Package‐Free Flexible Organic Solar Cells with Graphene top Electrodes

Package-free flexible organic solar cells are fabricated with multilayer graphene as top transparent electrodes, which show the highest power conversion efficiency of about 3.2% and excellent flexibility and bending stability. The devices also show good air stability, indicating that multilayer graphene is a promising environmental barrier that can protect the organic solar cells from air contamination. Flexible solar cells have promising applications in some emerging areas, such as portable or wearable electronics, synthetic skin, and conformal solar cells for building integration.1-3 Organic photovoltaic devices (OPVs), undergoing fascinating development recently, can be fabricated on flexible substrates by solution process and thus are excellent candidates for the above applications.1 One major challenge in developing flexible OPVs is the deformable transparent electrodes that cannot be achieved with traditional brittle transparent conductive oxides, such as indium tin oxide (ITO).4 It has been recognized that graphene is a promising material for flexible transparent electrodes because it shows ultrahigh carrier mobility, high transparency and excellent mechanical flexibility.5-9 In theory, highly doped graphene can have higher transparency and better conductance than ITO.7 Moreover, the fabrication of graphene electrodes is compatible with roll-to-roll process and mass production.7 It is notable that the recently developed chemical vapour deposition (CVD) methods for synthesising high-quality and large-area graphene have accelerated its practical applications especially for replacing the increasingly expensive ITO as transparent electrodes.5, 6 Acro et al. reported flexible OPVs (area: 0.75 mm2) with CVD graphene as the bottom transparent electrodes, which showed the maximum power conversion efficiency (PCE) of 1.18%.4 The flexible devices exhibited much better bending stability than a device with an ITO transparent electrode. Yin et al. reported flexible OPVs based on poly(3-hexylthiophene):phenyl-C61-butyric acid methyl ester (P3HT:PCBM) with highly reduced graphene oxide as transparent anodes prepared by facile solution process.10 The devices have the area of 1mm2 and the maximum PCE of 0.78%. After bending tests of 1600 cycles, the PCE of the device was decreased for about 30%. Huang et al. reported OPVs by using graphene/carbon nanotube composite films as anodes, which showed PCEs up to 1.27%.11 Then several approaches have been adopted recently to improve the performance of the devices by decreasing the sheet resistances of graphene electrodes. Wang et al. reported OPVs (area: 4 mm2) with multilayer HNO3-doped graphene electrodes, which had the PCEs up to 2.5%.12 Hsu et al. reported OPVs (area: 10 mm2) with sandwiched graphene/tetracyanoquinodimethane (TCNQ)/graphene stacked films as transparent electrodes, demonstrating the maximum PCE of 2.58%.13 We recently reported the semitransparent OPVs with highly doped single-layer graphene electrodes and showed the PCEs of 2.7%.14 It is notable that typical OPVs with P3HT:PCBM active layers and ITO electrodes have PCEs of 4 ∼ 5%.15, 16 Therefore, the devices with graphene electrodes have lower PCEs than expected. Another problem of the above devices is the small device area (≤ 10 mm2) that may limit their practical applications. As reported before, the PCE of an OPV that is smaller than 10 mm2 cannot represent the real efficiency of the device due to the edge effect.14, 17 All of the shortcomings can be attributed to the high sheet resistances of the graphene electrodes. Therefore, the preparation conditions of graphene electrodes need to be optimized. Graphene is impermeable to gas and liquid, which is another important property to many applications. For example, Geim's group recently reported the tunable permeation rate of water across graphene oxide membrane that is completely impermeable to other liquids, vapors, and gases.18 They observed that the permeation of water across the membrane was stopped when the average distance between two graphene layers is less than 0.7 nm. Compton et al. reported that the polystyrene–graphene nanocomposites have low O2 permeability, thus promising packaging materials.19 Chen et al. found that CVD graphene coated on the surface of Cu or Ni could protect the metal from oxidation.20 Therefore graphene-based materials are excellent environmental barriers.21 OPVs should be packaged in air to avoid performance degradation because organic semiconductors are sensitive to oxygen and moisture.22-24 Therefore, the OPVs with graphene top electrodes are naturally protected by the graphene layers although the protection effect has not been reported until now. In this paper, we report the fabrication of package-free flexible OPVs on polyimide (PI) substrates with highly doped multilayer CVD graphene as top transparent electrodes (anode) and P3HT:PCBM as active layer. The devices show the