Open Access
Add to BibSonomyAbstract
We report a growth of p-CuO nanowire arrays with a simple thermal oxidation and a fabrication of nanowire-based heterojunctions by coating the p-CuO nanowire arrays in an n-ZnO layer through a thermal decomposition method. Their optoelectronic properties and photovoltaic performance were investigated. Compared with the conductance in the dark, a 154% increase in photoconductance was obtained under a white light illumination of 100 mW/cm2. The heterojunctions exhibit an obvious photocurrent increment of 0.264 mA under the illumination of 141 mW/cm2. After annealing the heterojunctions at 100°C for 25 min, a larger open-circuit voltage of 0.37 V was obtained, the short-circuit current density increase to 0.63 mA/cm2 from original 0.49 mA/cm2. The overall power conversion efficiency is 0.1%.
©2011 Optical Society of America
1. Introduction
With gradual reduction of fossil fuels, low-cost and environmental-friendly photovoltaics, which convert sunlight into electrical energy, have attracted much attention and have been fabricated in bulk and low-dimensional materials [1–7]. Compared with the bulk materials, low-dimensional materials, such as semiconductor nanowires (NWs), have advantages in converting solar energy into electrical power due to reduced collection length for minority carriers and enlarged specific surface area for enhancing light absorption [8]. However, the common used semiconductor-based nanowires, such as ZnO, GaN, TiO2, etc., have wide bandgaps (large than 3.0 eV), which limit optical absorption in the visible region of the solar spectrum and lead to a low photovoltaic efficiency. So, high efficiency adsorption materials are still being developed for photovoltaic applications and NWs-based heterojunction structures formed by both wide and narrow bandgap semiconductors have been investigated with enhanced optical adsorption efficiency [9,10]. Since ZnO (bandgap 3.37 eV) [11] and CuO (bandgap 1.2−1.9 eV) [12] are environment-friendly oxide semiconductor materials and have complementary bandgaps for sunlight absorption with very good stability, photovoltaic effect of p-n heterojunctions formed by Al-doped CuAlO2 and ZnO films has been reported [13]. CuO nanorods and NWs have also been employed in dye-sensitized and hybrid solar cells [14,15]. In this work, we use the wide bandgap ZnO (transparent for visible lights) as an n-type material and the narrow bandgap CuO (strong absorption in visible region) as a p-type material to form heterojunctions. To absorb sunlight as much as possible, the p-CuO was synthesized to be nanowire arrays by a simple thermal oxidation without catalysts and was subsequently coated in the n-ZnO layer by a thermal decomposition method. Their optoelectronic properties were investigated and photovoltaic applications were demonstrated.
2. Experiment
In experiment, first, a 10 mm wide × 20 mm long × 0.5 mm high copper foil (99.9%) was cleaned in a diluted hydrochloric acid solution (1 mol/L) for 1~3 min as a substrate, followed by rinsing with deionized water and dried in air for several minutes. Then the cleaned copper foil substrate was placed on a quartz holder and immediately put on an electric oven. It was then heated to 500°C followed by fixing at 500°C for 3 hr. After that, the copper foil was cooled down to room temperature. As a result, CuO NW arrays were obtained on the surface of the copper substrate. Second, five droplets of saturated ethanol solution of zinc acetate were dropped onto the fabricated CuO NW arrays and dried with a blower in air for about 5 min. This dropping and drying steps were repeated 8-12 times to ensure that the CuO NW arrays were fully buried by the zinc acetate. After that, the sample was baked at 350°C for 20 min in air, which makes the zinc acetate decompose into ZnO. Then the sample was cooled to room temperature. So heterojunctions of the CuO NW arrays coated in the ZnO layer were obtained. Lastly, a 200 nm indium-tin oxide (ITO) was sputtered on the ZnO layer as a window for illumination and a conductive layer. Since the work function of Cu is 4.7 eV and the electron affinity of CuO is 4.07 eV, the contact between Cu and CuO is Ohmic contact. The contact between ITO (work function 4.7−4.9 eV) and Ag (work function 4.26 eV) is also Ohmic contact. Figure 1 schematically illustrates the fabrication process of the heterojunctions. To test the characteristics of the heterojunctions, a copper wire was adhered on the copper foil substrate by high quality silver conducting paint to serve as positive electrode. Another copper wire was adhered on the ITO layer to serve as a negative electrode.
