The Sun provides Earth with a staggering amount ofenergy—enough to power the great oceanic and atmosphericcurrents, the cycle of evaporation and condensation thatbrings fresh water inland and drives river flow, and the ty-phoons, hurricanes, and tornadoes that so easily destroy thenatural and built landscape. The San Francisco earthquake of1906, with magnitude 7.8, released an estimated 1017 joules ofenergy, the amount the Sun delivers to Earth in one second.Earth’s ultimate recoverable resource of oil, estimated at3 trillion barrels, contains 1.7 × 1022 joules of energy, whichthe Sun supplies to Earth in 1.5 days. The amount of energyhumans use annually, about 4.6 × 1020 joules, is delivered toEarth by the Sun in one hour. The enormous power that the

Sun continuously delivers to Earth, 1.2 × 105 terawatts,dwarfs every other energy source, renewable or nonrenew-able. It dramatically exceeds the rate at which human civi-lization produces and uses energy, currently about 13 TW.

The impressive supply of solar energy is complementedby its versatility, as illustrated in figure 1. Sunlight can be con-verted into electricity by exciting electrons in a solar cell. Itcan yield chemical fuel via natural photosynthesis in greenplants or artificial photosynthesis in human-engineered sys-tems. Concentrated or unconcentrated sunlight can produceheat for direct use or further conversion to electricity.1

Despite the abundance and versatility of solar energy, weuse very little of it to directly power human activities. Solar

© 2007 American Institute of Physics, S-0031-9228-0703-010-4 March 2007 Physics Today 37

Solar energy conversionGeorge W. Crabtree and Nathan S. Lewis

George Crabtree is a senior scientist at Argonne National Laboratory in Argonne, Illinois, and director of its materials science division.Nathan Lewis is a professor of chemistry at the California Institute of Technology in Pasadena, California, and director of the molecu-lar materials research center at Caltech’s Beckman Institute.

NC O

NN

NN

HHH N

C ON

N

NN

HHH

Holes

Electrons

H O2

H O2

CO2

CO2

O2

O2

Sugar

H , CHCH OH

2 4

3

,

Naturalphotosynthesis

(biomass)Artificial

photosynthesis

50–200 °CSpace, water

heating

400–3000 °CHeat engines,

electricity generation,industrial processes

Solar electric Solar fuel Solar thermal

Figure 1. Solar photons convert naturally into three forms of energy—electricity, chemical fuel, and heat—that link seam-lessly with existing energy chains. Despite the enormous energy flux supplied by the Sun, the three conversion routes supplyonly a tiny fraction of our current and future energy needs. Solar electricity, at between 5 and 10 times the cost of electricityfrom fossil fuels, supplies just 0.015% of the world’s electricity demand. Solar fuel, in the form of biomass, accounts forapproximately 11% of world fuel use, but the majority of that is harvested unsustainably. Solar heat provides 0.3% of theenergy used for heating space and water. It is anticipated that by the year 2030 the world demand for electricity will doubleand the demands for fuel and heat will increase by 60%. The utilization gap between solar energy’s potential and our use ofit can be overcome by raising the efficiency of the conversion processes, which are all well below their theoretical limits.

If solar energy is to become a practical alternative to fossil fuels, we must have efficient ways to con-vert photons into electricity, fuel, and heat. The need for better conversion technologies is a drivingforce behind many recent developments in biology, materials, and especially nanoscience.

38 March 2007 Physics Today www.physicstoday.org

electricity accounts for a minuscule 0.015% of world electric-ity production, and solar heat for 0.3% of global heating ofspace and water. Biomass produced by natural photosynthe-sis is by far the largest use of solar energy; its combustion orgasification accounts for about 11% of human energy needs.However, more than two-thirds of that is gathered unsus-tainably—that is, with no replacement plan—and burned insmall, inefficient stoves where combustion is incomplete andthe resulting pollutants are uncontrolled.

Between 80% and 85% of our energy comes from fossilfuels, a product of ancient biomass stored beneath Earth’ssurface for up to 200 million years. Fossil-fuel resources areof finite extent and are distributed unevenly beneath Earth’ssurface. When fossil fuels are turned into useful energythough combustion, they produce greenhouse gases andother harmful environmental pollutants. In contrast, solarphotons are effectively inexhaustible and unrestricted bygeopolitical boundaries. Their direct use for energy produc-tion does not threaten health or climate. The solar resource’smagnitude, wide availability, versatility, and benign effect onthe environment and climate make it an appealing energysource.

