FEATURE ARTICLE Meeting the Clean Energy Demand: Nanostructure Architectures for Solar Energy Conversion Prashant V. Kamat* Notre Dame Radiation Laboratory, Department of Chemistry & Biochemistry and Department of Chemical & Biomolecular Engineering, Notre Dame, Indiana 46556-5674 ReceiVed: October 23, 2006 In Final Form: December 6, 2006 The increasing energy demand in the near future will force us to seek environmentally clean alternative energy resources. The emergence of nanomaterials as the new building blocks to construct light energy harvesting assemblies has opened up new ways to utilize renewable energy sources. This article discusses three major ways to utilize nanostructures for the design of solar energy conversion devices: (i) Mimicking photosynthesis with donor-acceptor molecular assemblies or clusters, (ii) semiconductor assisted photocatalysis to produce fuels such as hydrogen, and (iii) nanostructure semiconductor based solar cells. This account further highlights some of the recent developments in these areas and points out the factors that limit the efficiency optimization. Strategies to employ ordered assemblies of semiconductor and metal nanoparticles, inorganic-organic hybrid assemblies, and carbon nanostructures in the energy conversion schemes are also discussed. Directing the future research efforts toward utilization of such tailored nanostructures or ordered hybrid assemblies will play an important task in achieving the desired goal of cheap and efficient fuel production (e.g., solar hydrogen production) or electricity (photochemical solar cells). The Energy Challenge The economic growth in many parts of the world during the past decade was able to be sustained because of the affordable energy prices. The dependence on oil and electricity has made energy a vital component of our everyday needs. The recent hike in oil and gas prices has prompted everyone to take a careful look at the issues dealing with our energy supply and demand. In the 20th century, the population quadrupled and our energy demand went up 16 times. The exponential energy demand is exhausting our fossil fuel supply at an alarming rate.1,2 About 13 terawatts (TW) of energy is currently needed to sustain the lifestyle of 6.5 billion people worldwide. By year 2050, we will need an additional 10 TW of clean energy to maintain the current lifestyle. The End of Cheap Oil. Emergence of more than a billion new consumers from 20 developing countries with a newly acquired spending capacity is increasing global CO2 emission at an alarming rate.3 For example, China and India, two countries with the largest numbers of new consumers have been adding new cars at an average annual increase of 19% and 14%, respectively. Motor vehicles alone account for more than 50% of air pollution in these countries.3 Whereas the consumer growth in the developing countries is a good marker for predicting a brighter global economic outlook, it undermines the demand for additional energy resources and the overall impact on the environment. According to Hubbert,5-7 the exponential growth we have seen in the last century is a transient phenomenon, and the fossil fuel production will follow the trend of a bell shaped curve. The peak oil production in the mainland U.S. seen in 1970 followed this predicted curve thus giving the credibility to his model. His projected growth for the worldwide oil prediction was modified (Figure 1) by Campbell4,8,9 and Deffeyes.10 According to this prediction, the peak production will be attained within the next decade. In the near term, we are not about to run out of oil, but the production will attain the peak. Of the 2000 billion barrels of net global oil reserve, we have already found 1800 billion barrels of which 875 billion barrels have been consumed.10 For every billion barrels of new oil discovered, we are consuming 4 billion barrels. Undoubtedly, new technologies can facilitate the extraction of oil from hard to obtain areas and sustain the production of each site longer, but it will have no impact on the reserves themselves. Extraction of oil from tar sands and shale will not be cheaper as it will demand additional energy for the extraction of oil from these sources. Coal and natural gas are likely to supplement the energy needs but this fossil fuel production will follow the Hubbert peak before the end of this century. The flow of energy supply by various sources in sustaining the population growth from 1600 to 2200 is shown in Figure 2. The role of oil and gas as per this model will have a significant impact as a major energy source for only a short duration. Diversification of our energy supply and polit- ical and social compromise for conservation will become inevitable if we need to maintain a healthy global economic growth.1,2,11,12 Green House Gas Emission. Another important consider- ation of increasing the energy production based on fossil fuels is its impact on the environment. Global warming from the fossil fuel greenhouse gases which contribute to the climate changes is becoming a major concern.13 Recent scientific reports point * E-mail: pkamat@nd.edu. Web site: http://www.nd.edu/���pkamat. 2834 J. Phys. Chem. C 2007, 111, 2834-2860 10.1021/jp066952u CCC: $37.00 �� 2007 American Chemical Society Published on Web 02/01/2007