The morphology of the sample was characterized using a scanning electron microscopy (SEM) and transmission electron microscopy (TEM) at an accelerating voltage of 15 and 200 kV, respectively. X-ray diffraction (XRD) patterns were collected using a Bruker Smart 1000 CCD diffractometer. X-ray photoelectron spectroscopy (XPS) measurements were performed on a Thermo Fisher Scientific ESCALAB 250 system with a monochromatic Al Ka X-ray source. All binding energies were calibrated to the C1s peak at 284.8 eV. I–V curves were obtained by using Keithley 6487.
3. Results and discussion
Figures 2a and b show SEM images of the fabricated CuO NW arrays in different magnification. The estimated density of the CuO NWs is 1.5 × 107 wires/mm2, the length of the NWs is in the range of 1−20 μm, while the diameters in the range of 50−220 nm. The calculated surface area of CuO NWs is about 25 times larger than the CuO film at same size. From TEM image of a single CuO NW (Fig. 2c), a 74 nm NW in diameter was measured. The high resolution TEM image (Fig. 2d) indicates that the CuO NW is high-quality single crystalline. The distance between two adjacent (110) crystal planes is 0.275 nm. Figure 2e shows the cross-section SEM image view of the CuO NW arrays coated by the n-ZnO shell, while Fig. 2f shows the cross-section SEM image view of the CuO NW arrays coated in the ZnO layer and Fig. 2g shows coated and totally covered by the ZnO layer. Figure 2e-g further shows that the CuO NWs were gradually coated by the ZnO with increasing dropping times. Figure 2h shows the top view SEM image of the CuO NWs coated in the ZnO layer and Fig. 2i shows its corresponding energy dispersive X-ray spectroscopy (EDS), which indicates the heterojunctions consisting of Cu, Zn, and O elements.
Fig. 2 Electron micrographs of CuO nanowires and the coated structure. (a) Top view scanning electron microscope (SEM) image of the CuO NW arrays fabricated by heating a copper foil at 500°C for 3 hr. (b) Cross section view SEM image of the CuO NW arrays. (c) Transmission electron microscope (TEM) image of a single CuO NW. (d) High resolution TEM lattice image of a single CuO NW selected from Fig. 2c. (e) Cross-section SEM image of the CuO NW arrays coated by the n-ZnO shell. (f) Cross-section SEM image of the CuO NW arrays coated in the ZnO layer. (g) Cross-section SEM image of the CuO NW arrays shows totally coated and covered by the ZnO layer. (h) Top view SEM image of CuO NWs coated in ZnO layer and (i) the corresponding EDS spectrum.
The XPS spectrum confirms Cu, Zn, and O appeared at the surface of sample (see Fig. 3a ). It should be mentioned that the C appeared in Fig. 3a comes from surface contamination. The binding energies of Cu 2p3/2, Zn 2p3/2, and O 1s are identified at 933.63, 1021.9, and 529.67 eV, respectively. The peaks of Cu 2p3/2 and Cu 2p1/2 appeared in Fig. 3b indicate that Cu is the II oxidation state. The peaks of Zn 2p3/2, Zn 2p1/2 indicate that Zn is also the II oxidation state (see Fig. 3c). The peak of O 1s appeared in Fig. 3d indicates that O is from the copper oxide and zinc oxide. The XPS measurements verify the structures consisting of CuO and ZnO.
Fig. 3 (a) Wide scan XPS spectrum and (b-d) Cu 2p, Zn 2p, and O 1s core level XPS spectra of the CuO NW arrays coated in ZnO layer.
Figure 4 shows the X-ray diffraction (XRD) spectra of the CuO NW arrays and the coated structures. It is noted that the Cu2O peak is from the thin-film precursor which was first formed before the growing of the CuO NWs. By comparing the spectra of Fig. 4a with 4b, four additional ZnO diffraction peaks were appeared and assigned to (100), (002), (101), and (110) surface of ZnO. The results of XPS showed in Fig. 3 and XRD showed in Fig. 4 indicate that the heterojunctions were formed by CuO and ZnO.