Raising efficiencyThe enormous gap between the potential of solar energy andour use of it is due to cost and conversion capacity. Fossil fuelsmeet our energy demands much more cheaply than solar al-ternatives, in part because fossil-fuel deposits are concen-trated sources of energy, whereas the Sun distributes photonsfairly uniformly over Earth at a more modest energy density.The use of biomass as fuel is limited by the production ca-pacity of the available land and water. The cost and capacitylimitations on solar energy use are most effectively addressedby a single research objective: cost effectively raising conver-sion efficiency.

The best commercial solar cells based on single-crystalsilicon are about 18% efficient. Laboratory solar cells basedon cheaper dye sensitization of oxide semiconductors aretypically less than 10% efficient, and those based on evencheaper organic materials are 2–5% efficient. Green plantsconvert sunlight into biomass with a typical yearly averagedefficiency of less than 0.3%. The cheapest solar electricitycomes not from photovoltaics but from conventional induc-tion generators powered by steam engines driven by solarheat, with efficiencies of 20% on average and 30% for the bestsystems. Those efficiencies are far below their theoretical lim-its. Increasing efficiency reduces cost and increases capacity,which raises solar energy to a new level of competitiveness.

Dramatic cost-effective increases in the efficiency ofsolar energy conversion are enabled by our growing abilityto understand and control the fundamental nanoscale phe-nomena that govern the conversion of photons into otherforms of energy. Such phenomena have, until recently, beenbeyond the reach of our best structural and spectroscopicprobes. The rise of nanoscience is yielding new fabricationtechniques based on self-assembly, incisive new probes ofstructure and dynamics at ever-smaller length and timescales, and the new theoretical capability to simulate assem-blies of thousands of atoms. Those advances promise the ca-pability to understand and control the underlying structuresand dynamics of photon conversion processes.

ElectricitySolar cells capture photons by exciting electrons across thebandgap of a semiconductor, which creates electron–holepairs that are then charge separated, typically by p–n junc-tions introduced by doping. The space charge at the p–n

junction interface drives electrons in one direction and holesin the other, which creates at the external electrodes a po-tential difference equal to the bandgap, as sketched in theleft panel of figure 1. The concept and configuration are sim-ilar to those of a semiconductor diode, except that electronsand holes are introduced into the junction by photon excita-tion and are removed at the electrodes.

With their 1961 analysis of thermodynamic efficiency,William Shockley and Hans Queisser established a milestonereference point for the performance of solar cells.2 The analy-sis is based on four assumptions: a single p–n junction, oneelectron–hole pair excited per incoming photon, thermal re-laxation of the electron–hole pair energy in excess of thebandgap, and illumination with unconcentrated sunlight.Achieving the efficiency limit of 31% that they established forthose conditions remains a research goal. The best single-crystal Si cells have achieved 25% efficiency in the laboratoryand about 18% in commercial practice. Cheaper solar cellscan be made from other materials,3 but they operate at sig-nificantly lower efficiency, as shown in the table above. Thin-film cells offer advantages beyond cost, including pliability,as in figure 2, and potential integration with preexistingbuildings and infrastructure. Achieving high efficiency frominexpensive materials with so-called third-generation cells,indicated in figure 3, is the grand research challenge for mak-ing solar electricity dramatically more affordable.

The Shockley–Queisser limit can be exceeded by violat-ing one or more of its premises. Concentrating sunlight al-lows for a greater contribution from multi-photon processes;that contribution increases the theoretical efficiency limit to41% for a single-junction cell with thermal relaxation. A cellwith a single p–n junction captures only a fraction of the solarspectrum: photons with energies less than the bandgap arenot captured, and photons with energies greater than thebandgap have their excess energy lost to thermal relaxation.Stacked cells with different bandgaps capture a greater frac-tion of the solar spectrum; the efficiency limit is 43% for twojunctions illuminated with unconcentrated sunlight, 49% forthree junctions, and 66% for infinitely many junctions.

The most dramatic and surprising potential increase inefficiency comes from carrier multiplication,4 a quantum-dotphenomenon that results in multiple electron–hole pairs for a single incident photon. Carrier multiplication was

*As verified by the National Renewable Energy Laboratory.Organic cell efficiencies of up to 5% have been reported in the literature.

