Introduction

Photovoltaic has been realized as a suitable generating electrical power by converting solar radiation into direct current electricity for the fulfillment of increasing world energy consumption with the least impact on the environment. Although solid state junction devices with high efficiency, usually made of silicon and compound semiconductors, dominate the commercial market, these photovoltaic technologies still receive constraints in market development due to both of expensive materials and complex manufacturing processes1,2,3,4. Consequently, the emerging photovoltaic technologies such as organic cells5,6,7, inorganic cells8,9, quantum dot cells3 and dye-sensitized solar cells (DSSCs)1,2,10,11,12,13, have attracted extensive attention because of their promising inexpensive techniques based on the solution-processed materials. However, these emerging photovoltaic technologies seem to be always associated with unsatisfactory efficiency.

In the past two years, methylammonium lead halide (CH3NH3PbX3, X = I, Cl or Br) and its mixed-halide crystals, corresponding to three-dimensional perovskite structures, have been used as light harvesters for mesoscopic heterojunction solar cells14. The advantages of the direct band gap, large absorption coefficient and high carrier mobility of methylammonium lead halide perovskite nanocrystals render them to be perfect light harvesters15,16,17. Most recently, a high efficiency about 12 ~ 15%, which is comparable to that of the commercial silicon solar cells, has been achieved in solid-state mesoscopic solar cells employing organic hole transporting materials such as Spiro-OMeTAD, PTAA and so on18,19,20,21,22,23,24,25. More importantly, these perovskite nanocrystals can be synthesized by simple and cheap techniques due to their self-assembling character, implying a great potential to bring down the cost of energy production. However, the counter electrodes (CEs) of these high performance photovoltaic devices still need noble mental such as Au or Ag, prepared by thermal evaporation under high vacuum condition15,18,19. Obviously, this high-cost metallic CE is an issue for its large-scale production. Meanwhile, the vacuum evaporation process is also highly energy consumptive and draws apart from its promising inexpensive techniques1. Therefore, the replacement of the costly metallic CEs would be a critical improvement for this kind of high-efficiency heterojunction photovoltaic cells based on CH3NH3PbI3 nanocrystalline light harvester.

Carbon is an abundantly available and low-cost material, which have been applied in DSSCs successfully26,27,28,29. In 1996, Kay and Grätzel firstly reported a new type of liquid monolithic DSSC employing carbon black/graphite composite CE and obtained a promising PCE of 6.7%26. This monolithic photovoltaic device permits printing paste layer by layer on a single FTO glass substrate by screen-printing technique, which offers more positive prospect for commercial production. Moreover, the work function of carbon (−5.0 eV) is close to Au (−5.1 eV). Although these advantages indicate that carbon may be an ideal material to substitute Au as a CE in CH3NH3PbI3 heterojunction photovoltaic cells, to date, none relevant research has been reported. In this communication, we assemble a monolithic CH3NH3PbI3 perovskite/TiO2 heterojunction solar cell based on Carbon black/Graphite CE and printable process. Due to the CH3NH3PbI3 perovskite be capable of acting as a light harvester and at the same time as a hole conductor15, herein we fabricate such heterojunction photovoltaic device without any organic p-type material for transporting positive charge carrier. The results indicates that the mesoscopic heterojunction solar cell with carbon black/spheroidal graphite counter electrode presents high stability and power conversion efficiency of 6.64%, which is higher than those of the flaky graphite based device and comparable to the conventional Au version.

Results

CH3NH3PbI3 has a perovskite structure, as shown in Figure 1a, which is composited of one Pb2+, one CH3NH3+ and three iodine anions in the unit cell. Hence, we prepared lead iodide perovskite precursor solution by mixing the CH3NH3I and PbI2 at 1:1 molar ratio to meet the atom ratio in CH3NH3PbI3. In the manufacturing steps of carbon black/graphite based monolithic device (see Figure 1b), the FTO glass substrates were firstly etched with a laser to form two detached electrode patterns before being cleaned ultrasonically with ethanol. Then, the patterned substrates were coated by a TiO2 dense layer by aerosol spray pyrolysis and a 1 μm nanoporous TiO2 layer was deposited by screen printing with the TiO2 slurry (PASOL HPW-18NR). After being sintered at 450°C for 30 min, a 1 μm ZrO2 space layer was printed on the top of the nanoporous TiO2 layer using a ZrO2 paste, which acts as a insulating layer to prevent short circuit. Finally, a carbon black/graphite CE was coated on the top of ZrO2 layer by printing carbon black/graphite composite slurry and sintering at 400°C for 30 min. After optimization (see Figure S1 ESI), the thickness of carbon black/graphite CE was controlled to about 10 μm.