To characterize rectifying behavior of the heterojunctions, a current-voltage (I−V) characteristic was measured at room temperature in the dark (Fig. 5a ). The I−V curve clearly shows the nonlinear increase of current and asymmetric I−V relation with a turn-on voltage of about 0.25 V. A rectification ratio of I +0.5/I −0.5 = 13.8 was calculated according to the forward current of 0.192 mA at +0.5 V and a reverse leakage current of 13.9 μA at −0.5 V. The rectifying behavior shows that the electrons flow from the n-ZnO to the p-CuO by applying a forward bias voltage across the heterojunctions. The heterojunctions exhibits a diode behavior in the dark. Under a white light illumination of 100 mW/cm2 (standard air mass (AM) 1.5 global solar spectra), an obvious photocurrent was appeared (Fig. 5b). The negative photocurrent at forward bias voltage (0~0.35 V) is consisted of two components. One is reverse photo-generated current which was caused by photo-generated voltage (V ph), the other is forward current which was caused by the bias voltage (V). The former is larger than the latter. Thus the photocurrent is negative. Figure 5c shows a relation of photocurrent versus applied bias voltage under different light illuminations. It can be seen that the negative photocurrent only changes a little when the bias voltage ranges from −1.0 to 0 V. The reason for this phenomenon is that the p-n heterojunctions are under a reverse voltage, the resistance of the heterojunctions (~14.5 KΩ) is larger than that (~1.28 KΩ) of under a forward bias voltage. At the bias voltage of 0−0.45 V, the negative photocurrent gradually decreases and approaches to zero with the increasing of the bias voltage. This is because the heterojunctions are under the forward bias voltage, the photo-generated current is still larger than the current caused by the bias voltage, and the direction of the former is opposite to that of the latter. With the increasing of bias voltage, the current caused by the bias voltage gradually increases, then it gradually equals to the photo-generated current, therefore, the negative photocurrent gradually decreases, and tends to zero. When the bias voltage increases from 0.45 to 1.0 V, the photocurrent begins to increase with the increasing of bias voltage. At this time, the heterojunctions are under forward bias voltage and the bias voltage is much larger than the photo-generated voltage, so the influence of the photo-generated voltage can be ignored, the photocurrent is mainly caused by the bias voltage. From Fig. 5c, we calculated that, at a bias voltage of 1.0 V, the conductances are 0.95, 1.04, 1.15, 1.39, and 1.57 mS under the illuminations of 11, 40, 65, 100, and 141 mW/cm2, respectively. The respective photoconductance is 0.04, 0.13, 0.24, 0.48, and 0.66 mS. Compared with the conductance in the dark at 1.0 V (0.91 mS), increments in photoconductance are 104%, 114%, 126%, 154%, and 173% for illuminations of 11, 40, 65, 100, and 141 mW/cm2, respectively. Figure 5d shows a relation of photocurrent versus light illumination intensity at different applied bias voltage. It can be seen that, at the bias voltage of 0.5 V, the photocurrent decreases slightly when increasing the light illumination intensity. This results from the increasing of photo-generated voltage with the increasing of the light illumination intensity, so V−V ph gradually decreases, which leads to the decreasing of the photocurrent. At the bias voltage of 0.6 V, the photocurrent keeps stable. In this case, the photo-generated voltage approaches to the bias voltage, photo-generated carriers caused by the light illumination are driven by both the bias voltage and the photo-generated voltage, so the photocurrent is almost the same under different light illumination. At the bias voltage of 0.7, 0.8, and 0.9 V, the photocurrent increases with the increasing of light illumination. This is because when the bias voltage is larger than the photo-generated voltage, the heterojunctions are under forward bias voltage, the photo-generated carriers are mainly driven by the bias voltage, so the photocurrent gradually increases with the increasing of light illumination. For the heterojunctions (grown at 500°C for 3 hr with a CuO NW density of 1.5 × 107 wires/mm2), the length of the CuO NWs influences optical absorption and the diameter of the CuO NWs influences collection of photo-generated carriers. With an increase of the NW length, more illuminated light can be absorbed for carrier generation. But more carrier traps and scattering centers will also be induced. These traps and scattering centers will cause carrier recombination. The optimal NW length should be 10−15 μm. Similarly, with a decrease of the NW diameter, carrier diffusion length will be decreased, this will benefit to carrier collection. But the surface state density will be increased. This increment is harmful to carrier collection. The optimal diameter of the NWs is ~200 nm. In this experiment, the total thickness of the ZnO layer filled in the gaps between the CuO NWs is 15–18 μm. Compared with our previous work (response time of 7.5 s) [16], a response time of 2.3 s has been improved at the falling edge of the photocurrent by using this structure. Moreover, this structure is convenient to control the thickness of the ZnO layer and to improve the collection efficiency of carriers. It should be pointed out that the absorption of ZnO for visible light is very weak, for example, its photocurrent is 10−5 mA under a 365-nm UV light illumination and only 10−8 mA under a 532-nm green light illumination [17]. For the ZnO layer of this work, its photocurrent is only 2.4 × 10−5 mA under a white light illumination of 141 mW/cm2.