Photovoltaic conversion efficienciesLaboratory

best*Thermodynamic

limit

Single junction 31%

Silicon (crystalline) 25%

Silicon (nanocrystalline) 10%

Gallium arsenide 25%

Dye sensitized 10%

Organic 3%

Multijunction 32% 66%

Concentrated sunlight (single junction) 28% 41%

Carrier multiplication 42%

discussed by Arthur Nozik in 2002 and ob-served by Richard Schaller and VictorKlimov two years later. Nanocrystals oflead selenide, lead sulfide, or cadmiumselenide generate as many as seven elec-trons per incoming photon, which sug-gests that efficient solar cells might bemade with such nanocrystals. In bulk-semiconductor solar cells, when an inci-dent photon excites a single electron–holepair, the electron–hole pair energy in ex-cess of the bandgap is likely to be lost tothermal relaxation, whereas in some nano-crystals most of the excess energy can ap-pear as additional electron–hole pairs. Ifthe nanocrystals can be incorporated intoa solar cell, the extra pairs could be tappedoff as enhanced photocurrent, whichwould increase the efficiency of the cell.

Hot-electron extraction provides an-other way to increase the efficiency ofnanocrystal-based solar cells: tapping offenergetic electrons and holes before theyhave time to thermally relax.5 Hot elec-trons boost efficiency by increasing theoperating voltage above the bandgap,whereas carrier multiplication increases the operating cur-rent. Femtosecond laser and x-ray techniques can provide thenecessary understanding of the ultrafast decay processes inbulk semiconductors and their modification in nanoscalegeometries that will enable the use of hot-electron phenom-ena in next-generation solar cells.

Although designs have been proposed for quantum-dotsolar cells that benefit from hot electrons or carrier multipli-cation, significant obstacles impede their implementation.We cannot attach wires to nanocrystals the way we do to bulksemiconductors; collecting the electrons from billions of tinydots and putting them all into one current lead is a problemin nanoscale engineering that no one has solved yet. A sec-ond challenge is separating the electrons from the holes, thejob normally done by the space charge at the p–n junction inbulk solar cells. Those obstacles must be overcome beforepractical quantum-dot cells can be constructed.5

Dye-sensitized solar cells, introduced by MichaelGrätzel and coworkers in 1991, create a new paradigm forphoton capture and charge transport in solar conversion.6 Ex-pensive Si, which does both of those jobs in conventionalcells, is replaced by a hybrid of chemical dye and the inex-pensive wide-bandgap semiconductor titanium dioxide. Thedye, analogous to the light-harvesting chlorophyll in greenplants, captures a photon, which elevates one of its electronsto an excited state. The electron is then quickly transferred tothe conduction band of a neighboring TiO2 nanoparticle, andit drifts through an array of similar nanoparticles to the ex-ternal electrode. The hole left in the dye molecule recombineswith an electron carried to it through an electrolyte from thecounter electrode by an anion such as I−. In addition to usingcheaper materials, the scheme separates the absorption spec-trum of the cell from the bandgap of the semiconductor, sothe cell sensitivity is more easily tuned to match the solarspectrum. The cell efficiency depends on several kinds ofnanoscale charge dynamics, such as the way the electronsmove across the dye–TiO2 and dye–anion interfaces, and theway charges move through the dye, the TiO2 nanoparticlearray, and the electrolyte. The development of new dyes andshuttle ions and the characterization and control of the dy-namics through time-resolved spectroscopy are vibrant and

promising research areas. An equally important researchchallenge is the nanoscale fabrication of dye-sensitized cellsto minimize the transport distances in the dye and semi-conductor and maximize the electron-transfer rate at the interfaces.

FuelOver the past 3 billion years, Nature has devised a remark-ably diverse set of pathways for converting solar photons intochemical fuel. An estimated 100 TW of solar energy go intophotosynthesis, the production of sugars and starches fromwater and carbon dioxide via endothermic reactions facili-tated by catalysts. Although plants have covered Earth ingreen in their quest to capture solar photons, their overallconversion efficiency is too low to readily satisfy the humandemand for energy. The early stages of photosynthesis are ef-ficient: Two molecules of water are split to provide four pro-tons and electrons for subsequent reactions, and an oxygenmolecule is released into the atmosphere. The inefficiency liesin the later stages, in which carbon dioxide is reduced to formthe carbohydrates that plants use to grow roots, leaves, andstalks. The research challenge is to make the overall conver-sion process between 10 and 100 times more efficient by im-proving or replacing the inefficient stages of photosynthesis.

