High efficiency inverted planar perovskite solar cells with solution ...

Research Article www.acsami.org

High Efficiency Inverted Planar Perovskite Solar Cells with SolutionProcessed NiOx Hole Contact Xuewen Yin,† Zhibo Yao,† Qiang Luo,† Xuezeng Dai,† Yu Zhou,† Ye Zhang,† Yangying Zhou,† Songping Luo,† Jianbao Li,‡,† Ning Wang,*,‡ and Hong Lin*,† †

State Key Laboratory of New Ceramics & Fine Processing, School of Materials Science and Engineering, Tsinghua University, Beijing 100084, P. R. China ‡ State Key Laboratory of Marine Resource Utilization in South China Sea, Materials and Chemical Engineering Institute, Hainan University, Haikou 570228, China S Supporting Information *

ABSTRACT: NiOx is a promising hole-transporting material for perovskite solar cells due to its high hole mobility, good stability, and easy processability. In this work, we employed a simple solution-processed NiOx film as the hole-transporting layer in perovskite solar cells. When the thickness of the perovskite layer increased from 270 to 380 nm, the light absorption and photogenerated carrier density were enhanced and the transporting distance of electron and hole would also increase at the same time, resulting in a large charge transfer resistance and a long hole-extracted process in the device, characterized by the UV−vis, photoluminescence, and electrochemical impedance spectroscopy spectra. Combining both of these factors, an optimal thickness of 334.2 nm was prepared with the perovskite precursor concentration of 1.35 M. Moreover, the optimal device fabrication conditions were further achieved by optimizing the thickness of NiOx hole-transporting layer and PCBM electron selective layer. As a result, the best power conversion efficiency of 15.71% was obtained with a Jsc of 20.51 mA·cm−2, a Voc of 988 mV, and a FF of 77.51% with almost no hysteresis. A stable efficiency of 15.10% was caught at the maximum power point. This work provides a promising route to achieve higher efficiency perovskite solar cells based on NiO or other inorganic hole-transporting materials. KEYWORDS: undoped nickel oxide, hole-transporting layer, inverted planar structure, perovskite solar cells, high efficiency

■

INTRODUCTION Methylammonium lead halide perovskites have been widely investigated as light absorption materials for efficient solutionprocessed solar cells due to their superior optoelectronic properties, such as suitable bandgap, ambipolar transport property, small exciton binding energy, broad range for light absorption with high extinction coefficient, long charge-carrier diffusion length, and lifetime.1−4 Nowadays, the certified efficiency of perovskite solar cells (PSCs) has already reached up to 22%, which proved to be a viable candidate for silicon solar cells and organic photovoltaics owing to its lower cost and simpler fabrication process.5−7 In the conventional normal n-i-p type PSCs, researchers have found that a high power conversion efficiency largely depends on the effectiveness of the hole-transporting material (HTM).8,9 Organic semiconducting HTMs such as 2,2′,7,7′© XXXX American Chemical Society

tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene (spiro-OMeTAD),10 poly[bis(4-phenyl)(2,4,6trimethylphenyl)amine] (PTAA),11,12 or poly(3-hexylthiophene-2,5-diyl) (P3HT)13 could be very expensive, which could largely hamper the development of perovskite photovoltaic technology in industry. Meanwhile, doping hygroscopic lithium salts or corrosive pyridine additives into HTMs may lead to instability and inferior performance of PSCs as the dopants are highly sensitive to moisture.14−16 There has been a growing interest in inverted planar device architectures typically employing a MAPbI3−PCBM ([6,6]phenyl-C61-butyric acid methyl ester) bilayer junction because Received: October 20, 2016 Accepted: December 28, 2016 Published: December 28, 2016 A

DOI: 10.1021/acsami.6b13372 ACS Appl. Mater. Interfaces XXXX, XXX, XXX−XXX

Research Article

ACS Applied Materials & Interfaces

Figure 1. (a) The schematic of NiOx-based perovskite solar cell; (b) the energy band level of solar cell; (c) the SEM image of perovskite layer; (d) the cross-sectional SEM image of the device.