Figure 1
figure 1

(a) The crystal structure of CH3NH3PbI3 perovskite and the corresponding energy levels of TiO2, CH3NH3PbI3 and Carbon. (b) A schematic structure of a carbon based monolithic device.

The synthesis of CH3NH3PbI3 and deposition on the monolithic device was carried out by drop-coating of a 30 wt% precursor solution onto the carbon black/graphite layer. Upon drying at 50°C, the films darkened in color, indicating the formation of CH3NH3PbI3 in the solid state, confirmed by X ray diffraction (XRD) spectroscopy (see Figure S2, ESI). In our device, when the CH3NH3PbI3 absorbing light, it generate electron on the conduction band (−3.93 eV) and hole on the valence band (−5.43 eV). Since the conduction band of TiO2 is at −4.0 eV and the conduction band of ZrO2 is at −3.27 eV, the electron on CH3NH3PbI3 conduction band could only inject into TiO2. Meanwhile, the hole on CH3NH3PbI3 valence band could inject into the carbon (−5.0 eV) (see Figure 1a). Under AM1.5 solar light of 100 mW·cm−2, the carbon black/flaky graphite based monolithic device produced Voc = 0.825 V and Jsc = 10.6 mA·cm−2, with FF = 0.46, corresponding to a PCE of 4.08%. For further improving the device performance, spheroidal graphite with better conductivity (see Table S1, ESI) and favorable morphology for pore-filling was used in the carbon composite CE to replace the flaky graphite. It could be found that the spheroidal graphite based device showed a much higher performance (Voc = 0.878 V, Jsc = 12.4 mA·cm−2, FF = 0.61 and a PCE of 6.64%) than that of the flaky graphite based device (See Figure 2a). The superiority of spheroidal graphite in device performance could be confirmed by the statistical data showed in Figure S3 and Table S2. These results indicate that the spheroidal graphite CE possess a potential capacity to replacing Au CE15.

Figure 2
figure 2

(a) Photovoltaic characteristics of CH3NH3PbI3 perovskite/TiO2 heterojunction solar cell based different carbon CEs. (b) IPCE as function of incident wavelength.

The incident photo to current conversion efficiency (IPCE) specifies the ratio of extracted electrons to incident photons at a given wavelength, which reflects the light response of the devices and is directly related to the Jsc. From the IPCE spectrum in Figure 2b, we can observe an excellent photocurrent response from 400 to 800 nm showed by the CH3NH3PbI3/TiO2 heterojunction devices. Moreover, the IPCE of spheroidal graphite based device reached a higher value than the flaky graphite based device in the full-scale visible region of the electromagnetic spectrum, in reasonable agreement with the measured values of J-V curves.

In order to investigate the huge difference in device performance of the two kinds of graphite based CEs, scanning electron microscopy (SEM) images have been observed from the cross section of the monolithic devices. Figure 3a and b show the different morphology of the spheroidal graphite and flaky graphite layers in monolithic devices. It' evident that spheroidal graphite possess a loose structure, attribute to the spheroidal morphology. However, in the flaky graphite based CE, large graphite sheets are stacked on the top of ZrO2 space layer. In view of the fact that the devices fabricated by spheroidal graphite perform much higher FF than the ones fabricated by flaky graphite, which deduce a better pore-filling in the TiO2 films of spheroidal graphite based monolithic devices. To confirm it, amplified SEM images of the cross section of TiO2 films have been obtained. From these images, we can find both of the TiO2 films in spheroidal graphite (SG-TiO2, Figure 3c) and flaky graphite (FG-TiO2, Figure 3d) based devices are coated by CH3NH3PbI3 apparently, compared to the bare TiO2 film (Figure 3e). It's notable that the CH3NH3PbI3 coated on SG-TiO2 are more uniform than that on FG-TiO2, corresponding a higher FF in spheroidal graphite based device. We impute the uneven distribution of CH3NH3PbI3 in FG-TiO2 to the arrangement style of flaky graphite, since the perovskite CH3NH3PbI3 coating on TiO2 surface prepared by drop-coating precursor solution onto the carbon black/graphite layer.