maximum PCE of ∼ 3.2% under the illumination of AM1.5 solar simulator. Because the devices are not packaged, they are very thin (∼50 μm) and extremely flexible, and exhibit excellent bending stability. Then we demonstrate pronounced packaging effect of multilayer graphene top electrodes on the OPVs in air for the first time. We find that air cannot diffuse across the narrow space between two graphene layers and thus the pores in the first layer of graphene are sealed by the second layer, which leads to excellent packaging effect on the devices by the multilayer graphene top electrodes. Figure 1a shows the structure of a flexible OPV with a graphene top electrode. The device was fabricated on a polyimide (PI) substrate with an inverted structure.14 A metal film was deposited on PI by thermal evaporation followed by coating a ZnO film as a hole blocking layer.25 Then a P3HT/PCBM active layer was spin coated on the ZnO film. A graphene electrode was transferred on the active layer from a copper foil by the typical transfer method using poly(methyl methacrylate) (PMMA).6 The PMMA/graphene electrode was doped with the composite thin film of PEDOT:PSS and Au nanoparticles to decrease the sheet resistance with the "floating method" reported before by us.14 All devices have the same area of 12 mm2. The fabrication details are described in the experimental section in the supporting information. a) Schematic diagram of an OPV with the inverted structure: PI/Metal/ZnO/P3HT:PCBM/PEDOT:PSS(Au)/Graphene/PMMA. b) The TEM image of Au nanoparticles in a PEDOT:PSS film. Inset: Selected Area Electron Diffraction Pattern (SAED) of Au nanoparticles in the PEDOT:PSS film. c) Comparison of sheet resistance of single- or multi-layer graphene before and after PEDOT:PSS and Au doping. d) UV-Vis transmittance spectra of graphene films with 1 to 4 layers. As reported before, graphene can be doped effectively by coating a layer of PEDOT:PSS on the surface.14 To further enhance the doping effect, we added HAuCl4 in PEDOT:PSS aqueous solution, which led to some Au nanoparticles in the solution. Figure 1b shows the TEM image and the selected area electron diffraction (SAED) pattern of the PEDOT:PSS film with Au nanoparticles. The red circles denote the position of Au nanoparticles with the average size of about 5 nm. We found that the single-layer CVD graphene film doped with PEDOT:PSS and Au composite has the sheet resistance of 158 ± 30Ω/ÿ while the one doped with PEDOT:PSS only shows the sheet resistance of about 285Ω/ÿ. Therefore, the Au nanoparticles in PEDOT:PSS can enhance the doping effect in graphene, which is due to the decreased Fermi energy in the graphene film because Au has relatively lower Fermi level than PEDOT:PSS.14 Another advantage of adding Au nanoparticles in PEDOT:PSS is the possible plasmatic enhancement for light absorption and efficiency reported very recently.26 To further decrease the sheet resistance of graphene electrodes, multilayer graphene films were prepared by layer-by-layer stacking method on Cu foils. The atomic force microscopy (AFM) images of single- and double-layer graphene films are show in Figure S1 (see supporting information). Then the graphene film was doped with PEDOT:PSS and Au on the bottom surface.14 The light absorption peak at ∼550 nm due to the plasmonic effect of the Au nanoparticles is shown in Figure S2 (see supporting information). It is notable that the PMMA film was not removed in the above process, which can keep the good quality of the underlying graphene film. Figure 1c shows the sheet resistance of doped or undoped multilayer graphene characterized in air for three times. The sheet resistance of four-layer graphene is decreased from 220 ± 22 to 68 ± 10Ω/ÿ after doping. All doped samples have very stable resistances even after being kept in air for 40 days, so the doped graphene has good air stability. Figure 1d shows the transmittance of the multilayer graphene films, which is reduced by about 2.3% in the visible region for an additional single layer. The four-layer graphene film shows the transmittance of about 90%, being comparable to ITO transparent electrodes.14 Three metals, including Al, Cu and Ag, were evaporated on PI substrates as the cathodes of flexible OPVs. Figure 2a shows the current density-voltage (J–V) characteristics of the OPVs with the three types of metal bottom electrodes and single-layer graphene top electrodes. The device with the Al electrode does not show any photocurrent which is due to the oxidation of Al surface when the device was fabricated in air.14 The device with the structure of PI/Cu/P3HT:PCBM/PEDOT:PSS/Graphene/PMMA has a short-circuit current density (Jsc) of 10.48 mA/cm2, an open-circuit voltage (Voc) of 0.38V, a fill factor (FF) of 34.5% and a PCE (η) of 1.39%. The device shows a low Voc owing to the energy mismatch between the work function of Cu (4.6 eV) and the LUMO of PCBM (3.7 eV),14, 27 which results in a Schottky contact between the active layer and the Cu electrode.28 