Matthew J.
Bierman
and
Song
Jin
*
Department of Chemistry, University of Wisconsin-Madison, 1101 University Avenue, Madison, Wisconsin, 53706, USA. E-mail: mbierman@chem.wisc.edu; jin@chem.wisc.edu; Web: jin.chem.wisc.edu
First published on 27th July 2009
Nanoscience and nanotechnology can provide many benefits to photovoltaic and photoelectrochemical applications by combining novel nanoscale properties with processability and low cost. Taking advantage of high quality, high efficiency, yet low cost nanomaterials could potentially provide the new and transformative approaches to enable the proposed generation-III solar technologies. Nanowires are interesting because they have a long axis to absorb incident sunlight yet with a short radial distance to separate the photogenerated carriers. In this perspective, we further suggest that more “complex” nanostructures, both in the form of hierarchically branching/hyperbranching nanowire structures and in the form of multi-component nanowire heterostructures of diverse materials, are potentially even more interesting for solar energy harvesting and conversion. The common bottom-up synthetic techniques to induce branching in nanowires to form hierarchical nanowire structures are reviewed. Several potential strategies for their incorporation into solar conversion devices are discussed and some fundamental issues and future directions are identified.
![]() | Matthew Bierman received his B.S. from the University of Wisconsin-La Crosse in 2003 and his Ph.D. from the University of Wisconsin-Madison in 2009. His work in nanoscience is motivated by materials development for energy applications including photovoltaics. His thesis work centered on nanowire growth mechanisms, including the introduction of the dislocation-driven nanowire growth mechanism, which was recognized with a 2009 Materials Research Society Graduate Student Award. He will be continuing his energy related research at the California Institute of Technology as a postdoctoral associate. |
![]() | Song Jin received his B.S. from Peking University in 1997 and his Ph.D. from Cornell University in 2002. He has been an assistant professor of chemistry at the University of Wisconsin-Madison since 2004. He is interested in the chemistry and physics of nanoscale materials. Dr Jin studies the fundamental formation mechanisms of nanowires, their novel physical properties, and applications in photovoltaic and thermoelectric energy conversion, nanospintronics, and nanomedicine. He has received NSF Career Award and ACS Exxon Mobil Solid State Fellowship, is a Sloan Research Fellow and a Research Corporation Cottrell Scholar, and was recognized as a TR35 by MIT Technology Review magazine in 2006. |
Broader contextSolar energy is clean, renewable, and nearly inexhaustible, and yet is not in widespread use. Solar conversion devices are not yet ubiquitous because of the relatively large cost of energy produced this way when compared to traditional energy sources such as fossil fuels. In order to bring the cost of solar energy in line with other sources, new science and new technology are needed. A variety of materials with at least two nanoscale dimensions (nanomaterials), from nanocrystals to nanowires, nanorods, and nanotubes, have been utilized to harvest solar energy that are cheaper than current technology while remaining efficient or could be even more efficient. Nanowires are interesting because they have a long axis to absorb incident sunlight yet with a short radial distance to collect the photogenerated carriers. We further suggest that hierarchically branching nanowires can be even more beneficial over conventional planar devices than simple nanowire arrays. In this perspective, we summarize the known mechanisms to control nanowire branching, identify the fundamental issues and suggest ways to incorporate them into solar conversion devices. |
The problems with conventional planar junction silicon photovoltaics are well documented. Silicon has an indirect bandgap and is inefficient at absorbing incident light. Therefore, a thickness of 125 µm silicon is required to absorb 90% of incident sunlight, in contrast to 900 nm of GaAs for the same efficiency.3 Because of this large thickness (Fig. 1a), even if a photon is absorbed, collected photocurrent is reduced since the minority carrier diffusion length is limited by various recombination processes within the absorber region. Increasing minority carrier diffusion length requires very pure, high quality, single crystalline silicon, which greatly increases the cost and complexity of functional silicon photovoltaic devices. Polycrystalline and amorphous silicon PV devices have a lower cost of fabrication but suffer in solar conversion efficiency due to the shorter minority carrier diffusion length.4,5 Multi-junction single-crystalline GaAs solar cells can be quite efficient but their materials cost and fabrication cost are very high. A variety of thin film PV devices based on non-silicon materials are inexpensive but their efficiency is not yet competitive with single crystalline silicon.4,5
![]() | ||
Fig. 1 Advantages of vertical nanowires and hyperbranched nanowires over planar solar energy conversion devices. Planar devices (a) must be thick enough to absorb incident light, which places materials quality restrictions due to limited minority carrier diffusion lengths. Nanowires (b) have a long axis to absorb incident light, and orthogonally, a short radial axis to separate charge, accommodating lower quality materials. Introducing branching (c) increases light scattering and therefore photocarrier generation. More complex morphologies such as hyperbranched nanowire heterostructures (d) and nanocrystal-sensitized (or catalyst-decorated) nanowires (e) can also be interesting for solar conversion. |
