Chapter 6    Revolution by Evolution

Wintertime’s arrival in England adds an unpleasant chill to already dreary, drizzly days. So, I wasn’t surprised at the resigned mood in the conference room as a motley crew of graduate students and postdocs shuffled in for a weekly lab meeting in 2011. Many of us hailed from much balmier climates—California, Spain, Italy, even the Caribbean islands. We were all here, amid the dreaming spires of Oxford University, to learn from Professor Henry Snaith, an up-and-coming scientist studying brand new materials for solar power.

Henry—he insists on being called by his first name—is not your average physics prof. A former rugby player at Cambridge University, he’s energetic, athletic, and competitive; once he took our lab skiing in the Alps, beat us down every run, and made fun of the slower folks (okay, just me). He’s also gregarious and charming, especially when discussing his work. And he’s equally comfortable in the lab or at a cocktail party. Most important, he is a brilliant scientist with a spooky intuition about how materials behave at the nanoscale, in the realm of atoms and electrons.

Not surprisingly, Henry earned attention from renowned scientists early in his career. He studied under Professor Sir Richard Friend at the famed Cavendish Laboratory at Cambridge, and he went on to pursue postdoctoral research in Switzerland under Professor Michael Grätzel, another venerated researcher. He was still a post-doc when our lab held its routine weekly meeting on that dreary day, a young scientist with a solid pedigree and fewer than 3,000 citations to his papers.¹ Over the next four years, Henry would amass ten times that many citations, ascending to become the second-most-cited scientist in the world in 2015.² As it turned out, that lab meeting was anything but routine.

Mike Lee, one of my fellow graduate students, unveiled a startling discovery. At Henry’s behest, he had flown to Japan to track down a chemical recipe that researchers at Tokyo University had invented for making novel solar cells. Then, back at Oxford and working late at night, a bleary-eyed Mike mistakenly flipped two of the chemical concentrations on his recipe. The mistake yielded a solar PV cell that was more than 10 percent efficient—able to convert more than 10 percent of the sun’s energy striking it into electricity.

Compared to the record silicon solar cell efficiency of more than 25 percent, Mike’s number was unremarkable. But it was a major jump for a material not already in commercial use. Our lab (as well as a scattered assortment of researchers around the world) was trying to get away from silicon’s limitations, such as the need for expensive manufacturing equipment and the brittle quality of silicon wafers that resulted in rigid, ugly panels. With the right material, we thought, a cheap inkjet printer might be able to produce rolls of flexible, efficient solar PV coatings. Flexible coatings would make it possible to put energy-producing materials on parts of buildings that cannot support solar panels and might increase solar’s aesthetic appeal as well. After more than a decade of trying, the best result that anyone had achieved was 7 percent efficiency. Mike had managed to get to more than 10 percent using a brand new material that he had stumbled on and not yet optimized—the equivalent of nailing a bulls-eye in darts while blindfolded. As the results sank in, our team of stunned researchers salivated, confident that concerted development of the material should make it possible to improve efficiency to 20 percent and well beyond.

The material Mike had tweaked to such dramatic effect is known as “perovskite” (“pear-AHV-skite”). Unlike their silicon counterparts, perovskite solar cells are simple to make at the lab bench. They are also remarkably versatile, enabling flexible, colorful, semitransparent, and lightweight variants. Once Mike’s results became public, the idea that they could have these properties and also be highly efficient at absorbing sunlight and converting it into electricity sparked a solar gold rush.

Our lab had a head start, as Henry smartly deployed the rest of us to learn as much about this material as we could while he and Mike prepared an announcement to the world for the prestigious pages of Science magazine.³ In those early days, our little group of researchers resembled kids in a playground, with perovskite all to ourselves. I was elated that perovskite acted like fairy dust on all my existing experiments, turning kooky solar cell designs into respectable devices. But Henry would gently prod me to recognize that perovskite was much more than a fairy-dust accessory to another experiment—it was the main event itself. Our monopoly over research into Henry’s perovskite ended a year later in 2012, when the rest of the scientific community read about the technology, and researchers around the world dumped their previous projects to pile onto the field.

This flurry led to flourishing international collaboration—but also fierce competition. The most intriguing subplot has been the rivalry between Henry’s lab and that of his former mentor, Michael Grätzel. Once, my colleague Sam Stranks discovered that Grätzel and his colleagues were racing to beat Sam to press with a research result that Sam had independently obtained weeks earlier and presented at a conference. Henry and Sam worked around the clock to prepare their own manuscript and win the scoop. Ultimately, Science published the two results side by side.

Beyond conferring credit to both teams, the simultaneous publication was a powerful signal that perovskites are such an important breakthrough that they merit the extra column inches in top journals that cover a myriad of other scientific fields.

In fact, just from 2012 to 2016, nearly thirty articles discussing perovskite solar were published in the top two scientific journals, Science and Nature. For a scientist, publishing in Science or Nature is akin to an actor receiving an Oscar nomination. And the surest way to publish a prestigious article relating to solar is to set a new efficiency world record. Responding to this incentive and the prospect of revolutionizing solar technology, scientists from around the world—most recently in South Korea—have pushed up the efficiency of perovskite solar cells faster than that of any other solar technology in history (reaching more than 22 percent by 2016).⁴

If you’re surprised that you haven’t heard of perovskite, don’t be. Outside the rarefied world of academia, where researchers confer celebrity status on the latest efficiency holder, very few people understand how remarkable this new material is. Some in the silicon-dominated solar industry may be aware of perovskite, but they are not worried about it challenging silicon’s primacy. At least in the short term, they are entirely justified. The frenzy in academic circles surrounding cell efficiency has very little to do with making a commercial product that will survive outside for years or even decades, as current silicon panels regularly do. In the lab, scientists attempt to break the efficiency record by cherry-picking the best result from tests of hundreds of tiny slivers of solar material a fraction of the size of your fingernail under a special lamp simulating perfect noontime sunshine.

Realizing what it will take to bring the technology closer to commercialization, academics have recently started to build bigger perovskite cells, subject them to real-world heat and moisture conditions, and test how long they survive. But the cells may never leave the lab unless investors become interested in them. Investors, however, are skittish about funding solar start-ups, having lost their shirts in the Silicon Valley clean technology boom and bust from 2006 to 2011. As long as investors and the solar industry continue to ignore academia, perovskite and other new solar technologies could well remain lab-bench novelties.

Such neglect could be disastrous for solar power’s long-term prospects. Chapters 4 and 5, on financial and business model innovation, explained how existing solar technology can flourish by accessing vast untapped pools of low-cost capital. That approach, on its own, might work for a decade, or maybe even two. But solar must do more than attract investment in existing technologies to meet the midcentury target of supplying a third of the world’s power. As described in chapter 3, solar will have to surmount its plummeting value as its penetration increases—the value deflation effect already manifest in Germany and California—as well as the strain on power grids that must integrate an intermittent generation source. Once solar power ascends to provide a double-digit percentage of global electricity, these obstacles could prevent it from delivering a larger share of the world’s electricity, even if the cost of silicon-based solar PV systems continues the rate of decline it has enjoyed over the last half-century.

