Improved efficiency of perovskite solar cells based on Ni-doped ZnO nanorod arrays and Li salt-doped P 3 HT layer for charge collection

We demonstrate regular-type perovskite solar cells based on vertically grown Nidoped ZnO nanorod arrays and doped P3HT for solar energy harvesting. PCBM was introduced between Ni-doped ZnO nanorod arrays and the perovskite layer to improve electron extraction, while P3HT doped with bis(trifluoromethane)sulfonimide lithium salt (LiTFSI) was used as hole transporting layer. Three types of perovskite materials, including MAPbI3, (MA)x(FA)1-xPbI3, and (MA)y(GA)1-yPbI3, were used as light harvesting layers to exploit conversion efficiency of photovoltaic devices. The optimized perovskite solar cell with the configuration of ITO/Ni-doped ZnO nanorods/PCBM/(MA)y(GA)1-yPbI3/P3HT + LiTFSI/Au revealed an open-circuit voltage of 0.83 V, a short-circuit current density of 23.73 mA/cm, a fill factor of 70%, and a power conversion efficiency of 13.79%. © 2016 Optical Society of America OCIS codes: (160.4236) Nanomaterials; (220.4241) Nanostructure fabrication; (040.5350) Photovoltaic; (350.6050)


Introduction
Perovskite solar cells based on hybrid organic-inorganic alkylammonium lead halides as light absorbers have attracted much attention recently because of superb photovoltaic performance, solution process, and low-cost technology.Several intrinsic properties including a wide visible absorption range from 300 to 800 nm, high absorption coefficient, and long exciton diffusion length make perovskite solar cells suitable for harvesting sunlight.The first attempt to use perovskite as a light harvester was implemented in liquid-type dye-sensitized solar cells, leading to low power conversion efficiency (PCE) values of 3.1-3.8%[1].The PCE of the perovskite-based devices was significantly enhanced to 9.7% by adopting a solid-state sandwiched architecture with the configuration of FTO/electron transporting layer (ETL)/perovskite/hole transporting layer (HTL)/Au [2].The perovskite absorber was sandwiched between the ETL and HTL.A hole-transporting material [2,2',7,7'-tetrakis(N,Ndi-p-methoxyphenylamine)-9,9'-spirobifluorene] (spiro-MeOTAD) was utilized as the HTL, and titanium dioxide (TiO 2 ) film was used as the ETL.The PCE value was further improved to 15.4% by interdiffusion of spin-coated stacking layers of lead iodide (PbI 2 ) and methylammonium iodide (MAI) [3], i.e., PbI 2 and MAI layers were prepared by a two-step spin-coating procedure.An alternative method, namely vapor-assisted solution process, was reported as the following procedure: PbI 2 was spin-cast into thin film, followed by dipping in MAI solution to form the final methylammonium lead iodide (CH 3 NH 3 PbI 3 or MAPbI 3 ) perovskite layer [4].A high PCE of 15% was obtained with the configuration of FTO/TiO 2 /MAPbI 3 /spiro-MeOTAD/Au.Most recently, an inspiring PCE up to 20.2% has been reported with the configuration of FTO/blocking TiO 2 layer/mesoporous TiO 2 layer/perovskite/poly[bis(4-phenyl)(2,4,6-trimethyl-phenyl)amine] (PTAA)/Au [5], using formamidinium lead iodide mixed with methylammonium lead bromide (FAPbI 3 ) 1- x (MAPbBr 3 ) x as the active layer.
In conventional perovskite solar cells, an n-type metal oxide film is usually required as the ETL on the transparent conductive electrode, e.g., indium tin oxide (ITO) and fluorine-doped tin oxide (FTO).Besides, this ETL can prevent direct contact between the oxide electrode and perovskite layer.Compact and/or mesoporous TiO 2 are commonly used as the ETL for constructing perovskite solar cells with high efficiency [2,4,5].However, TiO 2 has a lower electron transport mobility of 20 cm 2 /Vs compared to other metal oxides, such as zinc oxide (ZnO) with electron mobility of 230 cm 2 /Vs by the Hall effect measurement [6,7].The low intrinsic electron mobility of TiO 2 may lead to unbalanced charge transport