Figure 3
figure 3

Cross-sectional structure of the devices.

(a) Spheroidal graphite (SG) based monolithic device. (b) Flaky graphite (FG) based monolithic device. (c) CH3NH3PbI3 coated SG-TiO2 film. (d) CH3NH3PbI3 coated FG-TiO2 film. (e) Bare TiO2 film.

Finally, the long-term stability in the dark of the carbon black based monolithic CH3NH3PbI3 perovskite/TiO2 heterojunction solar cells with the initial efficiency of 6.64% was tested under conditions stored in dry air at room temperature without encapsulation and presented in Figure 4. It could be found that after 840 hours, although a slight decrease of Jsc was measured, the PCE value still remained above 6.5%. These impressive results are indicative of the superior stability of the CH3NH3PbI3 perovskite/TiO2 heterojunction devices.

Figure 4
figure 4

Long term stability at room temperature in the dark.

Inset: the changing characters of the device in 840 h after been fabricated.

Discussion

The present work established for the first time that carbon black/graphite can substitute noble metallic materials as an efficient CE in the application of hole-conductor-free CH3NH3PbI3 perovskite/TiO2 heterojunction solar cells. Impressive PCE values of exceeding 6.64% have been obtained with CH3NH3PbI3/TiO2 heterojunction device based on carbon black/spheroidal graphite CE. There still remains room for further greatly substantial improvement in the PCE, in particular by augmentation of the Jsc through control of the CH3NH3PbI3 perovskite crystallinity and the interface engineering. The fact that carbon black/graphite CE can prepared by screen printing permits the commercial production of high-efficiency CH3NH3PbI3 perovskite/TiO2 heterojunction solar cells. Furthermore, the simple fabrication process and long-term stability of such kind of carbon/graphite based monolithic CH3NH3PbI3/TiO2 heterojunction devices open up new avenues for future development of low-cost photovoltaic cells.

Methods

Fabrication of mesoscopic CH3NH3PbI3/TiO2 heterojunction solar cells

FTO glass plates with high transparency in the visible range were etched with a laser to form two detached electrode pattern before being ultrasonically cleaned with detergent, deionized water and ethanol successively. After that, the patterned substrates were coated with a roughly 100 nm compact TiO2 layer by aerosol spray pyrolysis at 450°C. After cooling down to room temperature naturally, a 1 μm TiO2 nanocrystalline layer (PASOL HPW-18NR) was deposited on top of the compact layer by screen printing and then sintered at 500°C for 30 min. Followed, a 1 μm ZrO2 spacer layer and a 10 μm mesoscopic carbon layer was printed on the top of the TiO2 nanocrystalline layer successively and then the films were sintered at 400°C for 30 min (See Figure S4 ESI). After cooling down, 5 μL of the CH3NH3PbI3 precursor (0.123 g CH3NH3I and 0.3625 g PbI2 were mixed in 1 mL γ-butyrolactone) was dipped on the top of the mesoscopic carbon layer. Then the devices were dried at 50°C on a hot plate under dark. During the drying procedure, the coated devices changed color from light yellow to dark brown, indicating the accomplishment of the solar cell.

Characterization

The cross section of the devices and the carbon films were imaged by a field-emission scanning electron microscope (FE-SEM). Photocurrent density-voltage (J-V) characteristics were measured using a Keithley 2400 source/meter and a Newport solar simulator (model 91192-1000) giving light with AM 1.5 G spectral distribution. A black mask with a circular aperture (0.125 cm2) smaller than the active area of the square solar cell (0.5 cm2) was applied on top of the cell. The incident photon conversion efficiency (IPCE) was measured using a 150 W xenon lamp (Oriel) fitted with a monochromator (Cornerstone 74004) as a monochromatic light source.