Strong Efficiency Improvements in Ultra-low-Cost Inorganic Nanowire Solar Cells


  • Kevin P. Musselman,

    Corresponding author
    1. Department of Materials Science, University of Cambridge, Pembroke St., Cambridge, CB2 3QZ (United Kingdom)
    • Department of Materials Science, University of Cambridge, Pembroke St., Cambridge, CB2 3QZ (United Kingdom).
    Search for more papers by this author
  • Andreas Wisnet,

    1. Department of Chemistry, Ludwig-Maximilians University, Butenandtstr. 11, Haus E, 81377 Munich (Germany)
    Search for more papers by this author
  • Diana C. Iza,

    1. Department of Materials Science, University of Cambridge, Pembroke St., Cambridge, CB2 3QZ (United Kingdom)
    Search for more papers by this author
  • Holger C. Hesse,

    1. Department of Physics and Center for Nanoscience (CeNS), Ludwig-Maximilians University, Amalienstr. 54, D-80799 Munich (Germany)
    Search for more papers by this author
  • Christina Scheu,

    1. Department of Chemistry, Ludwig-Maximilians University, Butenandtstr. 11, Haus E, 81377 Munich (Germany)
    Search for more papers by this author
  • Judith L. MacManus-Driscoll,

    Corresponding author
    1. Department of Materials Science, University of Cambridge, Pembroke St., Cambridge, CB2 3QZ (United Kingdom)
    • Department of Materials Science, University of Cambridge, Pembroke St., Cambridge, CB2 3QZ (United Kingdom).
    Search for more papers by this author
  • Lukas Schmidt-Mende

    Corresponding author
    1. Department of Physics and Center for Nanoscience (CeNS), Ludwig-Maximilians University, Amalienstr. 54, D-80799 Munich (Germany)
    • Department of Physics and Center for Nanoscience (CeNS), Ludwig-Maximilians University, Amalienstr. 54, D-80799 Munich (Germany)
    Search for more papers by this author


original image

Cu2O-ZnO nanowire solar cells are synthesized by electrodeposition from solutions near room temperature. A high-quality, self-assembling, nanostructured heterojunction is produced, which is found to dramatically improve the charge collection efficiency in these ultra-low-cost devices. Incident-photon-to-electron conversion efficiencies approaching unity are measured and considerable efficiency improvements are observed.

The need for sustainable power generation has encouraged research into a variety of photovoltaic materials and structures, with a greater emphasis being placed on a balance between performance and cost. The stability of many semiconducting oxides relative to other inexpensive solar cell technologies, such as organic1 and dye-sensitized2 cells, makes them an attractive alternative. Yet low-cost, non-toxic, inorganic solar cell technologies have received comparatively little attention. In a recent report, nine inorganic semiconductors were identified as having both the potential for annual electricity production in excess of worldwide demand and material extraction costs less than that of crystalline silicon.3 Further to materials costs, a recent study examined the high cost of modern vacuum deposition methods and highlighted the need for low-temperature, atmospheric, solution-based synthesis.4 Solution-based synthesis of several of the nine, promising inorganic materials has been demonstrated previously.5–7 Copper (I) oxide (Cu2O), in particular, has been synthesized extensively in polycrystalline form by electrodeposition from solutions near room temperature.5, 8, 9 Electrodeposition is easily scaled and can produce extremely uniform films on conductive substrates. The potential to synthesize inexpensive Cu2O photovoltaics on a variety of supporting substrates with minimal energy input is very attractive, not only for traditional solar panels, but also for the integration of solar cells into consumer products and building materials for energy harvesting applications. In this work, poor minority carrier transport is identified as the factor currently limiting the performance of electrodeposited Cu2O-ZnO solar cells. A nanowire architecture is successfully employed to improve minority carrier collection in these devices, resulting in a significant enhancement of the incident-photon-to-electron conversion efficiency (IPCE) (up to 85%). Key to the improvement is the design of a continuous nanowire heterojunction with an abrupt interface for the efficient separation of photogenerated charges.

