Tucked away in a cluttered Caltech chemistry lab filled with expensive, sophisticated equipment sits an ordinary $200 desktop printer. No paper ever passes through this printer and it’s always out of ink, but it may hold a key to solving the world’s energy crisis.

A little more than 1,000 miles northeast of Pasadena, a similar office printer is ensconced in a lab on the Fort Collins campus of Colorado State University. Both labs are run by Caltech alumni who think that the energy from the sun offers the best alternative to fossil fuels, with more potential than any other renewable energy resource. And they’re both betting that their relatively low-tech inkjet printers could play a big role in deriving clean-burning fuels from the sun.

The scientists are Nate Lewis ’77, MS ’77, Argyros Professor and professor of chemistry at Caltech, and Bruce Parkinson, PhD ’78, professor of chemistry at Colorado State. Although the two were only casual acquaintances in their Institute days, Parkinson’s best friend was fellow graduate student Kent Mann, PhD ’77, who recruited Lewis for assistance on a couple of solar energy experiments and was instrumental in turning Lewis’s focus from physics to solar photochemistry.

Caltech alumni and solar research colleagues Nate Lewis (at left) and Bruce Parkinson got together recently in Caltech’s Beckman Institute courtyard during a National Science Foundation site visit. Lewis adapted the standard inkjet printer at top to print out metal oxides after reading about Parkinson’s novel method to search for a way to produce solar energy.

At Caltech, both Lewis and Parkinson got training in electrochemistry, a discipline that is useful in the development of solar cells. In electrochemistry, energy can be converted when an electrode—typically a metal or semiconductor such as silicon—comes in contact with a solution of chemicals dissolved in water. After graduation, Lewis and Parkinson went their separate ways. Lewis went on to graduate school at MIT, joined the Stanford faculty in 1981, and then returned to Caltech as a professor in 1988. Parkinson conducted research in government and industry labs for several years before going to Colorado State in 1991.

For about 30 years, both Lewis and Parkinson have been involved in basic research on photoelectrochemistry, or “wet” solar cell technology, investigating the electrical energy that is produced when light shines on a semiconductor in a solution, and its application to solar energy conversion. In the past few years they have found themselves converging on what Lewis likes to laconically call “a large climate problem.” He’s referring of course to the planetary changes caused by humans’ near-insatiable consumption and burning of fossil fuels. As Lewis wrote in a recent issue of Caltech’s Engineering & Science magazine, “With population and GDP growth conspiring together, we can expect a tripling of energy demand by 2050,” which will lead to more CO2 pumped into the atmosphere, effecting even more changes than we see now. Without immediate action, he says bluntly, the world is headed for the biggest uncontrolled experiment humans have ever done. “The CO2 we produce over the next 40 years, and its associated effects, will last for a timescale comparable to modern human history. Within the next 20 years, we either solve this problem or the world will never be the same.”

Both Lewis and Parkinson are convinced that solar energy offers the best hope both for resolving civilization’s excessive reliance on fossil fuels and for averting a looming environmental catastrophe. “In one hour, more energy shines on the earth than all the energy consumed by humans in one year,” Lewis says. But there’s a big problem. “There’s plenty of sunlight, but you have to be able to capture, store, and distribute its energy to people when and where they want it. Just to make electricity, without the ability to store and transport it, doesn’t solve the problem. We don’t have the technology that can capture solar energy at low enough cost to make it economical at scale compared with fossil fuels.”

Many experts believe that one key to efficiently harnessing solar energy is to find an economical way of using sunlight to break up water molecules into their constituent oxygen and hydrogen atoms, so that the energy can be recovered from hydrogen when needed. This was a view that began to take shape when Parkinson and Lewis were at Caltech. In 1975, two Japanese scientists showed that ultraviolet light illuminating titanium oxide electrodes splits water into oxygen and hydrogen. Titanium oxide is a metal oxide that is the primary ingredient in white paint, so the fact that such a common material could break up water was big news. Since that time, says Parkinson, “Efficient and inexpensive production of hydrogen from water and sunlight has been the holy grail of photoelectrochemistry.”

Parkinson recalls that the Japanese breakthrough “had an impact on all the Caltech chemists on the third floor of Noyes. It probably had some influence on my choosing to do a postdoc in semiconductor photoelectrochemistry.” Lewis says that the Arab oil embargoes of the 1970s were also a factor. “There was an energy crisis at the time, and solar research was the thing to do,” he says. During his senior year, he was part of a Caltech team led by Harry Gray, the Beckman Professor of Chemistry, and Kent Mann that discovered a metal complex with hydrogen-producing properties similar to titanium oxide. “The compounds used visible light, but could not complete the water-splitting cycle robustly,” Lewis says. “TiO2 is robust, but uses ultraviolet light.”