of perovskite solar cells [8].Moreover, TiO 2 usually requires a high sintering temperature up to 450 °C to obtain a high-quality crystallization that hinders the development of flexible devices on plastic substrates.In contrast, ZnO is also an n-type semiconductor; it has higher electron mobility and a wide direct band gap (E g ) of 3.37 eV at 300 K. ZnO has been reported to be suitable for potential use in inorganic light-emitting diodes and solar cells [9,10].The sintering temperature of ZnO is also lower ~300 °C.Hence, ZnO has been considered as a good candidate to replace TiO 2 for the fabrication of perovskite solar cells.Kelly et al. introduced multi-coated layers of ZnO nanoparticles using room-temperature solution processing techniques as the ETL in perovskite solar cells [11].The device with the configuration of ITO/ZnO/MAPbI 3 /spiro-MeOTAD/Ag showed a PCE of 14.4%.The shortcircuit current density (J SC ) and PCE of the device were both enhanced as a result of the optimized ZnO thickness of 25 nm.Wang et al. employed a low-temperature-processed ZnOtin dioxide (SnO 2 ) nanocomposite as the electron collection layer for efficient planar perovskite solar cells [12].The average PCE of 14.0% was achieved from the devices with the configuration of ITO/ZnO-SnO 2 /MAPbI 3 /spiro-MeOTAD/Ag, which was higher than the one using pure ZnO nanoparticles (PCE = 13.6%).The PCE of the optimized device was further improved to 15.2% by introducing a mesoporous alumina (Al 2 O 3 ) layer between the ZnO-SnO 2 nanocomposite and perovskite layer.Li et al. reported an aluminum-doped zinc oxide (AZO) film as a highly stable electron collection layer for perovskite solar cells [13].
An average PCE of 11.8% was achieved from the devices with the configuration of AZO/MAPbI 3 /spiro-MeOTAD/Au.This kind of cell simplified device structure (without using FTO or ITO as electrode) and exhibited good stability in ambient environment compared with those of devices based on intrinsic ZnO.
In addition to nanoparticle-type ZnO, nanorod-type ZnO has also been applied as the ETL to construct perovskite solar cells.Park et al. employed non-doped ZnO nanorods with length of 1 μm for the fabrication of perovskite solar cells [14].The optimized device with the configuration of FTO/ZnO nanorods/MAPbI 3 / spiro-MeOTAD/Au showed a PCE of 11.13%.Compared to the perfect spectral response of ZnO nanorods, a reference perovskite solar cell based on TiO 2 nanorods with the same length exhibited lower open-circuit voltage (V OC ) of 869 mV and PCE of 10.02%.Meng et al. prepared AZO shells over the surface of ZnO nanorods with length of 450 nm for the application in MAPbI 3 -based perovskite solar cells [15].The V OC and PCE of the best device with the configuration of FTO/ZnO nanorods/AZO shell/MAPbI 3 /spiro-MeOTAD/Au reached 0.9 V and 10.7%, respectively, which were higher than those of the ZnO nanorod-based device without AZO shell (V OC = 0.77 V, PCE = 8.6%).The band position and electron density in AZO are suggested to be the origin for this positive effect of interfacial modification between ZnO nanorods and the perovskite layer.Mahmood et al. employed 1 μm-height ZnO nanorods/TiO 2 core-shell heterostructure as the ETL for superior light-harvesting and hysteresis free mesoscopic perovskite solar cells [16].Two types of ZnO nanorods with different aspect-ratio (AR) were prepared for comparison.The device with the configuration of FTO/ZnO nanorods/TiO 2 shell/MAPbI 3 /spiro-MeOTAD/Au showed average PCE values of 11.25% and 12.57% for low and high AR ZnO nanorods/TiO 2 shell-based perovskite solar cells.Superior light harvesting capability, larger surface area, prolonged charge-transport pathways, and lower recombination rate are responsible for better photovoltaic properties of the high AR ZnO nanorods/TiO 2 shell-based device.Apart from vertically grown ZnO nanorods, horizontally-arranged ZnO nanorods have also been applied in perovskite solar cells.Mahmood et al. demonstrated a double-layered ZnO film composed of nanosheets at the bottom and decorated with horizontal nanorods [17].This approach yields a maximum PCE of 10.35% from the device with the configuration of FTO/compact ZnO layer/double-layered ZnO film + MAPbI 3 /spiro-MeOTAD/Ag, which they attribute to the morphology and faster electron transport of the double-layered oxide film.