Cu2O is a p-type semiconductor with a direct bandgap around 2.0 eV and a theoretical maximum power conversion efficiency (PCE) of approximately 20% (Shockley-Queisser limit).5, 8 It has been employed in a number of heterojunction solar cells to date, with ZnO as the most successful n-type counterpart. ZnO is a widely-studied, wide-bandgap semiconductor (approx. 3.4 eV) with a high electron mobility that can also be synthesized cheaply from solution.10, 11 Bilayer Cu2O-ZnO heterojunction solar cells with simple planar interfaces and PCEs of 2%12 and 1.3%13 have been reported previously. In the first case, ZnO was sputtered onto Cu2O that was thermally oxidized by heating sheets of copper above 1000 °C. In the latter, ZnO and Cu2O were sequentially electrodeposited on conducting glass substrates.

In Cu2O-ZnO heterojunctions, ZnO acts as a window layer, such that the majority of absorption occurs in the Cu2O. Photogenerated holes are transported through the Cu2O to a collection electrode and electrons are transported through the Cu2O and ZnO as minority and majority carriers respectively. A built-in bias of approximately 0.6 to 0.7 V is expected for electrodeposited Cu2O-ZnO heterojunctions, which should approximate the achievable open-circuit voltage (VOC). For the thermally-oxidized Cu2O-ZnO heterojunctions mentioned before, a VOC of 0.595 V was obtained, in good agreement with the expected value.12 For electrodeposited bilayer cells, open-circuit voltages ranging from 0.19 to 0.59 V have been reported.13–15 These results suggest that while the VOC may be sensitive to the particular electrodeposition conditions used, values approaching the expected limit are possible.

The maximum theoretical short-circuit current density (JSC) attainable from a Cu2O absorbing layer can be estimated using reported absorption coefficient values.8, 16, 17 The values used are shown in Figure1a. The absorption coefficient is observed to increase slowly at energies just above the bandgap, as exciton transitions at these energies are parity forbidden and must be mediated indirectly by phonons.18 The first allowed transition occurs at approximately 2.6 eV, which is seen to result in a sharp increase in absorption at a wavelength around 475 nm. For various Cu2O thicknesses, the maximum JSC was calculated by multiplying the absorption profile by the standard AM 1.5G solar photon flux (100 mW cm−2) and integrating over all wavelengths. This relationship is shown in Figure 1b. Reflection effects have been neglected and the collection efficiency has been assumed to be 100%. The short-circuit current density approaches a limiting value of approximately 15 mA cm−2 for thick (i.e., 100 μm) Cu2O layers, and current densities greater than 10 mA cm−2 are possible for 10 μm active layers. To our knowledge however, the highest JSC previously reported for an electrodeposited bilayer Cu2O-ZnO heterojunction under AM1.5G illumination is 3.8 mA cm−2 (3 μm Cu2O layer).13 This clearly indicates that poor charge collection limits the performance of electrodeposited bilayer Cu2O-ZnO solar cells. If all photogenerated charges were collected from a 3 μm Cu2O layer, a JSC of approximately 7.8 mA cm−2 would be obtained.

Figure 1.

a) Reported absorption coefficient of Cu2O used for current-density calculation and corresponding LOD (thickness required to absorb 90% of incident radiation of a particular wavelength). b) Maximum theoretical JSC for a Cu2O absorbing layer under AM1.5G illumination. Inset: i) LC of minority carriers in the absorbing layer limits the useful device thickness to a value much less than LOD at most wavelengths. Absorption and charge collection can be decoupled into orthogonal directions by ii) coating the semiconducting layers onto a supporting nanostructure or iii) directly structuring the p-n junction.

Poor charge collection limits the useful device thickness. This is illustrated in inset (i) of Figure 1b, where for most wavelengths, the optical depth (LOD) of the absorbing layer (defined here as the thickness required to absorb 90% of incident radiation of a particular wavelength) is much larger than the charge collection length (LC) in the device. Photogenerated minority carriers created at distances greater than LC from the p-n junction recombine before reaching the interface.

