by Emma Foehringer Merchant
November 09, 2020

This story is the fourth installment in a GTM Squared series considering what the solar industry will look like in 2030. 


Silicon has been the semiconductor of choice in solar cells for decades. But it’s not all that great at doing the job.

Even under ideal conditions, scientists have found that silicon solar cells can only reach about 30 percent efficiency in converting sunlight to electricity. Most commercial models achieve less than 25 percent efficiency.

If solar is to expand in the way many hope it will — enough to lead the charge in decarbonizing the electricity sector — more efficient technologies will only help.

Enter perovskites, a class of compounds that share a common crystal structure that makes them highly efficient at converting light to electrical energy. The technology is still developing, but perovskite cells have already reached 25.5 percent efficiency according to the National Renewable Energy Laboratory.

“There’s a generational, transformational opportunity that the metal halide perovskites represent,” said Joseph Berry, a senior scientist at NREL who works on perovskite and hybrid solar cells.

Despite their potential, however, perovskite solar cells are less stable than their silicon counterparts. And researchers are still working to understand the materials science that would permit their widespread use in the solar industry.

With the first commercial perovskite production line expected to come online in 2021 and the Department of Energy continuing to pursue research, the next decade may show whether perovskite technology can handle more mainstream applications.

Perovskite vs. silicon

Monocrystalline passivated emitter and rear cell (PERC) silicon technology has become the standard in the solar industry. Existing knowledge about how silicon performs in semiconductors used for microelectronics helped silicon become a dominant material for solar, too. But it’s not the most efficient.

“At a fundamental level, the interaction of photons and silicon is not good,” said Berry. “It’s an indirect bandgap semiconductor, so that’s suboptimal.”

Though the industry is still working to squeeze as much efficiency as possible from this now-standard technology, those gains will only be able to continue for so long.

Perovskite cells are not the only “challenger technologies” contending for mono PERC’s mainstream status. Most major module producers now also make N-type, including TOPCon, cells, said Xiaojing Sun, a senior solar analyst at Wood Mackenzie. Those cells, also based on silicon, have a theoretical efficiency of just over 28 percent.

Though relatively new, those technologies do have an advantage because they are already somewhat commercialized, said Sun. But they don’t have the efficiency potential that perovskite does.  

“If you want to continue stepping up the curve to a 40 percent module, you have to move away from silicon alone,” said Jenya Meydbray, CEO at PV Evolution Labs, a solar testing lab. “Perovskites are clearly the leading candidate.”

Perovskites have been researched as a potential solar technology for less than two decades. In that time, they’ve been able to go from efficiency under 4 percent to over 25 percent, according to the Department of Energy. When combined in tandem with silicon cells, the thin-film technology boosts efficiency to 29.1 percent, according to NREL.

Though the technology is sensitive to degradation, its thin structure — it can be printed onto other materials — also makes it a promising candidate for futuristic applications in building-integrated PV or other installations where weight is a significant factor.

For now, no significant commercial production of perovskite solar exists, though U.K.-based startup Oxford PV is scaling up its first commercial line. Oxford estimates that between 10,000 and 20,000 scientists are working on perovskite technology worldwide. In addition to Oxford PV, WoodMac cites nine other companies working on commercializing perovskite. Most are in the precommercial stages of development.

Oxford uses tandem technology, putting perovskite on top of silicon cells. That allows the company to hedge “the inherent downsides of perovskite technology while tapping into a dominant technology,” said Sun. It also allows the company to rely on standard solar equipment.

Oxford, which was spun out of the university of the same name, bought a solar factory in Brandenburg, Germany (a short distance from Tesla’s German Gigafactory) in 2016 and plans to start up its first 100-megawatt production line in the second half of 2021.

“Our vision is, of course, that the world eventually will be driven by renewable energies,” said Oxford PV CEO Frank Averdung. “In that, we want to make perovskite the mainstream. We do believe that is possible within the decade.”

NREL's Berry said scientists still have work to do to understand the materials science that supports perovskite cells.

“The way we know how a silicon module performs over time in the field is by having field data, and nobody wants to wait 30 years to know that perovskites are going to be something that is going to be durable and can be deployed widely,” said Berry. “That’s what we’re focused on now, is to really try to understand the materials.”

Meanwhile, DOE in August offered up $20 million for perovskite research and manufacturing development. Up to half of that total will also go to establishing a center to validate the technology’s performance and bankability. Those awards will be decided next year.

The challenges of instability

The most significant barrier to widespread deployment of perovskite lies in its sensitivity to light and heat, an ironic but common issue for solar technologies. 

“If you have to encapsulate out light, you’re going to be in trouble if you’re a solar cell,” said NREL’s Berry. “While there are still open questions at extreme operational temperatures and under concentrated illumination, perovskites don’t appear to have fundamental issues that would preclude us [from] making very good and durable photovoltaics.”*

Perovskites can also be affected by moisture.  

“The technology itself still has this big hurdle to overcome, which is the stability of the perovskite cell,” said Sun. “It’s such a big paradox. On the one hand, it’s super sensitive to sunlight — that’s why it makes such a good solar module material. But on the other hand, it also disintegrates really quickly, so it loses its structural and electronic integrity very fast.”

Materials research is essential to ensuring the entire structure can remain stable. Perovskite cells can degrade quickly, losing output. Silicon cells, meanwhile, can last for decades. This year, startup Violet Power began to offer 50-year warranties on its crystalline silicon solar panels.

Bringing a new technology up to commercial speed also takes time and money. Once Oxford PV’s manufacturing line commences, the startup sees its most viable route to market to be residential sales in parts of Europe, Japan and California, where space constraints, and thus higher efficiencies, matter more. By 2024, Oxford PV plans to produce between 2 and 4 gigawatts of product and move into the commercial and industrial rooftop solar market.

So far Oxford has secured investments from oil giant Equinor, wind turbine maker Goldwind and PV producer Meyer Burger, among others. The company hopes to enter the utility-scale market before the end of the 2020s.

While WoodMac’s Sun says that trajectory is certainly possible, perovskite's mainstream success by 2030 will entirely depend on how quickly costs come down, as the utility sector is more sensitive to price variations per watt.

In the meantime, silicon will continue proliferating. But overall, NREL’s Berry sees thin-film technologies like perovskite as a “better strategy” because they require fewer materials than do silicon solar products.  

“The utility-scale challenge is a significant one, but I think it is one that is surmountable,” Berry said.

*This story has been updated to clarify that sensitivity to heat and light is a common issue among a variety of solar technologies.