To avoid a growth slowdown, solar PV will need to become still cheaper. And to do that, it has to become more efficient than silicon may ever allow. That’s why a group of leading solar scientists insisted in 2016 that solar research is not finished, arguing:

While [the] spectacular success [of solar photovoltaics (PV)] should be celebrated, many PV researchers have recently been dismayed to see shrinking public PV research and development (R&D) funding (both in the US and in other countries) … For higher penetration of the market, PV systems must cost even less to cover the additional costs of storage or transmission so that solar generation can be dispatched to cost-effectively meet electricity demand more broadly in both time and space.⁵

Fortunately, exciting research on many fronts has potential to improve the performance of solar PV and make it much cheaper. Future technologies lie on a spectrum from evolutionary to revolutionary.⁶ Evolutionary improvements to existing technologies have a better shot at making a commercial debut in the near term—remember that a successful innovation is one that makes it to market. And the most promising evolutionary option right now is for perovskite or other materials to literally piggyback right on top of existing silicon solar panels. Combining silicon with perovskite would boost the efficiency of existing panels, while largely leaving the existing production process and the finished product the same, maximizing the chances of market acceptance.

But down the road, revolutionary concepts that look nothing like today’s PV panels will be needed, and those advances require investment right now so they can ultimately bring massive benefits. The evolutionary route can lead to a revolution. For example, layering perovskite on top of silicon in the near term can eventually make it feasible to produce even more efficient and versatile all-perovskite solar coatings, smashing through the limits that constrain silicon technology.

Although perovskite is the lab leader, it does have potential drawbacks, and it isn’t the only promising solar material out there. Other approaches—including organic and quantum dot solar cells—come with their own unique advantages. They present the possibility of a virtually limitless range of color, transparency, and flexibility, as well as theoretically unmatched efficiency.

All these new technologies could enable a range of applications. They could coat weak shanty roofs in urban slums, power forward-deployed military operations, and give architects a diverse palette for designing beautiful, energy-efficient buildings. Highly efficient solar tarps might carpet deserts; Home Depot might sell solar paint as cheaply as wallpaper. And, new technologies could even make it feasible to realize an idea generally dismissed as science fiction: space-based solar power, which would harness a 24/7 supply of sunlight and beam power down to the Earth’s surface.

Right now, we don’t know which solar technologies and which applications will be needed to transform today’s fossil fuel–dominated power system. So, it makes eminent sense to pack our technological quiver with as many potent arrows as possible. Henry Snaith knows that tall barriers stand in the way of any upstart that aims to challenge silicon’s dominance over the solar industry. But that hasn’t stopped him from trying. He’s founded a start-up, Oxford PV, which has generated the most buzz of any solar start-up in years. He can’t do it alone, though. It will take scientists, entrepreneurs, investors, corporations, and—importantly—policymakers to help emerging solar technologies have their day in the sun.

From Photons to Electrons

To understand what it takes to maximize the efficiency of PV cells, consider how silicon cells convert incoming sunlight into electricity. Solar electricity production involves two of nature’s elementary particles: the photon and the electron. Photons are wispy particles of light, and a photon’s energy is directly related to what color it is. For example, blue photons have more energy than red ones; invisible ultraviolet photons have even higher energy, whereas infrared photons—also invisible—are low energy. Electrons are negatively charged particles that surround the positively charged nucleus of an atom.

A solar cell recruits both of these particles by transferring the energy from each incoming photon to an electron. Once endowed with an energy boost, the electron can break free from its host atom and exit the solar cell. The stream of electrons leaving the solar cell represents electricity, and the amount of energy pumped out by the solar cell depends on two values. First, the “electric voltage” is tied to how much energy is in each electron that flows out of the solar cell. Second, the “electric current” depends on the total number of electrons leaving the solar cell each moment. The power output of a solar cell—how much electric energy it pumps out every second—is just the voltage multiplied by the current.

So, if a solar cell could transfer all the energy from every single incoming photon to an electron, and make sure that every energized electron left the solar cell to do useful electrical work, it would maximize its electric voltage, current, and power output and be 100 percent efficient. But even the very best silicon cells can muster only a quarter of that figure. Why? Silicon is a semiconductor—a material that can switch back and forth between a conductor of electricity and an insulator, making it possible for electrons to toggle between staying put and moving around when photons strike them. And, semiconductors have a fundamental feature that limits their efficiency: the more photons the cell absorbs, the less energy per photon it can transfer to each electron. In other words, there is a trade-off between the maximum current and the voltage that a solar cell can produce, even though both quantities matter for the cell’s power output. The more current there is, the less voltage, and the more voltage, the less current. Because electrical output is simply voltage multiplied by current, 100 percent efficiency remains out of reach.

The voltage-current trade-off stems from a property of semiconductors called the “bandgap”: the amount of energy that an electron needs to break free from its host atom and contribute to the cell’s electric output. A photon with less energy than the bandgap will pass right through the solar cell. A photon with more energy than the bandgap can transfer only a bandgap’s worth of energy to an electron—the rest of the photon’s energy gets wasted as heat.

Think of a particular semiconductor’s bandgap as the amount of force needed to get ketchup flowing from a stuck bottle. Tapping the bottom too lightly won’t do anything to get the ketchup flowing (analogous to a low-energy photon passing right through the solar cell). Hitting it with the perfect amount of force will transfer just enough energy to get the ketchup flowing (like a photon with energy equal to the bandgap setting an electron free to flow out of the solar cell). And hitting the bottle with a sledgehammer will also get the ketchup flowing—but at the cost of expending a lot of energy to heave the hammer and wasting most of that energy (like a high-energy, ultraviolet photon transferring a bit of its energy to an electron and then dissipating a lot of heat).

To achieve high efficiency, scientists have to choose a material with an optimal bandgap. If the bandgap is too high, most photons will lack the energy needed to eject electrons and will pass right through the solar cell, failing to generate much electric current. If the bandgap is too low, most photons will set electrons free but only transfer a dribble of energy to each one, resulting in a low voltage. Silicon happens to have a decent bandgap somewhere in the middle, although it is a little lower than that of more ideal semiconductors, such as gallium arsenide, which is used to make more efficient solar cells but is too expensive to have commercial success. Theoretical models indicate that silicon’s highest efficiency is about 29 percent—a couple percentage points lower than that of gallium arsenide.

Despite more than a half-century of development, silicon has topped out at just above 26 percent efficiency, below its theoretical maximum. Why? The reason lies in the journey that an electron takes from being liberated by a photon to successfully exiting the solar cell. It turns out that the path is filled with perilous potholes, and electrons behave like drunk drivers. If a mobile electron runs into an obstacle or a trap, it loses as wasted heat some of or all the energy imparted by the photon. A solar material that is a perfect crystal lattice—that is, its atoms are perfectly arranged in a repeating pattern—will pose no obstacles to electrons careering across the lattice toward an external circuit. But real-life crystals tend to have defects or impurities that can impede an electron’s journey. So, it takes expensive equipment and temperatures above 1,000°C to produce high-purity silicon with very few defects. Those defects that remain, as well as ones at the interfaces between silicon and other layers of the finished solar cell, prevent some fraction of electrons from reaching the external circuit, reducing the efficiency of real cells below their theoretical limit.

This process is summarized in figure 6.1, which walks through the various losses that a silicon solar cell incurs in trying to convert sunlight into electricity. The x-axis shows the range of different photon energies, or colors, emitted by the sun. And for each color, the y-axis value sums up the energy from all the photons that pass through a 1-m² box every second, otherwise known as the “power density” for that color.

Figure 6.1

Harnessing sunlight with a silicon solar cell. This graph displays the total energy embodied in sunlight and the various losses incurred by a silicon solar cell in converting sunlight to electricity.

