Letter pubs.acs.org/NanoLett
Hole Selective NiO Contact for Efficient Perovskite Solar Cells with Carbon Electrode Xiaobao Xu,†,‡,§ Zonghao Liu,†,§ Zhixiang Zuo,† Meng Zhang,† Zhixin Zhao,† Yan Shen,† Huanping Zhou,‡ Qi Chen,‡ Yang Yang,‡ and Mingkui Wang*,† †
Michael Grätzel Centre for Mesoscopic Solar Cells, Wuhan National Laboratory for Optoelectronics, School of Optical and Electronic Information, Huazhong University of Science and Technology, Wuhan, Hubei 430074, P. R. China ‡ Department of Materials Science and Engineering, University of California, Los Angeles, California 90095, United States S Supporting Information *
ABSTRACT: In this study, we communicate an investigation on efficient CH3NH3PbI3-based solar cells with carbon electrode using mesoporous TiO2 and NiO layers as electron and hole selective contacts. The device possesses an appreciated power conversion efficiency of 14.9% under AM 1.5G illumination. The detailed information can be disclosed with impedance spectroscopy via tuning the interfaces between CH3NH3PbI3 and different charge selective contacts. The results clearly show charge accumulation at the interface of CH3NH3PbI3. The NiO is believed to efficiently accelerate charge extraction to the external circuit. The extracted charge could improve photovoltaic performance by shifting hole Fermi level down, achieving a high device photovoltage. A fast interfacial recombination at the interface of CH3NH3PbI3/electron selective contact layer (mesoporous TiO2), occurring in millisecond domains, is the critical issue for charge carrier recombination loss. KEYWORDS: Perovskite, metal oxide, solar cell, interface charge transfer, impedance
O
large amount photogenerated carrier loss in the vicinity of the MAPbX3/charge selective layer interface. The photocurrent is not only limited by light-harvesting and exciton dissociation, as well as space charge layer, but also by recombination of the electrons and holes in devices.19−21 Femtosecond transient adsorption spectroscopy and photoluminescence quenching investigations have shown an exceeding 100 nm transport length for photogenerated charges in MAPbX3 layer.22,23 This indicates that the electron/hole transport in their respective materials could eventually dominate the perovskite solar cell devices’ photovoltaic performance. For instance, the competition between the interfacial recombination and carrier transportation would determine the collection efficiency of charge carriers. This raises the importance of knowledge about those physical mechanisms that govern carrier transport and loss when various selective layers are used. This study communicates a strategy to address the aforementioned issues. We fabricated perovskite solar cell devices by using low-cost and robust mesoporous TiO2 and NiO layers as electron and hole selective contacts with doctorblade technique, respectively. In such devices, a mesoporous ZrO2 layer (insulator) was adapted as a space separator
rganic−inorganic lead halide perovskite based solar cells are considered as one of the most significant developments in the field of photovoltaics1,2 and have become one of the best choices for solar energy applications.3,4 Since the first report on solid-state devices using CH3NH3PbI3 as light harvester in 2012,5,6 power conversion efficiency (PCE) of this type of solar cell has already exceeded 15% by finely tuning interfacial energy levels and controlling thin film morphologies.7−10 Meanwhile, a wide range of architectures, such as mesoscopic heterojunction and planar heterojunction architectures, for highly efficient perovskite solar cell devices have been demonstrated benefiting from the remarkable CH3NH3PbX3 (MAPbX3, abbreviated as MA+ ion hereafter, and X = I, Br, and Cl) materials with excellent electrical and optical properties.11−14 For practical application of MAPbX3 in solar cells, there still remains some fundamental issue to be discussed. Organic chemicals, such as spiro-OMeTAD or PCBM, are commonly used for hole or electron transport purposes.15−17 However, these organic electronic components have to face disadvantages of poor crystallinity, low mobility, and possible degradation under environmental influences. Furthermore, simulation results have indicated that a photocurrent of 27.2 mA cm−2 could be reached for perovskite solar cell devices if total photons in the range of 280−800 nm can be fully converted to electricity.18 So far, most efficient perovskite solar cell devices only present a photocurrent of about 20 mA cm−2, indicating a © XXXX American Chemical Society
Received: December 8, 2014 Revised: March 25, 2015
A
DOI: 10.1021/nl504701y Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
Figure 1. Cross sectional SEM images of the three different hole transporter free perovskite solar cells. (a) Device A TiO2/ZrO2/carbon(MAPbI3); (c) device B TiO2/NiO/carbon(MAPbI3); and (e) device C TiO2/ZrO2/NiO/carbon(MAPbI3). Distribution of components in devices A and C obtained by the line-scan analysis of EDX map: (b) Ti, Zr, and Pb in device A, and (d) Ti, Zr, Ni, and Pb in device C, respectively.
Figure 2. Photocurrent density voltage (J−V) characteristics of the three different devices under AM 1.5G illumination (100 mW cm−2): (a) current−voltage characteristics of device A TiO2/ZrO2/carbon(MAPbI3) (black), device B TiO2/NiO/carbon(MAPbI3) (red), and device C TiO2/ ZrO2/NiO/carbon(MAPbI3) (blue); and (b) IPCE spectra of the three devices.
