Despite decades of research into new technologies and materials for solar energy conversion, the vast majority of the photovoltaic cells in use are based on crystalline silicon wafers. With few exceptions, other proposed solar-cell designs incorporate rare and costly chemical elements, require expensive fabrication techniques, or convert little of the incident light to electricity. The perfect balance between low cost and high performance has remained elusive.
Solar cells based on a class of materials called organometal halide perovskites have recently and rapidly emerged as one of the most promising contenders yet. Last year, just four years after the cells’ debut, two groups independently reported perovskite solar cells with power-conversion efficiencies of 15%. (Commercial silicon-wafer cells are about 20% efficient.) There’s no obvious reason to suspect that that’s anywhere near their efficiency limit. But much remains unknown about why the perovskites work so well.
Now Yanfa Yan, Wanjian Yin, and Tingting Shi, of the University of Toledo in Ohio, have used density functional theory to study the perovskites’ defect physics.1 They’ve uncovered some unusual properties that may help shed new light on the cells’ performance and how they can be improved.
High efficiency
A perovskite is any material with the formula ABX3, where A and B are cations and X is an anion, and with the cubic crystal structure shown in figure 1a. (The original perovskite, the mineral CaTiO3, was discovered in the Ural Mountains and named for Russian mineralogist Lev Perovski.) In the solar-cell perovskites, A is a polyatomic organic cation, usually methylammonium (CH3NH3+); B is a large atomic ion, usually lead; and X is a halogen—either chlorine, bromine, iodine, or some combination of the three.
Figure 1. Perovskite solar cells. (a) The crystal structure of the iodide perovskite CH3NH3PbI3, is composed of three ions: methylammonium (MA), lead, and iodine. (b) In a perovskite dye-sensitized solar cell, a 2- to 10-nm layer of perovskite (brown) coats a mesoporous film of titanium dioxide. (c) In a simpler thin-film architecture, the perovskite forms a flat film a few hundred nanometers thick. In both cells, sunlight enters through the optically transparent anode and is absorbed by the perovskite. Excited electrons are transported to the anode through the TiO2, and holes are conveyed to the cathode through the hole-transporting layer.
Figure 1. Perovskite solar cells. (a) The crystal structure of the iodide perovskite CH3NH3PbI3, is composed of three ions: methylammonium (MA), lead, and iodine. (b) In a perovskite dye-sensitized solar cell, a 2- to 10-nm layer of perovskite (brown) coats a mesoporous film of titanium dioxide. (c) In a simpler thin-film architecture, the perovskite forms a flat film a few hundred nanometers thick. In both cells, sunlight enters through the optically transparent anode and is absorbed by the perovskite. Excited electrons are transported to the anode through the TiO2, and holes are conveyed to the cathode through the hole-transporting layer.
All of those constituent elements are abundant. And the perovskites can be produced by solution processing, one of the cheapest methods available. Films of the iodide perovskite CH3NH3PbI3, for example, can be made from CH3NH3I and PbI2 dissolved in a common solvent.
The perovskites were originally used as a replacement for the dye in a dye-sensitized solar cell,2 a general solar-cell architecture first described in 1991 by Brian O’Regan and Michael Grätzel.3 As shown in figure 1b, a thin layer of perovskite coats a mesoporous film of titanium dioxide. Absorbed sunlight excites the perovskite’s electrons, which are then injected into the TiO2 conduction band and conveyed to the cell’s anode (made from the optically transparent but electrically conducting fluorine-doped tin oxide). The holes left behind are transported to the cathode.
In the first perovskite solar cells, as in the original dye-sensitized cells, the hole-transporting material was a liquid electrolyte solution; more recent devices employ an organic semiconductor. In July 2013 Grätzel and his group at the Swiss Federal Institute of Technology in Lausanne reported a 15% power-conversion efficiency from an iodide perovskite dye-sensitized cell.4
In that design, the perovskite’s only function is to absorb light and produce charge carriers. The perovskite layer is so thin—between 2 nm and 10 nm—that its charge-transport properties don’t come into play. But in 2012 Henry Snaith and colleagues at Oxford University reported on a cell they’d made in which the mesoporous TiO2 was replaced with mesoporous alumina, an electrical insulator.5 The Al2O3 just served as a scaffold to support the perovskite layer (Snaith and company used the mixed halide perovskite CH3NH3PbIxCl3−x), and electrons had to travel to the anode through the perovskite itself. Not only did the device still work, but its efficiency was slightly improved over the equivalent TiO2 cell.