Fig. 5 Current-voltage relations of the CuO NW arrays coated in ZnO layer. (a) I-V characteristics in the dark at room temperature. (b) I-V characteristics in the dark and a standard AM 1.5 global solar spectrum illumination (i.e. 100 mW/cm2). (c) Photocurrent-voltage curves under different illuminations of 11−141 mW/cm2. (d) Photocurrent-illuminations curves at different bias voltages.
Figure 6a shows a schematic energy band diagram of the CuO-ZnO p-n heterojunctions at thermal equilibrium condition. In Fig. 6a, V bi is the built in potential, E F the Fermi level, E g1 the bandgap of CuO, and E g2 is the bandgap of ZnO. Since the energy levels of the bottom of the conduction band (E c1) and the top of the valence band (E v1) of the CuO are E c1 = 4.07 eV and E v1 = 5.42 eV [15], while the energy levels of the ZnO are E c2 = 4.35 eV and E v2 = 7.7 eV [18], so ΔE c = 0.28 eV and ΔE v = [(E g2 − E g1) − ΔE c] = 1.72 eV. The intrinsic physical mechanism of photocurrent for the heterojunctions is that electrons in the valence band of the p-CuO and the n-ZnO will be excited to the conduction band when a white light illuminates the p-n junctions, as a result, holes are generated in the valence band (Fig. 6b). Under the forward bias voltage V with the light illumination, photons with energy less than the energy bandgap of ZnO but greater than the energy bandgap of CuO will transmit through the ZnO layer which acts as the transparent window, and be absorbed by the p-type CuO. According to the electron structure of the CuO [12], the 3d electrons in the valance band of the CuO are excited to the conduction band. Simultaneously, light with photon energy larger than the energy bandgap of ZnO will be absorbed by the ZnO layer. The holes and electrons generated in both sides of the heterojunctions are collected effectively and thus yield the photocurrent (schematically shown in Fig. 6b).
Fig. 6 (a) Schematic energy band diagram of the CuO-ZnO heterojunctions in the thermal equilibrium condition. (b) Schematic energy band diagram of the heterojunctions showing the photogenerated carrier transfer process under standard AM 1.5 illumination and forward bias.
To measure time-related photocurrent, the sample was illuminated by a white light of 141 mW/cm2 with an illuminating period of 30 s. The white light was turned On and Off repeatedly for five cycles. Each cycle is 60 s. The photocurrent was recorded by a picoammeter and shown in Fig. 7a . The rising and falling times obtained from 90%–10% of the maximum photocurrent at the rising and falling edges are 4.2 and 5.2 s, respectively. The relative slow response can be attributed to the following three reasons: (1) The transportation direction of photo-generated carriers in depletion region is opposite to the direction of the built in potential. The built in potential prevents the transportation of the generated carriers and causes a delay of time response; (2) The carrier mobility of the p-type CuO NW is relatively low (μ = 2–5 cm2 V−1s−1) [19], this is not benefit to carrier drift. The low drift velocity will lead to a slow response; (3) The larger area of depletion region in our heterojunctions brought a larger parasitic capacitance C. It caused a larger RC time constant which prolonged the time response. The response curve in Fig. 7a further shows that the photocurrent increases gradually and reaches a steady state at 210 s. After 210 s, the photo-generated current is saturated with a number of 0.264 mA. Compared with the peak photocurrent at 60 s, the increase of the photocurrent is mainly attributed to the Fermi-level pinning on the surface of the ZnO layer, which results in a recombination barrier. It influences the photo-generated carriers quick recombining when light was Off, while generation and recombination of carriers reach the equilibration after 210 s.
Fig. 7 (a) Time-related photocurrent of the CuO nanowire arrays coated in ZnO layer to the periodic light irradiation (Jlight = 141 mW/cm2, tOn/Off = 30 s) at a bias voltage of 0.5 V. (b) Current density-voltage curve (before and after annealing) in the standard AM 1.5 global solar spectrum illumination condition for the photovoltaic performance.