There are three routes to improving the efficiency of pho-tosynthesis-based solar fuel production: breeding or geneti-cally engineering plants to grow faster and produce more bio-mass, connecting natural photosynthetic pathways in novelconfigurations to avoid the inefficient steps, and using artifi-cial bio-inspired nanoscale assemblies to produce fuel fromwater and CO2. The first route is the occupation of a thrivingindustry that has produced remarkable increases in plantyields, and we will not discuss it further. The second and thirdroutes, which involve more direct manipulation of photo-synthetic pathways, are still in their early stages of research.

www.physicstoday.org March 2007 Physics Today 39

Figure 2. Novel conducting polymers enable solar cells thatare flexible, inexpensive, and versatile. The new materialscan be coated or printed onto flexible or rigid surfaces.(Image courtesy of Konarka Technologies.)

Nature provides many examplesof metabolic systems that convertsunlight and chemicals into high-energy fuels. Green plants use anelaborate complex of chlorophyllmolecules coupled to a reaction cen-ter to split water into protons, elec-trons, and oxygen. Bacteria use thehydrogenase enzyme to create hydro-gen molecules from protons and elec-trons. More than 60 species ofmethane-producing archaea, rem-nants from early Earth when the at-mosphere was reducing instead of ox-idizing, use H2 to reduce CO2 to CH4.Anaerobic organisms such as yeastsand bacteria use enzymes to fermentsugars into alcohols.

In nature, the metabolic path-ways are connected in complicatednetworks that have evolved for or-ganisms’ survival and reproduction,not for fuel production. The efficientsteps that are relevant for fuel pro-duction might conceivably be isolatedand connected directly to one anotherto produce fuels such as H2, CH4, oralcohols. Hybridizing nature in thatway takes advantage of the elaboratemolecular processes that biology hasevolved and that are still beyondhuman reach, while eliminating theinefficient steps not needed for fuelproduction. For example, the protonsand electrons produced in the earlystages of photosynthesis could link tohydrogenase to produce H2, and a fur-ther connection to methanogenicarchaea could produce CH4. The chal-lenges are creating a functional inter-face between existing metabolic mod-ules, achieving a competitive efficiency for the modifiednetwork, and inducing the organism hosting the hybrid sys-tem to reproduce. The ambitious vision of hybrids thatproduce energy efficiently sets a basic research agenda tosimultaneously advance the frontiers of biology, materialsscience, and energy conversion.

Artificial photosynthesis takes the ultimate step of usinginanimate components to convert sunlight into chemicalfuel.7,8 Although the components do not come from nature,the energy conversion routes are bio-inspired. Remarkableprogress has been made in the field.8 Light harvesting andcharge separation are accomplished by synthetic antennaslinked to a porphyrin-based charge donor and a fullerene ac-ceptor, as shown in figure 4. The assembly is embedded in anartificial membrane, in the presence of quinones that act asproton shuttles, to produce a light-triggered proton gradientacross the membrane. The proton gradient can do usefulwork, such as powering the molecular synthesis of adenosinetriphosphate by mechanical rotation of natural ATP synthaseinserted into the membrane. Under the right conditions, therequired elements self-assemble to produce a membrane-based chemical factory that transforms light into the chemi-cal fuel ATP, molecule by molecule at ambient temperature,in the spirit of natural photosynthesis.

Such remarkable achievements illustrate the promise ofproducing fuel directly from sunlight without the use of

biological components. Many fundamental challenges mustbe overcome, however. The output of the above energy con-version chain is ATP, not a fuel that links naturally to human-engineered energy chains. The last step relies on the naturalcatalyst ATP synthase, a highly evolved protein whose func-tion we cannot yet duplicate artificially. Laboratory approxi-mations of biological catalysts have catalytic activities thatare often orders of magnitude lower than those of their bio-logical counterparts, which indicates the importance of sub-tle features that we are not yet able to resolve or to reproduce.