That success raised the possibility of eliminating the mesoporous layer entirely and using the simpler, potentially cheaper thin-film architecture shown in figure 1c. And Snaith and colleagues did just that: In September 2013 they reported a 15%-efficiency cell with a 330-nm-thick film of mixed-halide perovskite.6 They’d made the film by vapor deposition rather than the cheaper solution processing because of the difficulty in creating a uniform flat film by solution-based methods.7
It’s unusual for a solution-processable material to be able to transport charge carriers more than 10 nm, let alone 330. Grätzel (in collaboration with Tze Chien Sum of Nanyang Technological University in Singapore) and Snaith both investigated the perovskite charge-transport properties more directly; they found transport lengths of about 100 nm for the iodide perovskite and a stunning 1 µm for the mixed halide perovskite.8 What makes the materials so good? And how can the devices be made better?
Point defects
Yan and colleagues had been studying the defect physics of other promising thin-film solar-cell materials, such as cadmium telluride and copper indium gallium selenide. Thin films of those materials are full of point defects that create electron energy levels near the middle of the semiconductor bandgap. Charge carriers that encounter those defects can lose energy, and electrons and holes can recombine, both of which hinder device performance. As Yin explains, “Our previous understanding of inorganic solar-cell materials led us to believe that the halide perovskites must exhibit unusual defect properties.”
There are only so many ways to make a point defect in a crystalline material: A lattice site can be vacant; an extra, or interstitial, ion can be present between lattice sites; or one ion can take the place of another. Focusing on the iodide perovskite (because the locations of I and Cl in the mixed-halide perovskite aren’t completely known), the researchers systematically looked at the 12 possible point defects. For each, they sought to calculate both the defect’s formation energy and the electronic energy levels that it creates.
The formation energy of a defect depends on the chemical potential μ of each of the constituent ions during film growth; the chemical potential depends in turn on the ion’s concentration or partial pressure. Stable growth of the iodide perovskite means that the chemical potentials of its constituent ions add up to the formation energy of the perovskite, so once two of the chemical potentials are specified, the third is known. Furthermore, large swaths of the chemical potential space shown in figure 2 are excluded because another phase—either PbI2 or CH3NH3I—forms before the perovskite does. That leaves a narrow, effectively one-dimensional range of perovskite growth conditions.
Figure 2. Growth of an iodide perovskite film can occur only in a narrow range (shaded in brown) of the chemical potentials μ of the constituent ions. As Yanfa Yan and colleagues found through density-functional calculations, one end of that range produces perovskite with p-type conductivity, and the other end produces n-type conductivity. (Adapted from ref. 1.)
Figure 2. Growth of an iodide perovskite film can occur only in a narrow range (shaded in brown) of the chemical potentials μ of the constituent ions. As Yanfa Yan and colleagues found through density-functional calculations, one end of that range produces perovskite with p-type conductivity, and the other end produces n-type conductivity. (Adapted from ref. 1.)
Yan and colleagues did their defect calculations at several points along that range. They found that wherever they looked, all the readily formed defects—such as Pb vacancies or interstitial methylammonium ions—created states with energies at the edges of the bandgap. Defects that create states near the middle of the bandgap—such as interstitial Pb ions or one ion substituting for another of the opposite charge—all had prohibitively high formation energies.
That’s an unusual coincidence among materials, but it offers an explanation for why the perovskites can conduct so well even when they’re riddled with defects. And based on what’s already known about perovskite conductivity, the result wasn’t unexpected. More surprising, though, was the finding that at one end of the range of formation conditions, all the readily formed defects produced states at the bottom of the bandgap, and at the other end, all the readily formed defects produced states at the top of the bandgap. That is, depending on formation conditions, the perovskite can have either p-type or n-type conductivity. If that variation can be better harnessed, it could pave the way for new device designs.
Weathering the elements
As perovskites continue on their meteoric rise, several challenges remain. Foremost among them is the stability of the cells. Good solar cells need to last for decades, even as they’re exposed to harsh weather conditions. But the perovskites sublimate at relatively low temperatures and dissolve in water. Although the cells are sealed against rainfall, they can leak. That not only damages the cell but introduces toxic lead into the environment. Some researchers are looking for similarly effective materials that replace the lead with the less harmful tin.