Simultaneous enhancement in stability and efficiency of low-temperature processed perovskite solar cells

Abstract

Mixed ion based perovskite solar cells (PSCs) have recently emerged as a promising photoactive material owing to their augmented electronic and light harvesting properties combined with stability enhancing characteristics. However, to date most of the high performing perovskite devices employ a high temperature (∼500° C) sintering process for depositing a conventional titanium oxide (TiO₂) based electron transport layer (ETL), which is a serious bottleneck towards roll-to-roll processing with flexible substrates, large scale manufacturability and also results in high energy consumption. The present work demonstrates simultaneous enhancement in efficiency and stability in the perovskite solar cell that is totally fabricated using low temperature methods with the synthesis process temperature not exceeding 150 °C at any stage. The perovskite devices, thus fabricated, exhibited high power conversion efficiency of ∼14.5% and device stability > 570 hours (normalized PCE to reach 80% of its original value), which is the first of this kind of accomplishment ever reported in entirely low temperature processed PSCs. It is noteworthy to mention that the presented devices utilize a ∼360 °C lower temperature than required for the conventional TiO₂ based PSCs to achieve similar enhancements in terms of stability and efficiency simultaneously. The high performing PSCs reported in this work incorporate mixed organic perovskite (MA_0.6FA_0.4PbI₃) as the light absorber and aluminium-doped zinc oxide (AZO) as the electron transport layer. Adding to the merits, the MA_0.6FA_0.4PbI₃/AZO devices exhibited a substantially low photocurrent hysteresis phenomenon. In order to examine the underlying causes of the efficiency and stability enhancements in AZO based devices, a low temperature processed MA_0.6FA_0.4PbI₃/ZnO device was also fabricated and comparatively studied. Investigations reveal that the improved dark carrier mobility and superior interfacial electronic properties at the perovskite/AZO interface are attributed to their enriched device performance. Slow perovskite decomposition rate/high device stability with AZO based perovskite devices was found to be associated with the more hydrophobic and acidic nature of the AZO surface and the related interfacial interactions with the adjacent perovskite layer.

---

1. Introduction

Mixed organic cation based perovskite solar cells have arisen as a promising replacement for their mono organic counterparts, owing to their augmented light-harnessing capability with red-shifted spectral response and relatively better stability.^1,2 The generic formula for these highly efficient mixed organic cation perovskite solar cells can be expressed as: ABX₃, where, A stands for a combination of methyl ammonium (MA) [CH₃NH₃⁺] and formamidinium (FA) [HN = CHNH₃⁺] organic cations, B denotes a divalent metal ion (usually Pb or Sn) and X symbolizes individual halogen element (I, Br or Cl) or a combination of them with a fixed molar ratio. Owing to the comparatively high ionic radius (1.9–2.2 Å) of the formamidinium cation than that of methyl ammonium ion (1.8 Å),¹ the formamidinium based perovskite material has a higher perovskite tolerance factor – which provides much better structural stability.³ However, the open circuit voltage (V_OC) is relatively low in such a pristine formamidinium perovskite device^4,5 resulting in a relatively lower device performance compared to that with methyl ammonium lead triiodide perovskite based PSCs.⁶ To optimize the trade-off between extended absorption and suppressed V_OC, the concept of mixed organic cation based perovskite has come into effect which tunes the perovskite bandgap by incorporating both the methylammonium and formamidinium organic cations in a single ABX₃ crystal structure.² Without any alteration in the inorganic constituent of perovskite structure, pure triiodide mixed organic MA_0.6FA_0.4PbI₃ (MA = methyl ammonium, FA = formamidinium) perovskite material has been reported² to be an excellent photo-harvesting material which renders an efficient perovskite solar cell (average PCE: 13.50%)² in conjunction with high temperature (∼500 °C) sintered titanium oxide (TiO₂) as electron transport layer (ETL). Aside from being a highly energy intensive process, the high temperature requirement with TiO₂ ETL is a rudimentary hindrance towards roll-to-roll processing of perovskite solar cells on flexible substrates like PET (polyethylene terephthalate) that necessitates sub-150 °C temperature processing^7,8 for proper functioning. In such scenario, low temperature processed zinc oxide (ZnO)^7,9,10 could be a promising alternative to TiO₂ ETL in a perovskite solar cell as it has nearly identical electron affinity (4.2 eV) as TiO₂ (ref. 9) along with nearly two times higher conductivity¹¹ and significantly larger (2000 times)¹² electron mobility. Besides, ZnO is endowed with overwhelming chemical and thermal stability,¹³ courtesy of its large energy band gap coupled with high exciton binding energy (60 meV). In addition to that, outstanding optical transparency of ZnO in the visible light spectrum¹⁴ also demonstrates it to be a worthy contender to TiO₂ ETL in normal structured perovskite solar cell. However, alkaline ZnO surface gives birth to deprotonation of methyl ammonium ion (CH₃NH₃⁺) when the perovskite film grown on ZnO is annealed at around 100 °C.¹⁵ Thus the perovskite crystal structure overlying the ZnO ETL undergoes thermal decomposition even at 100 °C temperature annealing. This thermal instability of perovskite film in conjunction with ZnO ETL is a major bottleneck towards practical application of perovskite solar cell involving continuous exposure to sunlight. In this regard, pre-heated (∼400 °C) ZnO film has been reported¹⁵ to provide better immunity against thermal instability of perovskite film, as it (pre-heated ZnO film) is devoid of surplus OH⁻ or CH₃COO⁻ radicals. Nevertheless, the pre-heat treatment of ZnO still cannot ensure identical thermal stability of perovskite film as obtained with TiO₂ ETL^15,16 and the process is not again compatible with flexible substrates for roll-to-roll process.

Of late, Zhao et al.¹⁶ have reported improved thermal-stability of mono organic cation based methyl ammonium lead triiodide (MAPbI₃) perovskite devices with Al doped ZnO (AZO) as ETL, achieving an average PCE of 11.80%. Moreover, they employed commercially available AZO patterned glass to fabricate the PSCs which hinders customisability/processability, as they are deposited using sophisticated sputter deposition techniques. In addition, the PSC devices reported in their work exhibit extremely high hysteresis phenomena due to unbalanced charge accumulation at the interfaces between perovskite and AZO interface, and it is well known that hysteresis is a serious issue in terms of evaluating the performance of PSCs^17,18 which leads to over or under-estimation of the device performance under real operating condition, thus raising a question regarding device reliability. Besides, the thick (∼1 μm thickness), sputter-deposited AZO ETL film utilized in the study¹⁶ has a high surface roughness (RMS surface roughness: 12.8 nm) that results in a considerably lower V_OC value (average V_OC: 910 mV) in comparison with previous studies with ZnO/MAPbI₃ perovskite devices.^15,19 Moreover, there are no studies till date that have probed into the electronic properties at perovskite/AZO interface and their relationship with device stability (i.e. electronic analysis for both fresh and aged devices).

In this work, for the first time ever, we have reported simultaneous enhancement in efficiency and stability with significant reduction in hysteresis in perovskite solar cells fabricated using low temperature methods. The demonstrated significance is achieved in mixed organic cation based planar perovskite (MA_0.6FA_0.4PbI₃) devices in conjunction with low temperature (<150 °C) solution processed AZO thin films as ETL using sol–gel technique (Scheme 1). The perovskite devices thus fabricated exhibited high power conversion efficiency (PCE) of ∼14.5% and device stability > 570 hours (normalized PCE to reach 80% of its original value). Conventional ZnO based PSCs were also fabricated for comparative study and to investigate the underlying origins of performance and stability improvements in MA_0.6FA_0.4PbI₃/AZO based devices. Besides, the present work is also the first to demonstrate the in-depth analysis into the charge transfer characteristics at the interface between mixed organic cation perovskite layer and adjacent charge selective interlayer (AZO). In addition to electronic studies, an elaborate surface morphology and surface topography measurements are carried out to probe into the superior performance/stability of the AZO based devices and the enhancements are partially associated with the recombination and current leakage processes. The suppressed hysteresis phenomenon in MA_0.6FA_0.4PbI₃/AZO based devices was elucidated in terms of electrode polarization process and found to exhibit significant reduction in interfacial capacitance. A month-long, systematic degradation study was also performed to discern the superior lifetime of MA_0.6FA_0.4PbI₃/AZO based devices corroborated by the extensive analysis of electronic parameters like interfacial contact resistance, recombination resistance, flat-band potential and low frequency regime interfacial capacitance (electrode polarization) of both the fresh and aged devices.

Scheme 1 The process of fabricating the perovskite solar cell as compared with the conventional methods is shown. (Top-left) Shows the conventional high temperature process for fabricating the ETL layer. (Bottom-left) Shows the proposed low temperature process following the route B and C. The perovskite solar cells thus fabricated using the low temperature (T < 150 °C) approach (B–C) resulted in 20% higher PCE and superior stability than that of those reported with conventional methods (T ∼ 500 °C) using the similar perovskite material system (MA_0.6FA_0.4PbI₃).