Solar fuels can be created in an alternate, fully nonbio-logical way based on semiconductor solar cells rather than onphotosynthesis. In photoelectrochemical conversion, thecharge-separated electrons and holes are used locally to splitwater or reduce CO2 at the interface with an electrolytic so-lution, rather than being sent through an external circuit todo electrical work.9 Hydrogen was produced at the elec-trode–water interface with greater than 10% efficiency byAdam Heller in 1984 and by Oscar Khaselev and John Turnerin 1998, but the fundamental phenomena involved remainmysterious, and the present devices are not practical. Apromising way to improve them is by tailoring the nanoscalearchitecture of the electrode–electrolyte interface to promotethe reaction of interest. A better understanding of how indi-vidual electrons negotiate the electrode–electrolyte interface

40 March 2007 Physics Today www.physicstoday.org

100 200 300 400 500

100

80

60

40

20P

ER

CE

NT

EF

FIC

IEN

CY

Shockley–Queisser limit

$0.10/Wp $0.20/Wp $0.50/Wp

$1.00/Wp

$3.50/Wp

III

III

COST IN DOLLARS PER SQUARE METER

Figure 3. The three generations of solar cells. First-generation cells are based on expensive silicon wafers and make up 85% of the current commer-cial market. Second-generation cells are based on thin films of materials suchas amorphous silicon, nanocrystalline silicon, cadmium telluride, or copperindium selenide. The materials are less expensive, but research is needed toraise the cells’ efficiency to the levels shown if the cost of delivered power is tobe reduced. Third-generation cells are the research goal: a dramatic increasein efficiency that maintains the cost advantage of second-generation materials.Their design may make use of carrier multiplication, hot electron extraction,multiple junctions, sunlight concentration, or new materials. The horizontalaxis represents the cost of the solar module only; it must be approximatelydoubled to include the costs of packaging and mounting. Dotted lines indicatethe cost per watt of peak power (Wp). (Adapted from ref. 2, Green.)

is needed before H2 can be producedwith greater efficiency or more com-plex reactions can be designed for re-ducing CO2 to useful fuels.

HeatThe first step in traditional energy con-version is the combustion of fuel, usu-ally fossil fuel, to produce heat. Heatproduced by combustion may be usedfor heating space and water, cooking,or industrial processes, or it may befurther converted into motion or elec-tricity. The premise of solar thermalconversion is that heat from the Sunreplaces heat from combustion; fossil-fuel use and its threat to the environ-ment and climate are thus reduced.

Unconcentrated sunlight canbring the temperature of a fluid toabout 200 °C, enough to heat spaceand water in residential and commer-cial applications. Many regions usesolar water heating, though in only afew countries, such as Cyprus andIsrael, does it meet a significant frac-tion of the demand. Concentration ofsunlight in parabolic troughs pro-duces temperatures of 400 °C, andparabolic dishes can produce temper-atures of 650 °C and higher.10,11 Powertowers, in which a farm of mirrors onthe ground reflects to a common re-ceiver at the top of a tower, can yieldtemperatures of 1500 °C or more.10,12

The high temperatures of solar powertowers are attractive for thermo-chemical water splitting and solar-driven reforming of fossilfuels to produce H2.11

The temperatures produced by concentrated sunlightare high enough to power heat engines, whose Carnot effi-ciencies depend only on the ratio of the inlet and outlet tem-peratures. Steam engines driven by solar heat and connectedto conventional generators currently supply the cheapestsolar electricity. Nine solar thermal electricity plants that usetracking parabolic-trough concentrators were installed inCalifornia’s Mojave Desert between 1984 and 1991. Thoseplants still operate, supplying 354 MW of peak power to thegrid. Their average annual efficiency is approximately 20%,and the most recently installed can achieve 30%.

Although those efficiencies are the highest for anywidely implemented form of solar conversion, they are mod-est compared to the nearly 60% efficiency of the best gas-firedelectricity generators. Achieving greater efficiency for solarconversion requires large-scale plants with operating tem-peratures of 1500 °C or more, as might be produced by powertowers. Another alternative, still in the exploration stage, is ahybrid of two conversion schemes: A concentrated solarbeam is split into its visible portion for efficient photovoltaicconversion and its high-energy portion for conversion to heatthat is converted to electricity through a heat engine.10

