In the past 150 years, burgeoning industrialization has increased carbon in the atmosphere by 40 percent and driven a continuing rise in global temperatures. The trend won't stop soon. Among the consequences: rising sea levels, increased air pollution, and more hurricanes, floods, and droughts. Meanwhile, the age of cheap oil and gas has come to an end.

In the short term humans urgently need to use energy more efficiently, and we need to stop putting carbon straight into the air. More important for the long term, we need to find or create ways to use energy that don't release any carbon at all.

Soon after taking over as director of Lawrence Berkeley National Laboratory in mid-2004, Steven Chu noted that one of the Lab's greatest strengths lay in its potential to mobilize multidisciplinary scientific programs to develop carbon-neutral sources of energy. In the spring of 2005 Chu convened a meeting of Lab researchers and expert guests whom he invited to tackle what he called the "hardest questions" confronting the effort to turn sunlight into abundant, dependable, cheap, and convenient chemical fuels and electricity for human use. Following are a sampling of just some of the ideas discussed at Solar to Fuel: Future Challenges and Solutions.

Some of the challenges concern matters of scale. Energy storage, for example: steel cylinders may be fine for the laboratory or the welding shop, but as yet there's no safe way to store hydrogen that's also convenient and cheap. Batteries are fine for flashlights and electric vehicles, but there's no good way to store a power plant's worth of electricity. If energy doesn't come as a solid or liquid, how do we store it?

A more fundamental challenge is catalysis. Whether it's splitting water into hydrogen and oxygen in a sort of reverse fuel cell, fermenting the cellulose in agricultural trash to make ethanol, or simply growing plants more cheaply by equipping them to fix nitrogen for themselves, there are many ways to enhance the conversion of sunlight to fuel — but none of these catalytic processes operate on a scale that as yet makes economical sense.

Doin' what comes naturally

Approaches to solar-to-fuel conversion may be divided broadly into those that modify nature — using the techniques of synthetic biology, for example — and those that start from scratch, using materials and chemical processes from the inorganic world. In the middle is a wide and promising range of tactics that borrow from nature and artifice both.

What the natural processes have in common is photosynthesis, a complex and very rapid sequence of reactions in plants and some bacteria that converts incident solar radiation into stored chemical energy. Two protein assemblies are involved. Photosystem II uses light to break molecules of water into oxygen, hydrogen ions (protons), and free electrons. From a plant's point of view the oxygen is waste, but the electrons and protons enable photosystem I to convert carbon dioxide to carbohydrates, fixing the carbon and energizing growth.

In terms of net productivity the process isn't fast or efficient, but it hardly matters: plants are cheap, and they're everywhere. Two examples serve to illustrate the many biologically based schemes for making their stored energy more useful to humans:

One approach is to design organisms that can produce fuels like ethanol or methane directly from sunlight, air, water, and soil. Ethanol from corn is already added to gasoline in this country, although as Tad Patzek of Berkeley Lab's Earth Sciences Division pointed out at the meeting, the energy content of the gasoline and other fossil fuels required to fertilize, cultivate, harvest, transport, and distill corn is significantly greater (at maximum corn-conversion efficiency, an extra five percent) than the energy content of corn ethanol itself. Not to mention the attendant pollution from all that extra fuel use. Clearly a bad bargain.

But take yeast, say — normally unresponsive to light — and modify it to contain bacteriorhodopsin, a bacterial relative of the "visual purple" protein that responds to light in the retina of the eye. Stimulated by sunlight, bacteriorhodopsin in yeast could make ATP, which would fix atmospheric CO 2 , which the yeast could convert directly to ethanol. Antón Vila-Sanjurjo and Carlos Bustamante of the Lab's Physical Biosciences Division (PBD) are at work on just such a synthetic organism.

A different route to biofuels is through more efficient conversion of biomass, including the great mass of trash from agriculture and forestry that now goes to waste. Half the carbon in biomass is tied up in various forms of cellulose, which forms the stiff walls of plant cells. Made of polymerized sugars, carbohydrates consisting almost entirely of hydrogen, carbon, and oxygen, cellulose is broken down by microbes that employ special enzymes — or protein machines, like the large, complex cellulosomes found in the membranes of some species of Clostridium bacteria.

No existing organism works fast enough, or with enough specificity, to efficiently produce hydrogen or methanol from cellulose for human use, however. Unlike yeast or E. coli , Clostridia are difficult to engineer. Nor can cellulosomes easily be transplanted to different species of microbes.

