlooked the material's potential use in solar cells.

Finally, in 2009, a group at Toin University turned a man-made version—a lead halide perovskite first synthesized in 1978—into a solar cell. The researchers dissolved selected chemicals in solution, then spun and dried that solution on a glass slide. The drying left behind a film of nanometer-scale perovskite crystals on top of the slide, much the way salt crystals emerge from evaporating tidal pools. This film generated electrons when it absorbed sunlight but not very well. The researchers added thin layers of material on either side of the perovskite nanocrystals to help them transfer the electrons to an external electrical circuit, supplying useful power.

The first tiny cells were only 3.8 percent efficient, and they were highly unstable, deteriorating within hours. Lee altered the perovskite's composition and replaced a problematic layer in the cell, pushing the efficiency beyond 10 percent. Another set of investigators, led jointly by Michael Grätzel of the Swiss Federal Institute of Technology in Lausanne and Nam-Gyu Park of Sungkyunkwan University in Korea, made a similar advance.

The recent march to 20 percent has been driven by some clever innovations. Creating a defect-free crystalline film requires tricky deposition methods, so a group headed by Sang Il Seok of the Korea Research Institute of Chemical Technology devised a multistep process that forced a more orderly crystal film to drop out of the spinning solution. By optimizing processing, Seok marched through three consecutive record efficiencies in 2014, from 16.2 to 20.1 percent.

Other scientists simplified the layering of added materials; the newest perovskite cells look more like a silicon cell—a simple stack of flat layers. In silicon's case, this design has made low-cost mass production possible. Recently perovskite researchers have also heated up the solution and the glass slide on which it is deposited, resulting in crystals that are several orders of magnitude bigger than those in the initial cells, an encouraging sign that the crystallinity is still improving.

Scientists are devising some novel traits, too. Varying the chemical ratio can create cells that have a gentle shade of yellow or a blush of crimson. Depositing perovskite on glass in islands instead of one thin layer can create films that are opaque or transparent or degrees in between. Together these options—refreshing choices over rigid, opaque, blue-black silicon cells—could help architects design skylights, windows and building facades that incorporate colorful perovskite solar films. Imagine a skyscraper with perovskite-tinted windows that shade the interior from hot sunlight by converting it into electricity, reducing the cooling bill while also providing power.

Long road to commercialization

Perovskites have a long way to go before they fulfill such visions. Although Korean and Australian researchers recently demonstrated printable cells that are 10 by 10 centimeters—large enough for commercially competitive products—the most efficient cells are still small prototypes. As labs and start-up companies scale up the devices, they must accomplish three prerequisites for commercialization: ensure that the cells are stable enough to produce electricity for decades, design a product that customers feel is safe to put in their homes and buildings, and satisfy critics who caution that the claims for perovskite efficiency levels are inflated.

The stability of the perovskite solar cell is arguably its Achilles' heel. Perovskites can degrade rapidly because they are sensitive to moisture, so they must be encased in a watertight seal. Cells fabricated by us in an inert atmosphere and encapsulated in epoxy have performed stably for more than 1,000 hours when exposed continuously to light. Researchers at the Huazhong University of Science and Technology in China, in collaboration with Grätzel, have also reached 1,000 hours even without encapsulation, and in recently published work they have deployed test panels outdoors in Saudi Arabia to show that their design will function in real-world conditions. At a recent Materials Research Society meeting in San Francisco, we disclosed results from Oxford Photovoltaics that demonstrate that perovskite cells can generate stable power output for more than 2,000 hours under full sunlight.

The industry convention for solar panels is a 25-year warranty, however. That equates to about 54,000 hours under constant, bright sunlight. Finding an effective moisture barrier that works for that long, over a wide temperature range, is crucial. Silicon manufacturers solved the problem by laminating the cells between glass sheets. This is perfect for large, ground-based installations. But because perovskite cells can be made as films that are much lighter and more flexible than cells on glass, alternative encapsulation strategies may open up broader applications, such as veneers for walls or windows that can generate electricity.

Fortunately, some progress has been made by companies trying to commercialize other flexible solar materials, such as the semiconductor made of copper indium gallium selenide. The encapsulation technologies work well, yet businesses have struggled to gain market share from silicon because the cells are less efficient and cost more. Perovskites, which should have higher efficiencies and lower processing costs, may be able to exploit the encapsulation advances.

Just as important as sealing out moisture is sealing in the cells' contents because of the tiny amount of lead added to the perovskite recipe. Lead is toxic, so the market will demand a high burden of proof that perovskite power is safe. For inspiration, researchers can again look to an alternative solar material, the only one besides silicon that has achieved significant commercial success: cadmium telluride.

Manufactured by First Solar, cadmium telluride panels have been deployed around the world and have exceeded safety standards despite the presence of an element far more toxic than lead: cadmium. First Solar has convinced communities that its panels are so well sealed that no cadmium could escape, even in a desert wildfire at 1,000 degrees C. The panels use a glass substrate, however, which precludes the flexibility and lower weight that perovskites promise. Yet perovskite companies can learn from First Solar's success in sealing and rigorously testing products.