Therefore, a hole blocking interfacial layer is needed between Cu and PCBM to enhance the open circuit voltage. However, Cu can be easily oxidized when a thin ZnO film was coated on the surface, which leads to much worse photovoltaic performance. It is notable that Ag is more suitable for the bottom electrodes of OPVs since Ag is more stable and conductive than Cu and has excellent optical properties (reflectance more than 90% for visible light).29, 30 a) J-V characteristics and b) EQEs of solar cells with different metals as cathode and single-layer graphene as anode. Inset: a) band structure of the device with Ag cathode. b) photograph of a flexible OPV on PI substrate. c) J-V characteristics of OPVs with different number of layers of graphene anode doped with PEDOT:PSS+Au or PEDOT:PSS only. d) EQEs of OPVs with different number of layers of graphene anodes. e) Normalized photovoltaic parameters of the flexible OPV with 2-layer graphene and f) normalized series resistance of the device and the sheet resistance of the graphene electrode as functions of bending cycles. The inset of Figure 2a shows the cascaded band structure of the device with Ag and doped graphene as cathode and anode, respectively. The device with the Ag bottom electrode exhibits Voc of 0.591 V, Jsc of 10.8 mA/cm2, FF of 45.7% and η of 2.91%. Figure 2b shows the external quantum efficiencies (EQE) of the devices with Cu or Ag as bottom electrodes. The device with the Ag electrode has higher EQE in the wavelength ranges of 320-600 nm due to the higher reflectance of the Ag film.30 The inset of Figure 2b demonstrates the excellent flexibility of one device subjected to bending with a radius of curvature of about 1.5 mm. Then multilayer graphene top electrodes were used in OPVs with Ag bottom electrodes. Figure 2c shows the performance of the OPVs with 1 to 4 layers of doped graphene. The device with single-layer graphene doped with PEDOT:PSS only was fabricated for comparison, which exhibits the lowest FF and PCE in all devices. The sheet resistance of the multi-layer graphene electrode decreases with the increase of layer number. However, more layers of graphene imply lower transmittance of electrodes and smaller short circuit currents of OPVs. Therefore, the optimum condition of the layer number is the compromise between the conductance and the transmittance of the graphene film. The photovoltaic parameters and the sheet resistances of the OPVs with different number of graphene layers were shown in Table S1 (see Supporting Information). The OPV with a 2-layer graphene electrode exhibits the best performance including Voc of 0.597 V, Jsc of 10.6 mA/cm2, FF of 50.1% and η of 3.17%. Figure 2d shows the EQEs of the OPVs with different layers of graphene electrodes. The EQE decreases with the increase of layer number, which is consistent with the short circuit currents shown in Figure 2c. The bending stability of the flexible OPV was then characterized. We chose the device with the 2-layer graphene electrode as described above and characterized all photovoltaic parameters of the device after the bending cycles with the radius of curvature of about 3 mm. The sheet resistance of the graphene electrode was also characterized by measuring the resistance between the two Au electrodes beside the OPV as shown in Figure 1a. Figure 2e shows the normalized photovoltaic parameters after different times of bending cycles. The efficiency was decreased for about 8% after 1000 times bending, which is the result of the degraded JSC (decreased for 3%) and FF (decreased for 4.4%). So the bending stability of our device is better than the flexible OPVs with graphene electrodes reported before.4, 10 Besides the quality of graphene film, the flexible substrate is another issue that may influence the bending stability of the device. It has been reported that PI is more thermally stable (smaller thermal shrinkage) and mechanically flexible (higher breaking strain) than PET used in some reported works.31 Figure 2f shows the relative change of the series resistance of the OPV and the sheet resistance of the graphene electrode after the bending tests. The sheet resistance of graphene increased for about 7% while the series resistance of the device increased for 16%. So the degradation of the flexible OPV can be attributed to the change in the graphene electrode as well as the active layer of the device. One major problem of OPVs is the poor air stability because organic semiconductors are sensitive to oxygen and moisture.22, 23 Therefore, OPVs need to be packaged to prohibit the diffusion of oxygen and moisture into the active layers, which may sacrifice the flexibility of the devices. Because graphene is an excellent material for environmental barrier that has been successfully used to suppress the transport of oxygen, helium, hydrogen and water vapour across the films containing it,18, 21 the graphene top electrode in our device can be regarded as a diffusion barrier to air. Therefore, all of