The area of “atmospheric losses” represents the power that is lost as sunlight travels from outer space through the Earth’s atmosphere, and photons bump into air molecules before reaching a solar cell on the Earth’s surface. Now the solar cell starts to rack up losses. Every photon below the bandgap of silicon, which is 1.1 eV (“eV,” or “electron-volt” is just a measure of energy appropriate for a tiny particle like an electron) passes right through the solar cell. Every photon above the bandgap transfers only a bandgap’s worth of energy to an electron, wasting the rest as heat. Thus, after these “thermalization losses,” the silicon solar cell’s electrons already have less than half the energy that the sun’s photons brought to the Earth’s surface. From there, electrons lose more energy as they zig and zag their drunken route toward the exit, bumping into crystalline defects and traps as they go. Those “voltage losses” further reduce the electric power output of the solar cell. What’s left—the area labeled “power output”—represents an efficiency of just 25 percent or less of the sunlight incident on the solar cell.

For decades, the received wisdom among experts was that it would take high-temperature processing and expensive equipment to prepare high-purity crystalline materials, as silicon is, to maximize the efficiency of solar cells. Getting to 30 percent efficiency and beyond was a commercial nonstarter because exotic approaches, such as stacking layers of semiconductor on top of each other to capture more of the solar spectrum, have traditionally been prohibitively expensive. But perovskite technology has shattered both of those conventional assumptions, forcing scientists to rethink whether silicon is indeed the best technology out there.

Say That Ten Times Fast

Even though it is a seemingly foreign, tongue-twisting term, perovskite isn’t anything new to scientists, especially not geologists. The term applies to a particular crystal structure; various minerals that take the perovskite structure are abundant below the Earth’s surface. What Mike fabricated at Oxford, though, was a synthetic perovskite, rather than a naturally occurring one. His creation incorporated inorganic atoms, as is usual in natural perovskites, with organic molecules. High-efficiency solar cells are normally made of inorganic materials. Such materials include silicon and gallium arsenide—traditional materials that are produced in wafers that can be turned into discrete solar cells—as well as materials such as cadmium telluride (CdTe) and copper indium gallium (di)selenide (CIGS) that can be deposited as continuous, thin films. Organic polymers are used in plastics, among other applications, and have historically been considered inefficient and unreliable solar materials, though they are easier to work with and cheaper than their inorganic counterparts. As it turns out, perovskite is a hybrid that combines the best elements of its inorganic and organic constituents.

One of the remarkable characteristics of perovskite is that it naturally coalesces into near-perfect crystals without much effort at all. To make perovskite materials in the lab, researchers will mix up the chemical constituents of the perovskite and dispense a few drops onto a spinning slide. Then, just like tide pools at the beach leave behind salt crystals, the evaporating solvents leave a film of perovskite crystals that are ready to guide drunken electrons safely to their off-ramps from a solar cell. Researchers are still working hard to improve perovskite crystal quality, and their improvements to date are a big reason why the efficiency of perovskite solar cells more than doubled from 2012 to 2016.⁷

Perovskites can be produced at low temperatures, which makes it possible to deposit them on flexible materials like plastics or ultrathin metal meshes.⁸ By contrast, making a silicon solar cell requires heating it to temperatures that rule out most flexible substrates. Flexibility is out of the question for current commercial silicon technology anyway, because silicon wafers are so brittle that they crack at the slightest stress. (I did a stint working on a silicon production line, and I snapped countless wafers when unpacking them, picking them up, and, in one instance, sneezing on one.)

Another advantage that perovskite has over silicon is an adjustable bandgap. Whereas silicon’s bandgap is fixed at a suboptimal, infrared 1.1 eV, scientists can adjust the perovskite bandgap by tweaking its chemical composition. That flexibility opens up the tantalizing possibility of cheap, multijunction solar cells that stack semiconductors of different bandgaps on top of one another to capture more of the solar spectrum than silicon can. Whereas a single material is limited to a maximum theoretical efficiency of 33 percent, a two-layer solar cell’s ceiling is 44 percent, and adding a third layer gets it all the way up to 50 percent.⁹

Here’s how such a multijunction cell works, considering just a two-layer, or tandem, cell for simplicity. The top layer has a high bandgap, so it absorbs visible and ultraviolet photons and harvests a large amount of energy per photon. The bottom layer has a lower bandgap, ideal for absorbing lower-energy photons. The infrared photons that pass right through the top layer get absorbed by the bottom layer. Because the high-energy photons have already been harvested, their energy does not get wasted as it would if the cell consisted of a low-bandgap layer alone. Thus, the layering mitigates the trade-off between absorbing more photons and harvesting as much energy per photon as possible.

Already, researchers have developed high-bandgap (about 1.7 eV) perovskites to layer on top of a silicon cell.¹⁰ Although the record efficiency for such a tandem device (nearly 24 percent as of 2017) was less than the record for a conventional silicon cell, these devices will almost certainly surpass silicon’s efficiency in the near future as researchers continue to optimize them.¹¹ Others have made a flexible tandem cell by combining perovskite and CIGS layers.¹² And in 2016, Dr. Tomas Leijtens at Stanford managed to combine two perovskites into a tandem device that was more than 20 percent efficient.¹³ (Tomas, a tall, blond Dutchman, was formerly a graduate student with me, so I can attest that behind his nonchalant surfer look is a brilliant scientific mind.) All-perovskite multijunction cells, with two, three, or even more layers, could shatter efficiency records in the future and convert more than a third of the sun’s energy to electricity.¹⁴

But in the nearer term, combining perovskites with silicon (figure 6.2) is the most promising research avenue—evolutionary, not revolutionary. Not everyone agrees that perovskites can provide a meaningful boost on top of silicon, but the results to date are promising.¹⁵ In press interviews, Henry has disclosed that his company, Oxford PV, can boost the efficiency of commercial silicon solar panels by a third with a perovskite coating.¹⁶ And although Oxford PV is out in front, other start-ups, such as in the United States and China, are racing to catch up.

Figure 6.2

How a tandem perovskite/silicon cell works.

Source: Image reprinted with permission from Sivaram, Stranks, and Snaith (2015).

If layering perovskite on top of silicon works, it could pay off handsomely. The factory equipment needed to deposit a perovskite layer would represent just a small fraction of the cost of a silicon solar cell production line. The gains would far outweigh that small cost. More efficient panels would be much cheaper per watt. Many associated costs of building solar projects, such as land and labor, would fall as well because fewer panels would be needed to produce the same power. Most important, the final product would look just like existing silicon solar panels that investors and customers are comfortable with. Overnight, this product would become the best deal on the market.

Perovskites are not quite there yet, and much work needs to be done to improve their commercial viability. Because of the skewed incentives of academia, most attention is focused on tweaking the chemical formula of perovskite to maximize its efficiency—new world records are guaranteed airing in prestigious journals.^17,18 But just as important is proving that they can last, and poor durability has been the Achilles’ heel of perovskite solar cells. Some argue that one of the elements found in many perovskite formulations—iodine—destabilizes the cell from within.¹⁹ Others have demonstrated that sunlight—yes, sunlight, the thing that a solar cell is supposed to receive—can degrade a perovskite solar cell (although it recovers in the dark).²⁰

Fortunately, scientists have started to design perovskite for durability as well as for peak efficiency. Several groups have demonstrated perovskite cells that remain stable during more than a month of testing. In one study, encasing perovskite in epoxy resin (similar to the glue used to bond together pieces of wood) safeguarded it from ordinarily deadly moisture over several months of storage in a hot and humid room.²¹ In 2017, researchers reengineered perovskites, using a brand new architecture that saw no degradation after more than a year.²² The efficiency of their device was just 11 percent, but the team’s durability-first attitude is promising, given that to attract significant investment, scientists will need proof that perovskite solar cells can survive in the real world for years, or even decades. As unglamorous as it is, extended field testing in harsh, outdoor environments will be a prerequisite for the technology to move toward commercialization.