out a higher global mean surface temperature and melting of arctic ice. The surface temperature of Atlantic Ocean today is higher than it has been for at least a millennium making the tropical storms and hurricanes stronger than ever. The United Nations Framework Convention on Climate Change calls for ������stabilization of greenhouse-gas concentrations in the atmo- sphere at a level that would prevent dangerous anthropogenic interference with the climate system 10 TW (10 �� 1012 watts) of carbon-emission-free power needs to be produced by the year 2050, almost equivalent to the power provided by all of today���s energy sources combined���.14,15 Meeting Clean Energy Demand. In order to meet the increasing energy demand in the near future, we will be forced to seek environmentally clean alternative energy resources.11,16,17 Three major options are at our disposal to tackle the 10 TW clean energy generation in the coming years. These include carbon neutral energy (fossil fuel in conjunction with carbon sequestration), nuclear power, and renewable energy. If we have to produce 10 TW energy using fossil fuels without affecting the environment, we need to find secure storage for 25 billion metric tons of CO2 produced annually (equal to the volume of 12500 km3 or the volume of Lake Superior!). Should nuclear power be the alternate source of energy, we will require construction of a new 1 GW (gigawatt)-electric nuclear fission plant everyday for the next 50 years somewhere on this planet. Renewable energy can be tapped from the available resources: hydroelectric resource (0.5 TW), from all tides & ocean currents (2 TW), geothermal integrated over all of the land area (12 TW), globally extractable wind power (2-4 TW), and solar energy striking the earth (120,000 TW). Among these options solar energy stands out as the most viable choice to meet our energy demand. Despite this vast resource, the energy produced from solar remains less than 0.01% of the total energy demand. Although renewable energy such as solar radiation is ideal to meet the projected demand, it requires new initiatives to harvest incident photons with greater efficiency.18,19 The single- crystal silicon based photovoltaic devices that are commercially available for installation deliver power with a 15% efficiency. These first generation devices suffer from high cost of manufacturing and installation. The second generation devices consisting of CuInGaSe2 (CIGS) polycrystalline semiconductor thin films can bring down the price significantly, but their efficiency needs to be enhanced in order to make them practically viable. Now being aimed are the third generation devices that can deliver high efficiency devices at an economi- cally viable cost. Our ability to design nanostructured semi- conductors, organic-inorganic hybrid assemblies, and molecular assemblies opens up new ways to design such third generation light energy conversion devices. Nanotechnology to the Rescue? During the past decade, nanomaterials have emerged as the new building blocks to construct light energy harvesting assemblies. Organic and inorganic hybrid structures that exhibit improved selectivity and efficiency toward catalytic processes have been designed. Size dependent properties such as size quantization effects in semiconductor nanoparticles and quan- tized charging effects in metal nanoparticles provide the basis for developing new and effective systems.20-26 These nano- structures provide innovative strategies for designing next generation energy conversion devices.27-32 Recent efforts to synthesize nanostructures with well-defined geometrical shapes (e.g., solid and hollow spheres, prisms, rods, and wires) and their assembly as 2- and 3-dimensional assemblies has further expanded the possibility of developing new strategies for light energy conversion.33-45 There are three major ways that one can utilize nanostructures for the design of solar energy conversion devices (Figure 3). The first one is to mimic photosynthesis with donor-acceptor molecular assemblies and clusters. The second one is the semiconductor assisted photocatalysis to produce fuels such as hydrogen. The third and most promising category is the nanostructure semiconductor based solar cells. This account highlights some of the recent developments in these areas and points out the factors that limit the efficiency optimization. Strategies to employ ordered assemblies of semiconductor and metal nanoparticles, inorganic-organic hybrid assemblies, and carbon nanostructures in the energy conversion schemes are discussed in the following sections. 