our OPVs were not packaged on the top. Consequently, the devices were only about 50 μm thick, which is the main reason for their excellent flexibility. Then the stability of the OPVs was characterized in air. The J-V curves of the devices are shown in Figure S3a-e (see supporting information). Figure 3 shows the photovoltaic parameters of the flexible OPVs with 1 to 4 layers of graphene electrodes after different periods in air. A control device with the standard inverted structure of Glass/ITO/ZnO/P3HT:PCBM/PEDOT:PSS/Au/PMMA was also fabricated and characterized at the same conditions. We coated a PMMA layer (300 nm) on top of the Au electrode of the control device to maintain the comparable condition to other devices. The change of Voc is negligible in all devices, whereas Jsc, FF and thus PCE of some devices degrade rapidly. As shown in Figure 3c, PCE of the control sample decreases very fast at the beginning and then tends to saturate after a certain period. The degradation of PCE represents the diffusion speed of oxygen and moisture into the active layer. We can define the degradation speed of PCE η as a function of time t by , where η(0) and η(t) are the initial PCE and the PCE at time t, respectively. Therefore the control device shows α = 2.75%/h at the beginning. The device with the single layer graphene electrode shows the initial degradation speed of α = 0.94%/h, indicating that the graphene layer has packaging effect although it is not as effective as expected. As shown in Figure 4a, there are inevitably a few holes existing in a single layer graphene, which provide the paths for the diffusion of air into the active layer and lead to device degradation. Evolution of photovoltaic parameters, including (a) short circuit current JSC, (b) Fill factor FF, and (c) PCE, of package-free OPVs with 1 to 4 layers of graphene or Au top electrodes measured in air. Schematic diagram of (a) single- and (b) double-layer graphene films as air (H2O, O2) barrier. It is interesting to find that α is decreased to only about 0.1%/h in the devices with 2 to 4 layers of graphene electrodes and the difference among the degradation speeds of the three devices can be hardly observed. So double-layer graphene is enough to well package the OPVs and further addition of graphene layer does not show better packaging effect. Figure 4b shows the schematic diagram of air diffusion across the double-layer graphene electrode. Oxygen or moisture needs to diffuse into one pore in the first graphene layer, pass through the space between the two graphene layers and diffuse out of another pore in the second layer. As reported before, the space between two graphene layers is impermeable to gas and water if the separation is less than 0.7nm.18 Therefore the thicknesses of single- or double-layer graphene films on Si substrates were characterized by AFM (See supporting information, Figure S4). The single-layer graphene is about 0.70 nm thick while the double-layer graphene is about 1.24 nm. So the average space between two stacked graphene layers is about 0.54 nm in agreement with earlier reports and thus impermeable to air and moisture.5, 18, 32 If the pores in the first layer graphene are not so many, each pore will be sealed by the second layer of graphene. Consequently, air cannot diffuse across the double-layer graphene film. So double-layer graphene electrodes show much better packaging effect than single-layer ones. The slow degradation of the devices with multilayer graphene electrodes in air is probably due to the change of active layer under light illumination and/or slow diffusion of air into the device.16, 22 In summary, package-free flexible OPVs have been fabricated on PI substrates with doped multilayer graphene as top electrodes and Ag films as bottom electrodes. The device with a double-layer graphene electrode shows the maximum PCE of 3.2% and excellent bending stability. It exhibits the relative degradation of PCE by about 8% after 1000 times bending cycles. More importantly, we demonstrate that 2 or more layers of graphene top electrodes can protect the OPV very well from the contamination of air because multilayer graphene films are impermeable to air. Therefore, stable OPVs with graphene top electrodes can be fabricated without package, which may simplify the device fabrication, enhance the flexibility and decrease the cost of the devices. These results indicate that graphene is an excellent material for the transparent electrodes of flexible OPVs as well as other organic devices especially some air sensitive ones. Supporting Information is available from the Wiley Online Library or from the author. This work is financially supported by the Research Grants Council (RGC) of Hong Kong, China (project number: PolyU5322/10E) and the Hong Kong Polytechnic University (project number: A-PL49, A-PK07 and 1-ZV8N). The authors would like to thank the support from Prof. X. L. Liang from Peking University. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.