Thermoelectric materials, which require no moving partsto convert thermal gradients directly into electricity, are an at-tractive possibility for reliable and inexpensive electricity pro-duction.13 Charge carriers in a thermal gradient diffuse fromhot to cold, driven by the temperature difference but creatingan electric current by virtue of the charge on each carrier. The

strength of the effect is measured by the thermopower, the ratioof the voltage produced to the applied temperature difference.Although the thermoelectric effect has been known for nearly200 years, materials that can potentially convert heat to elec-tricity efficiently enough for widespread use have emergedonly since the 1990s.13 Efficient conversion depends on mini-mizing the thermal conductivity of a material, so as not to short-circuit the thermal gradient, while maximizing the material’selectrical conductivity and thermopower. Achieving such acombination of opposites requires the separate tuning of sev-eral material properties: the bandgap, the electronic density ofstates, and the electron and phonon lifetimes. The most prom-ising materials are nanostructured composites. Quantum-dotor nanowire substructures introduce spikes in the density ofstates to tune the thermopower (which depends on the deriva-tive of the density of states), and interfaces between the com-posite materials block thermal transport but allow electricaltransport, as discussed by Lyndon Hicks and Mildred Dressel-haus in 1993.14 Proof of concept for interface control of thermaland electrical conductivity was achieved by 2001 with thin-filmsuperlattices of Bi2Te3/Sb2Te3 and PbTe/PbSe, which performedtwice as well as bulk-alloy thermoelectrics of the same materi-als. The next challenges are to achieve the same performance innanostructured bulk materials that can handle large amountsof power and to use nanodot or nanowire inclusions to controlthe thermopower. Figure 5 shows encouraging progress: struc-turally distinct nanodots in a bulk matrix of the thermoelectricmaterial Ag0.86Pb18SbTe20. Controlling the size, density, and dis-tribution of such nanodot inclusions during bulk synthesiscould significantly enhance thermoelectric performance.15

www.physicstoday.org March 2007 Physics Today 41

H CO3

H CO3

H CO3

H CO3

H CO3

H CO3OCH3

OCH3

OCH3

OCH3

NN

Zn NN

O

N NH

H C3

e–

Figure 4. An artificial antenna–reaction-center complex that mimics the earlystages of photosynthesis. The central hexaphenylbenzene core provides structureand rigidity for the surrounding wheel of five bis(phenylethynyl)anthraceneantennas that gather light at 430–475 nanometers. The energy is transferred toa porphyrin complex in 1–10 picoseconds (orange arrows), where it excites anelectron that is transferred to the fullerene acceptor in 80 ps; the resultingcharge-separated state has a lifetime of 15 nanoseconds. Complexes such asthe one shown provide the first steps in artificial photosynthesis. They have thepotential to drive further chemical reactions, such as the oxidation of water toproduce H2 or the reduction of CO2 to CH4, alcohol, or other fuel. (Adaptedfrom ref. 16.)

Storage and distributionSolar energy presents a scientific challenge beyond the effi-cient conversion of solar photons to electricity, fuel, and heat.Once conversion on a large scale is achieved, we must findways to store the large quantities of electricity and heat thatwe will produce. Access to solar energy is interrupted by nat-ural cycles of day–night, cloudy–sunny, and winter–summervariation that are often out of phase with energy demand.Solar fuel production automatically stores energy in chemi-cal bonds. Electricity and heat, however, are much more dif-ficult to store. Cost effectively storing even a fraction of ourpeak demand for electricity or heat for 24 hours is a task wellbeyond present technology.

Storage is such an imposing technical challenge that in-novative schemes have been proposed to minimize its need.Baseload solar electricity might be generated on constella-tions of satellites in geosynchronous orbit and beamed toEarth via microwaves focused onto ground-based receivingantennas. A global superconducting grid might direct elec-tricity generated in sunny locations to cloudy or dark loca-tions where demand exceeds supply. But those schemes, too,are far from being implemented. Without cost-effective stor-age and distribution, solar electricity can only be a peak-shaving technology for producing power in bright daylight,acting as a fill for some other energy source that can providereliable power to users on demand.

OutlookThe Sun has the enormous untapped potential to supply ourgrowing energy needs. The barrier to greater use of the solar

resource is its high cost relative to the cost of fossilfuels, although the disparity will decrease with the ris-ing prices of fossil fuels and the rising costs of mitigat-ing their impact on the environment and climate. Thecost of solar energy is directly related to the low con-version efficiency, the modest energy density of solarradiation, and the costly materials currently required.The development of materials and methods to improvesolar energy conversion is primarily a scientific chal-lenge: Breakthroughs in fundamental understandingought to enable marked progress. There is plenty ofroom for improvement, since photovoltaic conversionefficiencies for inexpensive organic and dye-sensitizedsolar cells are currently about 10% or less, the conver-sion efficiency of photosynthesis is less than 1%, andthe best solar thermal efficiency is 30%. The theoreticallimits suggest that we can do much better.