2. Experimental detail

2.1 Device fabrication

ITO/glass substrates were cleaned sequentially with Hellmanex III, DI water, acetone and isopropanol with duration of 10 minutes for each. For ZnO ETL, 0.48 M sol–gel ZnO precursor solution was prepared by dissolving zinc acetate dihydrate (Zn(CH₃COO)₂·2H₂O, Sigma-Aldrich, >99.0%) in 2-methoxyethanol (CH₃OCH₂CH₂OH, Sigma-Aldrich, 99.8%, anhydrous) with an additive ethanolamine (NH₂CH₂CH₂OH, Sigma-Aldrich, >99.5%) by stirring for 24 hours. For AZO ETL, a 0.05 M aluminum nitrate nonahydrate (Al(NO₃)₃·9H₂O, Sigma-Aldrich) solution in 2-methoxyethanol was prepared and stirred for 1 hour at 80 °C. A 15 wt% AZO precursor solution was prepared by mixing the as prepared aluminum nitrate nonahydrate solution with sol–gel ZnO precursor solution in required weight ratio. Both the sol–gel ZnO and AZO precursor solutions were spin coated on ITO/glass substrate at 4000 rpm for 60 s and then annealed on a hotplate at 140 °C for half an hour. 1 M PbI₂ solution in DMF (N,N-dimethylformamide, Sigma Aldrich, anhydrous) with 120 μL 4-TBP (4-tert-butylpyridine, Sigma Aldrich) was prepared at a temperature of 70 °C. The PbI₂ solution was spin cast on ZnO/AZO coated substrate at 3000 rpm for 30 s. Before spin coating, both the substrates and the solution were kept at 70 °C. Then the substrates were annealed at 100 °C for one hour and then dipped in a 2-propanol solution having a total iodide concentration of 0.0628 M with the mole fraction of CH₃NH₃I (MAI) and HN=CHNH₃I (FAI) to be 0.6 and 0.4 respectively. The devices were annealed for 10 minutes at 75 °C after the dipping. For HTL, 73.3 mg mL⁻¹ Spiro-OMeTAD (2,2′,7,7′-tetrakis (N,N-di-p-methoxyphenylamino)-9,9′-spirobifluorene, sublimed grade 99.8%, Borun chemicals) in chlorobenzene was doped with 17.5 μL Li-TFSI (520 mg mL⁻¹ in acetronitrile) and 28.8 μL 4-TBP. The Spiro-OMeTAD layer was spin coated on the perovskite layer with 1200 rpm for 30 s. Finally, 100 nm Ag layer was deposited on the Spiro-OMeTAD HTL coated substrate by thermal evaporation with an evaporation rate of 2 Å s⁻¹ under a vacuum condition of 1 × 10⁻⁶ mbar. The device area was fixed to be 0.045 cm² with the use of a metal mask. So, the overall device structures were: ITO/solgel ZnO or AZO/MA_0.6FA_0.4PbI₃ perovskite/Spiro-OMeTAD/Ag.

2.2 Device characterization

The current–voltage characteristics of the devices were measured with a NREL calibrated Keithley 2400 Source Meter under 100 mW cm⁻² (AM 1.5G) simulated sunlight. For optical characterization like transmittance, reflectance and absorbance measurement, a UV-VIS-NIR spectrometer (Perkin Elmer – Lambda 950) was used. X-ray diffraction (XRD) with CuKα radiation was performed by step-scanning with a step size of 0.02 degree. Surface topology and device cross sectional view were captured by Carl Zeiss AURIGA Cross Beam SEM (Scanning Electron Microscopy) and the surface roughness was measured with Bruker Dimension ICON SPM AFM (Atomic Force Microscopy) machine. The impedance analysis was conducted with an Autolab PGSTAT-30 equipped with a frequency analyzer module in the frequency range from 1 MHz to 10 Hz. AC oscillating amplitude was as low as 20 mV (RMS) to maintain the linearity of the response. Contact angle measurement was conducted using Ramé-hart contact angle goniometer (Model 200) and the contact angles were precisely determined using DROPimage advanced software.

3. Results and discussion

3.1 Material and morphological characterization of ZnO and AZO interlayers

In the present work, ZnO and Al doped ZnO thin films have been fabricated with a solution-route process at <150 °C temperature. The formation of basic ZnO structure and the incorporation of Al dopant into the ZnO crystal have been ensured from the XRD measurement and the characteristic bandgap calculation using Tauc plot. The XRD measurement of pristine ZnO (Fig. 1(A)) thin film exhibits characteristic (002) and (101) diffraction peaks, which corroborate to the formation of hexagonal wurtzite ZnO phase. The strong preferential growth along (002) plane indicates the orientation of the fabricated ZnO film to be along c-axis i.e. along the direction perpendicular to the ITO/glass substrate.²⁰ The absence of any distinct (001) characteristic peak substantiates to the fact that the produced ZnO film is non-oriented along a-axis i.e. the film orientation is not parallel to the underlying substrate.²⁰ A weak (101) diffraction peak is also visible in the XRD pattern of ZnO film which is anticipated with a ZnO precursor solution having the molar concentration in the range 0.3–0.6 M.²⁰ The XRD pattern of AZO film has been presented in Fig. 1(B). Despite exhibiting strong orientation along c-axis and weak alignment along a-axis, in an identical manner as ZnO film, the characteristic peak widths of AZO film are distinctive from those of pristine ZnO. Each of these characteristic diffraction peaks corresponds to a finite sized crystallite²¹ which collectively constitutes an individual grain by the concurrence of crystallite aggregation and gelation induced growth.²² The individual crystallite size is estimated using the Debye Scherrer equation:²³where, D, k, λ, β and θ denote the individual crystallite dimension, Scherrer constant (0.94 with cubic crystal), the X-ray wavelength used for XRD measurement (0.154 nm), full-width at half maximum (FWHM) for individual characteristic peak (peak width) and diffraction angle respectively. Table S1† lists the peak width (FWHM) and estimated dimension of individual crystallite of ZnO and AZO film calculated from the spectral fitting of XRD measurement data. As observed from Table S1,† both the (002) and (101) oriented crystallites in AZO have relatively small dimension compared to those of ZnO. The smaller individual crystallite size in AZO ascertains the incorporation of Al³⁺ ions in ZnO structure by replacing Zn²⁺ ions in active Zn ion sites or vacant sites,²⁴ since ionic radius of Al³⁺ (0.68 Å) is smaller than that (0.88 Å) of Zn²⁺ ion.²⁵ In line with the constituent crystallite dimension, individual grain size in AZO is expected to be lower than that of ZnO film which is evident from the SEM images of AZO and ZnO surface morphology presented in Fig. S1.† The smaller grain size of AZO can be ascribed to the AZO growth inhibition phenomena induced by Al₂O₃ precipitation on grain surface.²¹ However, the spherical nanostructured AZO grains are more closely packed compared to ZnO grains, which have mutually isolated crystallography with numerous voids at the grain boundary (Fig. S1†). Alongside the surface morphology analysis with SEM, we have also conducted surface topology characterization with AFM for both ZnO and AZO film (Fig. S2†). As observed from Fig. S2,† the surface topology of ZnO film is influenced by the incorporation of Al dopants. The RMS (root mean square) surface roughness of AZO film (3.44 nm) has been found to be lower than pristine ZnO film (3.64 nm). The variation in surface topography can be correlated to the increased grain packing density in AZO film²¹ by the substitution of Zn²⁺ ion with smaller radius Al³⁺ ion.

Fig. 1 XRD pattern of (A) ZnO and (B) AZO film (the * signs denote the characteristic peaks for ITO) and estimation of approximate material bandgap for low temperature processed (C) ZnO and (D) AZO thin film using Tauc plot.

Approximate material bandgaps of fabricated ZnO and AZO film have been calculated from Tauc plot. In general, the bandgap of a material can be approximated from the Tauc plot by relating its optical absorption with the photon energy using the following relation:

(αhν)² = A(hν − E_g) (2)

where, α, h, ν, A and E_g stand for absorption co-efficient, Planck's constant, frequency of light photon, proportionality constant and material bandgap respectively. The x-axis intercept of the extrapolated linear section of the Tauc plot gives a rough estimate of the material bandgap. Fig. 1(C) and (D) show the Tauc plots for pristine ZnO and AZO thin film respectively. From Fig. 1(C), the bandgap for pristine ZnO is found to be 3.46 eV which indicates the formation of basic ZnO structure. The bandgap of AZO is slightly higher (3.55 eV) than that of pristine ZnO (Fig. 1(D)).

In a normal structured device, the paramount optical role of ETL is to function as a highly transmitting, diminutively reflecting thin film layer to ensure maximum light absorption in photo-active perovskite film. In general, the absorbance of a thin film can be expressed as:²⁶

A = −log[β(1 − γ) × 10^−ρ + (1 − β)] (3)

where, A, β, γ and ρ correspond to thin film absorbance, substrate coverage ratio (thin film covered area/total substrate area), light scattering co-efficient (γ = 0, no scattering takes place at all; γ = 1, the incident lights are fully scattered from the thin film) and optical density of thin film respectively. Because of the smaller ionic radius of Al³⁺ ion in contrast with Zn²⁺ ion, the light scattering phenomena are expected to be lower²¹ from AZO film, rendering a lower scattering co-efficient, γ for AZO than that of ZnO. As conceived from eqn (3), the smaller value of γ contributes to comparatively higher light absorption in AZO film. Besides, as observed from the surface morphology characterization by SEM imaging (Fig. S1†), AZO film has more uniform surface coverage with fewer pinholes and voids implicating the surface coverage ratio, β for AZO film to be higher than that of ZnO film, which also contributes to relatively high absorption in AZO film. Comparatively high absorption in AZO film attributes to relatively low transmittance through it, compared to ZnO film, which is evident from the transmittance pattern of ZnO and AZO film presented in Fig. S3.† Nevertheless, both the ETL films exhibit above 85% transmittance over a wide range of spectrum extending from the blue to early red region of the visible spectrum (440–760 nm) and more than 80% transmittance up to the early near infra-red (IR) zone (∼1000 nm). Therefore, both ZnO and AZO films function as highly transmitting, less reflective ETL thin films that ensure high light absorption in the overlying photo-active material.

3.2 Material and morphological characterization of perovskite layer

On top of both ZnO and AZO ETL films, MA_0.6FA_0.4PbI₃ perovskite film was fabricated in an identical manner using two step dipping method. Fig. S4(A)† shows the XRD pattern of MA_0.6FA_0.4PbI₃ perovskite film along with PbI₂ XRD diffraction peaks for qualitative comparison of perovskite conversion. Characteristic diffraction peak of PbI₂ (001) at 12.50° is absent in the XRD pattern of MA_0.6FA_0.4PbI₃ perovskite film which ensures that the spin-coated PbI₂ layer has completely been converted into perovskite film. The full transformation of PbI₂ was assured by the addition of 4-tert-butylpyridine (4-TBP)²⁷ in PbI₂ precursor solution with DMF. As observed from XRD pattern in Fig. S4(A),† the presence of distinct (110), (220), (310), (224) and (314) characteristic diffraction peaks ensures the construction of tetragonal perovskite structure.²⁸ From the enlarged view (Fig. S4(B)†) of the XRD pattern, the (110) characteristic perovskite peak can be identified at a diffraction angle of 13.95° which is 0.15° right-shifted from pristine FAPbI₃ peak located at 13.80°.² This right shift in (110) diffraction peak is congruent with the previous report with mixed organic cation perovskite.² However, no distinct FAPbI₃ peak at 11.80° diffraction angle^1,2 was evident in the XRD pattern, which ascertains the formation of mixed organic cation perovskite in which both MAI and FAI organic cations have been incorporated in the same lattice frame.