Just as important, scientists must demonstrate that they can make perovskites that are much larger than the sliver-sized prototypes in the lab. They need, therefore, to come up with methods for depositing uniform, defect-free perovskites over larger areas. In 2016, an Australian group at Professor Martin Green’s PV center made a 12 percent efficient solar cell that was about as big as the palm of a hand.²³ Then in 2017, researchers in the U.S. National Renewable Energy Laboratory (NREL) made four fingernail-sized cells using a special perovskite ink and linked them together to make a 13 percent mini-panel.²⁴

These studies are promising; still, the efficiencies they achieved substantially lag the 20 percent efficiency that other scientists have managed to achieve on a single fingernail-sized cell.^25,26 But they are far from the last word. One of my former classmates from Stanford, Joel Jean, is leading a new initiative at the Massachusetts Institute of Technology (MIT) to help researchers transform their tiny perovskite cells into large-scale, lightweight, flexible, and highly efficient solar panels for use in the developing world.²⁷

A theoretical final drawback of perovskite may have little basis in reality, but could loom large in consumers’ imaginations. Because perovskite contains lead, it might be perceived as unsafe. This lead content is unlikely to pose any health hazard. After all, the amount in a perovskite solar panel is the same as in an equivalent volume of dirt.²⁸ Moreover, if perovskites make their debut layered on top of silicon, then the various sealing layers inside a conventional panel will virtually guarantee that no lead will be able to escape.

Nevertheless, the presence of a toxic element might well provoke public concern. A good example of what to expect comes from First Solar’s experience. That firm manufactures cadmium telluride solar panels containing the toxic element cadmium, and it has faced opposition from communities near the deserts that host its utility-scale installations.²⁹ To assuage their fears, First Solar has demonstrated that their panels would not discharge toxic cadmium even in a wildfire of 1,000°C. Down the road, any perovskite firm will have to do similar stress testing of their product, and any plans to seal in the perovskite with something flimsier than glass could complicate safety demonstrations.

One promising route might be to eliminate the lead from perovskite altogether, as Dr. Nakita Noel has proved can be done by replacing lead with tin (Nakita was also a fellow graduate student; I remember her best for her colorful Trinidadian slang and perfectionist experimental design).³⁰ Unfortunately, so far tin-based perovskites are less efficient than lead-based ones, so there remains a clear trade-off between high efficiency and use of nontoxic materials.³¹ Still, the meteoric rise of perovskite efficiencies and Henry’s inspired business decision to piggyback on silicon rather than confront it head on are reasons to be optimistic that perovskite will transition from invention to innovation—advancing from the lab into the world at large.

Alternatives to the Alternative

Even though perovskite is the undisputed leader in emerging PV technologies, its drawbacks create an opening for other contenders to shine. One is organic PV. Organic electronics have already made a splash in other fields. For example, your television or smartphone might have a screen lit up by vibrant organic light-emitting diodes. Today, organic solar cells cannot match perovskite cells in efficiency, but they are rapidly improving.

Organic solar cells also have some exciting advantages—at least in theory. They can be produced with no toxic materials at all. They are extremely versatile, able to take on the full range of colors and transparencies.³² (It’s possible to create perovskites of different colors as well, but organics rule when it comes to versatility.) Like perovskites, organic solar cells can be produced at low temperatures on flexible substrates, and they can be extremely lightweight.³³ They also have adjustable bandgaps, making organic tandem devices possible as well.³⁴

Organic solar cells have a long way to go to catch either perovskite or silicon, however. Recently, researchers have improved the notoriously poor stability of organic solar cells by introducing novel materials, but much more proof is needed that these cells can last in the real world.³⁵ And the best efficiency of organic solar cells as of 2017 was still only around 12 percent, in large part because organic materials make for quite treacherous terrain for an electron to travel across.

Researchers have tried to increase the surface area and decrease the volume of these materials—a move somewhat akin to replacing logs of firewood with kindling—but doing that just adds surface traps that can doom an electron’s journey. Researchers are toying with more radical ways to improve organic solar cell efficiencies—such as relying on a quantum effect in which electrons are teleported out of the cell, obviating the need to fix all the potholes in an electron’s path—but those prospects are way down the road.³⁶

Another exciting material for solar cells, “quantum dots,” could harness many more aspects of quantum mechanics. As the celebrated (and irreverent) physicist Richard Feynman quipped, “Nobody understands quantum mechanics.” Physics gets weird on the scale of nanometers (for reference, a nanometer is about the width of two silicon atoms)—and that weirdness is exactly what quantum dot solar cells would harness.

Quantum dots are semiconductor particles that are only a few nanometers in diameter. Many of their properties, including their bandgap, depend on their exact size, so scientists can manufacture quantum dots with a range of bandgaps just by making them bigger or smaller. And like perovskite or organic solar cells, it is possible to make quantum dot solar cells at low temperatures at the lab bench, opening up possibilities for flexible, lightweight solar coatings.³⁷ In fact, academics are even starting to combine technologies. For example, researchers recently created quantum dots out of perovskite to make a solar cell with greater than 10 percent efficiency.³⁸

Although the record quantum dot solar cell efficiency was only around 13 percent as of 2017, investigators have some tricks up their sleeves to raise that number. Here is where things get weird. First, quantum dots may be able to transfer the energy from a single photon to two or more electrons (a process known as “multiple exciton generation”), which would be less wasteful of photons with energy way above the bandgap. So instead of obliterating a single ketchup bottle with a sledgehammer, think of distributing the sledgehammer’s force across many ketchup bottles to get lots of ketchup flowing.

A second quantum way to avoid wasting high photon energies is to transfer more than a bandgap’s worth of energy from a photon to a single electron and extract the excess from the solar cell before it is released as heat (this is known as the “hot carrier” strategy). Some researchers have even proposed using a coating of quantum dots to absorb high-energy photons and emit twice as many low-energy photons to a solar cell waiting below in order to harness them more efficiently.³⁹ Although these effects have been observed in the lab, using them to improve cell efficiency is far off on the horizon.

A final way to take advantage of the weirdness of quantum dots is something called “plasmonic resonance.” If the quantum dots are just the right size, incoming light waves will jiggle their electrons at just the right frequency to transfer energy with very high efficiency.⁴⁰ Think of this as somewhat similar to an opera singer hitting the exact right note that shatters a champagne flute: very effective energy transfer—though a little messy.

All of these are exciting prospects for a new generation of solar technologies. Figure 6.3 plots the rise in efficiencies for several different solar materials side by side, showing the rapid progress across the board. Whereas the five plots on the left display the efficiencies of different solar cells, the two plots on the right show the efficiencies of full panels composed of many cells. The efficiencies of panels are lower than those of cells because of power losses that accompany putting cells together, and because bad cells drag down the performance of good ones. Even taking such losses into account, panel efficiencies could rise dramatically in the coming decades. The final plot on the right displays one study’s projection of how new technologies could increase the efficiency of commercial solar PV panels. That projection sees panel efficiency hitting 35 percent—roughly double the average panel efficiency in 2016—by midcentury.