Nanoscience and nanotechnology suggest new ways to solve the performance/cost tradeoff that has limited large-scale commercial applications of solar energy conversion devices. Chemically synthesized nanocrystals, nanowires, nanorods, nanotubes, and a variety of other nanostructures offer new opportunities for solar energy conversion that promise lower fabrication cost and higher conversion efficiency.6 As a result, there has been intensive research worldwide directed at taking advantage of nanomaterials to improve the applications of solar energy conversion. For example, solution grown colloidal semiconductor nanocrystals can be readily spin-coated or blended with semiconducting polymers to allow the facile fabrication of unconventional thin-film solar cells.7–10 Furthermore, the performance of light harvesting materials may actually be enhanced by various nanoscale mechanisms11 such as multiple exciton generation12,13 or the size-dependent control of the absorption wavelength of a quantum-confined particle.14
A variety of semiconductor nanowires can be synthesized in single-crystalline form with precisely controlled structures, diameters, lengths, chemical compositions and doping/electronic properties using a variety of processes such as the nanocluster catalyzed vapour–liquid–solid (VLS) growth mechanism.15 These one-dimensional (1-D) nanowires have great promise in solar energy conversion.16–21 Compared with conventional planar single crystalline or thin film PV or PEC cells (Fig. 1), in a nanowire, the light absorption and carrier separation occur in orthogonal directions (Fig. 1b) so solar light absorption occurs along the longer dimensions of the nanowire (microns or more) while the carrier separation occurs by diffusion across the short radial distance (tens of nanometres).22 As a result, overall device performance does not significantly suffer if lower quality (and less expensive) materials are used. Unconventional semiconductor materials or lower cost synthetic methods that create more defects and impurities might then become tolerable for large-scale solar applications.23 Even if a crystal defect exists within a nanowire, or indeed if an entire nanowire is defective, the performance of the entire device is not impeded as with single crystal devices.
In this perspective, we will build on these ideas and further expand the concept to argue that more “complex” nanostructures, both in the form of hierarchically branching/hyperbranching nanowire structures and in the form of multi-component nanowire heterostructures of diverse materials, are potentially even more interesting for the applications of solar energy harvesting and conversion. We offer a review of the rational synthetic strategies to create such complex nanostructures, and then suggest preliminary ideas and fundamental issues that need to be addressed in order to fully take advantages of such nanostructures for solar energy conversion. The use of nanomaterials, and especially branching nanowires, provides a pathway to reduce materials cost while ideally maintaining the efficiency of the host material, which represents the so-called Generation-III solar technologies24 that combine efficiency with cost.
![]() | ||
Fig. 2 Nanowire branching from sequential addition of VLS catalysts. Sequential catalyst application can be used to control branch density as shown with Si nanowires (a,b).27 Scale bars are 1 µm. This technique can be used to create an array of interconnected nanowires (c) by connecting patterned vertically aligned nanowires as shown in this plan-view electron micrograph.31 Images reproduced with permission ACS. |
This approach can also be adapted to integrate different materials into trunk/branch heterostructures. A clear example has been the growth of CdS branches grown sequentially from a ZnS trunk.37 After the second application of gold catalysts, the precursor was changed from cadmium dithiocarbamate to zinc dithiocarbamate. ZnO branches can also be grown from an In2O3 trunk.38 Another interesting heterojunction branching example is in the case of carbon nanotubes grown only from the tip of a SiC micropost, creating the world's smallest brush.39
The continuous in situ generation of a VLS catalyst during nanowire growth provides the source for multiple generations of branches to nucleate and grow at the same time (Fig. 3). This allows the growth of very dense hierarchical branched nanowire networks that have high quality epitaxial junctions, termed “hyperbranching”. The term “hyperbranched” can be perhaps traced back to the terminology in polymer literature, where “hyperbranched polymers” refer to non-linear polymers that have many branches and possibly many generations of branches. This is in contrast to “dendritic polymers” or “dendrimers”, which have the same number of branches at each branching point. Several recent studies have involved materials systems capable of the in situ formation of a catalyst.29,40–43 In the case of PbS (PbSe), lead metal itself serves as an in situ generated catalyst from the PbCl2 source formed in a reducing H2 atmosphere and can be rationalized from the Pb-S (Pb-Se) phase diagram. As shown in Fig. 3, the continuous introduction of catalyst randomly on the surface nucleates a dense, orthogonal network of nanowires and generates multiple generations of branches for as long as additional precursor is added. An analogous process could have been observed in the solution-phase synthesis of PbSe nanowires.44 Branching of PbSe nanowires was observed as the Pb : Se ratio approached 4 : 1. This lead-rich synthesis allowed Pb catalysts to condense on the growing PbSe nanowire, nucleating new branches at a right angle.