Figure 2. XRD patterns (a) and absorption spectra (b) of different concentration perovskite precursors.

films applied were deposited by an expensive method, which is not suitable for large scale fabrication. Therefore, developing a less costly solution-processed NiOx HTM for PSCs is highly desirable. Yang et al.30 adopted the sol−gel method to synthesize a 30−40 nm thick NiOx HTM layer and obtained a relatively low PCE of 9.11% due to low short-circuit current density (Jsc) and fill factor (FF), which can be attributed to the rough surface and relatively thin thickness of perovskite. Chen et al.34 prepared NiOx films by spray pyrolysis and achieved a relatively low PCE of 10.2% with no doping and interfacial modification due to low FF. As listed in Table S3, at present, the photovoltaic performance of PSCs using solution-processed NiOx as HTMs is still not satisfactory because of the low FF or Jsc. In this work, we focus on PSCs with an inverted structure of FTO/NiOx/MAPbI3/PCBM/Ag, in which NiOx films were prepared by a simple solution-processed method. A detailed investigation on the effect of the thickness of NiOx, MAPbI3, and PCBM on photovoltaic performance of PSCs was carried out, during which the interfacial charge transfer properties of

of the simple fabrication and relatively small hysteresis. The most popular p-type HTM used in inverted planar devices is usually poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS), which, however, is not suitable for long-term stability due to its high acidity and hygroscopicity.17 Compared to organic HTMs, p-type inorganic HTMs are usually chemically stable and cost-effective.18,19 There have been several reports on employing p-type inorganic materials as HTMs in PSCs, such as CuSCN,20−22 PbS quantum qots,23 CuI,24,25 (reduced) graphene oxide,14,26−28 Cu2O,29 and NiO.30,31 Among these materials, one of the most investigated inorganic candidates is nickel oxide. This benefits from its wide band gap and high conduction band edge, which is crucial for serving as electron blocking layer. To date, studies on PSCs using NiOx as HTM have made great progress. Seok et al.32 employed nanostructured NiOx by pulse laser deposition (PLD) as HTM in PSCs and achieved a very promising PCE of 17.3%. Seongrok Seo et al.33 prepared ultrathin and undoped NiO films by atomic layer deposition (ALD) and obtained a PCE of 16.4%. However, their NiOx B


Research Article


Figure 3. Voc (a), Jsc (b), FF (c), and PCE (d) distributions with different concentration perovskite precursors.

the perovskite MAPbI3 with an orthorhombic Pnma crystal structure, respectively. The peak intensity of perovskite behaved much more strongly with higher concentration of precursor due to the increase of the perovskite layer’s thickness, which can be seen in Figure S1; the thickness of MAPbI3 was 269.4, 334.2, 354.9, and 379.7 nm for MAPbI3 solution with concentration of 1.25 M, 1.35 M, 1.45 M, and 1.50 M, respectively. The light absorption from 400 to 750 nm also increased when the thickness of as-made perovskite films was increased, as shown in Figure 2b. To investigate the effect of perovskite thickness on the photovoltaic performance, Figure 3 shows photovoltaic performance distribution of PSCs based on different MAPbI3 thickness measured under AM 1.5 simulated sunlight (100 mW·cm−2). The curves were scanned from the open circuit voltage (Voc) to the short circuit current density (Jsc) at a scan rate of 0.076 V·s−1. As shown in Figure 3, PSCs fabricated with 269.4 nm thick MAPbI3 with concentration of 1.25 M displayed an average Jsc of 17.91 mA·cm−2, a Voc of 0.953 V, and a FF of 57.33%, resulting in a relatively low PCE of 9.83%. When increasing the thickness of MAPbI3 to 334.2 nm with concentration of 1.35 M, the average Jsc and FF of the device were increased to 20.23 mA·cm−2 and 66.09%, respectively, both of which were much higher than those of the devices with the other three concentrations of perovskite precursor. Moreover, the Voc was stabilized at 0.98 V, which was a little lower than the device with the perovskite concentration of 1.45 M. As a consequence, the highest PCE was achieved in the device with the perovskite precursor concentration of 1.35 M. The effect of perovskite thickness on the photovoltaic performance was characterized and analyzed by EIS and photoluminescence spectra, shown in the following section. IPCE spectra of PSCs with different MAPbI3 thickness were tested from 300 to 800 nm as shown in Figure 4. It is worth

NiOx/perovskite were analyzed by using electrochemical impedance spectroscopy (EIS) and photoluminescence measurements. By optimizing NiOx contact and the thickness of perovskite absorption layer and PCBM electron transporting layer, a promising PCE of 15.71% was obtained, which is comparable to the most state-of the-art high-performance PSCs using NiOx films as HTMs prepared by ALD33 or spray pyrolysis.34 This work provides a reasonable reference to fabricate high efficiency inverted PSCs based on solutionprocessed NiOx or other inorganic HTMs.