An encouraging development related to lead recently emerged from the Massachusetts Institute of Technology as well: Angela Belcher and her colleagues demonstrated that lead-acid car batteries can be recycled safely, with the lead content recovered to make perovskite cells. This result could be an environmental plus. Belcher estimates that the lead in a single car battery could enable production of around 700 square meters of perovskite cells, which at 20 percent efficiency would be enough to power 30 houses in a warm but sunny climate such as that in Las Vegas.

A different route would be to eliminate the lead altogether. Both our group and another one at Northwestern University have published preliminary reports on cells that use tin instead of lead. The efficiency and stability are worse, however, because tin tends to cause the perovskite to lose its crystalline structure over time, hampering an electron's ability to get out of the cells. A major advance would be needed for tin to match lead's long-term performance.

In addition to the issues listed here, researchers have to solve a smaller, quirkier problem. Critics have claimed that the efficiency numbers for perovskite cells might be inflated because of hysteresis—a jitter in the measurement that is likely caused by charged molecules migrating from one side of a cell to the other, which could create the appearance of greater current. This ion migration is very brief, however. Scientists are looking for ways to halt it, but in the near term, there is a simple remedy: wait out the migration and measure efficiency over a longer period. In most cases, this process renders efficiency readings that are similar to quick, initial measurements, but researchers may be tempted to report the higher of the readings. We are working with investigators worldwide to standardize the measurement process so that our results meet a high standard of scrutiny.

Finally, to succeed commercially, perovskite innovators need to provide a compelling economic narrative to attract the investment dollars required for scaling up production. Although materials for perovskites are abundant and cells can be processed at low temperatures into films that roll off inexpensive equipment, perovskite solar companies should not fall into the trap of competing on silicon's terms. There is little room to undercut silicon panels because most of the cost of an installation is not related to the panels but to what is called the “balance of system,” which includes installation materials and labor, permits and inspections, and other expenses related to system installation. An average U.S. residential solar installation in 2014 was priced at $3.48 per watt of electricity-generating capacity, yet the cost of the actual solar panel was only 72 cents per watt. Even if perovskite panels achieve the dirt-cheap 10 to 20 cents per watt that researchers think is possible, the improvement would reduce the final installed price by only a small percentage.

Perovskite companies can build on those small savings, though, by devising products that beat silicon's efficiencies. A highly efficient perovskite solar panel reduces the total installed cost per watt by requiring less land or roof space and therefore less labor and equipment. An even more imaginative example of changing the rules would be to sell perovskite products for applications that silicon cannot compete in, such as films that could be integrated right into building materials for walls, roofs and windows.

The hybrid solution

For now perovskites might have the best chance to reach the market as an ally rather than a competitor of silicon. Perovskites could literally piggyback off silicon's success, gaining entry to a $50-billion market.

An alliance could happen by adding a perovskite layer right on top of a silicon layer, creating a “tandem” solar cell. Perovskites are good at harnessing the higher-energy colors of sunlight, such as blue and ultraviolet, which silicon fails to capture, generating a much higher voltage in electrons. Researchers at Stanford University and M.I.T. recently stacked a perovskite cell on top of a sealed silicon cell, raising efficiency from the silicon's original 11 to 17 percent. They also assembled a tandem cell by layering perovskite on top of unsealed silicon, creating a single structure. The combination achieved just 14 percent efficiency, but that figure could surely go up with manufacturing refinements. Based on the two experiments, the researchers sketched out a scenario by which a tandem cell made with a state-of-the-art silicon component and a state-of-the-art perovskite device, combined using clever engineering, could surpass 30 percent efficiency without any radical change in either technology.

If a tandem solar panel could reach 30 percent efficiency, the impact on the balance-of-system cost could be enormous: only two thirds of the number of panels would be needed to produce the same amount of power as panels that are 20 percent efficient, greatly reducing the amount of roof space or land, installation materials, labor and equipment. Oxford Photovoltaics, Snaith's spin-off company, is partnering with traditional silicon manufacturers to boost silicon's efficiency with a perovskite coating over the silicon cell; the company is targeting prototypes of the tandem cells this year. Down the line, cheap solar coatings integrated into roofing or glazing materials could transform the entire cost structure of a solar-powered building.

Running in reverse

The quick rise of perovskite solar cells has inspired scientists and engineers to fabricate other types of prototype products that also might one day make it to market. Working with our colleagues at the University of Cambridge, we recently created light-emitting diodes (LEDs) and lasers using metal halide perovskites, which efficiently emit light (instead of absorbing it) through a process called luminescence.

This turnabout is not really surprising; when run in reverse, the world's most efficient solar cell, gallium arsenide, acts as an LED. Cheap, printable LEDs and lasers could lead to intriguing applications, from large-scale lighting to medical imaging.

Research into these novel products is very early, of course, but we think the work will become more popular. Perovskites make scientists feel like children in a candy shop; we have found a material whose properties fill almost every checkoff box on our wish list, including high efficiency, low cost, light weight, flexibility and aesthetic appeal. It will take a concerted, global effort by academia, industry and government to fully realize the potential perovskites have to move beyond the silicon era. But given the prize—cheap, clean energy and the next generation of electronics—we think perovskites are a good bet.