![]() | ||
Fig. 3 Hyperbranching nanowires through continuous in situcatalyst generation. Nanowires grow randomly with one generation growing epitaxially from previous generations (a), until a dense hyperbranched nanowire network is produced (b). This can be seen in nanowires of the cubic PbS and PbSe system (c, d, e) where lead metal self-catalyzes the nanowire growth.29Transmission electron microscopy (f) shows that the branches remain epitaxial. Images reproduced with permission ACS. |
The above examples of in situ catalysis are materials systems capable of self-catalytic VLS45 which most often have a low melting metal as one of the elements of the binary compounds, as in the case of the Ga-metal catalysis of GaN46 and the Pb-metal catalysis of PbS,28,29 but the catalyst material need not be self-similar as in the case of PbSe grown with In, Ga, or Bi40,47 or of InAs nanowires catalyzed by an additional Mn source.48 When the catalysts can be continuously fed into the reaction system, multiple generations of nanowire growth can possibly occur.
![]() | ||
Fig. 4 “Pine tree” nanowires of PbS (a) due to the combination of two nanowire growth mechanisms. Screw dislocation driven nanowire growth of the trunk nanowire (b) combined with branch VLS driven growth (c) combine to produce this morphology. Upright PbS pine tree nanowire forests grow on silicon substrates (d).28 Images reproduced with permission AAAS. |
![]() | ||
Fig. 5 Tetrapods of CdSe52 (a) and the formation mechanism of these and other tetrapods50 (b) with zinc-blende/wurtzite polymorphs. Electron micrographs (c) of the original ZnO tetrapods,54 called “fourlings.” Magnification is 40 k×. Inset magnification is 3500×.53 Images reproduced with permission ACS and AIP. |
Similar ZnO tetrapods, called “fourlings,” have been produced in the vapour phase (Fig. 5c) by combusting zinc in an oxygen atmosphere, as early as 1944.53,54 Their formation mechanism is similar to CdSe and involves a stacking fault between the central nucleus and each branch. It was originally suggested that a non-classical nucleation of a polycrystalline ZnO nucleus was the source of twinning, and was thought to be the fourling formation mechanism.54 The state of transmission electron microscopy then probably did not allow for lattice imaging and this growth is now thought to involve zinc-blende/wurtzite stacking faults.
So far, materials that form branched structures due to such polymorph instability are mostly confined to zinc blende/wurtzite structural types, but if multiple polymorphs could co-exist with similar stability, analogous behaviours could also occur in other material systems with other structural types and produce branching geometry different from tetrapods. One related example has been demonstrated in PbSe “merged-y” nanowires, where a twin plane in a [110] growth direction nanowire splits into two branching nanowires growing in <100> directions.44
![]() | ||
Fig. 6 Branching and hyperbranching in transition metal silicides nanowires. Hyperbranched FeSi nanowires synthesized viachemical vapour transport (a,b). FeSi nanowires growing off of a Si nanowire core (c), and the higher manganese silicides branching nanowires (d). |
Nanowire arrays aligned on a substrate are interesting for PV and PEC device applications as seen in the preliminary demonstrations of nanowire solar conversion devices. The greater possibility of heteroepitaxy discussed above presents more options in integrating nanowires into functional devices. Taking the example of PbQ nanowires that are grown in our lab, the [100] directions of the (100) surface (a faces) of rocksalt PbSe can match the [100] direction on the (001) surface (c face) of tetragonal rutile TiO2 with a supercell of 1.84 Å and only 0.03% mismatch (4 × TiO2c = 3 × PbSea). Similarly, the atomic spacing along the [110] direction on the (001) surface of rutile TiO2 (6.495 Å) are identical within 5% of the spacing along the [100] surface of PbSe (6.126 Å). Such benefits are confirmed by our preliminary success in epitaxial growth of PbS nanowires on single crystalline rutile TiO2,30 a wide bandgap semiconductor. Because of the interfacial strain relaxation, more complicated epitaxial relationships on unconventional substrates are allowed as shown in the epitaxial growth of hierarchical PbS nanowires on single crystal NaCl and mica (Fig. 7).30 This approach may allow more efficient charge transfer between absorber/host than, for instance, quantum dot sensitized solar cells coupled with an organic linker.10,78
![]() | ||
Fig. 7 Epitaxial growth of hyperbranched nanowires of PbS on (a) single crystalline NaCl, and (b) mica.30 Images reproduced with permission RSC. |
Besides the epitaxial nanowires on substrates, many types of unconventional heteroepitaxy have been realized in the nanowire geometry,79 for example, axial heterojunctions,69,71,72 core-shell nanowires,80 or quantum dots decorated nanowires. Some potential approaches to exploit nanowire heterostructures for solar conversion devices are using different materials for axial heterostructures, core/shell (see Fig. 1c) or trunk/branch nanowires (see Fig. 1d and e). Photovoltaic cell designs such as the dye-sensitized solar cells utilize narrow gap (molecular) light absorbers to produce photocarriers that are coupled to a wide bandgap collector material that transports one photocarrier back to the electrode. The idea of sequential branching can be used to couple narrow bandgap absorber branches to a wide bandgap collector trunk nanowire or substrate that would provide an efficient photocarrier collection pathway as illustrated in Fig. 8. Perhaps even several absorber materials with progressively smaller bandgaps can be used to more efficiently harvest the energy from the solar spectrum, akin to the multi-junction solar cells. In fact, due to the quantum confinement and the modification of bandgap, this may be even possible to realize using the same materials system with well controlled nanoscale dimensions.
