Nanoscale objects provide opportunities to revolutionize the conversion of solar energy by enabling highly efficient and low-cost devices. Challenges associated with demonstrating high efficiency and stability are now being addressed in the research community.
At a glance
Nanostructured materials are being investigated and developed as versatile components of optoelectronic devices with the ability to manipulate light (via plasmonic enhancement, photonic crystals, and so on) and control energy flow at nearly the atomic level. Nanostructured solar cells — a type of third- or next-generation solar cell1 — include those that are based on nanostructures and/or nanostructured interfaces such as nanowire, mesoscopic and quantum dot solar cells. They hold great promise towards new approaches for converting solar energy into either electricity (in photovoltaic devices) or chemical fuels. There are challenges to overcome but the potential benefits are worth the efforts.
The solar photovoltaic challenge
The commercial and domestic adoption of photovoltaics as an energy source depends on its costs to consumers in $/kWh (all costs are in US$), and on how it compares with that of conventional energy sources. The overall cost can be estimated from the rated cost per peak watt (in $/Wp), which includes a component related to the operation of the solar module ('module cost'), and one to its installation ('balance of system costs'). The former strongly depends on the power conversion efficiency (PCE), which is determined by the fraction of the total solar irradiance that a solar cell converts into electricity. The module cost is defined as the ratio of the cost per unit area to the product of both the PCE and the average solar irradiance at midday with no clouds (that is, 1,000 W m−2). The balance of system costs covers all other elements (wiring, supports, switches, inverter, and so on) of the installed photovoltaic system other than the module, and can be roughly approximated as equal to the module cost. The total cost per peak watt of a photovoltaic system is therefore roughly twice the module cost per peak watt, and scales inversely with PCE. Finally, to obtain the overall cost per kWh there are other factors to take into account: cloud coverage, the position of the Sun in the sky as it changes during the day, diurnal cycles, depreciation of the photovoltaic module and maintenance costs. As a rule of thumb, the numerical value of the cost per kWh is obtained by multiplying the total cost per peak watt by 0.05. With the current cost of photovoltaics approaching about 1 $/Wp, the average cost to the consumer is ~0.10 $/kWh, which is comparable to the average cost for grid electricity in the US (ref. 2) — this situation is called grid parity.
While recent cost reductions of modules are a step in the right direction, we believe that the costs are not yet low enough to guarantee substantial uptake of photovoltaics by consumers. In the US, for example, assuming a continued absence of a carbon tax, reaching grid parity is unlikely to displace a large fraction of energy utilization from fossil fuels because the fossil fuel infrastructure is huge and entrenched. For widespread adoption of solar, the costs of next-generation photovoltaics should probably be reduced to a half or a third of the average grid costs (that is, 0.05 to 0.03 $/kWh; Fig. 1, shaded region). This can be achieved, for example, by a module with PCE of 40% and a cost per unit module area of 160 $/m2, or a PCE of 30% with a module cost of 120 $/m2. Figure 1 shows the relationship between $/m2, rated module PCE and $/kWh. These targets assume long-term stability (10–15 years) and no environmental degradation or negative impact. According to the Shockley–Queisser detailed balance analysis, the PCE of photovoltaic devices with a single light-absorbing component is capped at ~33%3. The first silicon-based solar cell, introduced in 1941 by Russell Ohl4, had PCE of less than 1%; since then, owing to advances in materials processing/quality and device configurations, PCEs have asymptotically approached the Shockley–Queisser limit. The current laboratory record efficiencies for Si is 25.6%5, while for GaAs it is 28.8%5. These record efficiency cells employ bulk crystalline semiconductors.
How nanostructured solar cells help
Nanoscale systems exhibit different properties than bulk or thin films of the same compounds, and have allowed new ways of approaching solar energy conversion for electricity generation or fuels. The large surface-to-volume ratio of nanomaterials can provide various benefits, and, furthermore, objects with a size of ~1–20 nm can also exhibit quantization effects, which become more pronounced with decreasing size. Two broad approaches based on nanostructures are being explored for photovoltaics: (1) significant reduction in material usage and/or associated final costs; (2) photovoltaic devices with a higher limiting efficiency than that determined by the Shockley–Queisser analysis. Both approaches, individually or in combination, can lead to significantly lower costs per kWh.