The optical properties of constructed MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO films on top of ITO/glass substrates were investigated with absorbance measurement. Fig. S5(A)† illustrates the absorbance pattern of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO films on ITO/glass substrate. As perceived from Fig. S5,† the absorption onset for both the films were at ∼820 nm which is congruent with the red-shifted spectral response achieved with mixed organic cation perovskite (MA_0.6FA_0.4PbI₃) compared to mono organic methyl ammonium lead triiodide perovskite (MAPbI₃), having an absorption onset around ∼800 nm.⁷ The extended absorption in the early near infra-red (IR) zone is congruent with the slightly reduced bandgap (1.54 eV) of pristine MA_0.6FA_0.4PbI₃ perovskite compared to that (1.57 eV)⁶ of pristine MAPbI₃ perovskite, as estimated from the Tauc plot presented in Fig. S5(B).† As evident from Fig. S5(A),† absorption in MA_0.6FA_0.4PbI₃/AZO film is slightly lower compared to MA_0.6FA_0.4PbI₃/ZnO which is coherent with the as-discussed lower transmittance from AZO film compared to ZnO film.

Fig. 2(A–D) illustrate the SEM surface morphology image of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO films respectively on ITO/glass substrate. As observed from Fig. 2, the grain size of MA_0.6FA_0.4PbI₃ perovskite grown on AZO ETL film is larger compared to that grown on ZnO ETL film. The variation in the surface morphology of identically fabricated perovskite film on different ETL films is congruent with the earlier findings that the formation of perovskite film can be significantly modified based on the underlying layer, the perovskite grows on.^29,30 Some earlier literatures have attributed the enhanced surface morphology of perovskite to the better wettability of the underlying layer (on which the perovskite grows) with respect to perovskite precursor.^29,31 However, in our study, we have not found any significant difference between the wettability of ZnO and AZO thin film with respect to PbI₂ containing DMF precursor solution from contact angle measurement (Fig. S6†). In fact, on both ZnO and AZO films, DMF has a contact angle of ∼0 degree which resembles to the finding of a recent study by Miyaska et al.³² which demonstrates that the surface wettability of metal oxide ETL (with respect to perovskite precursor) does not demonstrate significant difference on macroscopic scale for perovskite surface morphology. In this regard, we can refer to an earlier study by Boyen et al.³³ which has reported that the inter-particle connectivity of perovskite grain and thus its surface morphology³⁴ can be enhanced by the removal of residual impurities from the underneath metal oxide ETL. It has already been reported that the residual impurities like hydroxyl (OH⁻) ions and organic CH₃COO⁻ radicals are significantly reduced in Al dopant incorporated ZnO film¹⁶ compared to pristine ZnO film. Thus the larger grain size of MA_0.6FA_0.4PbI₃ perovskite grown on AZO ETL film can be attributed to the enhanced inter-particle connectivity of perovskite owing to the significantly reduced residual impurity^33,34 at the underneath AZO film.

Fig. 2 Top view Scanning Electron Microscopy (SEM) images of (A) and (B) MA_0.6FA_0.4PbI₃ perovskite/ZnO and (C) and (D) MA_0.6FA_0.4PbI₃ perovskite/AZO on top ITO/glass substrate. Two and three dimensional Atomic Force Microscopy (AFM) images of MA_0.6FA_0.4PbI₃ perovskite film overlying (E) and (F) ZnO and (G) and (H) AZO ETL films on ITO/glass substrate.

We have also conducted atomic force microscopy (AFM) measurements of MA_0.6FA_0.4PbI₃ perovskite films deposited on top of ZnO and AZO films. Fig. 2(E–H) present the two and three dimensional AFM images of MA_0.6FA_0.4PbI₃ perovskite films on top of ZnO and AZO films respectively. From the statistical quantity analysis of AFM images, the RMS surface roughness of MA_0.6FA_0.4PbI₃/AZO film (55.7 nm) has been found to be lower than that (56.80 nm) of MA_0.6FA_0.4PbI₃/ZnO film. However, it is notable that the perovskite film grown with two step dipping method usually has higher RMS surface roughness compared to one step deposition technique,²⁷ as the film formation process involves the drying of 2-propanol over the perovskite film once the substrate is taken out of the dipping solution.³⁵ Nevertheless, the RMS values indicate that the perovskite surface topography is highly conducive for minimal leakage current and efficient charge extraction,^35,36 when applied as a photo-active layer in a fabricated solar cell.

3.3 Device performance of perovskite solar cells

In pursuance of assessing the device performance of MA_0.6FA_0.4PbI₃ perovskite in conjunction with low temperature processed ZnO and AZO ETLs, we have fabricated solar cells having the device structures-ITO/ZnO/MA_0.6FA_0.4PbI₃/Spiro-OMeTAD/Ag and ITO/AZO/MA_0.6FA_0.4PbI₃/Spiro-OMeTAD/Ag. Fig. 3(A) shows the schematic presentation of the two device structures under study. In our study, MA_0.6FA_0.4PbI₃ perovskite has been chosen as photo-active layer, since MA_0.6FA_0.4PbI₃ has been reported to render most optimum device performance among the mixed organic cation based perovskite devices having different compositional ratio of MAI to FAI.² During the optimization stage, we have also investigated the device performance of MA_0.6FA_0.4PbI₃ perovskite in conjunction with pristine ZnO and different concentration of AZO as ETL. Among them, MA_0.6FA_0.4PbI₃ perovskite has been found to render the optimum device performance in conjunction with 15 wt% (wt% of Al dopant in ZnO) AZO ETL (Table S2†). Different compositional ratios of MAI to FAI have also been investigated as photo-absorbing layer (along with pristine ZnO and optimized AZO ETL) to experimentally confirm the previously reported optimum device performance with MA_0.6FA_0.4PbI₃ perovskite² (Table S3†). For further reference in the remaining part of the paper, the pristine ZnO ETL and optimized AZO ETL devices will be termed as MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices. Table 1 contains the photovoltaic device performance of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices attained from I–V measurement. The best performing MA_0.6FA_0.4PbI₃/AZO device exhibits a PCE of 14.54% with the J_SC, V_OC and FF value being 20.2 mA cm⁻², 1014.160 mV and 71% respectively. The average values of J_SC, V_OC and FF for these devices are 20.18 mA cm⁻², 1004.46 mV and 68.89% respectively. To the best of our knowledge, the average PCE reported in our study is the highest ever average PCE attained with MA_0.6FA_0.4PbI₃ perovskite device surpassing the previous best average PCE of 13.50% with TiO₂ ETL,² even though the processing temperature of our perovskite devices are 300% lower than the previous study.² On the contrary, MA_0.6FA_0.4PbI₃/ZnO devices demonstrate comparatively lower average PCE of 11.82% with the average J_SC, V_OC and FF values being 21.44 mA cm⁻², 973.48 mV and 56.69% respectively. Even the best performing MA_0.6FA_0.4PbI₃/ZnO device shows relatively degraded photovoltaic performance (PCE: 13.40%, J_SC: 22.39 mA cm⁻² V_OC: 984.24 mV and FF: 60.66%) in contrast with the average performance exhibited by MA_0.6FA_0.4PbI₃/AZO devices. Fig. 3(B) and (C) show the IV curves of the best performing MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices respectively. The high reproducibility of both the device structures is evident from the statistical histograms (with normal distribution) of PCE, J_SC, V_OC and FF of 10 identically fabricated MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices, presented in Fig. S7 and S8† respectively. The values of PCE, J_SC, V_OC and FF of individual MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO device have been tabulated in Tables S4 and S5† for numerical comparison.

Fig. 3 (A) Schematic representation of fabricated MA_0.6FA_0.4PbI₃ perovskite solar cells having ZnO (device-A) and AZO (device-B) thin film as electron transport layer and I–V curve of the best performing (B) MA_0.6FA_0.4PbI₃ perovskite/ZnO and (C) MA_0.6FA_0.4PbI₃ perovskite/AZO devices.

Table 1 Average and best performance with ITO/solgel ZnO or Al doped ZnO (AZO)/MA_0.6FA_0.4PbI₃ perovskite/Spiro-OMeTAD/Ag device structures

ETL layer	Average/best value	Open circuit voltage, VOC (mV)	Short circuit current density, JSC (mA cm^−2)	Fill factor, FF (%)	Efficiency (%)	Series resistance, RS (Ω)	Shunt resistance RSH (Ω)
Pristine ZnO	Average	973.48	21.44	56.69	11.82	10.13	26 190
	Best	984.24	22.39	60.66	13.40	9.09	23 200
Al doped ZnO (AZO)	Average	1004.46	20.18	68.89	13.95	7.05	35 700
	Best	1014.160	20.20	71.00	14.54	5.81	38 600

---

As observed from Table 1, nearly 20% enhanced average PCE with MA_0.6FA_0.4PbI₃/AZO devices can be attributed to 21.5% increase in average FF along with 3% (30 mV) augmentation in average V_OC value in contrast with MA_0.6FA_0.4PbI₃/ZnO devices, which compensate about 6% decline in average J_SC with AZO devices. In general, FF of a photovoltaic device can be expressed using an empirical formula:³⁷

where, R_S, R_CH, V_OC and R_SH refer to series resistance of the cell, characteristic resistance , normalized open circuit voltage (, n and V_T are cell ideality factor and thermal voltage) and shunt resistance of the device respectively. Significant augmentation in FF with MA_0.6FA_0.4PbI₃/AZO devices can be explicated using eqn (4) in the light of more conforming surface morphology of mixed organic cation perovskite atop AZO film. The relevant detailed discussion has been portrayed in the following paragraph.