Spiro-MeOTAD is the most widely used HTL in conventional perovskite solar cells due to its hole transporting ability, excellent optical transparency in the visible range, and good thermal stability [2,12,13,[15][16][17].However, aside from its hole transporting feature, the complex synthesis, high cost, and low charge carrier mobility of spiro-MeOTAD limit its potential for commercialization [18].Besides, high doses of spiro-MeOTAD accompanying with small amounts of bis(trifluoromethane)sulfonimide lithium salt (Li-TFSI) and 4-tertbutylpyridine (TBP) are essentially required to achieve good hole transporting properties for efficient perovskite solar cells.Relative to small-molecule spiro-MeOTAD, poly(3hexylthiophene-2,5-diyl) (P3HT) is a regioregular semiconducting polymer with high hole mobility up to 0.1 cm 2 /Vs and good film-forming property [19].The price of P3HT is also lower than that of commercial spiro-MeOTAD, which is beneficial for commercialization and mass production.To improve the hole extraction ability, the Li-TFSI salt can be added into P3HT as HTL to increase the carrier density in the HTL and reduce recombination losses in perovskite solar cells [20,21].
As mentioned above, non-doped ZnO nanorods, ZnO nanorods/AZO core-shell, ZnO nanorods/TiO 2 core-shell, multi-layered ZnO nanoparticles, and ZnO-SnO 2 nanocomposites have been reported in the area of perovskite solar cells.However, nickel (Ni)-doped ZnO nanorods have not been applied as ETL for the fabrication of perovskite solar cells so far.The unique chemical stability of Ni on zinc sites recognizes it as one of the most efficient doping elements to improve and tune the optical and electrical properties of ZnO [22].Chen et al. reported large-scale Ni-doped ZnO nanowire arrays with improved electrical conductivity compared to the undoped ones [23].Cheng et al. reported Ni-doped ZnO nanorods by the hydrothermal synthesis at a high temperature up to 140 °C to be applied build components in spintronic devices [24].Zhang et al. demonstrated Ni-doped ZnO nanorods with higher photocatalytic activity than the undoped ZnO [25].Besides, Ni-doped ZnO films were utilized as gas sensors for the detection of ammonia, and they showed higher sensitivity compared with pure ZnO or nickel oxide [26].
Herein, we report on the low-temperature hydrothermal synthesis of non-doped and Nidoped ZnO nanorods for the fabrication of perovskite solar cells.The lengths of nanorods are reduced to 300 nm to increase transmittance of ZnO in the visible region, which is beneficial for incident photons to enter devices and to be absorbed by the active layer.Novel perovskite solar cells with the configuration of ITO/Ni-doped ZnO nanorods/ [6,6]-phenyl-C 61 -butyric acid methyl ester (PCBM)/perovskite/P3HT + Li-TFSI/Au were fabricated and evaluated.In addition to MAPbI 3 , two different kinds of perovskite materials, including methylammonium/formamidinium lead iodide (MA) x (FA) 1-x PbI 3 and methylammonium/guanidinium lead iodide (MA) y (GA) 1-y PbI 3 were utilized as the light harvesters.Our results revealed a maximum PCE of 13.79% with the device using Ni-doped ZnO nanorods as ETL and (MA) y (GA) 1-y PbI 3 as the active layer.Miscellaneous characterization techniques, including scanning electron microscopy (SEM), atomic force microscopy (AFM), X-ray diffraction (XRD) analysis, and optical measurements were also performed and demonstrated in this work.