Reduced material costs and/or usage. Front surface reflectivity in a solar cell leads to losses. If light is not absorbed by the active component, it cannot be converted into electricity. Crystalline Si reflects ~30% of the incoming light, and conventional cells employ anti-reflection layers to reduce these losses to about 3–4%, but at an added manufacturing cost. Nanostructured surfaces can reduce reflection losses when the size of the nanostructures is smaller than the wavelength of the incident light. If those nanostructures are also active components, then the need to deposit costly anti-reflection coatings and texturing can be avoided6. Furthermore, patterned nanostructures can be designed to capture even more light via light trapping so that less material is needed to absorb the solar flux7,8.
The low absorptivity of silicon requires the use of thick films in order to absorb all incident light. This poses a constraint on material quality, because high-purity materials, with long minority-carrier lifetimes, are needed to enable carrier collection over relatively large distances. Moreover, thicker films result in increased material costs. One approach to relaxing such requirements has been explored by using radial p–n junction Si nanowires7. A core–annular p–n junction (Fig. 2b) is fabricated along the length of the wire. When the incident light generates charge carriers at the junction, minority carriers only need to traverse the nanowire diameter in order to be collected. With less stringent requirements on the minority-carrier lifetimes, lower grade silicon can be used. By combining light-trapping effects and reduced carrier lifetimes, in nanowire solar cells both the amount and quality of raw materials used can be greatly reduced8.
In traditional solar cells fabricated from bulk semiconductors, electrons and holes are generated when light is absorbed in the semiconductor9. They then separate and migrate to different contacts to produce both a voltage V and current I, and thus power (P = I × V). Because of the large dielectric constants of these materials, electrons and holes are quickly screened from one another and do not interact. Nanostructuring is a way to bypass the requirement of high-dielectric-constant semiconductors and allows use of new classes of materials and device designs. For example, two dissimilar materials, where one is n-type (conducts electrons) and the other is p-type (conducts holes), can be intermixed with nanoscale morphology (Fig. 2c). Absorption of light produces an excited state (exciton) that undergoes rapid charge transfer producing electrons and holes in separate phases, making their interaction less probable. Two types of solar cell are based on this design: organic photovoltaic devices and dye-sensitized solar cells.
Another next-generation approach for photovoltaics is based on semiconductor nanocrystals. Their most important properties for photovoltaic applications are the strong size-dependence of the bandgap, and the large modification of the relaxation dynamics of photoexcited charge carriers that are created by the absorption of photons within energies larger than the bandgap. The bandgap of the absorber layer controls which photons can be absorbed and limits the output voltage of a solar cell9. Because only the radiation with higher energy than the bandgap is absorbed, narrower-bandgap materials absorb more solar photons, resulting in higher photocurrents. However, the output voltage is linearly proportional to the bandgap, and thus wider-bandgap materials allow higher voltages. In the Shockley–Queisser analysis there is an optimal bandgap that achieves the highest efficiencies, and ranges between 1.2 and 1.4 eV. Semiconductors with bandgaps lower than 1 eV are generally not employed in single-layer solar cells. Quantum confinement effects in quantum dots (QDs) can increase the bandgap by more than 1 eV compared with the bulk value10, expanding the range of semiconductor materials viable for photovoltaics. A prototypical example is PbS. Bulk PbS has a bandgap of 0.4 eV but PbS QDs can have bandgaps from ~0.6 to ~2 eV depending on their size.
Quantum dot solar cells (Fig. 2a) can be processed from colloidal solutions at ambient temperatures, enabling relatively low manufacturing costs. For example, PbS QDs can be synthesized and processed into films near ambient temperature11 without vacuum processing12. Efforts have focused on the specific deposition steps that can impact both the optical and electrical properties of the QD films13. Solar cell efficiencies are now approaching 9% using PbS QDs with a bandgap of 1.25–1.4 eV (refs 12,14). Other QDs of interest for photovoltaic applications are crystalline precursor inks, which after annealing produce continuous larger-grain thin films with bulk-like properties15. Examples include QDs made of CdTe, copper zinc tin sulfide (CZTS), copper indium gallium selenide (CIGS) and Si. Recent newcomers are solution-processed perovskite solar cells16, which evolved from nanostructured device concepts. However, like most other single-junction devices, their efficiency is limited by the Shockley–Queisser analysis, and will need to be incorporated into multijunction architectures in order to achieve higher efficiencies.