![]() | ||
Fig. 8 Branched nanowire heterostructures with well controlled electronic structures can improve solar energy harvesting. A narrow bandgap absorber that is matched to the solar spectrum is attached to a wide bandgap transport material like TiO2 to efficiently transport photocarriers to the collector base. |
The integration of multiple materials into one nanostructure is potentially significant. Can we seamlessly control diverse materials under continuous synthesis and catalysis conditions? Can we integrate and optimize several materials into a functional PV or PEC cell? Especially for PEC devices such as those for water splitting, several different materials are needed in one integrated functional device, including semiconductor(s) of different bandgaps and dopings, oxygen evolving and hydrogen evolving catalyst materials, all arranged in a nanoscale environment. This in fact mimics the natural photosynthetic systems which are remarkably engineered assemblies of nanostructures. Moreover, the materials involved must meet the different (and sometimes conflicting) requirements of electronic structures, materials compatibility, and chemical stability. For example, to create water compatible devices, many oxide semiconductors such as TiO2 and Fe2O3 with excellent stability are important to explore. Exploring these ideas to create “compositionally complex” multi-component nanostructures from the bottom-up will enable more sophisticated solar conversion device designs that are built on a new paradigm of bottom-up nanotechnology.
With many options to induce branching or hyperbranching in materials systems relevant to solar energy conversion, how do we best take advantage of these complex nanostructures? By incorporating hyperbranched nanowire materials into a photovoltaic cell, light absorption (and therefore cell efficiency) will be increased, while decreasing amounts of material compared to thin film designs could mean cost remains lower than in bulk photovoltaic cells. We suspect that the incorporation of such hierarchical nanostructures into a solar conversion device will be a dramatic departure from the conventional paradigm of planar junction solar cells and will be quite challenging. As a starter, making good electrical contact to the distributed 3-D structures can be a challenge. New paradigms are needed to take advantage of these novel structures for solar energy conversion, and we need to explore various unconventional device designs. We might imagine that for photovoltaic devices liquid junction electrochemical (regenerative) photovoltaic cells could be a promising approach. Suitable redox pairs have to be chosen based on the materials and band alignments to shuttle the holes (or electrons). Inorganic solution-deposited thin film hybrids with hierarchical nanostructures, and nanowire/organic-semiconductor composites might be interesting as well, since the deposition of one of the two active materials is conformal, thus allowing electrical integration with these 3-D structures. The initial practical demonstration of a hyperbranching nanowire photovoltaic cell will likely be a solution or polymer hybrid, or an excitonic solar cell with an active light absorbing dye or quantum dot, as illustrated schematically in Fig. 1e. Comparison to these types of cells sets a practical target for a successful demonstration of a hyperbranching effect, when compared with current cell efficiencies which range from 1.7–1.8% in P3HT/CdSe nanowire hybrid cells,8,81 4.2% in POHT/GaAs thin film hybrid cells,84 or 4.4% for axial heterojunction silicon nanowire cells.85
Enhanced solar energy harvesting due to hierarchical nanostructures applies to both photovoltaic applications where the electrons and holes are collected by electrodes and in solar fuel applications where the electrons and holes carry out redox reactions at catalytic centers. In fact, one may argue that chemically synthesized nanostructures, especially the hyperbranched nanostructures discussed in this perspective, might be intrinsically more suitable for photoelectrocatalysis (such as water-splitting) as opposed to photovoltaic application, because solar fuel production does not require a second solid-phase electrical contact but benefits from the high surface area characteristics of chemically synthesized nanostructures. Because of the high surface area, care must be taken to passivate the material surface for PV applications to prevent excess carrier recombination at nanostructured electrodes. Because of this consideration, perhaps branched nanowire materials could be better suited for use as nanostructured electrodes in a photoelectrochemical cell, effectively turning a limitation in a PV cell into a benefit in a PEC cell.