Materials
All chemicals were purchased from Alfa Aesar, Sigma-Aldrich or Acros and used without further purification.All solutions were filtered with 0.45 μm polytetrafluoroethylene (PTFE) filters before use.Methylammonium iodide (MAI) was synthesized by reacting methylamine and hydroiodic acid together for 4 hr, and the crude product was collected and purified by reprecipitation in diethyl ether for several times.The overall yield of MAI is 65%.Formamidinium Iodide (FAI) was synthesized by reacting formamidinium acetate and hydroiodic acid together for 4 hr, and the crude product was collected and purified by reprecipitation in diethyl ether for several times [27].Guanidinium Iodide (GAI) was purchased from Sigma-Aldrich.

Preparation of ZnO nanorod arrays
The non-doped and Ni-doped ZnO nanorod arrays were prepared by the hydrothermal method on the ITO substrates.The ITO/glass substrates (7 Ω/sq) were cleaned with detergent, acetone, and isopropyl alcohol (IPA) under ultrasonication, followed by nitrogen purge and drying in an oven.The preparation of ZnO nanorod arrays on the ITO substrates is shown in Fig. 1.The detailed fabrication process is described as follows.A solution of zinc acetate dihydrate (0.5487 g, 2.5 mmol) in 10 mL of IPA was stirred vigorously at 60 °C for 5 min.2-(Dimethylamino)ethanol (0.2228 g, 2.5 mmol) was slowly added to the above solution and stirred at the same temperature for an additional 2 hr to form a homogeneous precursor solution.This precursor solution was then spin-cast into thin film at 3,000 rpm for 30 s and calcinated at 200 °C in air for 30 min to form ZnO seed layers with thickness of 50 nm.
The next step is to prepare non-doped and Ni-doped ZnO nanorods on the seed layers.A solution consisting of zinc sulfate heptahydrate (0.287 g, 1 mmol) and ammonium chloride (2.14 g, 40 mmol) in 150 mL of de-ionized water was prepared.For Ni-doping, two different amounts of nickel(II) acetate hydrate (0.0025 g or 0.005 g) were added into the above zinc sulfate solution.The Ni doped proportion is labelled as 1% or 2%, respectively, equal to Ni / (Ni + Zn) (mol%).2M NaOH aqueous solution was then slowly added into three separate solutions to adjust their pH values to 10.7 to form the bath solution.The ZnO seed layers on the ITO substrates were immersed in the above bath solution (ZnO seed layers downward) and placed in a preheated oven (65 °C) for 45 min.The substrates were then washed with deionized water and IPA for 10 minutes to remove the residues, followed by calcinations at 300 °C in air for 30 min to form non-doped and Ni-doped ZnO nanorod arrays.The heights of those ZnO nanorods were estimated to be 300 nm by SEM cross-sectional observation.

Fabrication of perovskite solar cells
The fabrication process of perovskite solar cells based on ZnO nanorod arrays is illustrated in Fig. 2. The devices with the configuration of ITO/non-doped or Ni-doped ZnO nanorods/PCBM/perovskite/P3HT + Li-TFSI/Au were fabricated.The selection of the perovskite includes MAPbI 3 , (MA) x (FA) 1-x PbI 3 , and (MA) y (GA) 1-y PbI 3 , which were prepared by the two-step coating process.A PCBM solution in chlorobenzene (10 mg/mL) was spincoated onto the ZnO nanorod arrays at 3,000 rpm for 30 s and annealed at 100 °C for 5 min.First, a PbI 2 solution in N,N-dimethylformamide (450 mg/mL) was spin-coated onto ZnO nanorod arrays at 5,500 rpm for 30 s and annealed at 100 °C in a vacuum oven for 30 min.Second, three different solutions, namely MAI (40 mg/mL in IPA), MAI + FAI (12:1 in molar ratio), and MAI + GAI (6:1 in molar ratio), were loaded on PbI 2 layer for 60 s and then spincoated at 5,500 rpm for 30 s separately, followed by annealing at 100 °C in a vacuum oven for 30 min.A P3HT solution in chlorobenzene (20 mg/mL) was added to a lithium salt solution (0.54 mg of Li-TFSI in 6.8 μL of acetonitrile) and spin-coated on the perovskite layer at 1,000 rpm for 30 s, followed by annealing in a vacuum oven at 95 °C for 30 min.Finally, 80 nm-thick Au electrodes were deposited onto P3HT layer by thermal evaporation at base pressure of 10 −6 torr.The active area of devices is 1 mm 2 .The thickness of each layer is estimated from the cross-section SEM image (see the Appendix, Fig. 10).It is also found that PCBM is filled in the bottom of the ZnO nanorod arrays, as verified the SEM observation (see the Appendix, Fig. 11).