As observed from surface morphology characterization with SEM imaging (Fig. 2), MA_0.6FA_0.4PbI₃ perovskite crystal grown on AZO has larger grain size compared to the one fabricated on ZnO. In a recent study using temperature dependent PL (photoluminescence) spectra, Chen et al.³⁸ have disclosed that the distance between ionized acceptor and donor pair in large grain size perovskite is significantly higher than that in a smaller grain size perovskite crystal. In general, the photon energy of DAP (donor–acceptor pair) recombination can be expressed as:³⁸

where, E_g, E_A, E_D, q, ε₀, ε_r and R correspond to energy bandgap, acceptor binding energy, donor binding energy, elementary charge, free space dielectric constant, relative dielectric constant and the distance between ionized donor–acceptor pair respectively. As conceived from eqn (6), due to the comparatively large distance, R between ionized donor and acceptor pair in MA_0.6FA_0.4PbI₃ perovskite grown on AZO, the DAP recombination energy is lower in it compared to identical mixed organic cation perovskite fabricated atop ZnO. Besides, the binding energy of acceptor and donor (E_A and E_D) is expected to reduce substantially³⁸ with larger grain sized MA_0.6FA_0.4PbI₃ perovskite on AZO. Thus, from eqn (6), it can be inferred that the DAP recombination in MA_0.6FA_0.4PbI₃/AZO devices is significantly lower in comparison with MA_0.6FA_0.4PbI₃/ZnO devices. Besides, from Table 1, the average shunt resistance, R_SH for MA_0.6FA_0.4PbI₃/AZO devices is nearly 30% higher (26190 Ω for MA_0.6FA_0.4PbI₃/ZnO devices and 35700 Ω for MA_0.6FA_0.4PbI₃/AZO devices) compared to MA_0.6FA_0.4PbI₃/ZnO devices. Since, device shunt resistance, R_SH is an indirect implication towards the current leakage paths in a solar cell caused by crystal impurity and defects,³⁹ the MA_0.6FA_0.4PbI₃ perovskite grown on AZO ETL is expected to exhibit lower interfacial defects due to its reduced grain boundary area⁴⁰ and more oriented patterns of grain,³⁹ congruent with the SEM and AFM characterization presented in our work. As comprehended from eqn (4), higher value of device shunt resistance aided by the low DAP recombination and the mitigated current leakage phenomena in MA_0.6FA_0.4PbI₃/AZO devices contribute to their significant augmentation in average FF value compared to MA_0.6FA_0.4PbI₃/ZnO devices. Furthermore, higher perovskite crystallinity also corroborates⁴¹ to about 30% lower average series resistance, R_S value (10.13 Ω for MA_0.6FA_0.4PbI₃/ZnO and 7.05 Ω for MA_0.6FA_0.4PbI₃/AZO devices) in MA_0.6FA_0.4PbI₃/AZO devices, which connotes to the substantial improvement in their FF value according to eqn (4). In addition to these, owing to the aforementioned larger distance between donor and acceptor in MA_0.6FA_0.4PbI₃ perovskite atop AZO film, forming bound excitons is more difficult in it³⁸ compared to identically fabricated perovskite on ZnO. Thus, the DAPs in MA_0.6FA_0.4PbI₃ perovskite atop AZO film can be inferred to be separated more easily³⁸ than that on ZnO film, which correlates to excellent charge extraction⁴² and thus significantly boosted FF in MA_0.6FA_0.4PbI₃/AZO devices.

Along with the substantial FF augmentation, above 30 mV increase in average V_OC with MA_0.6FA_0.4PbI₃/AZO devices can also be explained in terms of the suppressed DAP recombination and low current leakage phenomena in contrast with MA_0.6FA_0.4PbI₃/ZnO devices. In principle, the V_OC of a photovoltaic device can be expressed as:³⁷

where, k, T and J₀ denote Boltzman's constant, absolute temperature and saturated reverse current density respectively. The saturated reverse current density, J₀ is anticipated to have lower value with MA_0.6FA_0.4PbI₃/AZO devices due to their larger value of shunt resistance³⁹ with regards to MA_0.6FA_0.4PbI₃/ZnO devices. As observed from eqn (7), the reduced saturated current density results in the augmentation in average V_OC value with MA_0.6FA_0.4PbI₃/AZO devices. On the contrary, the 6% decline in average J_SC with MA_0.6FA_0.4PbI₃/AZO devices is congruent with the slightly reduced optical absorption in MA_0.6FA_0.4PbI₃ perovskite film atop AZO owing to moderately low transmittance through the AZO film compared to pristine ZnO film (Fig. S3†). However, the decrease in J_SC with MA_0.6FA_0.4PbI₃/AZO devices is well-compensated with overwhelming enhancement in FF and V_OC, endowing those with an augmented device performance with reference to MA_0.6FA_0.4PbI₃/ZnO devices.

In order to understand the underlying mechanism behind the enhanced performance of MA_0.6FA_0.4PbI₃/AZO devices, we have conducted electrochemical impedance study (EIS) on both the device structures; since EIS is an efficient tool to reveal detailed information regarding the device electronic properties that govern the charge transfer and carrier recombination phenomena.^43,44 Fig. 4(A) displays the Nyquist plots for MA_0.6FA_0.4PbI₃/AZO and MA_0.6FA_0.4PbI₃/ZnO devices at 850 mV bias under dark. The experimental data were fitted with an equivalent circuit model previously reported with a perovskite device.⁴⁵ The equivalent circuit model consists of a series resistance, R_S in conjunction with a parallel R–C component and a parallel R-CPE (constant phase element) component (Fig. S9†). From the equivalent circuit displayed in Fig. S9,† the symbols R_S, R_C, C_C, R_Rec and CPE correspond to the resistance originating from the wire connection and metal contact, contact resistance at perovskite/ETL or perovskite/HTL interface, interfacial capacitance, recombination resistance and constant phase element representing the interface heterogeneity or surface roughness⁴³ respectively. Table 2 lists the fitted values of different electronic parameters extracted from the Nyquist plot from Fig. 4(A). In our study, the goodness of the fit has been ensured by attaining a very small (∼10⁻⁵) chi-squared (χ²) value associated with the variation in experimental and fitted data using Pearson's chi-squared test with Nova 1.11 Autolab fitting software. As comprehended from Table 2, the recombination resistance, R_Rec increases nearly 2.5 times in MA_0.6FA_0.4PbI₃/AZO devices compared to MA_0.6FA_0.4PbI₃/ZnO devices (3150 Ω for MA_0.6FA_0.4PbI₃/ZnO devices and 7690 Ω for MA_0.6FA_0.4PbI₃/AZO devices). Since, during the dark impedance measurement, the carriers are injected from external voltage source only, the substantial uplift in recombination resistance with MA_0.6FA_0.4PbI₃/AZO devices can be attributed to the reduced interfacial electron trapping phenomena²¹ at MA_0.6FA_0.4PbI₃ perovskite/AZO interface, which is congruent with the aforementioned smaller grain boundary area at perovskite film atop AZO ETL. Furthermore, MA_0.6FA_0.4PbI₃/AZO devices exhibit higher recombination resistance in contrast with MA_0.6FA_0.4PbI₃/ZnO devices irrespective of applied bias and impedance measurement condition (under light illumination or under dark) which are evident from Fig. 4(B) and (C) which illustrate the trend of recombination resistance as a function of applied bias under light illumination and under dark respectively. Tables S6 and S7† contain the bias dependent recombination resistance values of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices under dark and light respectively for numerical comparison. Significant augmentation in recombination resistance irrespective of bias and measurement condition corroborates to substantially subdued recombination phenomenon in MA_0.6FA_0.4PbI₃/AZO devices which is coherent with our initial observation of reduced DAP recombination³⁸ and current leakage phenomena³⁹ aided by larger perovskite grain on AZO film. The suppressed recombination phenomena in MA_0.6FA_0.4PbI₃/AZO devices result in a significant boost on device FF and V_OC (ref. 39) and thus contribute to the enhancement of overall device PCE. However, referring to Table 2, the contact resistance, R_C at perovskite/ETL and perovskite/HTL interface in MA_0.6FA_0.4PbI₃/AZO devices is slightly higher with reference to MA_0.6FA_0.4PbI₃/ZnO devices (78 Ω for MA_0.6FA_0.4PbI₃/ZnO devices and 86 Ω for MA_0.6FA_0.4PbI₃/AZO devices). Since, the HTL film remains unchanged for both the device structures; any variation in the contact resistance can be inferred to originate from the ETL layer. Moderately higher contact resistance in MA_0.6FA_0.4PbI₃/AZO devices explain the slight decline in their average J_SC value⁴⁵ which is counterbalanced with the substantial enhancement in FF and V_OC.

Fig. 4 (A) Nyquist plot of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO device at a bias of 850 mV under dark (B) recombination resistance as a function of applied bias for MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices extracted from Nyquist plot under light illumination and (C) under dark (D) dark carrier mobility as a function of applied bias voltage for MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices extracted from impedance measurement.

Table 2 Fitted values of different electronic parameters from Nyquist plot of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO perovskite devices at 0.85 V under dark

ETL layer	RS (Ω)	RC (Ω)	RRec (Ω)	CC (nF)	CPE-Y (nMho)	CPE-N
Pristine ZnO	28	78	3150	11.1	0.30	0.96
Al doped ZnO (AZO)	27.2	86	7690	17.2	0.28	0.96

---

In order to comprehend the monomolecular and bimolecular recombination phenomena in MA_0.6FA_0.4PbI₃/ZnO devices and MA_0.6FA_0.4PbI₃/AZO devices, the dark carrier mobility (calculated from EIS) of both the devices has been plotted as a function of applied bias in Fig. 4(D). Tables S8 and S9† represents the numeric values of bias conditions and relative dark carrier mobility obtained for MA_0.6FA_0.4PbI₃/ZnO devices and MA_0.6FA_0.4PbI₃/AZO devices respectively alongside the fitted electronic parameters extracted from Nyquist plots which are used to calculate the mobility values. The details of dark carrier mobility calculation have been provided in supporting info (see “dark carrier mobility calculation” section). In principle, monomolecular recombination prevails at or near short circuit condition, while bimolecular recombination dominates at or near open circuit voltage bias condition in a photovoltaic device having p–i–n configuration.^46,47 Monomolecular recombination is trap-centre mediated involving the encounter of a charge carrier with an occupied trap state of immobile charge having opposite polarity, while bimolecular recombination corresponds to generic electron–hole recombination regulated by the probability of both meeting each other. Usually, high carrier mobility is an indication towards suppressed recombination phenomena in a photovoltaic device.⁴⁸ As observed from Fig. 4(D), under low bias condition (0.4 V), MA_0.6FA_0.4PbI₃/AZO devices (μ = 2.61 cm² V⁻¹ s⁻¹) have ∼23% higher dark carrier mobility compared to MA_0.6FA_0.4PbI₃/ZnO devices (μ = 2.13 cm² V⁻¹ s⁻¹) which indicates substantial reduction in monomolecular recombination with MA_0.6FA_0.4PbI₃/AZO devices. The mitigated monomolecular recombination in MA_0.6FA_0.4PbI₃/AZO devices attributes to the significant reduction in interfacial trap states at MA_0.6FA_0.4PbI₃ perovskite/AZO interface compared to MA_0.6FA_0.4PbI₃/ZnO interface which is consistent with reduced grain boundary in MA_0.6FA_0.4PbI₃ atop AZO film (Fig. 2). However, both MA_0.6FA_0.4PbI₃ perovskite/AZO and MA_0.6FA_0.4PbI₃/ZnO devices have comparable dark carrier mobility values (μ = 0.82 cm² V⁻¹ s⁻¹ for MA_0.6FA_0.4PbI₃ perovskite/AZO devices and μ = 0.79 cm² V⁻¹ s⁻¹ for MA_0.6FA_0.4PbI₃ perovskite/ZnO devices) near open circuit voltage bias condition (0.9 V), indicating that the bimolecular recombination phenomenon is not significantly different in MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices.

