A new way to control the growth of crystalline materials called perovskites could lead to commercial solar cells that hit a sweet spot of high performance and low cost. Although individual perovskite cells have achieved promising results in the lab, until now it hasn’t been clear how they might be made in uniform batches.

Certain perovskites can harvest the energy of sunlight very efficiently because they strongly absorb both visible and infrared light. And unlike silicon films, which are made at high temperatures, perovskite films can be made from solution at much lower temperatures. It should be possible to make perovskite solar cells using low-cost, low-energy methods such as printing.

The first perovskite cells were made in 2009, but the best can already convert 17.9 percent of the energy in sunlight into electricity. That’s starting to be competitive with commercial thin-film cells like cadmium telluride and silicon, says Timothy Kelly, a chemist at the University of Saskatchewan in Saskatoon, Canada.

However, it has proven difficult to make high quality perovskite solar cells consistently. In the batches made so far there is a wide variation in how effectively individual cells can convert light into electricity. “When you make 10 different perovskite cells, you get 10 different efficiencies,” says Prashant Kamat, a chemist at the University of Notre Dame in Indiana. “It’s frustrating.”

The problem is caused by variation in the size of the crystals in different cells. To electrons in a solar cell, the boundaries between crystals are like walls, so larger crystals offer fewer impediments to the flow of electricity. New research published today in the journal Nature Nanotechnology could provide a way to solve that issue by showing how to control the growth of perovskite crystals.

The perovskite being developed for solar cells has an ingredient list that includes a hydrocarbon, ammonia, lead, and iodine. There are many perovskites—the name refers to the crystal structure of these materials—but this particular one is most promising for use in solar cells. The crystals are made in a two-step process that begins with coating a surface with lead iodide solution and allowing it to dry out. Then the surface is coated with a solution of methyl ammonium iodide. As that dries out, compounds from the two layers come together to form perovskite crystals.

Michael Grätzel, a chemist at the École Polytechnique Fédérale de Lausanne, Switzerland, and Nam-Gyu Park, a chemist at Sungkyunkwan University in Korea, have now worked out a recipe for taking control of that process. They found that by carefully controlling the concentrations of the starting solutions, and other processing conditions, they could consistently make perovskite films with the larger crystals needed for an efficient solar cell.

The Swiss and Korean groups used these methods to make perovskite solar cells with an average efficiency of 16.4 percent, and very little variation in efficiency between different cells.

Park says that now that it’s possible to make high-quality perovskite reliably, it’s time to deal with other problems with the material. One is that humidity causes the materials to break down and leak methyl ammonium. Park says that researchers either need to find a way to seal perovskite solar cells against humidity or find new versions of the materials. Another problem is that the materials are made using lead, which is toxic.

“Having learned from these materials, we should move to others, because lead is not environmentally benign, and this material is not stable,” says Mercouri Kanatzidis, a chemist at Northwestern University in Illinois. He and Northwestern materials scientist Robert Chang have been developing a lead-free perovskite that substitutes tin. It currently only converts light into electrical power with an efficiency of 6 percent. But they’re both optimistic, pointing to how the lead-based materials improved rapidly from about 3 percent in 2009 to about 18 percent today.

Meantime, Grätzel believes that the existing materials haven’t hit their upper performance limits yet. “I think 20 percent efficiency should be possible in the near term,” he says.