Characterization methods
The current density-voltage (J-V) characteristics of photovoltaic devices were taken using a Keithley 2400 source measurement unit under a simulated AM 1.5G exposure with a 1000 W Xenon Short Arc Lamp (USHIO UXL-10S).The simulated solar source was calibrated with a commercial silicon reference cell.The UV-vis absorption spectra of films were acquired at room temperature by a Princeton Instruments Acton 2150 spectrophotometer with a 150 W Xenon lamp (USHIO UXL-150S).Steady-state PL spectra were measured using a fluorescent spectrophotometer (Princeton Instruments Acton 2150) equipped with a 30 mW KIMMON KOHA He-Cd laser (325/442 nm) as excitation sources.For UV-Vis transmission and absorption measurements using a Xe lamp as the light source, the optical spot size is 3 × 3 mm 2 and the direction of the unpolarized incident light is perpendicular (θ = 90°) to the sample surface.For PL measurements using a He-Cd laser as the light source, the beam diameter is 0.94 mm and the direction of the linear polarized incident light is 30° to the normal direction of the sample surface.The surface morphology of films was measured by a Bruker Innova atomic force microscopy (AFM) instrument.Top-view and cross-section scanning electron microscopy (SEM) observations were performed with a JEOL JSM-6700F SEM instrument.The transmission electron spectroscopy (TEM) and high resolution TEM (HRTEM) were performed with a JEOL JEM-2010 TEM instrument.X-ray diffraction (XRD) patterns were obtained from a Rigaku D/MAX 2500 X-ray diffractometer.The Hall measurement was carried out with a BioRad HL5500 Hall Effect Measurement System.The external quantum efficiency (EQE) measurement was conducted using a PV Measurements QEX10 instrument.

Characterization of ZnO nanorod arrays
The side-view and top-view SEM micrographs of non-doped and Ni-doped ZnO nanorod arrays are shown in Fig. 3.It is seen that growth direction of ZnO nanorods is mainly vertical to the ITO substrate.The lengths of ZnO nanorods are estimated to be 280-300 nm from sideview SEM, and the diameters of ZnO nanorods are in the range of 35-60 nm from top-view SEM.Besides, there are still some spaces for perovskite materials to fill in.The AFM topographic and 3D images of ZnO nanorod arrays are shown in Fig. 4. It is clearly seen that the morphology of ZnO nanorods are consistent with SEM observation.The surface roughness (R a ) values of non-doped, 1% Ni-doped, and 2% Ni-doped ZnO nanorod arrays are calculated to be 26.6 nm, 18.4 nm, and 19.2 nm, respectively.The SEM and AFM observations of our non-doped ZnO nanorods arrays are similar to the previous reports [14,28].In addition, we notice that Ni-doped ZnO nanorods possess more aligned morphologies and smaller R a value compared with the non-doped ones, implying better transporting property for carriers.The TEM and HRTEM images of Ni-doped ZnO nanorods are shown in the supplementary information (see Fig. 12).The diameter and shape of a single ZnO nanorod are clearly observed.The UV-vis transmission and absorption spectra of non-doped and Ni-doped ZnO nanorods and ITO substrate are shown in Fig. 5(a) and 5(b), respectively.The transmittance of ZnO nanorod arrays is sufficiently high up to 90% in the visible range.Moreover, ZnO nanorods on the ITO reveal even higher transmittance than the bare ITO substrate.The reason to this phenomenon can be explained by the anti-reflection effect of ZnO, which has been reported in the previous literature [29,30].From Fig. 5(b) the absorption bands of non-doped and Ni-doped ZnO nanorods are located in the range of 250-380 nm that are in accordance with the previous report [31].Besides, it is noted that Ni-doped ZnO nanorods possess slightly higher absorption around 360 nm, which is attributed to NiO formation in ZnO nanorods, i.e., Ni-doping process [32].The absorption edges of both non-doped and Ni-doped ZnO nanorods are found at 380 nm, corresponding to their optical bandgaps of 3.26 eV.We conclude that the optical bandgap of ZnO nanorods was not affected by Ni-doping.Figure 6 shows the XRD patterns of the prepared ZnO nanorod arrays.The diffraction peaks of the ZnO nanorods are well matched with the ZnO standard (JCPDS 36-1451, see the Appendix, Fig. 13) [33].The highly oriented nature of our ZnO nanorods is connected with the intense diffraction peak at 2θ = 34.4°along (002) reflection [33,34].The sharp and strong (002) reflection also indicates preferred crystallization of ZnO nanorods along the c-axis direction, i.e., vertical to the substrate surface.Moreover, the incorporation of Ni atoms can be deduced from two increased peaks indexed (100) and (101) compared to the strong (002) peak, which is due to the formation of stress-induced ion size difference between Zn and Ni and the segregation of Ni in grain boundaries [35].The incorporation of Ni ions into the Zn site of the ZnO matrix is also verified by the energy dispersive X-ray (EDX) spectra (see the Appendix, Fig. 14).The presence of Zn, O, and Ni clearly indicates successful preparation of Ni-doped ZnO samples in this study.To further compare the conductivity of non-doped and Ni-doped ZnO nanorod arrays, simple sandwiched devices with the configuration of ITO/ZnO/Au were fabricated and their current-voltage characteristics were measured (see the Appendix, Fig. 15).It is observed that the device based on the 2% Ni-doped ZnO nanorod arrays show the highest electrical conductivity and lowest resistance among three samples, indicative of higher electron transporting properties for photovoltaic devices.The Hall measurement of non-doped and Nidoped ZnO nanorod arrays was also performed (see the Appendix, Fig. 16).It reveals that higher mobility of ZnO is obtained as the doping concentration of Ni is increased.