“There’s not much power in ultraviolet light,” Parkinson says. “Only a few percent of the photons emitted by the sun in this portion of the spectrum get through the atmosphere.” (It’s for this same reason that ultraviolet astronomical observatories are all space-based, where they can operate above the earth’s atmosphere.) Progress toward finding a viable successor to the first generation of metal oxides has been slow. Those that have been shown to work are “not efficient enough or don’t last long enough,” Lewis says.

For a metal oxide to work, it needs to meet several criteria, Lewis says. Ideally, it should be fairly abundant in nature, and able to absorb solar radiation across a wide portion of the electromagnetic spectrum so as to maximize the use of sunlight. It also has to be stable. Some compounds can split water across several wavelength ranges, says Lewis, but they essentially self-destruct in the process. (Sun-kissed silicon, for example, no sooner separates water into hydrogen and oxygen than it begins to corrode in the presence of the oxygen that’s released and breaks down ignominiously into sand.) Third, the electrons in the metal oxide have to be arranged in a suitable configuration for water to split into hydrogen and oxygen.

Parkinson says that while any viable candidate will most likely consist of several metals, there’s not enough theoretical knowledge to confidently determine which combinations of metals possess the requisite properties to perform efficient photoelectrolysis of water. So actual trial-and-error testing is needed to identify the best candidates.

For years, Lewis, Parkinson, and others have been studying metal oxides, using time-consuming methods that basically have involved analyzing one compound at a time. A couple of years ago, however, rapidly rising energy prices coupled with the looming environmental crisis made it clear that they had to find a way to speed up their investigations. With fuel consumption reaching unprecedented levels throughout the world, and signs of significant climate change appearing worldwide, even the most ardent cheerleaders for cheap oil and the most vocal global warming skeptics had begun to alter their views. “This was no longer just intellectual fun,” Parkinson says. “We needed to find a way to make hydrogen cheaply.”

Four years ago, Parkinson was attending a conference when he heard a research chemist talk about how he had used a standard inkjet printer to spew out different combinations of chemical compounds to then test for catalytic properties. He realized that he could use the same procedure to screen metal oxides for their suitability as solar energy photocatalysts. Part of his idea was to empty ink cartridges and load them with different combinations of chemicals to create different metal oxide
compounds.

Back in his Colorado lab, he and his graduate student Mike Woodhouse built a system using an inkjet printer in which they substituted dissolved metals for ink. They modified the printer to accept thin glass plates rather than paper, and programmed it to print out different combinations of the metals on the plates. Heating the printed plates in a furnace for two hours at 500°C converted the metal salts into metal oxides, after which the plates were submerged in a solution, while a laser, serving as a stand-in for the sun, irradiated the material. By checking the plates with an amplifier to measure whether any current was produced by each compound, they could determine whether water molecules were being split and energy was being stored.

“It was extremely simple and easy,” Parkinson says. In 2005, he and Woodhouse described the process in a paper published in the journal Chemistry of Materials.

When Lewis read about Parkinson’s procedure, he called him up and asked if he could get more information so he could develop his own system. “I interact with Bruce frequently,” Lewis says. “He is always very collegial and helpful in assisting people to advance the science of this area.” So, in the summer of 2006, he sent one of his graduate students, Jordan Katz, and an undergraduate, Todd Gingrich, to visit Parkinson and find out how his system worked.

“Bruce shared everything that he did with us and helped us with an initial setup method, which was great,” Lewis says. While still in the testing phase, “Bruce’s work helped legitimize screening as something one can do and actually get funded for, instead of just being a ‘fishing expedition,’ which it still partly is but hopefully can be done intelligently to find new materials.”

Once Katz and Gingrich came back from Colorado, they started building their own metal oxide analyzer, and had one up and running by the end of the summer of 2006. They spent much of the next year modifying some of Parkinson’s design features, including adding a method to analyze voltage, which makes it possible to obtain precise measurements of the electrons generated by the material. While it registers the presence of current when water is split, Parkinson’s technique is not designed to detect charges below this threshold.

Above left: Nate Lewis and Jordan Katz, PhD ’08, study one of the glass plates (shown in closeup, right) that are evaluated in the apparatus on the table to determine which metal oxides split water to produce energy. The glass plate from the Lewis lab is printed with different metal oxides that can then be evaluated for their ability to split water and produce energy when irradiated.