1. Donor-Acceptor Hybrid Assemblies In photosynthesis, light energy is converted into chemical energy by green plants. The essential roles of chlorophyll a, (Chla) are to capture solar energy, transfer the excitation energy to special locations, the reaction centers, and bring about the charge separation for the subsequent electron-transfer processes. Based on the principle of photosynthesis, a variety of donor- acceptor dyads and triads have been synthesized as light harvesting assemblies.46-53 Of particular interest has been the study of donor-acceptor systems containing Chla and porphy- rins that can mimic the photoinduced electron-transfer process of natural photosynthesis. Despite the success of achieving long- lived charge separation, the use of dyads and triads in solar cells is rather limited.48,54-56 A better understanding of the underlying physical principles of light absorption, energy transfer, radiative and nonradiative excited-state decay, electron Prashant V. Kamat is currently a Professor of Chemistry and Biochem- istry, a Senior Scientist at Radiation Laboratory, and a Concurrent Professor in the Department of Chemical and Biomolecular Engineering, University of Notre Dame. A native of Binaga, India, he earned his masters (1974) and doctoral degrees (1979) in Physical Chemistry from the Bombay University, and carried out his postdoctoral research at Boston University (1979-1981) and University of Texas at Austin (1981-1983). He joined Notre Dame Radiation Laboratory in 1983 and initiated the photoelectrochemical investigation of semiconductor nanoparticles. Dr. Kamat���s research has made significant contributions to three areas : (1) photoinduced catalytic reactions using semiconductor and metal nanoparticles, nanostructures and nanocomposites, (2) advanced materials such as inorganic-organic hybrid assemblies for utilizing renewable energy resources, and (3) environmental remediation using advanced oxidation processes and chemical sensors. He has directed DOE funded solar photochemistry research for more than 20 years. He has published more than 300 peer-reviewed journal papers, review articles, and book chapters. He has edited two books in the area of nanoscale materials. He was a fellow of Japan Society for Promotion of Science during 1997 and 2003 and was presented the 2006 Honda-Fujishima Lectureship award by the Japan Photochemical Society. Feature Article J. Phys. Chem. C, Vol. 111, No. 7, 2007 2835

transfer, proton-coupled electron transfer, and catalysis are important in designing molecular assemblies for energy conver- sion.57 New approaches have to be considered to harvest the separated charges in a donor-acceptor molecular system. Tailoring the optoelectronic properties of metal nanoparticles by organizing chromophores of specific properties and functions on gold nanoparticles can yield photoresponsive organic- inorganic nanohybrid materials. The organization of a densely packed photoresponsive shell encapsulating the nanoparticle core offers exciting opportunities for the design of novel photon- based devices for sensing, switching, and drug delivery58-61 Metal hybrids of organic molecules assembled as two- or three- dimensional architectures provide routes to the design of materials with novel electrical, optical, and photochemical properties having potential applications in nanophotonics, lithography, and sensing.30,62-69 The possibility of tailoring the optoelectronic properties of metal nanoparticles by organizing chromophores of specific properties and functions on gold nanoparticles prompts one to design photoresponsive organic- inorganic nanohybrids. Such an organization of a densely packed photoresponsive shell encapsulating the nanoparticle core offers exciting opportunities for the design of light energy conversion devices.29 Gold nanoparticles are widely used as probes for bimolecular labeling and as immunoprobes.70,71 Organized inorganic-organic