In recent years, photovoltaic materials have been nanostructured in an effort to reconcile short transport distances with thick absorbing layers. A cell, whose useful thickness is limited by carrier transport, can be folded to increase the effective absorbing thickness. This is achieved in inset (ii) of Figure 1b by coating the semiconducting layers on a supporting nanostructure, a strategy that has been employed in dye-sensitized solar cells.19 Alternatively, the semiconductors can be directly nanostructured, as illustrated in inset (iii). A variety of semiconducting nanowire and nanotube morphologies have been used to improve charge collection in hybrid organic-inorganic,11, 20 dye-sensitized,21 extremely thin absorber,22 and inorganic solar cells (Si,23, 24 CdS-CdTe,25 PbSe quantum dots,26 GaAs27). With the exception of toxic PbSe, the synthesis of all of these inorganic nanowire cells involved physical or chemical vapor deposition/etching in vacuum and high-temperature treatments.

Relatively little work has focused on the implementation of nanowire (NW) structures in inexpensive, stable, solution-processed, inorganic solar cells. Previous attempts to nanostructure Cu2O-ZnO solar cells required prohibitively long fabrication times or resulted in efficiencies of less than 0.1%.28–30 Hsueh et al. sputtered Cu2O onto ZnO NWs grown by a vapor-liquid-solid technique,28 and Yuhas and Yang drop-cast Cu2O nanoparticles into a hydrothermally-grown ZnO NW array with a 10 nm TiO2 coating.29 Sputtering of Cu2O onto ZnO NWs resulted in a club-like geometry, as the Cu2O did not fill the space between the wires nor uniformly coat their lengths. Drop-cast nanoparticles provided better filling of the NW array, but resulted in a discontinuous Cu2O-ZnO interface due to the random orientation and arrangement of the nanoparticles. In this work, high-quality interfaces are found to be vital for the production of efficient NW cells.

Cu2O-ZnO NW heterojunction solar cells were synthesized in this work by electrodepositing Cu2O onto arrays of freestanding ZnO NWs. The NW arrays were grown by a standard electrodeposition technique on Zn seed layers on indium tin oxide (ITO)/glass substrates.31 The wires were highly crystalline, grew in a variety of orientations, had nominal lengths on the order of 1 μm, and diameters ranging from approximately 20 nm to 150 nm, as indicated by scanning electron microscopy (SEM). Following electrodeposition of Cu2O onto the NW arrays, a Au or Ag top contact was evaporated. Typical synthesis times were 45 min and 30 min for the ZnO and Cu2O respectively.

Figure2a and b show SEM cross-sections of a Cu2O-ZnO NW heterojunction and a simple bilayer cell for comparison. The electrodeposited Cu2O was found to completely fill the space surrounding the nanowires and form continuous heterojunctions. This is clearly demonstrated by transmission electron microscopy (TEM) investigations (Figure 2c). No voids or gaps are observed between the adjacent Cu2O and ZnO. Energy dispersive X-ray (EDX) spectroscopy coupled with the TEM indicated that high-purity Cu2O and ZnO were deposited, with minimal impurities from the processing detected.

Figure 2.

Cross-sectional SEM images of a) NW (1250 nm nominal NW length) and b) bilayer Cu2O-ZnO heterojunction solar cells. c) TEM image of a Cu2O-ZnO NW heterojunction. Electrodeposition of Cu2O resulted in the formation of continuous heterojunctions with abrupt interfaces.