The solution may be to design a new organism from the chromosomes out, with custom-tailored enzymes, cellulose-degrading machinery, metabolic and signaling pathways, and genetic control mechanisms optimized for converting the cellulose in biomass, like sawdust or rice straw, to high-quality fuels. It's a multidisciplinary challenge of the kind that PBD's Jay Keasling has already practiced in the field of medicine, by engineering synthetic bacteria to produce valuable drug compounds that are usually made by plants.

It's not unnatural

Inorganic materials like metal and semiconductor compounds are proven catalysts for splitting water or carbon dioxide into their constituent elements — oxygen plus hydrogen or carbon, including ions and free electrons — when energized by ultraviolet or visible light. Life has made similar use of metals for a long time: Earth's first organisms probably fixed carbon through reactions catalyzed by iron and sulfur. In photosynthesis, higher plants and cyanobacteria use photosystem II instead, splitting water with a tiny molecular complex that incorporates manganese and calcium, and which may very well have been borrowed from minerals by early microbes.

Traditionally, scientists looking for a way to drive solar production of chemical fuels with the aid of solid catalysts haven't spent a lot of time studying plants and microbes; they are concerned with mounting the right inorganic catalyst in the right support structure to produce an efficient, stable system at the right price.

One approach, described by PBD's Heinz Frei, is to embed molecule-sized components for light harvesting and catalysis in a 3-D matrix structured on the nanometer scale, with different stages separated by only a few billionths of a meter. The properties of nanoparticles themselves aid the process; for example, the photon-to-electric-current efficiency of some nanocrystalline catalysts is hundreds of times that of the same materials in bulk. Such a 3-D nanoassembly would be mesoporous, penetrated by microscopic channels to carry off the gaseous products of catalysis.

Another promising idea comes by way of solar-cell research. Semiconductor photovoltaics, those that convert sunlight directly to a flow of electrons, are characterized by specific band gaps, a measure of the energy needed to boost an electron from a semiconductor's mostly full valence band into its mostly empty conduction band. While sunlight is made of photons with many different energies (many different colors), from energetic ultraviolet to lazy infrared, each semiconductor's band gap samples only a narrow part of that spectrum.

Wladek Walukiewicz of the Materials Sciences Division (MSD) described the discovery in 2001 by a group of Berkeley Lab researchers that the band gap of indium gallium nitride (known as a group-III nitride) can be varied across virtually the entire solar spectrum just by varying the proportions of indium and gallium in the alloy. The researchers' first impulse was to make a multijunction solar cell from several layers of these structurally similar alloys, each with a different band gap. While research continues on this idea, the notion of a variable-band-gap alloy has also given rise to a different suggestion: adjusting the band gap to match the optimum reaction potentials for splitting water.

The narrow band gaps of group-III nitrides like indium gallium nitride and gallium nitrogen arsenide come close to bracketing the energies needed to oxidize water (which frees oxygen atoms) and reduce it (which frees hydrogen ions); the band-gap edges can be precisely manipulated by varying the compositional proportions of the alloy. And group-III nitrides resist corrosion better than less efficient oxide semiconductors.

By using such materials to enclose an electrolyte, which could be a liquid, gel, or solid, between a sunlight-energized, oxidizing anode and a reducing cathode — or through other arrangements of the components — it may be possible to achieve what has yet to be demonstrated: stable, efficient conversion of water and sunlight to hydrogen by a photoelectrochemical cell.

No matter how efficient, catalysts made of expensive semiconductors, painstakingly assembled, or complex metal-oxide nanostructures may not be cost effective for a long time to come. An alternative approach is to take a page (or a leaf) from nature and make large amounts of inefficient but cheap, easy to process, flexible catalysts from plastic.

There are many kinds of conducting organic polymers, with band gaps sensitive to different parts of the solar spectrum. The catch is, it may take only a few hours' exposure to sunlight to degrade these polymers and squelch their photoelectric effect. MSD director Paul Alivisatos and his colleagues developed a prototype photovoltaic material by codissolving semiconductor nanorods, which function as positive poles, in a polymer that functions as the negative pole. At the Solar to Fuel meeting Alivisatos raised the question of designing polymers with built-in recycling capability, able to restore and maintain photosensitivity during exposure to sunlight. The challenge is whether this can be done without making the polymers or their manufacture too expensive.