The high surface area-to-volume ratio of nanostructures is often deleterious because it could cause carrier recombination that reduces performance and creates irreproducibility in device functions.86,87 At the same time, it provides an opportunity to effectively control the electronic processes in nanostructures through passivation or partial passivation. For instance, perhaps trunk nanowires can be passivated to protect current flow while branch nanowires remain exposed and catalytically active (see Fig. 1d).
We need to learn how to make hierarchical nanowire structures more efficiently and controllably. With the understanding of the growth mechanisms, we need to develop conditions for inducing certain degrees of branching to optimize the performance/cost tradeoff. Or we may be interested in a more regular spacing of branches. This is especially true of synthetic schemes that introduce catalysts continuously or that employ conditions to promote the in situ generation of metal catalysts (such as the flow of hydrogen gas in the case of CVD PbS nanowire synthesis). How much control can we have over in situ mechanisms? Also, the conditions necessary for combining distinct modes of nanowire growth (like those combining dislocation-driven and vapour–liquid–solid growth, producing “pine-tree” PbS nanowires) need to be fully investigated88 and possibly expanded to other materials systems to allow greater control of these mechanisms and to create more complex nanostructures.
We also note that careful control over the quality of trunk/branch junctions is critical to maintain the benefits of induced branching. If care is taken to ensure an epitaxial relation between trunk/branch junctions, there remains a continuous “fractal” carrier collection pathway to ensure that all photocarriers are collected by the device instead of recombining at the material's junctions. Perhaps this favors in situ catalysis where the growth atmosphere is never compromised. However, ballistic transport has recently been demonstrated around the trunk/branch junction in InAs branched nanowires grown from sequential nucleation,89 suggesting carriers will not be impeded from the addition of branches. Even more fundamentally, we need to understand the (possibly coherent) carrier separation at nanoscale junctions and the carrier transport through the hyperbranched or dendritic nanowire structures.
There are also other economic and practical considerations. Even though IV–VI and II–VI semiconductors commonly studied so far are nice model nanostructures for fundamental investigations, their practical applications will likely be limited due to toxicity, materials abundance, and photocorrosion. Many other less commonly studied semiconductor materials made of earth abundant elements and environmentally sustainable materials, such as Cu2S, Cu2O, or FeS2,90 may prove much more important in the long term when solar energy conversion technologies is widely adapted on a much larger scale than the current niche applications. For example, some inexpensive metal oxide materials such as Fe2O3 are stable against photo-oxidation in water and are particularly attractive for aqueous photocatalysis.91–93 While some of these generally less explored materials were previously considered to be “poor” semiconductors whose carrier density, mobility, and minority carrier diffusion length were too small to be useful for solar energy applications, they are now much more attractive9,83 because the new device design concepts based on nanowire structures may circumvent the problems. Another practical consideration is that if high temperature reactions are used to synthesize nanowires, there will be some constraints placed on compatible substrates. This can be helped by transferring the as-grown nanowires to other (perhaps unconventional) substrates.94,95 Alternatively, lower temperature solution synthesis pathways, such as the aqueous solution synthesis of ZnO and TiO2nanowires, should also be explored.6,96–98 In fact, the latter will also be less expensive and more amenable to scale up.
We suggest these techniques and the hierarchical nanowire structures will be central to the development of Generation-III solar materials that decrease the cost constraints of bulk, high quality materials while providing pathways for a more efficient conversion and collection of photocarriers. Several potential strategies for their successful incorporation into photovoltaic or photoelectrochemical solar conversion devices are discussed and some fundamental issues and future directions that need attention for furthering the applications of hierarchical nanostructures for better solar conversion devices are identified. It is also possible that the right applications that best utilize these hierarchical nanowire materials have not been discovered, and we may need to depart dramatically from our current paradigm of solar energy technologies to take full advantage of these novel materials. However, we are optimistic that sustained and dedicated research efforts in this promising area would provide us potentially transformative approach to solar energy conversion that is based on nanoscience and nanotechnology.
This journal is © The Royal Society of Chemistry 2009 |