■

RESULTS AND DISCUSSION

The Effect of the Thickness of Perovskite Layer. The device configuration and energy level diagram of the fabricated PSCs with an inverted planar structure are presented in Figure 1a,b. The commercially available FTO glass substrate was first covered with a thin NiOx layer by spin-coating NiOx precursor solution in ethanol. The light harvester, MAPbI3, was then prepared by a chlorobenzene assisted fast-crystalline method, which was widely employed to achieve a continuous, full coverage, and flat film (Figure 1c). After annealing, PCBM was used as ETM by spin-coating on perovskite in chlorobenzene. A typical cross-sectional SEM image of the device (MAPbI3 precursor solution concentration: 1.35 M) is demonstrated in Figure 1d, showing that the thickness of NiOx, MAPbI3, PCBM, and silver was approximately 49.4, 334.2, 68.5, and 93.2 nm, respectively. The photovoltaic performance of the device was largely influenced by the thickness of the light-absorption layer, which was carefully tuned by varying the concentration of perovskite precursors in this work. As depicted in Figure 2a, the strong Bragg peaks at 14.26°, 20.03°, 23.61°, 24.62°, 28.53°, 31.91°, 35.14°, 40.66°, and 43.48° could be ascribed to (110), (200), (211), (220), (320), (310), (321), (400), and (411) planes of C


Research Article


HTM layer.30 A smaller value of τ2 indicates a faster holeextraction process from perovskite to NiOx. In this work, the device with perovskite precursor concentration of 1.25 M had a minimum hole-extraction lifetime τ2 (30.35 ns) because of the thinnest perovskite thickness. EIS measurements were also performed in the dark to investigate the internal charge transporting and recombination through the devices. Figure 5b shows the Nyquist plots of PSCs at a bias of 0.7 V in the dark; the inset was the simulative equivalent circuit model. Only one semicircle could be distinguished from the Nyquist plots, corresponding to the interfacial transfer between NiOx and perovskite, which delivers information about the charge transfer resistance (Rct). It is obvious that an efficient charge transfer process will lead to a small Rct, which gives a small semicircle as observed.36 By fitting the Nyquist plots using the simulative equivalent circuit model, the total series resistances Rs of the devices were 10.64 Ω, 10.03 Ω, 11.75 Ω, and 9.68 Ω for 1.25 M, 1.35 M, 1.45 M, and 1.50 M, respectively. The closeness in Rs values in those devices was due to the similar device structure. The obviously different fitted Rct values of 153.9 Ω, 3126 Ω, 4575 Ω, and 12752 Ω corresponded to the devices with different perovskite thicknesses of 1.25 M, 1.35 M, 1.45 M, and 1.50 M, respectively. The smaller Rct implicates a faster charge transportation at the interface of NiOx/MAPbI3. In general, the Rct only depends on the perovskite thickness. Here, Rct as a function of perovskite thickness was plotted in Figure S2, which clearly shows that the increase of Rct was almost linearly dependent on the film thickness from thickness of 270 to 350 nm. The reason for the misleading results of Figure 5b was that the perovskite film thickness increase from 1.25 to 1.35 M was greatly larger than that from 1.35 to 1.45 M. Furthermore, the difference of Rct varied from 350 to 380 nm was obviously larger that than at thinner thickness before the critical point (thickness about 350 nm). As perovskite thickness further increased, the slope of Rct about thickness took a break, which means that a slight change in thickness will lead to a huge increase in Rct. This trend was also detected in PL decay measurement. As the absorption, PL spectra, and EIS results indicated, increasing the perovskite concentration would lead to the increased thickness of the perovskite layer, thus enhancing the light absorption and photogenerated carriers for the device. On the other hand, when increasing the thickness of MAPbI3, the electron and hole generated in MAPbI3 would come across a long road to transfer, resulting in a large charge transfer

Figure 4. IPCE spectra with different concentration perovskite precursors.