Characterization of perovskite films on ZnO nanorod arrays
The top-view SEM images of perovskite MAPbI 3 , (MA) x (FA) 1-x PbI 3 , and (MA) y (GA) 1-y PbI 3 films are displayed in Fig. 7(a)-7(c), respectively, indicative of homogeneous distribution of perovskite grains with sizes of 300-700 nm on non-doped ZnO nanorods.There are still small pinholes to be observed for MAPbI 3 and (MA) x (FA) 1-x PbI 3 on ZnO nanorods from SEM micrographs, which might affect the fill factor (FF) and performance of photovoltaic devices.Compared to the similar report of the perovskite film on 1000 nm-height ZnO nanorods, pinholes in this study were significantly reduced and the coverage ratio of the perovskite film was improved [14].Besides, those pinholes can be filled by depositing the next P3HT layer, preventing direct contact between top gold electrode and ZnO nanorods.It is noteworthy that the (MA) y (GA) 1-y PbI 3 represents a compact and pinhole-free film on ZnO nanorods among three perovskite films.The surface morphologies of perovskite films are observed from topographic AFM image in Fig. 7(d)-7(f).The grain distribution of these perovskite crystals from AFM is consistent with the SEM observation.The R a values of MAPbI 3 , (MA) x (FA) 1- x PbI 3 , and (MA) y (GA) 1-y PbI 3 films on ZnO nanorods were measured to be 14.1, 15.1, and 18.7 nm, respectively.The steady-state PL spectra of the perovskite MAPbI 3 on the bare glass, non-doped, and Ni-doped ZnO nanorods are shown in Fig. 8.The max PL emission of the perovskite is located at 770 nm under excitation of a He-Cd laser source, which is in accordance with the previous literature [13].The significant PL quenching of the perovskite is found when depositing on non-doped ZnO nanorods, implying charge extraction across the interface between the perovskite and ZnO [36].By depositing perovskite on Ni-doped ZnO nanorods, the PL emission of the perovskite is further prohibited, indicative of more effective carrier extraction and reduced recombination, which will help to improve J SC and performance of perovskite solar cells.It is known that the grain size of the perovskite could make an important attribution to PL spectra and thus the device performance.Hence, the AFM investigation of MAPbI 3 on bare glass, non-doped, and 2% Ni-doped ZnO nanorods was carried out.(see the Appendix, Fig. 17).It is seen that that the grain sizes of three perovskite samples are similar.Hence, the size effect for PL quenching experiments in Fig. 8 as well as device performance can be excluded.