A meeting in the middle

Some of the more intriguing ideas to come out of the workshop were also the more speculative. Among these were proposals for "biomimetic" structures, entirely artificial devices that nevertheless copy life's processes closely.

Dendrimers are branching polymers that derive their name from the Greek word for tree, except these trees can grow into spheroids with multiple branches radiating from a single root. Among the possible applications of these versatile structures are artificial photosystems that harvest light at the ends of the branches and transfer the energy to a central processor for splitting water or carbon dioxide. Jean Fréchet of MSD and the Chemical Sciences Division (CSD) acknowledges that artificial photosynthesis is still a distant target; nevertheless related processes, such as the absorption of light at one energy and its re-emission at a higher energy, have already been demonstrated.

Dendrimers are subject to the same drawbacks as other polymers, including lack of long-term photostability; if self-repairing dendrimers can be designed, the next step may be to attempt to build artificial manganese complexes, like those in photosystem II, incorporating inorganic metal compounds held in place by organic ligands — individual dendrimer molecules that can split water on the nanoscale.

Drawing slightly more distant inspiration from photosystem II, which he called "nature's paradigm of solar-to-fuel," was CSD's Chris Chang, who proposed purely synthetic nanostructures consisting of organic cages — or hybrid inorganic-plus-organic cages — surrounding metal-oxide or other catalysts. If equipped with the right suite of synthetic scaffolds, a single core could perform a variety of chemical reactions, just as in nature. But purely synthetic systems might avoid one problem all too common in nature: the huge time-scale difference between the ultrarapid actions of photosynthesis and the tediously slow conversion of photosynthetic products into biomass.

Chang and many others stressed the need for continuing fundamental research into the chemical processes of photosynthesis. Tom Moore, a professor in the Department of Chemistry and Biochemistry at Arizona State University, reminded the meeting participants that photosynthesis incorporates the most efficient catalysts and enzymes known for splitting water — and the reverse process: "Nature provides the basic paradigms for fuel-cell operation." Thus all approaches to carbon-neutral energy benefit from a better understanding of photosynthesis through computer modeling and new experimental techniques.

Aided by a Department of Energy INCITE grant of one million hours of supercomputing time at DOE's National Energy Research Scientific Computing Center (NERSC) based here, CSD's William Lester, Jr. and Berkeley Lab's Deputy Laboratory Director Graham Fleming have modeled the electronic structures of carotenoids that buffer excess photon energy in photosynthesis. In separate studies Fleming has used new spectrometry techniques to track the flow of excitation energy, in both time and space, in a light-harvesting molecular complex important to photosynthesis in green sulfur bacteria.

Fleming notes that "nature's system for harvesting light is exquisitely effective, wasting no heat because the reactions happen so fast. Unfortunately, current synthetic light-harvesting devices aren't following nature's model."

The Solar to Fuel meeting ended with the recognition by the participants that they'd succeeded in posing even more hard questions than they'd started with — among them, questions about how to balance scientific inspiration with economic reality; how to reconcile the near-perfect efficiency of photosynthesis on the molecular scale with its sluggish net efficiency on the human scale; how much of life's protein "furniture," as MSD's Philip Ross characterized it, needs to be mimicked faithfully in synthetic processes and how much can be discarded; and a reminder from Tad Patzek that no matter how good the tools scientists come up with, "they can't save the planet" — at best they are "wedges" to drive the changes needed to improve the human situation.

To Lab director Steven Chu, defining these questions was an essential first step. His advice to the researchers: "The solutions may not come this week or next month. Let these problems run in the background of your day-to-day thinking, so that you are subconsciously thinking about several big and important problems" — problems whose solutions, whatever they turn out to be, will benefit by being examined from divergent points of view.

Additional information

• More about the computational methods used by Graham Fleming and William Lester, Jr . to investigate photosynthesis
• More about Tad Patzek's research
• More about the research of Carlos Bustamante and his colleagues
• More about Jay Keasling's group
• More about Heinz Frei's research
• More about the work Wladek Walukiewicz and his colleagues have done with the remarkable alloy indium gallium nitride (InGaN)
• More about the hybrid polymer-nanocrystalline solar cell developed by Paul Alivisatos and his colleagues
• More about Jean Fréchet's research
• More about the Chris Chang group
• More about Tom Moore and his work on solar energy conversion
• More about Photosystem II
• Read coverage of the Solar to Fuel meeting by the Berkeley Lab View