As an important application, photovoltaic performance of the CuO NWs coated in the ZnO layer was tested under an AM 1.5 illumination. Figure 7b shows the measured current-voltage (J–V) curve. The measured open-circuit voltage V oc is 0.36 V and the short-circuit current density J sc is 0.49 mA/cm2. It should be mentioned that the relatively low short-circuit current density is mainly due to a relatively larger sheet resistance of the ITO conductive layer (~200 Ω/sq). The filling factor (FF) is about 35%, which was calculated according to FF = I m V m/I sc V oc [20], where I m and V m are the current and voltage, respectively, at the maximum power point of the I–V curve. Further experiment indicates that the V oc, J sc, and FF can be improved to 0.37 V, 0.63 mA/cm2, and 36.9%, respectively by annealing the structure at 100°C for 25 min. The overall power conversion efficiency is 0.1%, which was calculated according to η = V oc×J sc×FF/P [21], where P is the illumination intensity. It should be pointed out that the low power conversion efficiency is due to the large sheet resistance of the ITO conductive layer and the low mobility in ZnO layer caused by ZnO nanoparticles. The NW-based heterojunctions offer advantage of separating the process of light absorption and carrier collection. In addition, due to the diameter of the CuO NW (50−220 nm) is close to the minority diffusion lengths (100−200 nm). Thus the carrier collection efficiency can be improved by using the NWs compared to that of the thin film counterpart.
4. Conclusion
In conclusion, nanowire-based p-n heterojunctions were fabricated by coating p-CuO nanowires arrays into an n-ZnO layer, their optoelectronic properties and photovoltaic performance were investigated under white light illumination with different intensity. Compared with the conductance in the dark, an about 154% increment in photoconductance was obtained under a white light illumination of 100 mW/cm2. The heterojunctions exhibit a photo-generated current of 0.264 mA after three cycles under the illumination of 141 mW/cm2. Under the white light illumination of 100 mW/cm2, the heterojunctions show a larger open-circuit voltage of 0.37 V and filling factor of 36.9%. The overall power conversion efficiency is 0.1%. These results show that the heterojunctions could be useful for durable and nontoxic photovoltaic components.
Acknowledgments
The authors thank A/Prof. Yong Liu for assistance in photovoltaic measurements and Prof. Xiang Zhou for assistance in experiment. This work was supported by the National Natural Science Foundation of China (Grants 60625404 and 10974261).
References and links
1. S. H. Park, A. Roy, S. Beaupré, S. Cho, N. Coates, J. S. Moon, D. Moses, M. Leclerc, K. Lee, and A. J. Heeger, “Bulk heterojunction solar cells with internal quantum efficiency approaching 100%,” Nat. Photonics 3(5), 297–302 (2009). [CrossRef]
2. N. Haberkorn, J. S. Gutmann, and P. Theato, “Template-assisted fabrication of free-standing nanorod arrays of a hole-conducting cross-linked triphenylamine derivative: toward ordered bulk-heterojunction solar cells,” ACS Nano 3(6), 1415–1422 (2009). [CrossRef] [PubMed]
3. S. Ren, N. Zhao, S. C. Crawford, M. Tambe, V. Bulović, and S. Gradečak, “Heterojunction photovoltaics using GaAs nanowires and conjugated polymers,” Nano Lett. 11(2), 408–413 (2011). [CrossRef]
4. L. E. Greene, M. Law, B. D. Yuhas, and P. Yang, “ZnO-TiO2 core-shell nanorod/P3HT solar cells,” J. Phys. Chem. B 111, 18451–18456 (2007).