Increasing efficiency limits
New photovoltaic technologies should have the potential not only to reduce module costs but also to achieve PCE beyond ~33%. Two critical assumptions of the Shockley–Queisser analysis are: (1) photons with energy less than the semiconductor bandgap are not absorbed and thus cannot contribute to PCE; (2) energetic electrons created by high-energy photons immediately relax to the band edge (that is, the fraction of energy of photons with energy greater than the bandgap is immediately lost as heat). Approaches to achieve higher limiting efficiencies attempt to either use the high-energy photons more efficiently, or recover the low-energy photons normally not converted. Nanostructures are being explored to eliminate these losses.
Multijunction solar cells commonly employ a stack of p–n junctions where the bandgap of the light-absorbing semiconductor decreases in successive layers. In the limit of an infinite number of junctions, the theoretical PCE reaches 68% at 1-sun intensity17. A 1-sun PCE of 34.1% has been achieved in a triple-junction solar cell based on III–V semiconductors5. Under optical concentration, the efficiency increases, and the record PCE for any photovoltaic cell is 44.4% when measured with light equivalent to 302 suns (ref. 5). However, these high-efficiency cells are far too expensive. Nanomaterials are being explored as part of multiple-junction solar cells18, 19, because the same material can be used to produce layers with different bandgaps, and the multiple junctions can be solution-processed at significantly lower cost.
Other examples of circumventing assumptions of the Shockley–Queisser analysis using nanostructures include intermediate-band solar cells20 and solar cells based on multiple exciton generation (MEG)1. In addition, nanocrystals can be components of up and/or down conversion layers and fluorescent concentrators21, all of which attempt to reach higher efficiencies at lower cost. Multiple exciton generation, or carrier multiplication, allows a single photon to produce multiple excitons (electron–hole pairs)22. This increases the photocurrent and can permit the PCE to exceed the limit imposed by the Shockley–Queisser analysis23, which assumes that each photon only produces one electron. Multiple exciton generation is greatly enhanced in QDs compared with bulk semiconductors because the competition between the relaxation pathways of the MEG and of phonon emission, as well as the hot-carrier relaxation24, can be modified to favour MEG. Quantum confinement also enhances the Coulombic interaction that drives MEG. In addition to single-component nanostructures, multicomponent nanocrystals with selectively arranged individual domains that allow complex functionalities offer new opportunities for photovoltaic research. Such nanoheterostructures allow for further tailoring of the excited-state relaxation dynamics, resulting in additional enhancements of the MEG efficiency25.
Importance of efficiency certification
While nanostructures offer significant potential advantages for both increasing the PCE and reducing manufacturing costs, the technology is not yet mature, and the efficiency of proof-of-principle devices lies well below those of established technologies. Furthermore, it is not always possible to independently verify the reported efficiency of such devices.
The PCE encloses information on the performance of a solar cell in a single number, and thus allows a direct comparison of devices across many technologies. However, photovoltaic research is a highly competitive and fast-moving field, and device characterization can be problematic at the lab scale; thus, there is a need to benchmark the measurements. Researchers can 'certify' their device efficiencies by submitting completed devices to one of several institutions that can provide standardized PCE measurements26. Certified cells must have a well-defined illumination area and account for (or block) collection of photogenerated carriers from outside that area. Efficiency certification is an essential component of the research and development of new photovoltaic technologies that can be used to lend credence to new approaches. A table of certified record efficiencies is published twice a year with updated efficiency values5. In these reports, Martin Green, Keith Emery and co-workers consider cells with active area ≥1 cm2 (for measurements at 1 sun). For exploratory device concepts, researchers generally work with cells with smaller active areas (about 0.1 cm2), which thus are not included in these tables. Small device areas eliminate series resistance, reduce non-uniformity issues, enable more devices per area for statistical purposes, and thus remove issues that can be addressed at later stages of development. While PCE of small-area devices are not included in the tables, certified efficiencies of proof-of-concept cells are still valuable as benchmarks and the National Renewable Energy Laboratory (NREL) provides a chart that tabulates record efficiencies as a function of time across both established and emerging technologies27.