Figure 6.3

Comparison of solar PV efficiency across different technologies. The first six panels chart the progress in record efficiencies for solar cells and panels made of different materials between 2010 and 2017. The rightmost panel projects the efficiency for the best, commercially available solar panel by 2050 (it is likely that this solar panel will consist of multiple semiconductor layers; it might also combine emerging and existing technologies).

Source: Historical record efficiency data from NREL and industry reports. Future projection from Albrecht and Rech (2017).

But are academics barking up the wrong tree? Is higher efficiency even necessary? Does the world really need rolls of flexible and lightweight solar coatings? Or are today’s silicon panels already up to the task of providing cheap solar power in the twenty-first century? Answering these questions requires delving into the possible applications of new technologies and understanding whether they can add value to existing markets and unlock brand new ones.

One Size Doesn’t Fit All

The assertion that silicon solar panels are all the world needs for solar to flourish implicitly makes a huge bet: that silicon panels can be cheap, modular, and versatile enough to serve a wider range of uses than they do today. But this is unlikely. Even in solar’s core markets today—utility-scale and rooftop installations—higher-efficiency panels than those sold now could substantially reduce costs, which is crucial to securing solar’s competitiveness when value deflation sets in at high penetrations. And there are other markets for which today’s silicon PV panels are inadequate, such as in places in both the developed and developing worlds where roofs are weak, in areas where military operations are carried out, and even in outer space.

For today’s solar PV markets, however, it isn’t obvious why silicon needs replacing, not least because silicon solar panels keep getting cheaper. For decades, they have fallen in cost regularly by roughly 20 percent for every doubling of total production, as producers incrementally shave their costs as they gain scale and experience. This pattern makes it possible to roughly forecast costs well into the future by making some assumptions about the continued growth of solar production.⁴¹ A 2017 study forecasts that the total cumulative solar capacity around the world could hit 8 terawatts (or 8,000 gigawatts) in 2030. By then, the average solar panel would cost $0.25 per watt of rated capacity, thanks to lower-cost production, incrementally better efficiencies, and longer panel lifetimes.⁴²

On top of this, there will likely be declines in the associated costs to install a solar PV project, including other hardware, land, labor, and soft costs such as permitting and financing. Recently, the U.S. Department of Energy (DOE) released a road map incorporating all those reductions. In 2030, the study found, electricity from a utility-scale solar PV plant might cost just 3 cents per kilowatt-hour (figure 6.4), roughly half its cost in 2016.⁴³

Figure 6.4

The DOE SunShot 2030 cost road map for solar power. Comparison of the cost of a fully installed utility-scale solar PV installation in 2016 with one in 2030, based on projected declines in the costs of PV system components and maintenance as well as longer PV panel lifetimes.

Source: DOE (2016).

The problem, my colleague Shayle Kann and I have argued, is that even halving solar PV’s costs probably won’t be enough. That would still leave the cost of electricity from solar PV roughly double the level needed to outpace value deflation, which sharply reduces the economic value of each additional kilowatt-hour of electricity from solar PV as more of it connects to the grid.⁴⁴

Some observers contend that replacing silicon solar panels with some other material won’t solve the devaluation problem because the panel represents only a fraction of the costs of a utility-scale solar PV installation. Indeed, as the MIT Energy Initiative has reported, “Reducing the cost of [solar panels] by half only reduces estimated costs by about 15% for the utility-scale projects we analyze, and 9% for the residential-scale projects.” The Initiative’s report concluded that trying to improve solar panel efficiency beyond 15 percent is just wasted effort.⁴⁵

But others disagree strongly. In fact, Martin Green—who, remember, predicted the dominance of silicon solar back when that was a very lonely view—argues that higher efficiencies can make a big difference to solar’s long-term prospects.⁴⁶ In the near term, he predicts that the industry will try to max out the efficiency of silicon solar panels, using a design that he invented in Australia decades ago. But down the road, firms that can increase the efficiency even more, for example, by using perovskite-silicon or even perovskite-perovskite tandem cells, will have a big advantage. In addition to lowering panel costs, higher panel efficiencies enable solar installations to pump out more power with fewer panels. And needing fewer panels reduces the land, labor, and equipment costs of a solar project. This math applies to utility-scale applications and is even more important for installations on rooftops with limited areas, which put space at a premium.

Higher efficiencies are not the only advantage that new technologies offer. Shattering the existing paradigm of heavy, rigid panels would also broaden solar’s appeal. Historically, installers often had to choose performance over flexibility. Thin films of “amorphous” silicon, which has low crystallinity, were good enough for powering calculators or camping equipment, but not for much else because they were very inefficient. But in being flexible, lightweight, and highly efficient, perovskites and other emerging technologies could extend the uses of solar in existing markets and make inroads into new ones.

The new materials—perovskites, organics, and quantum dots—all share the prospect of being printed on flexible substrates en masse, much as reams of newspapers are spooled through high-volume printers. The ease of transporting solar rolls would reduce logistics and shipping costs across the industry. Then, the ability to unroll solar tarps out in the desert, with minimal equipment to secure them, could slash the cost of utility-scale solar.

The elastic solar materials that have already been demonstrated in the lab could be deployed in new types of terrain, such as shifting landfills or uneven hillsides. In distributed markets, solar paint could be far cheaper and more aesthetic than existing rooftop panels, as could solar materials integrated into buildings in new ways. Tesla has already announced solar roof shingles that cleverly conceal silicon cells in expensive roofing materials. If it can switch to the high efficiencies and stunning versatility of new materials in the lab, it could make even more appealing and affordable products for the mass market.

The first new market for innovative materials might be rooftop solar for the developing world. An estimated 1 billion people in the developing world live in urban slums, mostly under roofs too weak to support heavy silicon solar panels.⁴⁷ In the best cases, these roofs are made from thin sheets of metal, and in the worst cases, they are flimsy textiles that barely offer shelter. Most of these people also lack access to reliable energy. Therefore, flexible, elastic rolls of highly efficient solar coatings could make it possible for these populations to power their own needs and perhaps even feed surplus power into the grid.⁴⁸ Importantly, researchers in academia and industry will need to develop sealants to protect the solar rolls from moisture, dust, and wear, and they will need to develop ways to anchor flexible PV to makeshift roofs and durably connect electrical cables. All of this will require engineering innovation beyond just the core solar material innovation that academic labs are currently pursuing.

The versatility of new technologies could also enable the integration of solar and building materials in a variety of ways. In densely populated urban areas, roof space for solar panels is constrained. But skyscraper windows and building facades receive plenty of direct and diffuse sunlight over the course of the day. Making it possible to convert some of that sunlight—which otherwise might heat up building interiors and increase power demand for running air conditioners—into power could offset a building’s energy use. Already, researchers in the lab have developed semitransparent perovskite cells that produce electricity with 13 percent efficiency while keeping out 85 percent of the heat that would pass through a window in the absence of the cells.⁴⁹ And because new materials can take on a range of colors and transparencies, building-integrated PV could enhance, rather than constrain, an architect’s palette—think of power-generating stained glass, for instance.