noting that the IPCE was similar to those previously reported for inverted PSCs.35 The perovskite film showed the highest IPCE value of 82.78% at 550 nm, which was consistent with the highest transmittance of FTO at 550 nm, showing the maximum photon-to-electron conversion efficiency. The loss from 300 to 450 nm could be ascribed to the absorption of NiOx films. However, the increase from 650 to 800 nm could be ascribed to the enhancement of perovskite thickness, which prevented the transmittance of long wavelength light. The integrated current density derived from the IPCE spectra corresponded well to the measured value from I−V measurement under simulated sunlight. The steady-state photoluminescence spectroscopy of MAPbI3 interfaced with NiOx HTM was tested to assess the effect of thickness of MAPbI3 on the charge transfer of NiOx/ MAPbI3. As illustrated in Figure 5a, the perovskite MAPbI3 showed a luminescence peak at 772 nm, and the MAPbI3 film with thickness of 269.4 nm on NiOx had the lowest photoluminescence intensity among those investigated films. The photoluminescence intensity was enhanced with the increased concentration of MAPbI3, which should be attributed to the increased thickness of absorption layer MAPbI3.We fitted the dynamic PL decay time curve to a biexponential decay function: y = A1 exp( − t /τ1) + A 2 exp( − t /τ2)

(1)

The fitting results are shown in Table S1. In this function, the small decay time constant τ1 reflects the diffusion of the photogenerated excitons into defects. The large time constant τ2 is associated with the radiative exciton lifetime of MAPbI3 on

Figure 5. Photoluminescence spectra (λex = 460 nm) (a) and EIS Nyquist plots obtained under dark condition at −0.7 V bias voltages (inset: shows equivalent circuit employed to Nyquist fitting) (b) of perovskite films with different perovskite concentration precursors. D


Research Article


Figure 6. (a) XRD patterns of nickel oxide on slide glass (blue) and standard cubic Fm3m nickel oxide (red); (b) transmittance spectra of different nickel oxide layers; (c) SEM image of 49 nm thick nickel oxide (inset: EDS test area); (d) EDS spectra of 49 nm thick nickel oxide.

Figure 7. AFM images of NiOx films fabricated by spin-coating different times on FTO: bare FTO (a, d), 2 times (b, e), and 4 times (c, f). (Inset: Ra, surface roughness; RMS, root-mean-square roughness).

by the thickness of the hole-transporting layer. In this study, we prepared the HTM layer by spin-coating method. The film thickness could be controllably tuned by varying the concentration of precursor or the spin-coating times, which could induce the change of photovoltaic properties. As shown in Figure S3, the thicknesses of NiOx films with different spin

resistance and a long hole- extraction process in the device. Both factors combined to result in the best photovoltaic performance device with 334.2 nm thick perovskite layer prepared with perovskite precursor concentration of 1.35 M. The Effect of the Thickness of NiOx Layer. The photovoltaic performance of the device is also largely affected E


Research Article

ACS Applied Materials & Interfaces coating times were 40, 49, 59, and 71 nm for one, two, three, and four times, respectively. As displayed in the XRD spectra in Figure 6a, the Bragg peaks at 37.24°, 43.28°, and 62.86° could be assigned to the typical diffractions of (111), (002), and (220) planes of the NiOx with a cubic Fm3m crystal structure (PDF: 47-1049), respectively. The optical properties of the NiOx films on the different thick samples were determined with UV−vis absorption spectra, as shown in Figure 6b. It is clearly seen that NiOx film showed high transmittance of over 70% from 400 to 800 nm, only a ∼3% decrease comparing to bare FTO glass, which was highly desirable for high performance PSCs because it ensured the incoming photons arriving to the neighboring absorber layer. The morphology and microstructure of NiOx films were observed by SEM (Figure 6c), showing a continuous and fully-covered NiOx film formed by many small nanoparticles with size of several nanometers. The energy dispersive spectroscopy (EDS) spectrum of the NiOx deposited on top of FTO is shown in Figure 6d, where nickel and oxygen elements could be obviously observed. To quantitatively characterize the surface roughness of NiOx films with different thickness, AFM was employed under tapping mode. Figure 7a−c shows comparative two-dimensional AFM images of the NiOx films with different thickness on FTO. Figure 7d,e shows corresponding three-dimensional plots. The surface roughnesses (Ra) and its root-mean-square (RMS) for NiOx films are listed in Figure 7. RMS values were 34.73, 30.98, and 29.32 nm for NiOx films with thickness of 0, 59, and 71 nm, respectively. Compared with that on bare FTO, the Ra and RMS values decreased with the increase of spincoating times, indicating that the films turned to be flatter with the increase of spin-coating times, which could also be verified in SEM images of NiOx films with different thickness in Figure S4. We also tested the steady-state photoluminescence of MAPbI3 on the NiOx films with different thickness. As illustrated in Figure S5, the quenching effect was observed on interfaces of NiOx/perovskite with different NiOx film thickness. In detail, the glass/FTO/NiO x /CH 3 NH 3 PbI 3 samples exhibited 56.6%, 45.3%, 49.7%, and 57.4% of the PL intensity of glass/CH3NH3PbI3 for 40, 49, 59, and 71 nm thick NiOx films, respectively. To further confirm the charge transport processes, time-resolved PL was also conducted and the corresponding results are summarized in Table S2. The PL lifetime was fitted with a biexponential decay function containing a fast decay and a slow decay process explained as mentioned above. The sample with 49 nm thick NiOx (spincoating two times precursor) had a minimum hole-extraction lifetime τ2 (32.85 ns). As the thickness of the NiOx layer increased from 0 to 49 nm, both the PL intensity and τ2 decreased due to the enhancement of hole-extraction capability. When the thickness of NiOx layer increased from 49 to 71 nm, however, both the PL intensity and τ2 increased due to the reduction of charge transfer from CH3NH3PbI3 to NiOx, which can be seen in Table 1. Based on static and decay PL spectra, the sample with 49 nm thick NiOx film had the fastest holeextraction effect. The photovoltaic performances with different NiOx thickness-based PSCs are summarized in Figure 8 and Table 1. When increasing the thickness of NiOx from 40 to 49 nm, the Jsc, Voc, and FF of the device increased to 20.48 mA· cm−2, 0.978 V, and 69.83%, resulting in an improvement in PCE, appropriately due to the suitable HTL thickness and modified surface roughness. As listed in Table 1, further