One issue with emerging photovoltaic technologies made at the lab scale involves the stability of the materials or devices. Devices with poor stability provide a complication for certification purposes, due to different measurement conditions (atmosphere, temperature, humidity, light source), contact probes, transportation, and so on. Thus, the certified results may not necessarily reflect the true potential of a given technology. The stability issues can be explored at later stages of research and development, if the promise of a particular technology warrants such efforts. Thus, even in those cases where certification is not possible, reports of high PCE are still valuable to the research community. Such results should still be reported in the literature, but researchers should refrain from claiming record efficiencies, explain why certification was not possible and verify their measurement practices and equipment against standard solar cells.
Device stability is obviously an important consideration for photovoltaic technologies as the longer the photovoltaic system operates, the lower is the total cost. Many of the same features that make nanostructured solar cells attractive also introduce additional challenges. Nanostructured materials have larger surface area; the surface properties of photovoltaic films and electronic devices in general require attention. For nanostructured devices, improper passivation of internal surfaces can hinder long-term stability. Stability issues are being addressed by chemistry11, 14, or device configuration12. Encapsulation is also a route towards making materials commercially viable, provided that the overall cost and efficiency of the cell satisfy the requirements discussed above.
No consensus exists among the research community on how to properly characterize the stability. Low-maintenance, long-lifetime solar simulators are not readily available, and lifetime tests inherently take a long time to perform. Therefore, some approaches involve accelerated lifetime testing in harsh environments, such as 85 °C at 85% humidity, but that translation to real-world stability may not be the same for all technologies28.
In conclusion, next-generation solar cells must meet stringent requirements in terms of PCE, cost and long-term stability (10–15 years). We believe that nanostructured solar cells have the potential to achieve these objectives.
- Acc. Chem. Res. 46, 1252–1260 (2013). , , &
- U. S. Energy Information Administration; http://www.eia.gov
- J. Appl. Phys. 32, 510–519 (1961). &
- Prog. Photovolt. 17, 183–189 (2009).
- Prog. Photovolt. 22, 701–710 (2014). , , , &
- Nature Nanotech. 7, 743–748 (2012). , &
- Nature Mater. 9, 239–244 (2010). et al.
- Nano Lett. 10, 1082–1087 (2010). &
- 2009). Physics of Solar Cells: From Basic Principles to Advanced Concepts (Wiley,
- J. Chem. Phys. 80, 4403–4409 (1984).
- ACS Nano 8, 614–622 (2013). , , , &
- Nature Mater. 13, 796–801 (2014). , , &
- 2014). , , , & in Advanced Concepts in Photovoltaics (eds Nozik, A. J., Conibeer, G. & Beard, M. C.) Ch. 11 (Royal Society of Chemistry,
- Nature Mater. 13, 822–828 (2014). et al.
- Curr. Opin. Colloid Interface Sci. 14, 245–259 (2009). &
- 2014). in Advanced Concepts in Photovoltaics (eds Nozik, A. J., Conibeer, G. & Beard, M. C.) Ch. 7 (Royal Society of Chemistry,
- J. Phys. D 13, 839–846 (1980).
- Adv. Mater. 23, 3144–3148 (2011). et al.
- Nature Photon. 5, 480–484 (2011). et al.
- 2014). , , in Advanced Concepts in Photovoltaics (eds Nozik, A. J., Gonibeer, G. & Beard, M. C.) Ch. 13 (Royal Society of Chemistry,
- Nature Photon. 8, 392–399 (2014). et al.
- Physica E 14, 115–120 (2002).
- Nano Lett. 9, 836–845 (2009). et al.
- Annu. Rev. Phys. Chem. 52, 193–231 (2001).
- Nature Commun. 5, 4148 (2014). et al.
- Prog. Photovolt. 20, 606–614 (2012). , , , &
- Solar Energ. Mater. Solar Cells 95, 1253–1267 (2011). et al.
This material is based on work supported by the Office of Science, Office of Basic Energy Sciences; the Solar Photochemistry program supports work on photochemistry of semiconductor nanocrystals, and the Center for Advanced Solar Photophysics within the Energy Frontier Research Centers program supports work on quantum dot solar cells.