NiO mesoscopic layer was further deposited onto the insulating ZrO2 layer to form p−i−n layer architecture in device C (Figure 1d). In a p−i−n layer junction, the carrier separation depends on the drift presenting within the electric field of the junction. Therefore, this would inhibit the geminate recombination and enhance the charge collection.33 A homogeneous distribution of MAPbI3 in the devices is illustrated by the energy dispersive X-ray spectroscopy (EDX) image. As shown in Figure 1b,d, the intensity of EDX for lead signal in the profile has a sharp increase near the TiO2/FTO glass interface, then presenting homogeneous across the layers. The concave curve of Ti, Zr, and Ni concentration distribution suggests the successful deposition of various oxides layers. For comparison purposes, a NiO layer in device B (Figure 1c) was used instead of the ZrO2 seen in device A.28 In devices A−C, the mesopores within the oxide layers are filled with MAPbI3.8,26,27 Some preliminary photovoltaic experiments were conducted to evaluate the performance of devices by varying the TiO2, ZrO2, and NiO film thicknesses. It was found that the film thickness played a vital role on the devices performance (Table S1). Therefore, a 480 nm-thick TiO2 combined with 500 nmthick ZrO2 and 480 nm-thick NiO is adopted in the following optimizations and investigations. Figure 2a exhibits J−V characteristics of the mesoscopic junction in the devices under standard AM 1.5G illumination at 100 mW cm−2. The photovoltaic parameters, open circuit voltage (VOC), fill factor (FF), short circuit current density (JSC), and PCE are tabulated in Table 1. Device A (TiO2/ZrO2/carbon(MAPbI3)) exhibited
between TiO2 (n-type semiconductor) and NiO (p-type semiconductor) layers to realize a p-i-n layer configuration. Recently, NiO has been used in perovskite heterojunction solar cells with efficiency in which NiO acts as an electron blocking material and hole transport material.24 The mesoscopic NiO layer acted as a hole conductor or an electron blocking layer to suppress charge recombination and facilitate the hole extraction.24 This can be proved with photoinduced absorption spectroscopy and photoluminescence measurements.25 In this study, three devices are investigated aiming to obtain a systematic comparison of interfacial recombination. Devices A, B, and C are denoted to TiO2/ZrO2/carbon(MAPbI3),26,27 TiO2/NiO/carbon(MAPbI3),28 and TiO2/ZrO2/NiO/carbon(MAPbI3), respectively. The NiO interlayer promises to greatly enhance the device performance by extracting hole efficiently and enlarge the difference between electron/hole Fermi levels of selective materials, thus improving the device VOC. This resulted in efficient perovskite solar cell devices based on mesoscopic TiO2 and NiO layer selective contacts, showing a high PCE of 14.9% with an average of 13.7% under AM 1.5G illumination testing conditions. Electronic impedance spectroscopy was applied to understand the internal electrical processes in such devices.29−31 Figure 1 presents the cross-section SEM images of perovskite solar cell devices A−C using mesoscopic inorganic oxide layers. Device A (Figure 1a) with TiO2/ZrO2/carbon(MAPbI3) has been accepted as a p−n junction, in which a depletion region is formed at the MAPbI3/TiO2 junction.31,32 Herein, a p-type B
DOI: 10.1021/nl504701y Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
Unfortunately, the direct contact of the TiO2 to NiO in device B results in lower photovoltaic performance, compared with that of device C, possibly originating from an anomalous recombination at the TiO2/NiO interface. N2 adsorption− desorption isotherms and the corresponding pore size distribution curves of TiO2 and NiO are shown in Figure S1. The amount of absorbed N2 increased/decreased at a high P/P0 (P/P0 > 0.9) for the two samples (Figures S1a and 1b). Surface area and pore volume were determined to be 95.17 m2 g−1 and 0.47 cm3 g−1 for TiO2, and 45.99 m2 g−1 and 0.16 cm3 g−1 for NiO, respectively. The pore size analyzed by the Barrett− Joyner−Halenda method indicated a narrow distribution for both samples (5−50 nm). The high surface area and pore volume of the oxides allow efficient contacts between the oxides and MAPbI3, which play important roles for the improved performance. Considering similar film thickness for the TiO2 and NiO layers, the ratio of the surface area for the TiO2 and NiO layer can be estimated to be about 5:4. This result coincides with the EDX analysis in Figure 1e, showing that the concentration of Pb in TiO2 and NiO films almost keeps the same. Electrical impedance spectroscopy (IS) measurements were conducted to investigate the internal electrical properties of perovskite solar cell devices. Figure 3 shows the Nyquist plots and the corresponding Bode phase plots in the frequency range
Table 1. Photovoltaic Parameters of Devices A−C under AM 1.5G Illumination; the Obtained RCE of Devices Were Extracted from the IS (under Illumination) at −0.55 V
device A device B device C
VOC [V]
JSC [mA cm−2]
FF
PCEmax [%]
PCEavg [%]
RCE [Ω]
0.846 0.865 0.917
14.3 17.23 21.36
0.66 0.7 0.76
8.00 10.4 14.9
7.2 9.8 13.7
30 16 25