Table 1. Device Performances of NiOx Films with Different Thickness time

thickness (nm)

Jsc(mA·cm −2 )

Voc(V)

FF (%)

PCE (%)

Rs(ohm·cm 2 )

1 2 3 4

40 49 59 71

17.33 20.48 18.35 18.05

0.968 0.978 0.968 0.897

63.42 69.83 65.57 64.61

10.64 13.99 11.64 10.46

4.06 3.31 4.08 4.44

Figure 8. I−V curves of the devices with different NiOx precursor spin-coating times.

increase in the HTL thickness would lead to the increase of series resistance Rs from 3.31 ohm·cm2 (the device with 49 nm thick NiOx) to 4.44 ohm·cm2 (the device with 71 nm thick NiOx), which could largely hamper Jsc and Voc. This was probably due to the increased recombination and a higher resultant series resistance when dissociated carriers travel a longer distance to reach the FTO electrode. The Effect of PCBM Concentration. We also investigated the effect of the thickness of PCBM layer on the photovoltaic performance of PSCs. As shown in Figure 9a, PSCs fabricated with 16 mg·mL−1 PCBM as ETM displayed a Jsc of 19.03 mA· cm−2, resulting in a relatively low PCE of 13.50%. When the PCBM concentration increased to 18 mg·mL−1, the device showed the best PCE of 14.71% with a Jsc of 19.67 mA·cm−2. Further increasing the PCBM concentration to 20 mg·mL−1, the Jsc decreased to 17.38 mA·cm−2, thereby obtaining a lower PCE of 11.77%. When PCBM concentration was relatively low (16 mg·mL−1), too thin ETL made leakage current occur due to the direct contact between perovskite layer and Ag back electrode. Furthermore, the direct contact between Ag and perovskite also led to the formation of the AgI compound, which would worsen the device performance.37,38 The thickness of the PCBM layer increased with the concentration of PCBM, thus avoiding contact between perovskite and Ag back electrode as well as the formation of AgI compound. However, relatively thick PCBM layer (20 mg·mL−1) also resulted in a decrease in Jsc and PCE. This was probably due to the increased recombination and a higher resultant series resistance when dissociated carriers travel a longer distance to reach the Ag electrode.39 In order to further investigate the effect of the series resistance Rs on the photovoltaic characteristics, the J−V characteristics were analyzed by using the following expression: J = Jph − J0 [exp(q(V + R sJ )/(AkBT )) − 1]

(2)

where J is the density of the current passing through the external circuit and Jph is the light induced current density, which is considered equal to Jsc. J0 is the dark saturate current F


Research Article


Figure 9. I−V curves (a) and plots of − dJ and (Jph − J)−1 (b) of the device with different concentration PCBM. dV

Figure 10. I−V curves (a), IPCE spectrum (b), steady-state photocurrent measurement (c), and I−V curves stored in air (d) of the best performance device.

density, q is the elementary charge, V is the applied voltage, Rs is the series resistance, A is the ideality factor (typically 1