Device fabrication and evaluation
The J-V characteristics of the solar cells using non-doped and Ni-doped ZnO nanorods as the ETL are shown in Fig. 9(a).The optimized perovskite solar cell based on non-doped ZnO nanorods demonstrated a J SC of 21.33 mA/cm 2 , a V OC of 0.81 V, a FF of 0.6 and a PCE of 10.37% under the AM 1.5G illumination.A similar perovskite solar cell based on non-doped ZnO nanorods with length of 1 μm was reported by Park et al. [14], revealing J SC , V OC , FF, and PCE values of 20.08 mA/cm 2 , 0.99 V, 0.56, and 11.13%, respectively.It is seen that the performance of our devices is comparable with the previous result in the literature.By using 1% Ni-doped ZnO nanorods as the ETL, the device parameters J SC , FF, and PCE were improved to 22.46 mA/cm 2 , 0.67, and 12.19%, respectively, while the V OC remained unchanged.By using 2% Ni-doped ZnO nanorods as the ETL, the optimized perovskite solar cell achieved a J SC of 23.18 mA/cm 2 , a V OC of 0.81 V, a FF of 0.68 and a PCE of 12.77% under the same illumination.The device parameters of perovskite solar cells based on nondoped and Ni-doped ZnO nanorods are summarized in Table 1.We conclude that the V OC of devices was not affected by Ni-doping process.The significant increase in J SC and FF can be explained by better extraction of electrons and reduced series resistance (R s ) for the devices based on Ni-doped ZnO nanorods.As mentioned previously, more effective carrier extraction between the perovskite and Ni-doped ZnO nanorods has been verified by PL quenching experiments.The R s values can be estimated from the slope of the J-V characteristics at the V OC point that is commonly employed in the previous literature [37,38].The R s value of the device using non-doped ZnO nanorods as the ETL was calculated to be 5.91 ohms-cm 2 , and it was reduced to 4.69 and 3.47 ohms-cm 2 for the devices using 1% and 2% Ni-doped ZnO nanorods as the ETL.Previous works have been focused on the modification of ZnO nanorods by covering with other materials such as TiO 2 and AZO to form core-shell heterojunctions [15,16].In this research, we demonstrate Ni-doped ZnO nanorods as ETL that has not been reported in the development of perovskite solar cells so far.
To further improve the efficiency of Ni-doped ZnO nanorods-based perovskite solar cells, two different kinds of perovskite materials, including (MA) x (FA) 1-x PbI 3 and (MA) y (GA) 1- y PbI 3 , were utilized as the active layers.It is reported that high-quality mixed-organic-cation perovskite (MA) x (FA) 1-x PbI 3 exhibits photovoltaic performance superior to that of MAPbI 3 owing to the broader light-harvesting spectrum and the collection of longer-wavelength photons [27].Another perovskite material (MA) y (GA) 1-y PbI 3 also possesses better device performance compared with MAPbI 3 , which is due to significantly enhanced carrier lifetimes and V OC [39].In this study, a mixture of MAI and FAI (or GAI) with molar ratio of 12:1 (or 6:1) were prepared, and perovskite layers were obtained by the same two-step coating process.Besides, the 2% Ni-doped ZnO nanorod arrays were selected as the ETL for the fabrication of perovskite solar cells, since they demonstrated better photovoltaic properties compared with non-doped or 1% Ni-doped ones.The J-V curves of perovskite solar cells using MAPbI 3 , (MA) x (FA) 1-x PbI 3 , and (MA) y (GA) 1-y PbI 3 as light absorbers are depicted in Fig. 9(b).The optimized device based on (MA) x (FA) 1-x PbI 3 produced a J SC of 23.88 mA/cm 2 , a V OC of 0.82 V, and a FF of 0.64, yielding a PCE of 12.53%.By replacing with (MA) y (GA) 1- y PbI 3 as the active layer, the optimized device showed even higher photovoltaic performance, including a J SC of 23.73 mA/cm 2 , a V OC of 0.83 V, a FF of 0.7, and a PCE of 13.79%.The device parameters of perovskite solar cells using different perovskite layers on the 2% Nidoped ZnO nanorods are summarized in Table 2.It is concluded that the perovskite solar cell composed of (MA) y (GA) 1-y PbI 3 and the 2% Ni-doped ZnO nanorod arrays reveals the highest efficiency in this research.The EQE spectra of the devices based on MAPbI 3 , (MA) x (FA) 1- x PbI 3 , and (MA) y (GA) 1-y PbI 3 on the 2% Ni-doped ZnO nanorod arrays are shown in Fig. 9(c), which is similar to the previous work using ZnO film as ETL [37].The (MA) y (GA) 1-y PbI 3 device revealed the highest EQE values among three devices, with the EQE maximum approaching 80% around 500 nm.The EQE of the (MA) x (FA) 1-x PbI 3 device was higher than that of the MAPbI 3 one in the range of 600-800 nm due to better conversion of longwavelength photons [27].The EQE trend of the three devices is in good accordance with their photovoltaic performance, as revealed in Fig. 9(b) and Table 2.   b The R s of the perovskite solar cells was calculated using (dJ/dV) −1 at V = V OC .

Fig. 1 .
Fig. 1.Schematic illustration of fabrication process of the ZnO nanorod arrays on the ITO substrate.

Fig. 2 .
Fig. 2. Fabrication procedure of perovskite solar cells based on the ZnO nanorod arrays.

Table 1 . Performance of perovskite solar cells based on non-doped and Ni-doped ZnO nanorods.
a Average PCE values were obtained from 18 devices.

Table 2 . Performance of perovskite solar cells using different perovskite layers on Ni- doped ZnO nanorods.
Average PCE values were obtained data from 18 devices. a