The military would also be an eager customer for new solar technologies. Whereas a gallon of diesel fuel might cost consumers $3 at the pump, it can cost the military fifteen times as much in the field once the cost of fuel convoys and their protection is factored in.⁵⁰ If, rather than refueling diesel generators to power a base in Afghanistan, the military could instead easily transport lightweight rolls of solar film that it could unroll and deploy, it would save money—and maybe lives, too.

But for me—a science fiction junkie—the most exciting prospect of all is truly out of this world. Recall that the strongest solar radiation is actually in outer space, before the molecules in the Earth’s atmosphere absorb a chunk of the energy in sunlight. And, obviously, there is no such thing as nighttime in outer space—the sun shines 24/7. So, there is ten times as much solar energy to harness outside the Earth’s atmosphere compared with at the surface, taking into account the lack of seasons, weather, variable daylight, or atmospheric losses.⁵¹ Would it be possible to construct massive solar farms in outer space and somehow send the energy back down to Earth, providing massive amounts of 24/7 power?

This idea is not as far-fetched as you might imagine. In fact, Japan is very serious about building an orbital solar power satellite. Its goal is to lead an international consortium to build a satellite that can pump out 1 GW—which would weigh more than 10,000 tons and measure several kilometers across—in the 2030s. As hard as setting up the satellite would be, Japan is focusing on the even tougher challenge of beaming power back down to Earth via microwave radiation, which a several-hundred-meter receiving station on Earth would turn into electricity. Elon Musk is dismissive of the idea because of the conversion losses from sunlight to power to microwaves to power again. But Japan’s space agency believes that the abundance of 24/7 power in space could more than compensate for the losses.⁵²

New solar materials could go a long way toward making an orbital solar power satellite viable. Lightweight, highly efficient, and foldable or rollable solar film could be much easier to launch into space. In addition to solar material innovation, there are at least five other categories of innovation needed to realize the solar power satellite vision. These include “wireless power transmission, space transportation, construction of large structures in orbit, satellite attitude and orbit control … and power management.”⁵³

Perhaps these challenges are collectively too daunting to surmount, especially in an affordable way. But the Japanese attitude is the right one when it comes to imagining the future of solar technology: remarkable advances become possible by pouring attention and resources into innovation.

Escaping Lock-In

Unfortunately, the industry as a whole takes a dim view of innovation. In other industries, such as the semiconductor industry from which solar descended, Intel and other large corporations routinely spend up to 20 percent of their revenue on R&D. In solar, that figure is more like 4 percent to 7 percent for a U.S. firm, such as First Solar, and around 1 percent for Chinese firms. Recently, both Chinese corporate and government funding for solar technology innovation has risen, but it remains well below U.S. levels.⁵⁴

As long as China retains its stranglehold on the solar industry, the country is unlikely to abandon the silicon technology that has vaulted its firms to the forefront of the global industry. Indeed, most in the industry view silicon solar panels as the end point of a half-century of innovation. Now, the industry is focused on ruthless cost-cutting, from the upstream production of polysilicon to the downstream deployment of silicon panel–based installations. And because the factories that make solar panels, including the facilities and equipment to produce high-purity silicon and PV cells, carry a high price tag, firms are reluctant to change course once they sink large amounts of capital into the ground.⁵⁵

This tendency puts the industry at risk of suffering a technology lock-in, entrenching the dominance of silicon solar PV. The economic theory of lock-in explains that an incumbent, dominant technology gets an advantage against emerging technology upstarts. Even if the upstarts might have the potential to cost less and perform better upon further development and scale production, they may flounder in a free market that favors first movers.⁵⁶ By 2030, the cost of electricity from silicon solar PV projects could halve, making silicon an even more formidable incumbent. Although far superior technologies would be even cheaper, efficient, and more versatile if they were commercialized, the world might get stuck with only incrementally better silicon panels. Those silicon panels might not be up to the task of defeating the most problematic form of lock-in—namely, the world’s dependence on fossil fuels.

Might lock-in be happening right now? There is no way to tell. Silicon may fall more rapidly in cost than currently anticipated, and value deflation might turn out to be less severe than today’s best simulations suggest. And if indeed solar is destined for lock-in, that will become clear only in retrospect. That was the case for nuclear power, for which it’s now possible to point out the fateful decision that led to disastrous lock-in.

Following World War II, U.S. Navy admiral Hyman Rickover chose one of several potential nuclear reactor designs being investigated at the time, the light-water reactor, to power American submarines. Because the design worked, he then chose it again to power aircraft carriers and, ultimately, civilian nuclear power plants.⁵⁷ The rest of this story of path-dependence is history. Today over 90 percent of all nuclear plants around the world are light-water reactors, owing to aggressive U.S. export and nonproliferation policies in the 20th century. Never mind that these reactors can melt down and are expensive to build, nor that several of the other designs that Admiral Rickover passed over might be better options but have been stymied to date because of the dominance of the light-water reactor.⁵⁸

Some signs suggest that solar power could be headed toward a nuclear-like lock-in. The fall in silicon solar panel prices is beneficial in the short term. But the drop makes it harder for emerging technologies that might not be cost-competitive before economies of scale kick in at mass production to break into the market in the long run.

What’s more, the financial innovation discussed in chapters 4 and 5 could aid this lock-in. Silicon solar projects might soon be able to tap into public capital markets and access cheap finance, thanks to the existence of decades of performance data and thirty-year manufacturer warranties. That ability would add another cost advantage to silicon over emerging technology competitors that public investors, having just gotten comfortable with silicon technology, would be loath to bet on. As a result, financial innovation could act as a barrier, rather than a bridge, to technological innovation.

Public policy can exacerbate lock-in as well. In the nuclear case, by applying rules that were customized for light water reactors, U.S. nuclear regulators have made it very difficult for firms to deploy new reactor designs. And in solar, policies that are ostensibly technology-neutral and subsidize all solar technologies actually implicitly tilt the playing field against emerging technologies. For example, developers building solar projects in the United States are much more likely to use federal tax credits to deploy silicon solar PV rather than try to build projects based on emerging technologies that may not yet be ready for prime time. And state-level renewable energy mandates are monopolized by silicon solar at the expense of emerging technologies. These policies expand silicon’s advantage and make it even harder to break into the market.

They also create political constituencies that attempt to entrench the first-generation technology. In the case of silicon, public policies such as federal tax credits have bred armies of lobbyists who aim to support the extension of those policies.⁵⁹ They were successful in 2015, when Congress extended the tax credits for both solar and wind power. To be sure, political coalitions that support silicon solar PV and other clean energy technologies are important forces behind a transition away from fossil fuels. Some scholars have argued that they might be the crucial players to pressure governments to pass carbon pricing policies, which enjoy widespread support from economists.⁶⁰ But in advocating narrow policy interventions that support mature clean energy technologies over emerging ones, these coalitions could also contribute to technology lock-in.

Vanquishing lock-in will take a combination of public policy interventions and private-sector ingenuity. Intervening in the market is difficult, both technically and politically—just ask the officials who disbursed a loan guarantee to Solyndra. Chapter 10, on U.S. policy recommendations, will delve into ways that the government can reduce lock-in barriers to innovation most effectively. These include ramping up spending on R&D, funding first-of-their-kind field demonstration projects, and making it easier for private firms to use public research facilities to reduce the costs of developing new technologies.