Solar conversion is a young science. Its majorgrowth began in the 1970s, spurred by the oil crisisthat highlighted the pervasive importance of energy toour personal, social, economic, and political lives. Incontrast, fossil-fuel science has developed over morethan 250 years, stimulated by the Industrial Revolu-tion and the promise of abundant fossil fuels. The sci-ence of thermodynamics, for example, is intimately in-tertwined with the development of the steam engine.The Carnot cycle, the mechanical equivalent of heat,

and entropy all played starring roles in the development ofthermodynamics and the technology of heat engines. Solar-energy science faces an equally rich future, with nanoscienceenabling the discovery of the guiding principles of photonicenergy conversion and their use in the development of cost-competitive new technologies.This article is based on the conclusions contained in the report1 of theUS Department of Energy Basic Energy Sciences Workshop on SolarEnergy Utilization, April 18–21, 2005. We served as chair (Lewis) andcochair (Crabtree) of the workshop and were principal editors of thereport. We acknowledge the US Department of Energy for support ofboth the workshop and preparation of the manuscript.

References1. N. S. Lewis, G. W. Crabtree, eds., Basic Research Needs for Solar

Energy Utilization: Report of the Basic Energy Sciences Workshop onSolar Energy Utilization, April 18–21, 2005, US Department ofEnergy Office of Basic Energy Sciences (2005), available athttp://www.sc.doe.gov/bes/reports/abstracts.html#SEU.

2. W. Shockley, H. J. Queisser, J. Appl. Phys. 32, 510 (1961); M. A.Green, Third Generation Photovoltaics: Advanced Solar Energy Con-version, Springer, New York (2003); Y. Hamakawa, ed., Thin-FilmSolar Cells: Next Generation Photovoltaics and Its Applications,Springer, New York (2006); P. Würfel, Physics of Solar Cells: FromPrinciples to New Concepts, Wiley, Hoboken, NJ (2005).

3. J. Liu et al., J. Am. Chem. Soc. 126, 6550 (2004).4. A. J. Nozik, Physica E (Amsterdam) 14, 115 (2002); R. D. Schaller,

V. I. Klimov, Phys. Rev. Lett. 96, 097402 (2006); J. E. Murphy et al.,J. Am. Chem. Soc. 128, 3241 (2006).

5. A. J. Nozik, Inorg. Chem. 44, 6893 (2005).6. B. O’Regan, M. Grätzel, Nature 353, 737 (1991); M. Grätzel,

Nature 414, 338 (2001). 7. M. R. Wasielewski, J. Org. Chem. 71, 5051 (2006).8. D. Gust, T. Moore, A. Moore, Acc. Chem. Res. 34, 40 (2001).9. N. S. Lewis, Inorg. Chem. 44, 6900 (2005); A. Heller, Science 223,

1141 (1984); O. Khaselev, J. A. Turner, Science 280, 425 (1998).10. D. Mills, Sol. Energy 76, 19 (2004).11. A. Steinfeld, Sol. Energy 78, 603 (2005).12. C. Dennis, Nature 443, 23 (2006).13. R. Service, Science 306, 806 (2004).14. L. D. Hicks, M. S. Dresselhaus, Phys. Rev. B 47, 12727 (1993); 47,

16631 (1993); R. Venkatasubramanian et al., Nature 413, 597 (2001).15. E. Quarez et al., J. Am. Chem. Soc. 127, 9177 (2005).16. G. Kodis et al., J. Am. Chem. Soc. 128, 1818 (2006). �

42 March 2007 Physics Today www.physicstoday.org

Figure 5. A nanodot inclusion in the bulk thermoelectricmaterial Ag0.86Pb18SbTe20, imaged with high-resolution trans-mission electron microscopy. Despite a lattice mismatch of2–5%, the nanodot (indicated by the dotted line) is almostperfectly coherently embedded in the matrix. The arrowsshow two dislocations near the interface, and the white boxindicates the unit cell. The nanodot is rich in silver andantimony relative to the matrix. (Adapted from ref. 15.)