Figure3a displays IPCE data (thick lines) for NW heterojunctions with a variety of nominal NW lengths and approximately 3 μm absorbing Cu2O layers. The IPCE of the devices are similar at wavelengths less than approximately 460 nm, but differ significantly for longer wavelengths. For wavelengths below 460 nm, the optical depth of the Cu2O is less than 150 nm (as shown in Figure 1a), such that all photons are absorbed close to the Cu2O-ZnO interface and photogenerated electrons are easily collected regardless of the NW length. For wavelengths greater than 460 nm, however, the optical depth increases to values of many microns. Minority electrons resulting from the absorption of these photons are created too far from the interface to be efficiently collected by a traditional bilayer junction. At these longer wavelengths, a clear enhancement in IPCE is observed for cells employing longer ZnO nanowires, indicating that charges are collected from further within the Cu2O layer. For the cell with a nominal NW length of 1250 nm, an IPCE as high as 85% was measured. A marked decrease in IPCE is observed at wavelengths below 375 nm. This wavelength corresponds to the onset of absorption in the ZnO. The low IPCE suggests that recombination is significant in the ZnO layer; many holes created near the ITO-ZnO interface by photon absorption in the ZnO recombine before they can reach the ZnO-Cu2O interface.

Figure 3.

a) IPCE measurements of Cu2O-ZnO NW solar cells with 3 μm Cu2O layers and a variety of nominal NW lengths (thick lines), and the theoretical maximum IPCE expected for a 3 μm Cu2O absorbing layer (thin, dotted line). Enhancement of electron collection from the Cu2O layer and the resulting JSC is observed with increasing NW length. b) Absorbance measurements of ZnO NW arrays with different NW lengths: increased scattering from longer nanowires reduces the fraction of transmitted light. Inset: The enhanced charge collection in NW cells with longer nanowires is due to both minority electron collection from further within the Cu2O layer (larger LC) and a reduction in the optical depth LOD of the Cu2O, arising from light scattering by the ZnO nanowires.

The maximum expected IPCE for a 3 μm Cu2O absorbing layer is included in Figure 3a as a thin, dotted line for reference. It was calculated using the absorption coefficient values from Figure 1a, again ignoring reflection and assuming a collection efficiency of 100% for photogenerated carriers. For wavelengths greater than 475 nm, the IPCE of the cell with a nominal NW length of 1250 nm is seen to approach, and in some cases exceed, the theoretical maximum. This indicates enhanced light absorption by the NW cells, beyond that expected for a planar Cu2O layer. The enhanced absorption has been attributed to the scattering of incoming light by the ZnO NWs. Light scattering from ZnO NWs has been shown previously to increase with NW length and result in an enhancement of the optical path length and absorption in a surrounding material.32 Figure 3b shows absorbance data for ZnO NW arrays of different length; a clear decrease in transmitted light is observed for longer nanowires. Thus the IPCE at wavelengths greater than 460 nm increases with ZnO NW length both because of electron collection from further within the Cu2O layer and because of enhanced photon absorption near the Cu2O-ZnO interface, arising from light scattering by the ZnO nanowires. The inset of Figure 3b illustrates this behavior; longer ZnO nanowires reduce LOD (more light is absorbed) and increase LC (a greater percentage of photogenerated carriers are collected).

For these preliminary NW devices, brief annealing was required to observe rectification and a photovoltaic effect. All NW devices in Figure 3a were annealed for 2 h at 100 °C on a hot plate unless otherwise noted. IPCE data for a NW device (1000 nm nominal length) after both 2 and 4 h anneals is included in Figure 3a. A clear increase in IPCE is observed at longer wavelengths with annealing. X-ray diffraction peaks for Cu2O were observed to sharpen and increase in intensity when devices were annealed, indicating that the enhanced IPCE is due to an improvement in the crystallinity of the electrodeposited Cu2O with annealing, and a corresponding increase in the electron transport length.

The JSC of the devices in Figure 3a under AM 1.5G conditions (100 mW cm−2) are listed in the figure. The IPCE enhancement over the photon-rich 450 to 600 nm wavelength range results in an increase in JSC with increasing NW length. This result conclusively demonstrates that the limited charge collection associated with inorganic solar cells electrodeposited from low-temperature solutions can be addressed by nanostructuring of the active layers.