3.4 Mott Schottky analysis (capacitance vs. voltage) and conductivity

Alongside the elaborate investigation into the device recombination phenomena by means of impedance spectroscopy, we have conducted Mott–Schottky analysis of MA_0.6FA_0.4PbI₃ perovskite/AZO and MA_0.6FA_0.4PbI₃/ZnO devices to divulge further into the device electronic properties contributing to augmented PCE with AZO ETLs. Fig. 5(A) and (B) illustrate the Mott–Schottky plots of MA_0.6FA_0.4PbI₃ perovskite/ZnO and MA_0.6FA_0.4PbI₃/AZO device respectively at 10 kHz frequency under dark. In general, the x-axis intercept of the extrapolated linear section of the Mott–Schottky curve provides the flat band potential. The dopant density can also be calculated from the slope of the linear portion of the curve using the following expression:⁴⁹where, N, V_fb, V, C, A and ε correspond to dopant density, flat-band potential, applied potential, junction capacitance originating from depletion width modulation, device area and perovskite dielectric constant respectively. Depletion width at zero applied bias, W can also be calculated using the following mathematical expression:

Fig. 5 Mott–Schottky curve at 10 KHz frequency with (A) MA_0.6FA_0.4PbI₃/ZnO and (B) MA_0.6FA_0.4PbI₃/ZnO devices under dark.

Table S10† contains the values of dopant density, flat-band potential and zero bias depletion width with of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices. From Table S10,† the flat-band potential in MA_0.6FA_0.4PbI₃/AZO device (V_fb = 1.13 V) is 170 mV higher compared to that in MA_0.6FA_0.4PbI₃/ZnO device (V_fb = 0.96 V). Flat-band potential can be defined as the potential required to compensate the energetic difference between the quasi Fermi level of perovskite and that of ETL/cathode electron selective contact⁴⁴ and mathematically expressed as:

V_fb = E_Fn − E_P (10)

where, E_Fn and E_P correspond to quasi Fermi level of ETL/cathode contact and perovskite film respectively. Since the variation in recombination mechanism between the two device structures is dominated by trap-mediated monomolecular recombination rather than the generic bimolecular recombination originating from the bulk properties (as comprehended from dark carrier mobility measurement in Fig. 4(D)), the disparity in flat band potentials can be attributed to the dissimilar quasi Fermi levels of the ETL layers in lieu of ascribing it to the quasi Fermi levels of identically fabricated MA_0.6FA_0.4PbI₃ perovskite films in both devices. In general, quasi Fermi level is a combined property of charge carriers, strongly modulated by the energetic disorder in the material.⁵⁰ Despite being an inorganic semiconductor, low temperature, ambient air annealed ZnO film has a high order of energetic disorder at the grain boundaries owing to its chemisorption of oxygen.⁵¹ Consequently, low temperature processed ZnO film exhibits large density of deep trap states, even extending up to 0.2 eV below the conduction band.⁵² Thus, any electron from these unoccupied deep trap states can contribute to the Fermi level⁵⁰ in low temperature processed ZnO that causes a spatial variation in quasi Fermi level⁴⁶ resulting in its downward shift compared to the quasi Fermi level position with high temperature processed ZnO, having higher crystallinity.⁵³ On the contrary, Al incorporation in ZnO causes the replacement of Zn²⁺ ion by Al³⁺ ion in the ZnO host lattice, which contributes to one free electron per atom that increases the overall n-type carrier concentration. In general, band filling model suggests that the lowest states in the conduction band of a semiconductor can be filled with the electrons if the n-type carrier concentration exceeds the conduction band of density, N_C (ref. 54) which can be expressed as:where, m* (0.38m₀ in this case⁵⁴), k, T and h stand for effective mass of electron, Boltzman's constant, absolute temperature and Planck's constant respectively. From eqn (11), the minimum n-type carrier concentration required for the lowest conduction band state filling in ZnO is found to be 6 × 10¹⁸ cm⁻³. Since, the Al doping concentration in our study is higher (3.13 × 10²¹ cm⁻³) than the calculated value, it can be inferred that the free electrons contributed from the Al³⁺ ions will fill the deep trap states at the conduction band in low temperature processed AZO, which will attribute to the upshift in quasi Fermi level position along the conduction band in accordance with Burstein–Moss effect.^21,55,56 According to eqn (11), this escalation in quasi Fermi level corroborates to the higher flat-band potential with MA_0.6FA_0.4PbI₃/AZO devices, as observed in Mott–Schottky curve in Fig. 5(A) and (B). The higher flat-band potential in MA_0.6FA_0.4PbI₃/AZO devices also elucidates the significant improvement in average V_OC compared to MA_0.6FA_0.4PbI₃/ZnO devices. In general, the device V_OC can be attributed to the energetic offset between the quasi Fermi level of ETL/cathode contact, E_Fn and the quasi Fermi level of HTL/anode contact, E_Fp and can be mathematically expressed as:⁵⁷

V_OC = E_Fn − E_Fp (12)

Since, the HTL film (Spiro-OMeTAD) remains constant for both the device structures, the augmented V_OC in MA_0.6FA_0.4PbI₃/AZO devices can be attributed to the upshift of quasi Fermi level in them owing to Burstein–Moss effect. Fig. 6 illustrates a schematic presentation of the trap assisted downshift of quasi Fermi level in low temperature processed ZnO in MA_0.6FA_0.4PbI₃/ZnO device (device-A) along with the Burstein–Moss effect assisted upshift in quasi Fermi level in low temperature processed AZO film in MA_0.6FA_0.4PbI₃/AZO device (device-B) demonstrating the aftermath of such Fermi level shifting on device performance. In device-A, the electrons from perovskite LUMO (lowest unoccupied molecular orbital) hop between the deep trap states⁵⁸ present in low temperature processed ZnO film, increasing the probability of recombining with holes before being collected by the ITO cathode. Thus the deep quasi Fermi position below the conduction band hinders the efficient charge extraction by ITO which can be connoted to comparatively low electrical conductivity previously reported for pristine ZnO film based devices compared to doped ones.^16,21 Conversely, the free electrons contributed by the substituent Al³⁺ ions occupy the trap states and uplifts the quasi Fermi level in low temperature processed AZO film that impede the acute electron hoping phenomena in device-B. The reduced electron hoping renders efficient charge extraction capability to the ITO cathode and enhances the electrical conductivity likewise. The higher electrical conductivity thus again enunciates the augmented FF and PCE in MA_0.6FA_0.4PbI₃/AZO devices with reference to MA_0.6FA_0.4PbI₃/ZnO devices (Fig. 6).

Fig. 6 Schematic presentation of (A) deep trap state assisted downshift of quasi Fermi level in low temperature processed pristine ZnO film attributing to lower flat-band potential, lower carrier conductivity and higher charge carrier recombination phenomena in MA_0.6FA_0.4PbI₃/ZnO device and (B) Burstein–Moss effect assisted upshift in quasi Fermi level of AZO film contributing to higher flat-band potential, higher carrier conductivity and low recombination phenomena in MA_0.6FA_0.4PbI₃/AZO devices. Device-B exhibits higher FF, V_OC and PCE compared to device-A. (E_P, E_Fn, and E_F denote quasi Fermi level of perovskite, ETL layer and ITO respectively. E_C and E_V stand for conduction band and valence band levels respectively for ETL layers (ZnO and ETL).

3.5 Interface polarization (capacitance vs. frequency) and hysteresis phenomenon

In addition to the detailed investigation into the factors contributing to the enhanced device performance with MA_0.6FA_0.4PbI₃/AZO devices, we have also probed into the photocurrent hysteresis behaviour⁵⁹ of the two device structures under study. Photocurrent hysteresis phenomena over or underestimate the real device performance of perovskite solar cells under operating condition. Thus hysteresis has arisen as one of the major bottlenecks towards commercial production of perovskite solar cells. Exploring the photocurrent hysteresis behaviour in a perovskite solar cell is thus of paramount importance. Fig. 7(A) and (B) demonstrate the hysteresis phenomena in MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices respectively relative to scan rate. The corresponding PCE, J_SC, V_OC and FF values of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices at two different scan rates have been tabulated in Table S11.† As observed from Fig. 7(A), (B) and Table S11,† the scan rate related hysteresis phenomenon is less in MA_0.6FA_0.4PbI₃/AZO devices compared to MA_0.6FA_0.4PbI₃/ZnO devices. However, it is noteworthy that the device hysteresis behaviour in the present study is substantially subdued in contrast with the only reported earlier literature incorporating sputtered AZO ETL and MAPbI₃ perovskite as light-harvester.¹⁶ In the present study, we have analysed frequency dependent capacitive response for both MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices that unravels conducive information regarding the photocurrent hysteresis behaviour in the devices.⁶⁰ Fig. 7(C) illustrates the frequency dependent capacitance plot at zero applied bias for MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices. A recent photocurrent hysteresis study conducted by Belmonte et al.⁶⁰ has proposed that two concurrent polarization mechanisms take place in perovskite solar cells. Electrode polarization is one of them which is primarily evolved from the considerable coulombic interactions in the vicinity of metal and ionic conductor interface.⁶¹ Origin of electrode polarization in perovskite devices has been connoted to the interfacial phenomena related with migration of ions having slow kinetics^17,60 and charge accumulation at the interfacial trap states at perovskite/ETL or perovskite/HTL interface. Thus, the trapped ionized carriers at the vicinity of perovskite and charge selective contacts modulate the local electric field in the device, giving rise to electrode polarization. Contemplating the ion continuity and Poisson electrostatic relation, electrode polarization effect accounts for the confinement of interfacial space charge within the ionic Debye length, L_D, that can be formulated at zero applied potential as following:⁶⁰where, ε₀, ε_r, V_T and n are free space permittivity, relative permittivity for perovskite, thermal voltage, and state of density for the mobile ions founding the interfacial space charge respectively.