5. J. J. Cole, X. Wang, R. J. Knuesel, and H. O. Jacobs, “Integration of ZnO microcrystals with tailored dimensions forming light emitting diodes and UV photovoltaic cells,” Nano Lett. 8(5), 1477–1481 (2008). [CrossRef] [PubMed]
6. W. Wei, X.-Y. Bao, C. Soci, Y. Ding, Z.-L. Wang, and D. Wang, “Direct heteroepitaxy of vertical InAs nanowires on Si substrates for broad band photovoltaics and photodetection,” Nano Lett. 9(8), 2926–2934 (2009). [CrossRef] [PubMed]
7. S. Dayal, N. Kopidakis, D. C. Olson, D. S. Ginley, and G. Rumbles, “Photovoltaic devices with a low band gap polymer and CdSe nanostructures exceeding 3% efficiency,” Nano Lett. 10(1), 239–242 (2010). [CrossRef]
8. Z. Fan, D. J. Ruebusch, A. A. Rathore, R. Kapadia, O. Ergen, P. W. Leu, and A. Javey, “Challenges and prospects of nanopillar-based solar cells,” Nano Res. 2(11), 829–843 (2009). [CrossRef]
9. K. P. Musselman, A. Wisnet, D. C. Iza, H. C. Hesse, C. Scheu, J. L. MacManus-Driscoll, and L. Schmidt-Mende, “Strong efficiency improvements in ultra-low-cost inorganic nanowire solar cells,” Adv. Mater. (Deerfield Beach Fla.) 22(35), E254–E258 (2010). [CrossRef]
10. B. D. Yuhas and P. Yang, “Nanowire-based all-oxide solar cells,” J. Am. Chem. Soc. 131(10), 3756–3761 (2009). [CrossRef] [PubMed]
11. J. Yan, X. Fang, L. Zhang, Y. Bando, U. K. Gautam, B. Dierre, T. Sekiguchi, and D. Golberg, “Structure and cathodoluminescence of individual ZnS/ZnO biaxial nanobelt heterostructures,” Nano Lett. 8(9), 2794–2799 (2008). [CrossRef] [PubMed]
12. D. Wu, Q. Zhang, and M. Tao, “LSDA+U study of cupric oxide: electronic structure and native point defects,” Phys. Rev. B 73(23), 235206 (2006). [CrossRef]
13. K. Tonooka, H. Bando, and Y. Aiura, “Photovoltaic effect observed in transparent p–n heterojunctions based on oxide semiconductors,” Thin Solid Films 445(2), 327–331 (2003). [CrossRef]
14. S. Anandan, X. Wen, and S. Yang, “Room temperature growth of CuO nanorod arrays on copper and their application as a cathode in dye-sensitized solar cells,” Mater. Chem. Phys. 93(1), 35–40 (2005). [CrossRef]
15. Q. Bao, C. M. Li, L. Liao, H. Yang, W. Wang, C. Ke, Q. Song, H. Bao, T. Yu, K. P. Loh, and J. Guo, “Electrical transport and photovoltaic effects of core-shell CuO/C60 nanowire heterostructure,” Nanotechnology 20(6), 065203 (2009). [CrossRef] [PubMed]
16. X. H. Zhao, P. Wang, and B. J. Li, “CuO/ZnO core/shell heterostructure nanowire arrays: synthesis, optical property, and energy application,” Chem. Commun. (Camb.) 46(36), 6768–6770 (2010). [CrossRef]
17. H. Kind, H. Yan, B. Messer, M. Law, and P. Yang, “Nanowire ultraviolet photodetectors and optical switches,” Adv. Mater. (Deerfield Beach Fla.) 14(2), 158–160 (2002). [CrossRef]
18. Z. Guo, D. Zhao, Y. Liu, D. Shen, J. Zhang, and B. Li, “Visible and ultraviolet light alternative photodetector based on ZnO nanowire/n-Si heterojunction,” Appl. Phys. Lett. 93(16), 163501 (2008). [CrossRef]
19. L. Liao, Z. Zhang, B. Yan, Z. Zheng, Q. L. Bao, T. Wu, C. M. Li, Z. X. Shen, J. X. Zhang, H. Gong, J. C. Li, and T. Yu, “Multifunctional CuO nanowire devices: p-type field effect transistors and CO gas sensors,” Nanotechnology 20(8), 085203 (2009). [CrossRef] [PubMed]
20. J. A. Czaban, D. A. Thompson, and R. R. LaPierre, “GaAs core-shell nanowires for photovoltaic applications,” Nano Lett. 9(1), 148–154 (2009). [CrossRef] [PubMed]
21. Z. Fan, H. Razavi, J.-W. Do, A. Moriwaki, O. Ergen, Y.-L. Chueh, P. W. Leu, J. C. Ho, T. Takahashi, L. A. Reichertz, S. Neale, K. Yu, M. Wu, J. W. Ager, and A. Javey, “Three-dimensional nanopillar-array photovoltaics on low-cost and flexible substrates,” Nat. Mater. 8(8), 648–653 (2009). [CrossRef] [PubMed]