Figure4 shows the J–V characteristics of NW cells with a nominal NW length of 1000 nm and various Cu2O thicknesses under AM 1.5G illumination. For 1000 nm nanowires, a Cu2O thickness of approximately 2 μm was found to produce the highest JSC. The highest JSC, VOC, fill factor (FF), and PCE reported previously for a nanostructured Cu2O-ZnO heterojunction under standard illumination are 2.35 mA cm−2, 0.13 V, 29%, and <0.1% respectively.28 For the electrodeposited NW cell with a 2 μm Cu2O layer in Figure 4, a JSC of 5.4 mA cm−2, VOC of 0.21 V, FF of 32%, and PCE of 0.36% were obtained, all representing significant increases. This dramatic improvement has been attributed to the quality of the Cu2O-ZnO heterojunction formed. The electrodeposited Cu2O was found to completely fill the ZnO NW array, forming an abrupt interface over the entire ZnO nanowire surface for the efficient transfer of photoexcited charges. In particular, the JSC of 5.4 mA cm−2 observed for the NW device is the highest value ever reported for an electrodeposited Cu2O-ZnO heterojunction under AM 1.5G illumination, and approaches the value of 6.8 mA cm−2 reported for heterojunctions employing highly-crystalline Cu2O oxidized at temperatures above 1000°C.12

Figure 4.

J–V measurements of NW Cu2O-ZnO solar cells under AM 1.5G illumination. All cells were composed of 1000 nm ZnO nanowires and various Cu2O thicknesses. Cells indicated with solid lines were annealed for 4 h at 100 °C.

When special care was taken to limit the formation of impurities at the Cu2O-ZnO interface by modifying the Cu2O deposition solution used, the VOC and PCE were further improved, as indicated by the dashed curve in Figure 4. The PCE of approximately 0.5% measured for this device is five times higher than any value previously reported for a nanostructured Cu2O-ZnO solar cell under standard solar illumination. As well, this efficiency is similar to that of early silicon NW solar cells23 and recent GaAs NW solar cells,27 both which were synthesized by expensive, high-temperature, vacuum deposition methods. The improved interface also eliminated the need for annealing, whereas all other NW devices shown in Figure 4 were annealed at 100 °C for 4 h. The steps taken to limit impurity formation will be discussed in more detail elsewhere.33 By optimizing properties such as NW dimensions and spacing,34 Cu2O thickness, trap densities, carrier concentrations, and interfacial quality, further improvements in the efficiency of Cu2O-ZnO NW solar cells will be possible.

Cu2O-ZnO heterojunctions synthesized by electrodeposition from solutions near room temperature are extremely attractive. The materials are abundant, stable, and have low toxicities and the potential for efficient photovoltaic conversion. The fabrication method is simple, scalable, and requires minimal energy input. In this work, poor charge collection was identified as the factor currently limiting the performance of electrodeposited bilayer Cu2O ZnO solar cells and it was demonstrated that a NW architecture can be used to enhance the minority carrier collection in electrodeposited Cu2O-ZnO heterojunctions. Electrodeposition of Cu2O was found to result in complete filling of self-assembling ZnO NW arrays to produce continuous junctions with abrupt interfaces. The NW architecture reduced the optical depth of the absorbing layer by scattering the incident light and collected charges from further within the absorbing layer, resulting in IPCEs of up to 85%. With these simple improvements to the properties of the heterojunctions, record short-circuit current densities (5.4 mA cm−2) and power conversion efficiencies (0.5%) were obtained under AM 1.5G illumination. This work represents significant progress in the development of ultra-low-cost, stable, inorganic solar cells. Scalable, low-temperature, solution-based methods have been developed, which permit control of the nanoscale properties of inexpensive photovoltaic materials and enable considerable efficiency enhancements in this exciting class of devices.

Experimental Section

Commercial substrates (Praezisions Glas & Optik, 14 mm × 14 mm × 0.7 mm) were used and consisted of an approximately 200 nm thick ITO layer (sheet resistance less than 10 Ω sq−1) on soda glass. All substrates were thoroughly cleaned in an ultrasonic bath with acetone and iso-propanol for 20 min prior to use.