But it will also take approaches like the business model of Henry’s firm, Oxford PV, to gradually bring new technologies into the solar industry. Henry’s approach—to put a perovskite layer on top of existing silicon cells, adding a minor step to an otherwise undisturbed production process—is an evolution en route to a revolution. In the near term, Oxford PV could succeed at wooing Chinese solar panel manufacturers to use its technology and boost silicon panel efficiency by one-third. But once firms amass experience at manufacturing perovskite layers at scale, they could then consider manufacturing all-perovskite solar cells. And that could be a viable commercial route toward the high-efficiency, flexible, lightweight, and aesthetically pleasing solar rolls that academics dream of.

I hope Henry succeeds. (In case you are wondering, I have no financial interest in his venture.) But it will probably take many more companies like his to improve the odds of commercial technology advancing. Fortunately, there is no shortage of great ideas out there. To reach its massive potential, solar will require them to win the support they deserve.

Notes

1.  Henry Snaith, Google Scholar Statistics, https://scholar.google.com/citations?user=I2D3pUMAAAAJ&hl=en.

2.  Thomson Reuters IP & Science, “Thomson Reuters Announces the World's Most Influential Scientific Minds,” PR Newswire, January 13, 2016, http://www.prnewswire.com/news-releases/thomson-reuters-announces-the-worlds-most-influential-scientific-minds-300204299.html.

3.  M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami, and H. J. Snaith, “Efficient Hybrid Solar Cells Based on Meso-Superstructured Organometal Halide Perovskites,” Science 338, no. 6107 (2012): 643–647, doi:10.1126/science.1228604.

4.  Juan-Pablo Correa-Baena et al., “The Rapid Evolution of Highly Efficient Perovskite Solar Cells,” Energy & Environmental Science 10, no. 3 (2017): doi: 10.1039/c6ee03397k.

5.  Sarah Kurtz et al., “Solar Research Not Finished,” Nature Photonics 10, no. 3 (2016): 141–142, doi:10.1038/nphoton.2016.16.

6.  Martin Green and Stephen Bremner, “Energy Conversion Approaches and Materials for High-Efficiency Photovoltaics,” Nature Materials 16, no. 1 (2016): 23–34, doi: 10.1038/nmat4676.

7.  Giles Eperon and David Ginger, “Perovskite Solar Cells: Different Facets of Performance,” Nature Energy 1, no. 8 (2016): 16109, doi:10.1038/nenergy.2016.109.

8.  Yaowen Li et al., “High-Efficiency Robust Perovskite Solar Cells on Ultrathin Flexible Substrates,” Nature Communications 7 (2016): 10214, doi: 10.1038/ncomms10214.

9.  Steve Albrecht and Bernd Rech, “Perovskite Solar Cells: On Top of Commercial Photovoltaics,” Nature Energy 2, no. 1 (2017): 16196, doi:10.1038/nenergy.2016.196.

10.  David Mcmeekin et al., “A Mixed-Cation Lead Mixed-Halide Perovskite Absorber for Tandem Solar Cells,” Science 351, no. 6269 (2016): 151–155, doi:10.1126/science.aad5845.

11.  Kevin Bush et al., “23.6%-efficient Monolithic Perovskite/Silicon Tandem Solar Cells with Improved Stability,” Nature Energy 2 (2017): 17009, doi:10.1038/nenergy.2017.9.

12.  Fan Fu et al., “High-Efficiency Inverted Semi-Transparent Planar Perovskite Solar Cells in Substrate Configuration,” Nature Energy 2, no. 1 (2016): 16190, doi:10.1038/nenergy.2016.190.

13.  Giles Eperon et al., “Perovskite-Perovskite Tandem Photovoltaics with Optimized Band Gaps,” Science 354, no. 6314 (2016): 861–865, doi:10.1126/science.aaf9717.

14.  Albrecht and Rech, “Perovskite Solar Cells: On Top of Commercial Photovoltaics.”

15.  Zhengshan Yu, Mehdi Leilaeioun, and Zachary Holman, “Selecting Tandem Partners for Silicon Solar Cells,” Nature Energy 1, no. 11 (2016): 16137, doi:10.1038/nenergy.2016.137.

16.  Kathryn Nave, “The Simple, Cheap Trick That Makes Solar Panels More Efficient,” Wired UK, January 31, 2017, http://www.wired.co.uk/article/solar-panels-boosting-efficiency.

17.  Donqin Bi et al., “Efficient Luminescent Solar Cells Based on Tailored Mixed-Cation Perovskites,” Science Advances 2, no. 1 (2016): doi:10.1126/sciadv.1501170.

18.  Michael Saliba, Wolfgang Tress, Antonio Abate, and Michael Grätzel, “Incorporation of Rubidium Cations into Perovskite Solar Cells Improves Photovoltaic Performance,” Science 354, no. 6309 (2016): 206–209, doi:10.1126/science.aah5557.

19.  Regan Wilks and Marcus Bär, “Perovskite Solar Cells: Danger from Within,” Nature Energy 2, no. 1 (2017): 16204, doi:10.1038/nenergy.2016.204.

20.  Wanyi Nie et al. “Light-Activated Photocurrent Degradation and Self-Healing in Perovskite Solar Cells,” Nature Communications 7 (2016): 11574, doi:10.1038/ncomms11574.

21.  Angèle Reinders, Pierre Verlinden, Wilfridus Sark, and Alexandre Freundlich, Photovoltaic Solar Energy: From Fundamentals to Applications, (Chichester, U.K.: Wiley, 2017).

22.  Giulia Grancini et al., “One-Year Stable Perovskite Solar Cells by 2D/3D Interface Engineering.” Nature Communications (2017), doi: 10.1038/ncomms15684.

23.  Wilson Da Silva, “Trendy Solar Cells Hit New World Efficiency Record,” UNSW Newsroom, 2016, http://newsroom.unsw.edu.au/news/science-tech/trendy-solar-cells-hit-new-world-efficiency-record.

24.  Mengjin Yang et al., “Perovskite Ink with Wide Processing Window for Scalable High-Efficiency Solar Cells,” Nature Energy 2 (2017): https://www.nature.com/articles/nenergy201738.

25.  Xiong Li, Dongqi Bi, and Chenyi Yi, “A Vacuum Flash-Assisted Solution Process for High-Efficiency Large-Area Perovskite Solar Cells,” Science 353, no. 6294 (2016): 58–62, doi: 10.1126/science.aaf8060.

26.  Hairen Tan et al., “Efficient and Stable Solution-Processed Planar Perovskite Solar Cells via Contact Passivation,” Science 355, no. 6326 (2017): 722–726, doi:10.1126/science.aai9081.

27.  See Grid Edge Solar, a research program at the Massachusetts Institute of Technology (MIT): http://www.gridedgesolar.com.

28.  Nam-Gyu Park, Michael Grätzel, Tsutomu Miyasaka, Kai Zhu, and Keith Emery, “Towards Stable and Commercially Available Perovskite Solar Cells,” Nature Energy 1, no. 11 (2016): 16152, doi:10.1038/nenergy.2016.152.

29.  Herman Trabish, “The Lowdown on the Safety of First Solar’s CDTE Thin Film,” Greentech Media, March 12, 2012, https://www.greentechmedia.com/articles/read/how-safe-is-first-solars-cdte-thin-film.

30.  Nakita Noel et al., “Lead-Free Organic–Inorganic Tin Halide Perovskites for Photovoltaic Applications,” Energy & Environmental Science 7, no. 9 (2014): 3061, doi: 10.1039/c4ee01076k.