Fig. 7 Scan rate related photocurrent hysteresis phenomena in (A) MA_0.6FA_0.4PbI₃/ZnO and (B) MA_0.6FA_0.4PbI₃/AZO device (C) device capacitance at zero applied bias as a function of frequency for MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO device (the inset shows the enlarged view of low frequency capacitance phenomena).

The mathematical expression for interfacial space charge capacitance, C_SC can be deduced by equating the ionic Debye length, L_D to the space charge width at zero potential, w. The expression of C_SC can be written as:⁶⁰

As comprehended from eqn (14), interfacial trap-mediated electrode polarization phenomenon in perovskite device is exhibited by excess capacitance specially at low frequency range in the frequency dependent capacitance spectra.⁶⁰ The excess low frequency capacitance alters the required time period for attaining steady state condition, bringing forth photocurrent hysteresis phenomena.⁶² As observed from Fig. 7(C) MA_0.6FA_0.4PbI₃/ZnO devices exhibit about 18% higher interfacial capacitance (C = ∼200 nF for MA_0.6FA_0.4PbI₃/ZnO devices and C = ∼170 nF for MA_0.6FA_0.4PbI₃/AZO devices) in the low frequency region (1 Hz) compared to MA_0.6FA_0.4PbI₃/AZO devices. Relatively small low frequency capacitance in MA_0.6FA_0.4PbI₃/AZO devices indicates low photocurrent hysteresis phenomena as depicted in Fig. 7(A) and (B). Thus the lower hysteresis in MA_0.6FA_0.4PbI₃/AZO devices can be corroborated to the large grain-assisted, reduced interfacial charge trapping in MA_0.6FA_0.4PbI₃ perovskite atop AZO ETL, as congruent with surface morphology characterization (Fig. 2). However, it is worth mentioning that the dipolar polarization effect, associated with the variation in intrinsic perovskite property via crystalline phase transition is almost identical in MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices, as enunciated by the nearly analogous capacitance plateau (Fig. 7(C)) at the intermediate frequency zone (100–1000 Hz).⁶⁰

3.6 Device stability

We have also carried out a systematic degradation study of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices to assess the device stability acquired with these two device structures. Of late, Zhao et al.¹⁶ have reported substantial thermal and photo-stability of MAPbI₃ perovskite solar cells incorporating sputtered AZO ETL compared to those having pristine ZnO ETLs. However, the study¹⁶ did not divulge into the dark stability phenomena of the perovskite devices^63–65 in conjunction with AZO ETLs that can unravel conducive information regarding the perovskite decomposition mechanism in an aged device. Aiming at closing this knowledge gap, in the present work, un-encapsulated MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO samples were stored for 30 days in a N₂ filled glovebox following the protocols mentioned in earlier perovskite degradation studies under dark^64,65 and device performance was evaluated at regular intervals at a humidity controlled (35–40% relative humidity) ambience at room temperature. Fig. 8(A)–(D) lift up the normalized PCE, J_SC, V_OC and FF values of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices respectively from the day of fabrication up to 30 days. Day wise PCE, J_SC, V_OC and FF values of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices have been tabulated in Tables S12 and S13† respectively. Fig. S10 and S11† show the IV curves of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices respectively throughout the time period of degradation study (30 days) with certain intervals. As observed from Tables S12 and S13,† the highest PCE for AZO or ZnO ETL based devices can be obtained after 4-5 days, which is consistent with the identical trend reported in previous perovskite literatures.^16,63,66 The initial rise (for the first 4–5 days since the day of fabrication) in PCE of the reported devices can be attributed to the substantial rise in FF value (Fig. 8(D)) which is related to the oxidation of Spiro-OMeTAD HTL in ambient condition for improved hole conductivity^67–69 and thus more efficient charge extraction. As comprehended from Fig. 8(A), MA_0.6FA_0.4PbI₃/AZO devices retain nearly 70% of their initial PCE even after 1 month, whereas for MA_0.6FA_0.4PbI₃/ZnO, PCE drops down to 36% of the commencing PCE during the same time range which denotes nearly two times higher stability with MA_0.6FA_0.4PbI₃/AZO devices. As observed from Fig. 8(B) and (C), about two-fold enhancement in device stability with MA_0.6FA_0.4PbI₃/AZO devices originates from two times higher J_SC and about 8% higher V_OC values retained in MA_0.6FA_0.4PbI₃/AZO devices compared to MA_0.6FA_0.4PbI₃/ZnO in course of 1 month long degradation study. Normalized FF (Fig. 8(D)) values of the two device structures, however, have comparable values (82.4% and 81.8% of the initial FF values for MA_0.6FA_0.4PbI₃/AZO and MA_0.6FA_0.4PbI₃/ZnO devices respectively after 30 days) and have less substantial impact on the variation in device stability compared to the other two IV parameters (J_SC and V_OC). To unravel the underlying reasons behind the ameliorated device stability with MA_0.6FA_0.4PbI₃/AZO devices, we have conducted impedance characterization of 30 day-aged MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices. Fig. 9(A) shows the Nyquist plots of degraded (1 month) MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices at 850 mV bias under dark. The fitted values of various electronic parameters using equivalent circuit model in Fig. S9† have been tabulated in Table S14.† As evident from Table S14,† the contact resistance, R_C at the perovskite/ETL or perovskite/HTL interfaces for both the aged MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices has increased compared to their commencing values. However, as observed from Table 2 and S14,† the contact resistance has increased 63% (78 Ω to 127 Ω) in MA_0.6FA_0.4PbI₃/ZnO device in course of 1 month degradation study whereas the contact resistance value has risen only 14% (86 Ω to 98 Ω) in degraded AZO based devices within the same time frame. Lower contact resistance in degraded MA_0.6FA_0.4PbI₃/AZO device is also coherent with nearly double J_SC (J_SC = 17.53 mA cm⁻² for MA_0.6FA_0.4PbI₃/AZO and J_SC = 9.51 mA cm⁻² for MA_0.6FA_0.4PbI₃/ZnO device); although the initial J_SC was higher with MA_0.6FA_0.4PbI₃/ZnO device (J_SC = 20.39 mA cm⁻² for MA_0.6FA_0.4PbI₃/AZO and J_SC = 21.95 mA cm⁻² for MA_0.6FA_0.4PbI₃/ZnO device) (see Tables S12 and S13†). Since, the HTL layer (Spiro-OMeTAD in both cases) remains constant for both the device structures, the variation in device contact resistance can be attributed to the disparity in degradation phenomena of the ETL layers (ZnO and AZO) and their concomitant influence on the decomposition of neighbouring MA_0.6FA_0.4PbI₃ perovskite film. In general, perovskite film decomposes rapidly in the presence of water vapour.⁷⁰ Since, ZnO film is hygroscopic, it traps water at its surface¹⁵ and with aging, ZnO becomes saturated with moisture. The absorbed moisture in ZnO layer may contribute to the transformation of MA_0.6FA_0.4PbI₃ film into MA_0.6FA_0.4PbI₃·H₂O and eventually may convert MA_0.6FA_0.4PbI₃·H₂O into (MA_0.6FA_0.4)₄PbI₆·2H₂O and PbI₂ (ref. 70) and thus the perovskite structure is fallen apart. The possible conversion of MA_0.6FA_0.4PbI₃ film into (MA_0.6FA_0.4)₄PbI₆·2H₂O and PbI₂, is, however a conjecture based on the in situ experimentation conducted by Kelly et al.⁷⁰ and further experimentation is required for the confirmation of the proposed perovskite degradation phenomena. However, AZO film is less hydrophilic compared to pristine ZnO as indicated by the higher contact angle of water droplet (49.67°) on AZO film compared to the angle (29.05°) on ZnO from contact angle measurement (Fig. S12†). Due to the more hydrophobic nature of AZO film, the trapped moisture-assisted perovskite decomposition rate is expected to be lower in MA_0.6FA_0.4PbI₃/AZO devices, which explicates higher J_SC and lower contact resistance⁶³ retained in the MA_0.6FA_0.4PbI₃/AZO devices which eventually lead to their higher stability compared to MA_0.6FA_0.4PbI₃/ZnO devices. Apart from the lower moisture assisted degradation, slower perovskite decomposition rate in stable MA_0.6FA_0.4PbI₃/AZO devices may also be attributed to abated ion migration rate of methyl ammonium iodide (CH₃NH₃I) from perovskite crystal structure.⁶⁵ In an aged perovskite device, small voids present in Spiro-OMeTAD HTL film trigger the migration of CH₃NH₃I ion through them to react with silver cathode and produce AgI (silver iodide) which results in the decomposition in perovskite layer.⁶⁵ Since, the alkaline surface of pristine ZnO film has been reported to be responsible for the deprotonation of CH₃NH₃I ion in perovskite film,¹⁵ the ion migration process of CH₃NH₃I is anticipated to be highly accelerated in an aged MA_0.6FA_0.4PbI₃/ZnO device. On the contrary, more acidic nature of AZO surface¹⁶ contributes to more sluggish CH₃NH₃I ion migration phenomena and slower rate of perovskite decomposition in MA_0.6FA_0.4PbI₃/AZO devices, which again corroborates to the higher normalized J_SC and enhanced device stability (Fig. 8(A) and (B)) with MA_0.6FA_0.4PbI₃/AZO devices in contrast with MA_0.6FA_0.4PbI₃/ZnO devices.