“Knowing the voltage is important because that figure tells us just how close we are to the needed number,” Lewis says. “It takes 1.23 volts of potential to split water.” If we only get 1.2 volts, says Katz, “while that material could not split water as is, it would still be interesting and worth further investigation, because it would be awfully close to giving enough voltage” to split water and perhaps could be modified to work effectively.

Lewis’s system also makes it possible to test up to 250 separate compounds on each plate, whereas Parkinson’s system produces a continuous gradient pattern of different ratios of three or four metals at a time. “We want to mass-produce thousands of compounds as quickly as possible,” Katz says. “The printer is like pouring from the beaker, using microliter spritzes of ink. Printers are good at spraying liquid and mixing in proportions. In a few minutes, we can put 250 dots on a plate and test them. In an hour, we can do 1,000 just using a commercial $200 printer.”

In reconfiguring Parkinson’s system to evaluate dozens of compounds per plate, Katz says that one of the challenges was writing the software to instruct the printer to print the dissolved metals in different combinations and concentrations. He and Gingrich also had to create software to keep track of the photoelectrochemical results.

“It’s not a trivial process,” Katz says. “You have to fabricate the glass so that there are no interfering signals between metal oxides, you have to print the compounds on the glass, you have to bake it, and then screen it. Each process takes time.”

Common sense dictates which candidates are chosen for analysis. “We first consider the abundant metals in the crust of the earth,” Katz says. “For any successful electrode to have an impact on a global scale, we need a colossal amount of the appropriate material. We don’t want to use a metal oxide that costs $5,000 a gram. We probably won’t use radioactive elements, either.” He sees promise in metal oxide candidates whose current utility is limited by how selectively they absorb light.

“To solve the problem, you can take a compound that you know works and try to extend its light-absorbing region to absorb more sunlight” by gradually adjusting the ratios of the metals that compose it to see if you can optimize the electrical response. “Or you can try to make an unstable compound that only works up to a point more stable. Or you can try to find a new and completely different compound that just works.”

Says Katz, “There’s no reason why we can’t find a stable metal oxide that absorbs visible light that doesn’t degrade and that can produce energy. If the third compound that we test works, that’s great, but it could be the ten millionth compound. It’s unlikely we’ll find something quickly, but there’s only one way to find out.”

Katz, who recently left Caltech for a postdoctoral fellowship at UC Berkeley, figures that once the screening system is out of the test phase in a month or two, it will be able to analyze about 20,000 metal oxides a week. Parkinson says that he’s tested about 500 metal oxides, with no major breakthroughs so far. He says that he only recently read Katz’s thesis describing the Lewis lab’s screening system and so hasn’t decided yet whether it is worth copying. But he is convinced that to have any chance of finding a metal oxide that works, he has to expand the project.

To that end, Parkinson hopes to enlist dozens of university labs around the country, using simple but effective testing kits operated by an army of undergraduates. “I want to use undergraduate students’ hands and brains,” he says. “The kits should be inexpensive and easy to use.” With that in mind, he taught a class last year in which he had students build metal oxide screening kits out of Lego Mindstorms robotic kits, using a laser pointer. The prototype kits worked, but before he can ramp up production, he needs to test them further.

Undergraduate students at Colorado State University examine one of the metal oxide analyzers that they built on a shoestring partly out of Legos.

Parkinson recently received a grant from the Camille and Henry Dreyfus Foundation to send kits to nine undergraduate universities that have offered to recruit students to test them. He hopes to expand the project later this year by joining up with a National Science Foundation–funded Caltech initiative called the Chemical Bonding Center, in which investigators, including Lewis, are seeking ways to efficiently store solar energy in the form of chemical bonds. “The more people who do this, the better,” Parkinson says. “There has to be a national commitment here. It’s a difficult project, but the payoff is so big and so important.”

Although metal oxide devices offer a promising route to capturing and utilizing solar energy on a global scale, other alternatives do exist, some of which have been around for eons. We call them plants. More than two billion years ago, algae evolved the trick of splitting sunlight through photosynthesis. Could this ancient biological breakthrough be the basis of new solar-energy technologies today? It’s a nice idea, Lewis says, but “plants are not the optimum color to absorb sunlight efficiently. In addition, a significant amount of the energy of a living system that it derives from the sun is used just to regenerate and keep the system living. We know that metal oxides already work. They just have the problem that the ones that work well are unstable and the ones that are stable don’t work well. We know it can be done, we just don’t yet know how we can do it under commercially feasible conditions.” Cost is key, says Katz. “We’re trying to beat out coal. No one is going to be interested in exploiting solar energy for 10 times the cost of burning coal, just because we have global warming.”