nanohybrids, with hierarchical superiority in architecture, can be developed by assembling monolayers of organic molecules containing functional groups, such as amines, thiols, isothiocynate, and silanes, on to the three- dimensional surface of metal nanoparticles.59,72 Such monolayer protected metal clusters (MPCs) prepared by adopting the ���two- phase extraction��� procedure73,74 can be functionalized with chromophores by ���place exchange��� reactions.75-77 For example, porphyrin-alkanethiolate monolayer protected-gold nanoclusters (H2PCnMPC) form spherical shape clusters that can be em- ployed as light harvesting antenna (Figure 4). They exhibit efficient light-harvesting capability and suppress undesirable energy transfer quenching of the porphyrin singlet excited-state by the gold surface relative to the bulk gold. a. Excited State Interactions. The close vicinity of a metal nanocore alters the excited deactivation pathways of the surface- bound molecules. For example, Drexhage and co-workers have observed a distance-dependent quenching of excited states of chromophores on metal surfaces.78 One of the noticeable properties of the fluorophore molecules when bound to metal surfaces is the decrease in singlet lifetime as a result of energy transfer from excited dye molecules to bulk metal films.79-81 Total quenching of the singlet-excited-state of the chromophores can limit the application of chromophore-labeled metal nano- particles in optoelectronic devices and photonic materials. Interestingly, recent studies on the photophysical properties of chromophore-linked gold nanoparticles from our group82-87 and others88-94 have suggested a dramatic suppression in the quenching of the singlet-excited-state when these chromophores are densely packed on Au nanoparticle surfaces. A better understanding of the excited-state processes will enable effective utilization of chromophore-functionalized metal nanostructures for light-harvesting and other specialized applications. Possible deactivation pathways of the photoexcited fluorophore bound to a gold nanoparticle, viz., (A) intermolecular interactions, (B) energy transfer, (C) electron transfer, and (D) emission from the chromophores bound on the metal nanoparticles, are summarized in Figure 5. In the case of hybrid assemblies having metal nanoparticles as the core, the energy transfer depends critically on the size and shape of the nanoparticles, the distance between the dye Figure 1. Left: Discovery trend. Oil discovery peaked in the 1960s, when we were finding more than we used. Now, the situation is reversed, meaning that the historic trend of growth at about 2% a year cannot be maintained as we consume our inheritance from past discovery. Right: World production of oil. Production has to mirror discovery, starting and ending at zero, with a peak in between at the halfway point. Production matched the theoretical curve well until the oil shock of the 1970s meaning that peak is lower and later than would otherwise have been the case, but decline is inevitable given a finite total. (From ref 4. Reprinted with permission from Springer.) Figure 2. Sustaining the population with different energy resources. Each source of energy supports a corresponding population. The impact on population of oil and gas has been dramatic but is short-lived. (From ref 4 Reprinted with permission of Springer.) 2836 J. Phys. Chem. C, Vol. 111, No. 7, 2007 Kamat

molecule and the nanoparticle, the orientation of the molecular dipole with respect to the dye-nanoparticle axis, and the overlap of the chromophore emission with the nanoparticle absorption.95 Energy transfer processes were found to dominate the deactiva- tion of the excited-state in fullerene-functionalized gold clusters (C60-R-S-Au).96 Clusters of C60-R-S-Au can be visualized as antenna systems containing a gold nanoparticle as the central nanocore and appended fullerene moieties as the photoreceptive hydrophobic shell. C60-R-SH emission is totally quenched when the fullerene is anchored to the gold nanocore. Near field optical microscopy further aids in elucidating the photoinduced energy transfer to metal particles.97-99 The low singlet as well as triplet yields of the fullerene moiety in excited C60-R-S-Au nanohybrids confirm that most of the excited-state energy is quickly dissipated to the Au core via energy transfer. Dulkieth et al.95 isolated the resonant energy transfer rate from the decay rates of excited lissamine dye molecules by chemically attaching them to gold nanoparticles of different size. The increase in lifetime with decreasing particle size (particle