the highest PCE of 8.0% with a JSC of 14.3 mA cm−2, a VOC of 0.846 V, and a FF of 0.66. The replacement of ZrO2 with NiO resulted in a significant increase in VOC, JSC, and FF, achieving a PCE of 10.4% for device B. Device C with TiO2/ZrO2/NiO/ carbon(MAPbI3) configuration showed a further performance improvement, achieving the highest PCE of 14.9% with a VOC of 0.917 V, a JSC of 21.36 mA cm−2, and a FF of 0.76. Devices B and C present higher FF compared to that of device A, indicating smaller series resistance and larger shunt resistance of these devices. Considering a similar TiO2 thickness of 480 nm in three devices, the p-type NiO is believed to accelerate the hole transport and collection. Figure 2b presents the IPCE spectra of these devices. A better IPCE performance in devices B and C compared to device A in the range from 460 to 800 nm also indicates better charge collection in these devices, while the three devices have similar light-harvesting abilities.34
Figure 3. Electronic impedance spectroscopy characteristics of the three devices under illumination (10 mW cm−2). Electronic impedance spectrum in the form of Nyquist plots (left) and Bode phase plots (right) for device A TiO2/ZrO2/carbon(MAPbI3) (black), device B TiO2/NiO/ carbon(MAPbI3) (red), and device C TiO2/ZrO2/NiO/carbon(MAPbI3) (blue) measured under illumination with a bias at (a) −0.8 V, (b) −0.55 V, and( c) −0.3 V. The real lines are the fitting results. C
DOI: 10.1021/nl504701y Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
Figure 4. Derived equivalent circuit components obtained from impedance measurements under light for three different devices: (a) the recombination resistance Rct1 at the MAPbI3/TiO2 interface; (b) the corresponding capacitance C1; and (c) the charge recombination lifetime τe1 for device A TiO2/ZrO2/carbon(MAPbI3) (black), device B TiO2/NiO/carbon(MAPbI3) (red), and device C TiO2/ZrO2/NiO/carbon(MAPbI3) (blue).
charge recombination resistance Rct1 exponentially decreases with the bias for the three devices after −0.75 V. Similar results were observed in planar heterojunction solar cells, which was ascribed to the eliminate of the depleted region width by increasing the bias on devices.40 Figure 4b presents the capacitance C1−voltage characteristic of various devices, showing the same tendency with Rct1. These evidence lead us to propose that there are two capacitances in series connection controlling of the apparent capacitance C1 in impedance study. When the applied bias is less than the built-in potential (i.e., about −0.75 V in this study), the space charge from the TiO2/ MAPbI3 junction makes a great contribution to the capacitance C1. As the applied potential exceeds the built-in potential, the obtained capacitance C1 would mainly originate from a charge accumulation (i.e., chemical capacitance) at the MAPbI3/TiO2, showing an exponential dependence on the bias.23 The interfacial charge recombination rate (k) for the MAPbI3/ TiO2 interface can be informed by the charge lifetime (τe1, k ≈ (τe1)−1), which is determined from the recombination resistance (Rct1) and the carrier accumulated capacitance (C1) by τe1 = Rct1·C1. Figure 4c shows the calculated lifetime τe1 as a function of the applied bias. The interfacial recombination in the middle frequency region shows a millisecond time scale. Generally, the classical spectral features of the injected electrons/holes in solar cells can be explained with the transport/recombination model. The former is evidenced by a straight line associated with the carrier transport in semiconductor oxides. This is observed only when the transport resistance is smaller than the recombination resistance in impedance spectroscopy.31 In this case, the charge collection is efficient, which can be determined by ηcoll =1/(1 + (d/Ld)2), where d2 is the mean square displacement necessary for an electron to reach the MAPbI3/TiO2 interface from the point where it is photogenerated, and Ld is the diffusion length. The diffusion length can be realized by Ld = l(Rct/Rt)1/2, where l is the TiO2 film thickness. However, when the recombination model appears, the carriers would be consumed by recombination before they are collected at the current collector. As shown in Figure 3b,c, at lower bias (less than −0.55 V), the straight-line characteristic is visible in devices B and C. Unlike the usual observation of a linear Warburg impedance feature in the intermediate-frequency range corresponding to the trans-
from 1 MHz to 10 mHz for devices A−C under illumination at a bias of −0.8, −0.55, and −0.3 V, respectively. The frequency analysis shows three separated semicircles in the Nyquist diagram and three peaks in the Bode plot for devices B and C under high bias (−0.8 V, Figure 3a), while only two such characteristics are observed for device A. Upon the lower bias, a linear impedance feature appears as shown in the inset of Figure 3b,c. We can conclude that the peak in the high frequency regime (104 to 106 Hz) is related to the electron transfer process on the carbon electrode surface.29 However, the interpretation for the response in intermediate (104 to 10 Hz) and low frequencies (