31.  K. P. Marshall, M. Walker, R. I. Walton, and R. A. Hatton, “Enhanced Stability and Efficiency in Hole-Transport-Layer-Free CsSnI3 Perovskite Photovoltaics,” Nature Energy 1 (2016): 16178, doi:10.1038/nenergy.2016.178.

32.  Prakash Singh and O. S. Kushwaha, “Progress Towards Efficiency of Polymer Solar Cells,” Advanced Materials Letters 8, no. 1 (2016): 2–7, doi:10.5185/amlett.2017.7005.

33.  Guglielmo Lanzani, Annamaria Petrozza, and Mario Caironi, “Organics Go Hybrid,” Nature Photonics 11, no. 1 (2017): 20–22, doi:10.1038/nphoton.2016.260.

34.  Miaomiao Li et al., “Solution-Processed Organic Tandem Solar Cells with Power Conversion Efficiencies >12%,” Nature Photonics 11, no. 2 (2016): 85–90, doi:10.1038/nphoton.2016.240.

35.  Sarah Holliday et al., “High-Efficiency and Air-Stable P3HT-Based Polymer Solar Cells with a New Non-Fullerene Acceptor,” Nature Communications 7 (2016): 11585, doi: 10.1038/ncomms11585.

36.  Jean-Luc Brédas, Edward Sargent, and Gregory Scholes, “Photovoltaic Concepts Inspired by Coherence Effects in Photosynthetic Systems,” Nature Materials 16, no. 1 (2016): 35–44, doi: 10.1038/nmat4767.

37.  Mingjian Yuan, Mengxia Liu, and Edward Sargent, “Colloidal Quantum Dot Solids for Solution-Processed Solar Cells,” Nature Energy 1, no. 3 (2016): 16016, doi:10.1038/nenergy.2016.16.

38.  Abhishek Swarnkar et al., “Quantum Dot-Induced Phase Stabilization of CsPbI3 Perovskite for High-Efficiency Photovoltaics,” Science 354, no. 6308 (2016): 92–95, doi:10.1126/science.aag2700.

39.  Hongbo Li, Kaifeng Wu, Jaehoon Lim, Hyung-Jun Song, and Victor Klimov, “Doctor-Blade Deposition of Quantum Dots onto Standard Window Glass for Low-Loss Large-Area Luminescent Solar Concentrators,” Nature Energy 1 (2016): 16157, doi:10.1038/nenergy.2016.157.

40.  Prashant Kamat, “Quantum Dot Solar Cells: The Next Big Thing in Photovoltaics,” Journal of Physical Chemistry Letters 4, no. 6 (2013): 908–918, doi:10.1021/jz400052e.

41.  Francois Lafond et al., “How Well Do Experience Curves Predict Technological Progress? A Method for Making Distributional Forecasts,” Journal of Technological Forecasting and Social Change. In Press. 2017, https://arxiv.org/pdf/1703.05979.pdf.

42.  Nancy M. Haegel et al., “Terawatt-Scale Photovoltaics: Trajectories and Challenges,” Science 356, no. 6334 (April 14, 2017): doi: 10.1126/science.aal1288.

43.  U.S. Department of Energy (DOE) Office of Energy Efficiency & Renewable Energy, “The SunShot Initiative’s 2030 Goal: 3₵ per Kilowatt Hour for Solar Electricity,” December 2016, https://energy.gov/sites/prod/files/2016/12/f34/SunShot%202030%20Fact%20Sheet-12_16.pdf.

44.  Varun Sivaram and Shayle Kann, “Solar Power Needs a More Ambitious Cost Target,” Nature Energy 1, no. 4 (April 7, 2016): doi:10.1038/nenergy.2016.36.

45.  “The Future of Solar Energy,” MIT Energy Initiative, 2015, https://energy.mit.edu/wp-content/uploads/2015/05/MITEI-The-Future-of-Solar-Energy.pdf.

46.  Martin Green, “Commercial Progress and Challenges for Photovoltaics,” Nature Energy 1, no. 1 (2016): 15015, doi:10.1038/nenergy.2015.15.

47.  UN-Habitat, Slum Almanac 2015–2016 (Nairobi, Kenya: UNON, Publishing Services Section, 2016).

48.  Darren Lipomi and Zhenan Bao, “Stretchable, Elastic Materials and Devices for Solar Energy Conversion,” Energy & Environmental Science 4, no. 9 (2011): 3314, doi: 10.1039/c1ee01881g.

49.  Hoyeon Kim et al. “Empowering Semi-Transparent Solar Cells with Thermal-Mirror Functionality,” Advanced Energy Materials 6, no.14 (2016), doi: 10.1002/aenm.201502466.

50.  Moshe Schwartz, Katherine Blakeley, and Ronald O’Rourke, Department of Defense Energy Initiatives: Background and Issues for Congress (CRS R42558) (Washington, DC: Congressional Research Service, 2012), https://fas.org/sgp/crs/natsec/R42558.pdf

51.  Daisuke Sato, Noboru Yamada, and Koji Tanaka, “Thermal Design of Photovoltaic/Microwave Conversion Hybrid Panel for Space Solar Power System,” IEEE Journal of Photovoltaics 7, no. 1 (2017): 374–382, doi:10.1109/jphotov.2016.2629843.

52.  Paul Marks, “Star Power,” New Scientist 229, no. 3060 (2016): 38–41, doi: 10.1016/s0262-4079(16)30325-6.

53.  Susumu Sasaki, “How Japan Plans to Build an Orbital Solar Farm,” IEEE Spectrum: Technology, Engineering, and Science News, April 24, 2014, http://spectrum.ieee.org/green-tech/solar/how-japan-plans-to-build-an-orbital-solar-farm.

54.  Jeffrey Ball, Dan Reicher, Xiaojing Sun, and Caitlin Pollock, The New Solar System: China’s Evolving Solar Industry and Its Implications for Competitive Solar Power in the United States and the World, Report, Steyer-Taylor Center for Energy Policy and Finance, Stanford University, March 20, 2017.

55.  Douglas M. Powell et al., “The Capital Intensity of Photovoltaics Manufacturing: Barrier to Scale and Opportunity for Innovation,” Energy and Environmental Science 12 (2015): http://pubs.rsc.org/en/Content/ArticleLanding/2015/EE/C5EE01509J#!divAbstrat.

56.  W. Brian Arthur, “Competing Technologies, Increasing Returns, and Lock-In by Historical Events,” The Economic Journal 99, no. 394 (1989): 116-131, doi: 10.2307/2234208.

57.  Robin Cowan, “Nuclear Power Reactors: A Study in Technological Lock-in,” Journal of Economic History 50, no. 3 (1990): 541-567, http://www.jstor.org/stable/2122817?origin=JSTOR-pdf.

58.  Varun Sivaram, “Unlocking Clean Energy,” Issues in Science and Technology 33, no 2. (Winter 2017): http://issues.org/33-2/unlocking-clean-energy/.

59.  David G. Victor and Kassia Yanosek, “The Crisis in Clean Energy: Stark Realities of the Renewables Craze,” Foreign Affairs, July 2011, https://www.foreignaffairs.com/articles/2011-06-16/crisis-clean-energy.

60.  Jonas Meckling et al., “Winning Coalitions for Climate Policy,” Science 349 (2015), doi: 10.1126/science.aab1336.