Fig. 8 Normalized device performances of MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO device as a function of sample storage time in a N₂ filled glove box. (A) Normalized PCE, (B) normalized J_SC (C) normalized V_OC and (D) normalized FF (normalized parameter value = value of a parameter at a particular day/corresponding value on day 0).

Fig. 9 (A) Nyquist plot of 1 month degraded MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO device at a bias of 850 mV under dark, Mott–Schottky curve at 10 KHz frequency with 30 days degraded (B) MA_0.6FA_0.4PbI₃/ZnO and (C) MA_0.6FA_0.4PbI₃/AZO device under dark (D) capacitance of aged device at zero applied bias as a function of frequency for MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO device (the inset shows the enlarged view of low frequency capacitance phenomena).

Apart from the high normalized J_SC owing to slow perovskite decomposition, the enhanced stability for MA_0.6FA_0.4PbI₃/AZO devices also originates from their high normalized V_OC, as comprehended from Fig. 8(C). To comprehend relatively enhanced normalized V_OC in degraded MA_0.6FA_0.4PbI₃/AZO devices, we have conducted Mott–Schottky analysis of 1 month aged MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices. Fig. 9(B) and (C) show the Mott–Schottky curves of 30 day-old MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices respectively at 10 kHz frequency under dark. The flat-band potential, dopant density and zero bias depletion width of both the aged devices have been listed in Table S15.† As observed from Fig. 9(B), (C), Tables S10 and S15,† flat-band potentials of both the aged MA_0.6FA_0.4PbI₃/ZnO and MA_0.6FA_0.4PbI₃/AZO devices have decreased compared to their commencing value on the day of fabrication. However, the flat-band potential decreases to ∼35% of its initial value in (0.96 V to 0.63 V in 1 month) MA_0.6FA_0.4PbI₃/ZnO devices; whereas, the reduction in flat-band potential for MA_0.6FA_0.4PbI₃/AZO devices is 15% (1.13 V to 0.96 V in 1 month). Low flat-band potential in aged MA_0.6FA_0.4PbI₃/ZnO devices indicates that the energetic offset between the quasi Fermi levels of perovskite and ZnO reduces significantly with device aging (from eqn (10)), which insinuates towards increased trap states both in the bulk and at the interface⁴⁴ between MA_0.6FA_0.4PbI₃ perovskite and ZnO ETL. Conversely, the higher flat-band potential in aged MA_0.6FA_0.4PbI₃/AZO devices corroborates to superior interfacial property in them in comparison with aged MA_0.6FA_0.4PbI₃/ZnO devices. About 10% smaller low frequency (∼1 Hz) capacitance in aged MA_0.6FA_0.4PbI₃/AZO devices (Fig. 9(D)) also substantiates to the conjecture that the interface trapping phenomenon in MA_0.6FA_0.4PbI₃/AZO devices is significantly less⁶⁰ with aging effect compared to MA_0.6FA_0.4PbI₃/ZnO devices. Reduced interface trapping phenomena in MA_0.6FA_0.4PbI₃/AZO devices can be attributed to the reduced perovskite grain boundary (which may act charge trapping sites^39,71,72) aided by large grain atop AZO film and trap state filling in AZO associated with Burstein–Moss effect.^21,54 Subdued interfacial trapping endows with suppressed recombination phenomenon in aged MA_0.6FA_0.4PbI₃/AZO devices which is also congruent with 95% higher recombination resistance (5940 Ω for MA_0.6FA_0.4PbI₃/AZO and 3050 Ω for MA_0.6FA_0.4PbI₃/ZnO devices) from Table S14† and 69% higher shunt resistance (20100 Ω for MA_0.6FA_0.4PbI₃/AZO and 11900 Ω for MA_0.6FA_0.4PbI₃/ZnO devices) compared to MA_0.6FA_0.4PbI₃/ZnO devices from Tables S12 and S13.† Thus, the higher recombination and shunt resistance in aged MA_0.6FA_0.4PbI₃/AZO devices contribute to their higher V_OC value (from eqn (6) and (7)) which lead to enhanced device stability in contrast with MA_0.6FA_0.4PbI₃/ZnO devices.

So, all the device characterizations conducted in the present study are congruent with augmented power conversion efficiency, enhanced device stability and subdued hysteresis phenomena in MA_0.6FA_0.4PbI₃/AZO devices compared to MA_0.6FA_0.4PbI₃/ZnO devices.

4. Conclusion

In summary, we have reported a highly efficient (∼14.5%) mixed organic cation based pure triiodide (MA_0.6FA_0.4PbI₃) perovskite device fabricated in conjunction with a low temperature processed Al doped ZnO (AZO) ETL in a planar device structure. Besides 20% improvement in average PCE, the MA_0.6FA_0.4PbI₃/AZO based PSCs demonstrate two fold increase in device stability and substantially subdued scan-rate dependent hysteresis phenomena when compared to MA_0.6FA_0.4PbI₃ perovskite devices fabricated along with pristine ZnO ETL. Moreover, large grain size and concomitant reduction in MA_0.6FA_0.4PbI₃ perovskite grain boundary overlying the AZO layer ensure superior charge transfer across the interface thereby lowering the donor–acceptor pair (DAP) recombination and significantly suppressing current leakage phenomena, which correlate to their efficient charge extraction and enhanced fill factor. Nearly 25% enhanced dark carrier mobility obtained from electrochemical measurements corresponds to the reduced monomolecular (trap-induced) recombination phenomenon and which is translated into augmented FF value in MA_0.6FA_0.4PbI₃/AZO devices. In addition, the deep trap state passivation with the free electrons from ionized Al dopants and concomitant upshift of quasi Fermi level in AZO film aided by the Burstein–Moss effect result in a high flat-band potential in MA_0.6FA_0.4PbI₃/AZO devices. Moreover, enhanced flat-band potential coupled with high recombination resistance from impedance measurement explains the larger open circuit voltage observed in MA_0.6FA_0.4PbI₃/AZO devices and its implication on high PCE. Additionally, lower electrode polarization effect exhibited by MA_0.6FA_0.4PbI₃/AZO devices describes the reason for suppressed photocurrent hysteresis behaviour compared to their ZnO counterparts. A one month-long systematic degradation study reveals that the perovskite decomposition rate is slow in aged MA_0.6FA_0.4PbI₃/AZO devices owing to the reduced water adsorption in hydrophobic AZO film or sluggish migration of CH₃NH₃I ions due to more acidic nature of AZO surface. Electronic investigations of the aged devices reveal substantially reduced interfacial charge trapping phenomena in MA_0.6FA_0.4PbI₃/AZO devices which enunciate their enhanced device lifetime compared to MA_0.6FA_0.4PbI₃/ZnO devices. Overall, the present work demonstrates simultaneous enhancement in device performance and stability in PSCs fortified with low hysteresis and opens up a new research pathway towards the fabrication of less energy extensive, flexible substrate compatible, efficient perovskite solar cells exhibiting superior device performance, stability and reduced photocurrent hysteresis behaviour for reliable operation aimed at long term practical applications.

Acknowledgements

The authors gratefully acknowledge the financial support provided by Future Solar Technologies Pty. Ltd. for this research work. The authors would also like to acknowledge the endless support from the staffs of Photovoltaic and Renewable Energy Engineering School, Electron Microscope Unit (EMU) and Solid State and Elemental Analysis Unit under Mark Wainwright Analytical Center, UNSW. The authors would also like to express their gratitude to Dr Vinicius Goncales for helping with the contact angle measurement.