diameter range of 1-30 nm) was indicative of decreased efficiency of energy transfer in smaller size particles. Heeger and co-workers100 investigated the role of energy as well as electron transfer in the quenching of emission of conjugated polymers in the presence of gold nanoparticles of varying size. They concluded that resonance energy transfer dominates when the diameter of Au nanoparticle is 2 nm. Using the principle of energy transfer, attempts are being made to develop biosen- sors.101,102 b. Photoinduced Electron Transfer. Semiconductor nano- particles are known to accept electrons from an excited sensitizer and transfer the electrons to another acceptor molecule bound to the surface. The demonstration of semiconductor particle mediated electron transfer between donor and acceptor mol- ecules bound to its surface was demonstrated in our early studies.103-105 The nonmetallic property of ultrasmall metallic particles can also be utilized to capture electrons from an excited sensitizer and thus mediate a photoinduced electron-transfer process. In polar solvents, pyrene-linked Au nanoparticles (Py-R1- S-Au) exhibit noticeably lower yields compared to unbound pyrene thiol (Py-R1-SH).82 Transient absorption experiments using pulsed laser irradiation (337 nm) of Py-R1-S-Au nanoparticle, in polar solvents such as tetrahydrofuran or acetonitrile confirm the electron transfer between gold nano- particle and pyrene Figure 6). The charge-separated states in Py-R1-S-Au assemblies are fairly long-lived as indicated by the longer lifetime of the pyrene cation radical (4.5 ��s). These observations demonstrate the ability of gold nanoparticles as electron acceptors. Figure 3. Strategies to employ nanostructured assemblies for light energy conversion. Figure 4. Examples of gold nanoparticles functionalized with (a) porphyrin, (b) C60, and (c) pyrene (from ref 29). Feature Article J. Phys. Chem. C, Vol. 111, No. 7, 2007 2837

Controlled charging of the Au nanoassembly enables one to modulate the excited-state interaction between the gold nanocore and a surface-bound fluorophores.83,87 For example, a bifunc- tional surface-linking molecule such as mercaptopropionic acid was used to link the gold nanoparticle to the TiO2 surface (thiol group to gold and carboxylic group to TiO2). Spectroelectro- chemical experiments carried out using a thin layer electro- chemical cell showed the emission spectra of pyrene modified gold particles which were linked to TiO2 film cast on an optically transparent electrode (OTE/TiO2/-OOC-R2-S- (Au)-S-R1-Py) and subjected to different applied potentials (Figure 7). As the electrode is biased to negative potentials, an increase in the emission yield was observed. The overall shape of the emission band remains the same suggesting that the photoactive molecule contributing to the emission is unper- turbed. At potentials more negative than -1.0 V, 90% of the quenched emission is restored by charging the gold nanopar- ticles. The quantized charging effects studied with organic- capped gold nanoparticles suggest that the potential shift amounts to about 0.1 V per accumulated electron.106 The electron transfer from excited pyrene molecules to gold nanocore experiences a barrier as we charge them with negative electro- chemical bias. The photoinduced electron-transfer mechanism in chlorophyll a bound gold nanoparticles was also confirmed from the electrochemical modulation of fluorescence of Chla.87 In the absence of an applied bias, Chlorophyll a cast on gold particulate films, as a result of electron transfer, exhibits a very weak fluorescence emission. However, upon negatively charging the gold nanocore by external bias, the fluorescence intensity increases. Charging the gold nanoparticles increases the energy barrier and thus suppresses direct electron transfer. This suppressed electron-transfer pathway at negative bias increases in radiative process. In addition to this indirect evidence for the electron transfer between excited chlorophyll a and gold nanoparticles, direct evidence for electron transfer was also Figure 5. Photoexcitation of the chromophore bound to gold nanoparticles followed by its deactivation via energy transfer, electron transfer, and intermolecular interactions (from ref 29). Figure 6. Modulation of photoinduced electron transfer between excited chromophore and gold nanoparticles (from ref 83). Figure 7. Modulation of photoinduced charge transfer in a pyrene modified gold particles linked to TiO2 film cast on an optically transparent electrode (OTE). (From ref 83.) 2838 J. Phys. Chem. C, Vol. 111, No. 7, 2007 Kamat