For the growth of ZnO nanowires, a Zn seed layer approximately 50 nm thick was sputtered onto the substrates using an Emitech sputter coater. Alternatively, this coating can be replaced by an electrodeposited seed layer,31 such that these devices can be synthesized entirely at ambient pressure from low-temperature solutions, which makes them suitable for deposition onto a conductive surface with any form factor.

Electrodepositions of Cu2O and ZnO were performed using a standard three-electrode cell in aqueous electrolytes of dissolved precursors. A Princeton Applied Research Model 363 Potentiostat/Galvanostat was used to supply a constant deposition potential or current. The substrates were electrically connected with an insulated wire and Ag paste, which was masked using Kapton polyimide tape. A Ag/AgCl reference electrode in a saturated aqueous KCl solution was employed, with a 6.25 cm2 inert platinum counter electrode. The sample and platinum electrodes were placed approximately 1 cm apart in the deposition solution and the reference electrode was placed immediately adjacent to the sample surface. The depositions were performed under computer control and the amount of charge collected at the working electrode was used to estimate the nominal thickness of material deposited.

ZnO nanowires were potentiostatically deposited at −1.0V vs. Ag/AgCl from a simple ZnCl2 (5 × 10−4M)/KCl (0.1 M) aqueous solution at 78 °C, following from previous reports.31 Oxygen was bubbled in the solution throughout the deposition to ensure oxygen saturation. Cu2O was deposited galvanostatically at −1.0 mA cm−2, following the method of Izaki.13

The Cu2O deposition solution consisted of CuSO4 (0.4 M)/lactic acid (3 M) at 40 °C, to which NaOH (4 M) was added to adjust the pH to approximately 12.5. The chemicals used were reagent grade, and the water purified (resistivity greater than 16 MΩ cm). Typical ZnO NW depositions took approximately 45 min, and 33 min was required to deposit a Cu2O film approximately 3 μm thick.

Gold or silver contacts (0.125 cm2 active area) were evaporated onto the Cu2O using a BOC Edwards resistance evaporator to form an ohmic contact.

J–V measurements were performed using a Keithley 2400 SourceMeter with a custom-made LabView program. A solar simulator equipped with AM 1.5G filters was used at 100 mW cm−2 intensity. IPCE measurements were performed following the J–V measurements, again using the Keithley SourceMeter under computer control and a 150 W Xenon light source focused onto an Omni 150 (LOT-Oriel) dual-grating monochromator with appropriate bandpass filters (incident power of 0.3 mW cm−2 at 490 nm). Both IPCE and J–V measurements were calibrated using an ISE Fraunhofer institute certified silicon reference diode equipped with a KG5 filter to minimize spectral mismatch errors. Absorbance measurements were performed using a Agilent/HP 8453 UV-Vis spectrometer. Scanning electron micrographs were obtained using a LEO VP-1530 field emission SEM. TEM measurements were performed with a Jeol JEM-2011 operated at a 200 kV acceleration voltage and with a FEI Titan operated at an acceleration voltage of 300 kV. The Titan is equipped with an EDX system which was used for chemical analysis. Film crystallinity was examined using a Bruker D8 theta/theta XRD system with Cu Kα radiation (λ = 0.15418 nm) and a LynxEye position sensitive detector.


KPM and JLMD would like to acknowledge the International Copper Association, Peterhouse (Cambridge), the Natural Sciences and Engineering Research Council of Canada, and the Higher Education Funding Council for England, who have funded this work. LSM, HH, and CS would like to thank the German research foundation (DFG) for funding in the Cluster of Excellence “Nanosystems Initiative Munich (NIM)”. This work was also enabled by an Academic Research Collaboration Grant from the British Council Germany and DAAD. The authors would like to thank S. Schmidt and M. Döblinger for technical support on the TEM.