References

1. N. J. Jeon, J. H. Noh, W. S. Yang, Y. C. Kim, S. Ryu, J. Seo and S. I. Seok, Nature, 2015, 517, 476–480.
2. N. Pellet, P. Gao, G. Gregori, T.-Y. Yang, M. K. Nazeeruddin, J. Maier and M. Grätzel, Angew. Chem., Int. Ed., 2014, 53, 3151–3157.
3. T. M. Koh, K. Fu, Y. Fang, S. Chen, T. C. Sum, N. Mathews, S. G. Mhaisalkar, P. P. Boix and T. Baikie, J. Phys. Chem. C, 2014, 118, 16458–16462.
4. S. Pang, H. Hu, J. Zhang, S. Lv, Y. Yu, F. Wei, T. Qin, H. Xu, Z. Liu and G. Cui, Chem. Mater., 2014, 26, 1485–1491.
5. S. Aharon, A. Dymshits, A. Rotem and L. Etgar, J. Mater. Chem. A, 2015, 3, 9171–9178.
6. I. E. Castelli, J. M. García-Lastra, K. S. Thygesen and K. W. Jacobsen, APL Mater., 2014, 2, 081514.
7. D. Liu and T. L. Kelly, Nat. Photonics, 2014, 8, 133–138.
8. M. H. Kumar, N. Yantara, S. Dharani, M. Graetzel, S. Mhaisalkar, P. P. Boix and N. Mathews, Chem. Commun., 2013, 49, 11089–11091.
9. J. Zhang and T. Pauporté, J. Phys. Chem. C, 2015, 119, 14919–14928.
10. J. Zhang, P. Barboux and T. Pauporté, Adv. Energy Mater., 2014, 4, 1400932.
11. M. Law, L. E. Greene, J. C. Johnson, R. Saykally and P. Yang, Nat. Mater., 2005, 4, 455–459.
12. A. K. Chandiran, M. Abdi-Jalebi, M. K. Nazeeruddin and M. Grätzel, ACS Nano, 2014, 8, 2261–2268.
13. Y.-Z. Gu, H.-L. Lu, Y. Geng, Z.-Y. Ye, Y. Zhang, Q.-Q. Sun, S.-J. Ding and D. W. Zhang, Nanoscale Res. Lett., 2013, 8, 1–5.
14. Y. Sun, J. H. Seo, C. J. Takacs, J. Seifter and A. J. Heeger, Adv. Mater., 2011, 23, 1679–1683.
15. J. Yang, B. D. Siempelkamp, E. Mosconi, F. De Angelis and T. L. Kelly, Chem. Mater., 2015, 27, 4229–4236.
16. X. Zhao, H. Shen, Y. Zhang, X. Li, X. Zhao, M. Tai, J. Li, J. Li, X. Li and H. Lin, ACS Appl. Mater. Interfaces, 2016, 8, 7826–7833.
17. H. J. Snaith, A. Abate, J. M. Ball, G. E. Eperon, T. Leijtens, N. K. Noel, S. D. Stranks, J. T.-W. Wang, K. Wojciechowski and W. Zhang, J. Phys. Chem. Lett., 2014, 5, 1511–1515.
18. B. Chen, M. Yang, S. Priya and K. Zhu, J. Phys. Chem. Lett., 2016, 7, 905–917.
19. Y. Zhou, M. Yang, W. Wu, A. L. Vasiliev, K. Zhu and N. P. Padture, J. Mater. Chem. A, 2015, 3, 8178–8184.
20. L. Znaidi, Mater. Sci. Eng., B, 2010, 174, 18–30.
21. A. Aprilia, P. Wulandari, V. Suendo, Herman, R. Hidayat, A. Fujii and M. Ozaki, Sol. Energy Mater. Sol. Cells, 2013, 111, 181–188.
22. L. Spanhel and M. A. Anderson, J. Am. Chem. Soc., 1991, 113, 2826–2833.
23. H. Oh, J. Krantz, I. Litzov, T. Stubhan, L. Pinna and C. J. Brabec, Sol. Energy Mater. Sol. Cells, 2011, 95, 2194–2199.
24. A. De Sio, K. Chakanga, O. Sergeev, K. von Maydell, J. Parisi and E. von Hauff, Sol. Energy Mater. Sol. Cells, 2012, 98, 52–56.
25. R. Shannon, Acta Crystallogr., Sect. A: Cryst. Phys., Diffr., Theor. Gen. Crystallogr., 1976, 32, 751–767.
26. Y. Tian and I. G. Scheblykin, J. Phys. Chem. Lett., 2015, 6, 3466–3470.
27. H. Zhang, J. Mao, H. He, D. Zhang, H. L. Zhu, F. Xie, K. S. Wong, M. Grätzel and W. C. H. Choy, Adv. Energy Mater., 2015, 5, 1501354.
28. M. Lv, X. Dong, X. Fang, B. Lin, S. Zhang, J. Ding and N. Yuan, RSC Adv., 2015, 5, 20521–20529.
29. D. Zhao, M. Sexton, H.-Y. Park, G. Baure, J. C. Nino and F. So, Adv. Energy Mater., 2015, 5, 1401855.
30. Q. Wang, C. Bi and J. Huang, Nano Energy, 2015, 15, 275–280.
31. W. Ke, G. Fang, J. Wan, H. Tao, Q. Liu, L. Xiong, P. Qin, J. Wang, H. Lei, G. Yang, M. Qin, X. Zhao and Y. Yan, Nat. Commun., 2015, 6, 6700.
32. Y. Numata, Y. Sanehira and T. Miyasaka, ACS Appl. Mater. Interfaces, 2016, 8, 4608–4615.
33. B. Conings, L. Baeten, T. Jacobs, R. Dera, J. D'Haen, J. Manca and H.-G. Boyen, APL Mater., 2014, 2, 081505.
34. T. Salim, S. Sun, Y. Abe, A. Krishna, A. C. Grimsdale and Y. M. Lam, J. Mater. Chem. A, 2015, 3, 8943–8969.
35. K. Wang, C. Liu, P. Du, H.-L. Zhang and X. Gong, Small, 2015, 11, 3369–3376.
36. C. Liu, K. Wang, P. Du, T. Meng, X. Yu, S. Z. D. Cheng and X. Gong, ACS Appl. Mater. Interfaces, 2015, 7, 1153–1159.
37. M. A. Green, Solar cells: Operating principles, technology, and system applications, Englewood Cliffs, N. J., Prentice-Hall, Inc., 1982.
38. J. Chen, T. Shi, X. Li, B. Zhou, H. Cao and Y. Wang, Appl. Phys. Lett., 2016, 108, 053302.
39. J. Lian, Q. Wang, Y. Yuan, Y. Shao and J. Huang, J. Mater. Chem. A, 2015, 3, 9146–9151.
40. S. Bag and M. F. Durstock, ACS Appl. Mater. Interfaces, 2016, 8, 5053–5057.
41. C. Bi, Y. Shao, Y. Yuan, Z. Xiao, C. Wang, Y. Gao and J. Huang, J. Mater. Chem. A, 2014, 2, 18508–18514.
42. V. D'Innocenzo, G. Grancini, M. J. P. Alcocer, A. R. S. Kandada, S. D. Stranks, M. M. Lee, G. Lanzani, H. J. Snaith and A. Petrozza, Nat. Commun., 2014, 5, 3586.
43. A. Pockett, G. E. Eperon, T. Peltola, H. J. Snaith, A. Walker, L. M. Peter and P. J. Cameron, J. Phys. Chem. C, 2015, 119, 3456–3465.
44. A. Guerrero, J. You, C. Aranda, Y. S. Kang, G. Garcia-Belmonte, H. Zhou, J. Bisquert and Y. Yang, ACS Nano, 2016, 10, 218–224.
45. D. Liu, J. Yang and T. L. Kelly, J. Am. Chem. Soc., 2014, 136, 17116–17122.
46. W. Yang, Y. Yao and C.-Q. Wu, J. Appl. Phys., 2015, 117, 095502.
47. S. R. Cowan, A. Roy and A. J. Heeger, Phys. Rev. B: Condens. Matter Mater. Phys., 2010, 82, 245207.
48. T. M. Clarke, J. Peet, A. Nattestad, N. Drolet, G. Dennler, C. Lungenschmied, M. Leclerc and A. J. Mozer, Org. Electron., 2012, 13, 2639–2646.
49. G. Garcia-Belmonte, A. Munar, E. M. Barea, J. Bisquert, I. Ugarte and R. Pacios, Org. Electron., 2008, 9, 847–851.
50. J. Bisquert and G. Garcia-Belmonte, J. Phys. Chem. Lett., 2011, 2, 1950–1964.
51. S. Major, A. Banerjee, K. L. Chopra and K. C. Nagpal, Thin Solid Films, 1986, 143, 19–30.
52. M. Smirnov, Journal of Advanced Research in Physics, 2010, 1.
53. R. Lin, M. Miwa, M. Wright and A. Uddin, Thin Solid Films, 2014, 566, 99–107.
54. K. G. Saw, N. M. Aznan, F. K. Yam, S. S. Ng and S. Y. Pung, PLoS One, 2015, 10, e0141180.
55. Y. Imai and A. Watanabe, J. Mater. Sci.: Mater. Electron., 2004, 15, 743–749.
56. A. P. Roth, J. B. Webb and D. F. Williams, Solid State Commun., 1981, 39, 1269–1271.
57. S. Ryu, J. H. Noh, N. J. Jeon, Y. Chan Kim, W. S. Yang, J. Seo and S. I. Seok, Energy Environ. Sci., 2014, 7, 2614–2618.
58. P. P. Boix, J. Ajuria, I. Etxebarria, R. Pacios, G. Garcia-Belmonte and J. Bisquert, J. Phys. Chem. Lett., 2011, 2, 407–411.
59. Y. Zhang, M. Liu, G. E. Eperon, T. C. Leijtens, D. McMeekin, M. Saliba, W. Zhang, M. de Bastiani, A. Petrozza, L. M. Herz, M. B. Johnston, H. Lin and H. J. Snaith, Mater. Horiz., 2015, 2, 315–322.
60. O. Almora, I. Zarazua, E. Mas-Marza, I. Mora-Sero, J. Bisquert and G. Garcia-Belmonte, J. Phys. Chem. Lett., 2015, 6, 1645–1652.
61. A. Serghei, M. Tress, J. R. Sangoro and F. Kremer, Phys. Rev. B: Condens. Matter Mater. Phys., 2009, 80, 184301.
62. H.-S. Kim and N.-G. Park, J. Phys. Chem. Lett., 2014, 5, 2927–2934.
63. J. You, L. Meng, T.-B. Song, T.-F. Guo, Y. Yang, W.-H. Chang, Z. Hong, H. Chen, H. Zhou, Q. Chen, Y. Liu, N. De Marco and Y. Yang, Nat. Nanotechnol., 2015, 11, 75–81.
64. C.-G. Wu, C.-H. Chiang, Z.-L. Tseng, M. K. Nazeeruddin, A. Hagfeldt and M. Gratzel, Energy Environ. Sci., 2015, 8, 2725–2733.
65. Y. Kato, L. K. Ono, M. V. Lee, S. Wang, S. R. Raga and Y. Qi, Adv. Mater. Interfaces, 2015, 2, 1500195.
66. L. Q. Zhang, X. W. Zhang, Z. G. Yin, Q. Jiang, X. Liu, J. H. Meng, Y. J. Zhao and H. L. Wang, J. Mater. Chem. A, 2015, 3, 12133–12138.
67. J. Carrillo, A. Guerrero, S. Rahimnejad, O. Almora, I. Zarazua, E. Mas-Marza, J. Bisquert and G. Garcia-Belmonte, Adv. Energy Mater., 2016, 6, 1502246.
68. W. H. Nguyen, C. D. Bailie, E. L. Unger and M. D. McGehee, J. Am. Chem. Soc., 2014, 136, 10996–11001.
69. L. Badia, E. Mas-Marzá, R. S. Sánchez, E. M. Barea, J. Bisquert and I. Mora-Seró, APL Mater., 2014, 2, 081507.
70. J. Yang, B. D. Siempelkamp, D. Liu and T. L. Kelly, ACS Nano, 2015, 9, 1955–1963.
71. Q. Chen, H. Zhou, Z. Hong, S. Luo, H.-S. Duan, H.-H. Wang, Y. Liu, G. Li and Y. Yang, J. Am. Chem. Soc., 2014, 136, 622–625.
72. Z. Liang, S. Zhang, X. Xu, N. Wang, J. Wang, X. Wang, Z. Bi, G. Xu, N. Yuan and J. Ding, RSC Adv., 2015, 5, 60562–60569.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra14186b

---